United States
Environmental Protection
Ageney
Health Assessment of
1,3-Butadiene
-------
EPA/600/P-98/001F
October 2002
Health Assessment
of 1,3-Butadiene
National Center for Environmental Assessment-Washington Office
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC
-------
DISCLAIMER
This document has been reviewed in accordance with the U.S. Environmental Protection
Agency's peer and administrative review policies and approved for presentation and publication.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
ABSTRACT
This assessment was conducted to review the new information that has become available
since EPA's 1985 health assessment of 1,3-butadiene.
1,3-Butadiene is a gas used commercially in the production of styrene-butadiene rubber,
plastics, and thermoplastic resins. The major environmental source of 1,3-butadiene is the
incomplete combustion of fuels from mobile sources (e.g., automobile exhaust). Tobacco smoke
can be a significant source of 1,3-butadiene in indoor air.
This assessment concludes that 1,3-butadiene is carcinogenic to humans by inhalation,
based on the total weight of evidence. The specific mechanisms of 1,3-butadiene-induced
carcinogenesis are unknown, however, it is virtually certain that the carcinogenic effects are
mediated by genotoxic metabolites of 1,3-butadiene.
Animal data suggest that females may be more sensitive than males for cancer effects;
nevertheless, there are insufficient data from which to draw any conclusions on potentially
sensitive subpopulations.
The human incremental lifetime unit cancer (incidence) risk estimate is based on
extrapolation from leukemias observed in an occupational epidemiologic study. A twofold
adjustment to the epidemiologic-based unit cancer risk is then applied to reflect evidence from
the rodent bioassays suggesting that the epidemiologic-based estimate may underestimate total
cancer risk from 1,3-butadiene exposure in the general population.
1,3-Butadiene also causes a variety of reproductive and developmental effects in mice; no
human data on these effects are available. The most sensitive effect was ovarian atrophy
observed in a lifetime bioassay of female mice. Based on this critical effect and using the
benchmark concentration methodology, an RfC (i.e., a chronic exposure level presumed to be
"without appreciable risk" for noncancer effects) was calculated.
In summary, the EPA's conclusions about the health effects of 1,3-butadiene are:
• 1,3-Butadiene is carcinogenic to humans by inhalation.
• The unit cancer risk estimate is 0.08/ppm (based primarily on linear modeling and
extrapolation of human data). This incorporates an adjustment factor of 2 to address
concerns for sensitive populations. The corresponding estimate of the chronic
exposure level of 1,3-butadiene resulting in extra cancer risk of 10"6 (i.e., 1 in a
million) is 0.01 ppb.
• A chronic RfC (0.9 ppb), an acute reference value (7 ppb), and a subchronic reference
value (7 ppb) are presented for noncancer effects.
Preferred citation:
U.S. Environmental Protection Agency (EPA). (2002) Health assessment of 1,3-butadiene. National Center
for Environmental Assessment, Washington, DC: EPA/600/P-98/001F. Available from: National Technical
Information Service, Springfield, VA, .
ii
-------
CONTENTS
LIST OF TABLES ix
LIST OF FIGURES xiii
PREFACE xiv
AUTHORS, CONTRIBUTORS, AND REVIEWERS xv
1. INTRODUCTION 1-1
1.1. BACKGROUND 1-1
1.2. OVERVIEW OF THE SCOPE OF THIS HEALTH ASSESSMENT 1-1
1.3. SUMMARY OF EPA'S CARCINOGEN ASSESSMENT (U.S. EPA, 1985) 1-3
2. OVERVIEW OF EXPOSURE TO 1,3-BUTADIENE 2-1
2.1. PHYSICAL/CHEMICAL PROPERTIES 2-1
2.2. PRODUCTION AND USE 2-1
2.2.1. SB Latex and Rubber Production 2-2
2.2.2. Polybutadiene Production 2-2
2.2.3. Neoprene Rubber Production 2-3
2.2.4. Acrylonitrile-Butadiene Resin Production 2-3
2.2.5. Nitrile Elastomer Production 2-3
2.2.6. Adiponitrile Production 2-3
2.3. SOURCES AND EMISSION 2-4
2.3.1. Mobile Sources 2-4
2.3.1.1. On-Road Mobile Sources 2-4
2.3.1.2. Nonroad Mobile Sources 2-6
2.3.1.3. Aircraft 2-6
2.3.2. Miscellaneous Sources 2-7
2.3.2.1. Miscellaneous Chemical Production 2-7
2.3.2.2. Secondary Lead Smelters 2-7
2.3.2.3. Petroleum Refining 2-7
2.3.2.4. Combustion Sources 2-8
2.4. AMBIENT CONCENTRATION OF 1,3-BUTADIENE 2-9
2.4.1. Air 2-9
2A.I.1. Ambient Monitoring Data 2-9
2.4.1.2. Ambient Source Apportionment 2-19
2.4.2. Indoor Exposure to 1,3-Butadiene 2-24
2.4.3. Inside Automobile 2-25
2.4.4. Water 2-25
2.4.5. Food 2-25
2.5. MODELED ON-ROAD MOBILE SOURCE BUTADIENE EXPOSURE
ESTIMATES 2-25
2.6. CONCLUSIONS ABOUT PATHWAYS OF EXPOSURE 2-26
in
-------
CONTENTS (continued)
3. PHARMACOKINETICS 3-1
3.1. OVERVIEW OF PHARMACOKINETIC STUDIES 3-1
3.2. ABSORPTION 3-2
3.3. BIOTRANSFORMATION 3-4
3.3.1. Pathway Elucidation 3-5
3.3.2. Enzyme Identification 3-11
3.3.3. Rates of Metabolism 3-13
3.3.3.1. In Vitro 3-14
3.3.3.2. Enzymatic Activities 3-14
3.3.3.3. In Vivo 3-27
3.4. DISTRIBUTION 3-31
3.4.1. Blood and Tissue Concentrations 3-31
3.4.1.1. Species Differences 3-31
3.4.1.2. Gender Differences 3-39
3.4.1.3. Chronic Exposures 3-39
3.4.2. MolecularDosimetry 3-43
3.4.2.1. Hemoglobin Adducts 3-44
3.4.2.2. DNA Adducts 3-47
3.5. EXCRETION 3-49
3.5.1. Pathways of Excretion 3-50
3.5.2. Identification of Metabolites 3-52
3.6. PHARMACOKINETICS OF STRUCTURALLY RELATED COMPOUNDS . . . 3-56
3.7. DISCUSSION 3-61
4. MUTAGENICITY 4-1
4.1. 1,3-BUTADIENE 4-1
4.1.1. Gene Mutations 4-1
4.1.2. Cytogenetic Effects 4-15
4.1.2.1. Somatic Cells 4-15
4.1.2.2. Germ Cells 4-17
4.2. METABOLITES 4-20
4.2.1. Gene Mutations 4-21
4.2.2. Cytogenetic Effects 4-22
4.2.2.1. Somatic Cells 4-22
4.2.2.2. Germ Cells 4-24
4.3. DISCUSSION 4-25
5. REPRODUCTIVE AND DEVELOPMENTAL EFFECTS 5-1
5.1. REPRODUCTIVE EFFECTS 5-1
5.1.1. Carpenter et al. (1944) 5-1
5.1.2. Owen et al. (1987); Owen and Glaister (1990) 5-1
5.1.3. NTP (1984) 5-4
5.1.4. NTP (1993) 5-4
IV
-------
CONTENTS (continued)
5.1.5. Bevan et al. (1996) 5-10
5.1.6. Hackett et al. (1988a) 5-10
5.1.7. Pacchierotti et al. (1998b) 5-11
5.2. DEVELOPMENTAL EFFECTS 5-11
5.2.1. IISRP(1982) 5-11
5.2.2. Hackett et al. (1987a) 5-12
5.2.3. Hackett et al. (1987b) 5-14
5.3. STRUCTURE-ACTIVITY RELATIONSHIPS 5-18
5.4. PLAUSIBLE MECHANISMS OF REPRODUCTIVE TOXICITY 5-30
5.5. SUMMARY AND CONCLUSIONS 5-34
6. TOXICITY IN ANIMALS 6-1
6.1. SUBCHRONIC TOXICITY 6-1
6.2. CHRONIC TOXICITY 6-2
6.2.1. Rat Study 6-2
6.2.2. Mouse Study 6-2
6.3. CARCINOGENICITY 6-10
6.3.1. Rat Study 6-10
6.3.2. Mouse Studies 6-12
6.3.2.1. 2-Year Study (NTP, 1993) 6-12
6.3.2.2. 2-Year Stop-Exposure Study (NTP, 1993) 6-23
6.3.2.3. Summary of NTP (1993) Study 6-30
6.3.2.4. 1-Year Study (Irons et al., 1989; Irons, 1990) 6-31
6.4. STUDIES OF 1,3-BUTADIENE METABOLITES 6-32
6.5. RELATED COMPOUNDS 6-32
6.6. DISCUSSION AND CONCLUSIONS 6-34
7. EPIDEMIOLOGIC STUDIES OF CARCINOGENICITY 7-1
7.1. MONOMER PRODUCTION 7-1
7.1.1. Texaco Cohort 7-1
7.1.1.1. Downs et al. (1987): Mortality Among Workers at a
Butadiene Facility 7-1
7.1.1.2. Divine (1990): An Update on Mortality Among Workers at a
1,3-Butadiene Facility—Preliminary Results 7-4
7.1.1.3. Divine et al. (1993): Cancer Mortality Among Workers at a
Butadiene Production Facility 7-6
7.1.1.4. Divine and Hartman (1996): Mortality Update of
Butadiene Production Workers 7-8
7.1.2. Shell Oil Refinery Cohort 7-11
7.1.2.1. Cowles et al. (1994): Mortality, Morbidity, and
Hematological Results From a Cohort of Long-Term
Workers Involved in 1,3-Butadiene Monomer Production 7-11
7.1.3. Union Carbide Cohort 7-12
-------
CONTENTS (continued)
7.1.3.1. Ward et al. (1995): Mortality Study of Workers in
1,3-Butadiene Production Units Identified From a
Chemical Workers Cohort; Ward et al. (1996c): Mortality
Study of Workers Employed in 1,3-Butadiene Production Units
Identified From a Large Chemical Workers Cohort 7-12
7.2. POLYMER PRODUCTION 7-13
7.2.1. Cohort Identified by Johns Hopkins University (JHU) Investigators 7-13
7.2.1.1. Matanoski and Schwartz (1987): Mortality of
Workers in Styrene-Butadiene Polymer Production 7-13
7.2.1.2. Matanoski etal. (1989): Epidemiologic Data Related
to Health Effects of 1,3-Butadiene; Matanoski et al. (1990):
Mortality of a Cohort of Workers in the Styrene-Butadiene
Polymer Manufacturing Industry (1943-1982) 7-17
7.2.1.3. Matanoski et al. (1989): Epidemiologic Data Related to
Health Effects of 1,3-Butadiene; Santos-Burgoa et al. (1992):
Lymphohematopoietic Cancer in Styrene-Butadiene
Polymerization Workers 7-19
7.2.1.4. Matanoski et al. (1993): Cancer Epidemiology
Among Styrene-Butadiene Rubber Workers 7-22
7.2.1.5. Matanoski et al. (1997): Lymphohematopoietic
Cancers and Butadiene and Styrene Exposure in
Synthetic Rubber Manufacture 7-22
7.2.2. Cohort Identified by University of Alabama (UAB) Investigators 7-24
7.2.2.1. Delzell et al. (1996): A Follow-Up Study
of Synthetic Rubber Workers 7-24
7.2.2.2. Macaluso et al. (1996): Leukemia and Cumulative
Exposure to Butadiene, Styrene, and Benzene Among
Workers in the Synthetic Rubber Industry 7-27
7.2.2.3. Sathiakumar et al. (1998): Mortality from Cancer and
Other Causes of Death Among Synthetic Rubber Workers 7-29
7.3. POPULATION STUDY 7-30
7.3.1. Loughlin et al. (1999): Lymphatic and Haematopoietic Cancer
Mortality in a Population Attending School Adjacent to
Styrene-Butadiene Facilities, 1963-1993 7-30
7.4. SUMMARY AND DISCUSSION 7-37
7.4.1. Monomer Production 7-37
7.4.2. Polymer Production 7-39
7.4.3. Relevant Methodologic Issues and Discussion 7-41
7.4.4. Criteria of Causal Inference 7-48
VI
-------
CONTENTS (continued)
8. HAZARD CHARACTERIZATION 8-1
8.1. INTRODUCTION 8-1
8.2. HAZARD ASSESSMENT 8-2
8.2.1. Metabolism and Pharmacokinetics 8-2
8.2.2. Mutagenicity 8-3
8.2.3. Characterization of Reproductive and Developmental Effects 8-4
8.2.4. Cancer Characterization 8-5
8.2.4.1. Cancer Evidence From Experimental Animals 8-6
8.2.4.2. Cancer Evidence in Humans 8-6
8.2.4.3. Cancer Characterization Conclusion 8-14
8.2.4.4. Recent Evaluations of 1,3-Butadiene by Other Organizations ... 8-15
8.3. SPECIAL SUBPOPULATIONS 8-16
8.3.1. Sensitive Subpopulations 8-16
8.3.2. Highly Exposed Subpopulations 8-18
8.4. FUTURERESEARCHNEEDS 8-19
8.4.1. Epidemiology 8-19
8.4.2. Toxicology 8-20
8.4.3. Molecular Biology 8-21
8.5. SUMMARY AND CONCLUSIONS 8-21
9. PHARMACOKINETIC MODELING 9-1
9.1. INTRODUCTION 9-1
9.2. PBPK MODELS FOR 1,3-BUTADIENE 9-2
9.2.1. Hattis and Wasson (1987) 9-2
9.2.2. Hallenbeck (1992) 9-2
9.2.3. Kohn and Melnick (1993) 9-4
9.2.4. Johanson and Filser (1993) 9-8
9.2.5. Evelo et al. (1993) 9-12
9.2.6. Medinsky et al. (1994) 9-14
9.2.7. Elaine Presentations 9-18
9.2.7.1. Kohn and Melnick (1996) 9-18
9.2.7.2. Johanson and Filser (1996) 9-19
9.2.7.3. Csanady et al. (1996) 9-19
9.2.7.4. Sweeney et al. (1996) 9-20
9.2.8. Reitz et al. (1996) 9-20
9.2.9. Sweeney et al. (1997) 9-28
9.2.10. Kohn (1997) 9-35
9.3. SUMMARY 9-40
9.4. CONCLUSIONS 9-43
10. QUANTITATIVE RISK ASSESSMENT FOR 1,3-BUTADIENE 10-1
10.1. EPIDEMIOLOGICALLY BASED CANCER RISK ASSESSMENT 10-1
10.1.1. Exposure-Response Modeling 10-1
vn
-------
CONTENTS (continued)
10.1.2. Prediction of Lifetime Excess Risk of Leukemia Mortality 10-3
10.1.3. Sources of Uncertainty 10-14
10.1.3.1. Power Calculation for Lung Cancer for the Delzell et al. Cohort 10-16
10.1.4. Summary and Conclusions 10-18
10.2. CANCER RISK ESTIMATES BASED ON RODENT BIOASSAYS 10-23
10.2.1. Rat-Based Estimates 10-23
10.2.2. Mouse-Based Estimates 10-24
10.2.2.1. Quantal 10-24
10.2.2.2. Time-to-Tumor 10-27
10.2.3. Discussion 10-32
10.3. DOSE-RESPONSE MODELING OF REPRODUCTIVE
AND DEVELOPMENTAL TOXICITY 10-35
10.3.1. Introduction 10-35
10.3.2. Prenatal Developmental Toxicity Studies 10-35
10.3.3. Dominant Lethal Studies 10-40
10.3.3.1. Anderson et al. (1993, 1995, 1996) 10-40
10.3.3.2. Anderson et al. (1998) 10-45
10.3.3.3. Brinkworth et al. (1998) 10-48
10.3.3.4. Analysis of Combined Dominant Lethal Studies 10-52
10.3.4. Chronic Studies 10-53
10.3.5. Summary and Conclusions 10-62
10.4. REFERENCE CONCENTRATIONS FOR REPRODUCTIVE AND
DEVELOPMENTAL EFFECTS 10-63
10.4.1. Introduction 10-63
10.4.2. Calculation of RfCs 10-63
10.4.3. Discussion 10-67
10.4.4. Conclusions 10-69
10.5. CONCLUSIONS ON QUANTITATIVE RISK ESTIMATES 10-69
11. DOSE-RESPONSE CHARACTERIZATION 11-1
11.1. QUANTITATIVE RISK ESTIMATION FOR CANCER 11-1
11.2. QUANTITATIVE ESTIMATION (RfC) FOR NONCANCER EFFECTS 11-7
11.3. SUMMARY AND CONCLUSIONS 11-8
REFERENCES R-l
Vlll
-------
LIST OF TABLES
2-1. 1,3-Butadiene on-site and off-site releases (in pounds) 2-5
2-2A. Summary of 1,3-butadiene ambient data from the EPA
Aerometric Information Retrieval System (AIRS) for 1988 to 1991 2-11
2-2B. Summary of 1,3-butadiene ambient data from the EPA Aerometric
Information Retrieval System (AIRS) for 1992 to 1994 2-13
2-3. Summary of 1,3-butadiene ambient data from the Urban Air Toxics
Monitoring Program (UATMP) 2-17
2-4. Summary of outdoor urban data from the National Ambient Volatile
Organic Compounds (NAVOC) Database 2-18
2-5. Summary of air monitoring program results for 1,3-butadiene 2-20
2-6. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information
Retrieval System (AIRS) (Table 2-2) based on sampling locations 2-22
2-7. Summary of 1,3-butadiene data from Table 2-3 based on sampling locations 2-22
2-8. Summary of 1,3-butadiene data from Table 2-4 based on sampling locations 2-23
2-9. Summary of the relative contributions to ambient 1,3-butadiene
emissions given as percentage of total mg/year 2-24
2-10. Annual-average exposure results (|ig/m3) for 1,3-butadiene, total population -
all on-road vehicles, base control scenario 2-27
2-11. Annual average exposure results (|ig/m3) for 1,3-butadiene in Chicago by
demographic group - all on-road vehicles, base control scenario 2-27
3-1. Tissue:blood partition coefficients 3-3
3-2. Identification of urinary metabolites in Figure 3-1 3-7
3-3. Identified 1,3-butadiene metabolites 3-8
3-4. In vitro Michaelis-Menton kinetic parameters for mouse liver 3-15
3-5. In vitro Michaelis-Menton kinetic parameters for rat liver 3-15
3-6. In vitro Michaelis-Menton kinetic parameters for human liver 3-16
3-7. In vitro Michaelis-Menton kinetic parameters for mouse lung 3-17
3-8. In vitro Michaelis-Menton kinetic parameters for rat lung 3-17
3-9. In vitro Michaelis-Menton kinetic parameters for human lung 3-17
3-10. Estimated intrinsic clearance for liver (Vmax/Km) 3-18
3-11. Tissue and blood concentrations (pmol/g tissue) 3-37
3-12. Tissue:blood concentration ratios 3-37
3-13. Mouse:rat concentration ratios 3-38
3-14. Female:male concentration ratio 3-40
3-15. Blood concentrations of reactive metabolites (pmol/g tissue) 3-41
3-16. 3,4-Epoxybutene tissue:blood concentration ratios 3-42
3-17. l,2,3,4-Diepoxybutanetissue:blood concentration ratios 3-42
3-18. Mouse:rat concentration ratios 3-43
4-1. Summary of earlier mutagenicity testing of 1,3-butadiene, 1,3-butadiene
monoepoxide and l,2:3,4-diepoxybutane (derived from tables in IARC, 1999) .... 4-2
4-2. Summary table of genetic effects of 1,3-butadiene 4-5
ix
-------
LIST OF TABLES (continued)
4-3. Summary table of genetic effects of l,2-epoxy-3-butene (EB) 4-7
4-4. Summary of genetic effects of l,2:3,4-diepoxybutane (DEB) 4-9
4-5. Summary table of genetic effects of 3,4-epoxy-l,2-butanediol (EBD) 4-11
5-1. A summary of the reproductive and developmental effects of 1,3-butadiene 5-2
5-2. Reproductive tract lesions in female B6C3FJ mice exposed to 1,3-butadiene
by inhalation 5-6
5-3. Reproductive tract lesions in male B6C3FJ mice exposed to 1,3-butadiene by
inhalation 5-8
5-4. Maternal toxicity in Sprague-Dawley CD rats exposed to 1,3-butadiene by
inhalation 5-13
5-5. Maternal toxicity in pregnant CD-I mice exposed to 1,3-butadiene by inhalation . 5-15
5-6. Developmental toxicity in CD-I mice exposed to 1,3-butadiene by inhalation .... 5-16
5-7. Malformations and variations in CD-I mice exposed to 1,3-butadiene by
inhalation 5-17
5-8. A summary of reproductive and developmental toxicity effects of chemicals
structurally similar to 1,3-butadiene 5-19
6-1. Relative organ weights (g/100 g body weight) in Sprague-Dawley rats 6-3
6-2. Incidence of nephropathy in male Sprague-Dawley rats 6-3
6-3. Survival of male and female B6C3FJ mice exposed to
1,3-butadiene by inhalation for 103 weeks 6-5
6-4. Incidence of hyperplasia in male and female B6C3FJ mice exposed to
1,3-butadiene by inhalation for 103 weeks 6-9
6-5. Incidence of hyperplasia in male B6C3Fj mice exposed to
1,3-butadiene by inhalation in the stop-exposure study 6-10
6-6. Incidences of tumors in Sprague-Dawley rats exposed to
1,3-butadiene by inhalation 6-11
6-7. Incidence of primary neoplasms in male B6C3Fj mice
exposed to 1,3-butadiene by inhalation for 103 weeks 6-13
6-8. Incidence of primary neoplasms in female B6C3Fj mice
exposed to 1,3-butadiene by inhalation for 103 weeks 6-15
6-9. Incidence of primary neoplasms in male B6C3Fj mice exposed
to 1,3-butadiene by inhalation for 9 months and 15 months 6-17
6-10. Incidence of primary neoplasms in female B6C3Fj mice exposed
to 1,3-butadiene by inhalation for 9 months and 15 months 6-18
6-11. Incidence of primary neoplasms in male B6C3Fj mice exposed to 1,3-butadiene
by inhalation in the stop-exposure study 6-27
6-12. Significant positive primary tumor incidences in mice and rats exposed to
1,3-butadiene for 2 years 6-35
7-1. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—
monomer production 7-14
x
-------
LIST OF TABLES (continued)
7-2. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—
polymer production 7-31
8-1. Summary of key epidemiologic studies 8-8
8-2. Carcinogenicity assessments of 1,3-butadiene 8-17
9-1. Parameter values used in the Hattis and Wasson (1987) PBPK model 9-3
9-2. Parameter values used in the Kohn and Melnick (1993) PBPK model 9-5
9-3. Parameter values used in the Johanson and Filser (1993) PBPK model 9-9
9-4. Parameter values used in the Evelo et al. (1993) PBPK model 9-13
9-5. Parameter values used in the Medinsky et al. (1994) PBPK model 9-16
9-6. Parameter values used in the Reitz et al. (1996) PBPK model 9-25
9-7. Parameter values used in Sweeney et al. (1997) PBPK model 9-29
9-8. Parameter values used in Kohn (1997) PBPK models 9-38
10-1. Results from exposure-response models of cumulative exposure to 1,3-butadiene
using alternative structural forms reported by Health Canada (1998) and Delzell
et al. (1995) 10-4
10-2. Maximum likelihood estimates (MLEs) of excess leukemia mortality risk with
95% upper confidence limits (95% UCL) for continuous lifetime exposures to
varying concentrations of 1,3-butadiene based on several models reported by
Health Canada (1998) and U.S. population rates 10-11
10-3. Maximum likelihood estimates (MLEs) and 95% lower confidence
limits (95% LCL) of continuous exposure concentrations associated with
varying excess risk levels based on relative rate results of several models
reported by Health Canada (1998) and U.S. population rates 10-12
10-4. Maximum likelihood (ECp) and 95% lower-bound (LECp) estimates of the
continuous exposure concentrations associated with varying levels
of excess risk (p) of leukemia mortality 10-13
10-5. Cancer potency (unit risk) estimates for leukemia mortality based on linear
extrapolation from the LEC01 or EC01 calculated from the models presented by
Health Canada (1998) 10-22
10-6. Dose-response data for linearized multistage model 10-25
10-7. Parameter estimates for multistage Weibull time-to-tumor
model based on female mouse tumor incidence, without 625 ppm group 10-29
10-8. Parameter estimates for multistage Weibull time-to-tumor
model based on male mouse tumor incidence, without 625 ppm group 10-29
10-9. Human unit cancer risk estimates (extra risk, computed for risks of
10"6) based on female mouse tumor incidences, without 625 ppm group,
using multistage Weibull time-to-tumor model 10-31
XI
-------
LIST OF TABLES (continued)
10-10. Human unit cancer risk estimates (extra risk, computed for risks of
10"6) based on male mouse tumor incidences, without 625 ppm group,
using multistage Weibull time-to-tumor model 10-31
10-11. Unit potency estimates (extra risk) summed across tumor sites 10-33
10-12. Fetal weight data from prenatal (developmental) toxicity study 10-37
10-13. Fetal weight modeling 10-39
10-14A. Male-mediated developmental toxicity, prenatal component 10-43
10-14B. Male-mediated developmental toxicity, postnatal component 10-44
10-15. ECs and LECs for male-mediated developmental toxicity 10-45
10-16. Male-mediated developmental toxicity 10-46
10-17. ECs and LECs for male-mediated developmental toxicity 10-47
10-18. Male-mediated developmental toxicity 10-49
10-19. ECs and LECs for male-mediated developmental toxicity 10-50
10-20. ECs and LECs for male-mediated developmental toxicity from the
combined studies 10-51
10-21. NTP chronic study (1993) 10-55
10-22. ECs and LECs for ovarian, uterine, and testicular atrophy using the
quantal Weibull and log-logistic models 10-58
10-23. Parameters for Weibull time-to-response model used to model
reproductive effects observed in mice based on ppm butadiene exposure 10-61
10-24. Human benchmark 1,3-butadiene exposure concentrations
calculated for reproductive effects observed in mice using a Weibull
time-to-response model (extra risk) 10-61
10-25. Points of departure and sample chronic RfC calculations for reproductive and
developmental effects of 1,3-butadiene 10-65
11-1. Estimates of upper bounds on human extra unit cancer risk (potency)
from continuous lifetime exposure to 1,3-butadiene based on animal
inhalation bioassays 11-3
11-2. Reference concentrations (RfCs) for 1,3-butadiene, based on mouse data 11-9
xn
-------
LIST OF FIGURES
3-1. Schematic of 1,3-butadiene metabolism 3-6
3-2. Comparison of blood concentrations of EB in male mice and rats 3-33
3-3. Comparison of blood concentrations of DEB in male mice and rats 3-34
6-1. Kaplan-Meier survival curves for male B6C3FJ mice
exposed to 1,3-butadiene by inhalation for 103 weeks 6-6
6-2. Kaplan-Meier survival curves for female B6C3FJ mice
exposed to 1,3-butadiene by inhalation for 103 weeks 6-7
6-3. Kaplan-Meier survival curves for male mice in the
stop-exposure inhalation of 1,3-butadiene 6-24
9-1. Reactions modeled in the PBPK model of Sweeney et al. (1997) 9-31
10-1. Predicted rate ratios from Health Canada (1998) Poisson regression models 10-5
10-2. Excess risk and 95% upper confidence limit estimates for leukemia mortality
based on the log-linear model reported by Health Canada (1998) 10-7
10-3. Excess risk and 95% upper confidence limit estimates for leukemia mortality
based on the power model reported by Health Canada (1998) 10-8
10-4. Excess risk and 95% upper confidence limit estimates for leukemia mortality
based on the linear model reported by Health Canada (1998) 10-9
10-5. Excess risk estimates for leukemia mortality based on the shape model reported
by Health Canada (1998) 10-10
10-6. Log-logistics model for fetal weight data, with data points and
standard error bars 10-41
10-7. Log-logistic model for data on all prenatal deaths from the three dominant
lethal studies, excluding the 1,250 ppm exposure group, with data points and
standard error bars 10-54
10-8. Ovarian atrophy (groups 1-5) using log-logistic model 10-59
10-9. Uterine atrophy (groups 1-5) using quantal Weibull model 10-59
10-10. Testicular atrophy (groups 1-6) using quantal Weibull model 10-60
Xlll
-------
PREFACE
This Health Assessment of 1,3-Butadiene was prepared to serve as a source document for
Agency-wide use. This document was developed primarily for use by the U.S. Environmental
Protection Agency's (EPA's) Office of Transportation and Air Quality (OTAQ), previously
known as the Office of Mobile Sources (OMS), to support decision making regarding Section
202(1)2 of the Clean Air Act Amendments of 1990. The exposure information included here is
an overview of the ambient exposures and exposure to populations adjacent to emission sources,
without any actual exposure assessment as such.
In the development of this assessment, relevant scientific literature has been incorporated
from the period July 1, 1985, through January 31, 2000. Key studies were evaluated to
qualitatively describe and, where possible, quantitatively identify the noncancer and cancer
effects of 1,3-butadiene. The assessment also includes a summary, conclusions, and hazard and
dose-response characterization. Measures of dose-risk relationships relevant to ambient air
exposures are discussed so that the adverse health effects can be placed in perspective with
possible exposure levels.
This final document reflects a consideration of all comments received on an External
Review Draft dated January 1998 (EPA/600/P-98/001 A) provided by the Environmental Health
Committee (EHC) of the EPA's Science Advisory Board (at a public meeting held on April 30
and May 1, 1998) and the public.
xiv
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
This document was prepared by the National Center for Environmental Assessment-
Washington Office (NCEA-W) of EPA's Office of Research and Development. A preliminary
draft of this report was prepared by Oak Ridge National Laboratory (ORNL) under Interagency
Agreement No. DW89937638-01-0. Aparna M. Koppikar1 served as the Project Manager,
providing overall direction and technical assistance.
AUTHORS
Chapter 1: Aparna M. Koppikar1
Chapter 2: ChiehWu1
Chapter 3: Teresa Leavens2
Chapter 4: Lawrence Valcovic1
ChapterS: Carole Kimmel1
Chapter 6: Rosmarie Faust3, Jennifer Jinot1
Chapter 7: Aparna M. Koppikar
Chapter 8: Jennifer Jinot and Aparna M. Koppikar
Chapter 9: Jennifer Jinot
Chapter 10: Jennifer Jinot and Carole Kimmel
Chapter 11: Jennifer Jinot
CONTRIBUTORS
Thomas M. Crisp1
Kowetha Davidson3
Carol Forsyth3
Robert Young3
Dharm Singh1
Steven Bayard1 (now at OSHA)
Milton Siegal1
John Schaum1
Leslie Stayner4
National Center for Environmental Assessment-Washington Office.
2National Health and Environmental Effects Research Laboratory (RTF)
3Oak Ridge National Laboratory.
4National Institute for Occupational Safety and Health.
xv
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
Stephen Gilbert4
Randall Smith4
Kim Hoang1 (now at Region 9)
Bruce Allen (RAS Association, Chapel Hill)
REVIEWERS
David Bayliss (NCEA-W)
Robert Beliles (NCEA-W)
Pam Brodowicz (QMS)
Robert Bruce (NCEA-Cin)
James Cogliano (NCEA-W)
Richard Cook (QMS)
Michael Dellarco (NCEA-W)
Gary Kimmel (NCEA-W)
William Pepelko (NCEA-W)
Woodrow Setzer (NHEERL-RTP)
This final document reflects a consideration of all comments received on an External
Review Draft dated January 1998 (EPA/600/P-98/001 A) provided by the Environmental Health
Committee (EHC) of the EPA's Science Advisory Board at a public meeting held on April 30 and
May 1, 1998, and public comments. The members of the EHC were as follows:
Chair
Mr. Mark J. Utell, University of Rochester Medical Center, Rochester, NY
Members
Dr. Cynthia Bearer, Case Western Reserve University, Cleveland, OH
Dr. John Doull, University of Kansas Medical Center, Kansas City, KS
Dr. David G. Hoel, Medical University of South Carolina, Charleston, SC
Dr. Abby A. Li, Monsanto Company, St. Louis, MO
Dr. Michele Medinsky, Chemical Industry Institute of Toxicology, RTF, NC
Dr. Lauren Zeise, California Environmental Protection Agency, Berkeley, CA
xvi
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS (continued)
Consultants
Dr. Richard Albertini, University of Vermont, Burlington, VT
Dr. Elaine Faustman, University of Washington, Seattle, WA
Dr. Karl Kelsey, Harvard School of Public Health, Boston, MA
Dr. R. Jeff Lewis, Exxon Biomedical Sciences, Inc., East Millstone, NJ
Dr. Judith MacGregor, Toxicology Consulting Services, Rockville, MD
Dr. David Parkinson, L.I. Occupational and Environmental Health Center, Port Jefferson, NY
Dr. Roy Short, New York University Medical School, New York, NY
Dr. James Swenberg, University of North Carolina, Chapel Hill, NC
Special Advisors
Dr. Genevieve Matanoski, The Johns Hopkins University, Baltimore, MD
Dr. Ronald Melnick, National Institutes for Environmental Health Sciences, RTF, NC
ACKNOWLEDGMENTS
The authors would like to acknowledge the contributions of several people who have
made this report possible:
• Terri Konoza of NCEA-W, who was responsible for coordinating and managing the
production effort.
• The CDM Group, Inc., under the direction of Kay Marshall, who was responsible for
editing, word processing, and literature searches.
The authors also would like to acknowledge the contributions of the following individuals
for providing additional data:
• Diana Anderson, BIBRA Toxicology International, UK
• William Eastin, NIEHS, RTF, NC
• Martin H. Brinkworth, Institute of Reproductive Medicine of the University,
Domagkstrasse, Germany
• Mike Walkers, Health Canada
xvn
-------
1. INTRODUCTION
This chapter presents background information about 1,3-butadiene, including an overview
of the scope of this health assessment and a summary of the Environmental Protection Agency's
(EPA's) previous health assessment (U.S. EPA, 1985).
1.1. BACKGROUND
1,3-Butadiene (CH2=CH-CH=CH2, CAS No. 106-99-0) is a colorless gas produced by
three different processes: (1) oxidative dehydrogenation of n-butene (the Oxo-D or O-X-D
process), (2) catalytic dehydrogenation of n-butane and n-butene (the Houdry process), and (3)
recovery from the C4 coproduct (by-product) stream from the steam-cracking process used to
manufacture ethylene (the ethylene coproduct process). This noncorrosive gas has a boiling
point of -4.4°C and a vapor pressure of 1,790 mm Hg at 20°C (Santodonato, 1985). 1,3-
Butadiene is highly volatile and has low solubility in water; thus, environmental release results
primarily in atmospheric contamination. Atmospheric decomposition of 1,3-butadiene occurs
primarily by photoinitiated reactions. A significant amount of decomposition also occurs by the
gas-phase reaction with ozone and by reaction with nitrate radicals at nighttime, particularly in
urban areas (U.S. DHHS, 1992). The major photooxidation products of 1,3-butadiene are
acrolein and formaldehyde (Maldotti et al., 1980).
Approximately 12 billion pounds of 1,3-butadiene are produced annually worldwide,
including 3 billion pounds produced in the United States (Morrow, 1990; USITC, 1990). 1,3-
butadiene is used as an intermediate in the production of polymers, elastomers, and other
chemicals. Its major uses are in the manufacture of styrene-butadiene rubber (SBR) (synthetic
rubber) and of thermoplastic resins. Elastomers of butadiene are used in the manufacture of tires,
footwear, sponges, hoses and piping, luggage, packaging, and a variety of other molded products.
In addition, 1,3-butadiene is used as an intermediate to produce a variety of industrial chemicals,
including the fungicides captan and captfol. The primary way the 1,3-butadiene is released into
the environment is via emissions from gasoline- and diesel-powered vehicles and equipment.
Lesser releases occur from the combustion of other fossil fuels and biomass. Minor releases
occur in production processes, tobacco smoke, gasoline vapors, and vapors from the burning of
plastics as well as rubber (Miller, 1978).
1.2. OVERVIEW OF THE SCOPE OF THIS HEALTH ASSESSMENT
EPA first published a health assessment of 1,3-butadiene in 1985. The purpose of this
current assessment is to review new information on 1,3-butadiene and update the conclusions
from the earlier assessment. This reassessment is intended to serve as a source document for risk
1-1
-------
assessors inside and outside the Agency, although its development was specifically prompted by
a request from EPA's Office Transportation and Air Quality (OTAQ), previously known as
Office of Mobile Sources (OMS), to support decision making regarding Section 202(1)2 of the
Clean Air Act Amendments of 1990. As requested by OTAQ, the document focuses primarily
on the following major health effects: carcinogenicity, mutagenicity, and
reproductive/developmental toxicity. A detailed exposure assessment was not requested and was
not conducted.
The remainder of Chapter 1 reviews the 1985 health assessment. For background
purposes, Chapter 2 provides a brief overview of 1,3-butadiene exposure. Chapter 3 reviews the
data on metabolism and pharmacokinetics, while Chapter 4 discusses the evidence on
mutagenicity and cytogenetic effects, including dominant lethal effects.
Toxicity of 1,3-butadiene in experimental animals is reviewed in Chapters 5
(reproductive and developmental effects) and 6 (other toxicity, primarily carcinogenicity). From
the available data, 1,3-butadiene appears to have low acute toxicity. Most data on acute toxicity
precede 1985 and are summarized by Himmelstein et al. (1997); however, they are not reviewed
in this document. From available short-term and subchronic studies, the major effects are
reproductive (e.g., dominant lethal effects) and developmental (e.g., reduced fetal weight), which
are reviewed in Chapter 5, with the exception of dominant lethal effects, which are reviewed in
Chapter 4. Some hematological/immunological effects also have been observed in subchronic
studies and are discussed briefly in Chapter 6. The major effects observed in chronic
experimental studies are carcinogenicity, reviewed in Chapter 6, and ovarian, uterine, and
testicular atrophy, reviewed in Chapter 5.
Evidence of carcinogenicity from epidemiologic studies is reviewed in Chapter 7. The
epidemiologic studies focus on workers involved in the production of either 1,3-butadiene
monomer or styrene-butadiene polymer. Chapter 8 provides a summary and characterization of
the hazard attributable to 1,3-butadiene exposure with the scope of the health effects review.
Chapter 9 reviews the physiologically based pharmacokinetic (PBPK) models that have been
developed for 1,3-butadiene. Chapter 10 presents quantitative risk estimates for 1,3-butadiene
effects. Cancer potency estimates are derived from the epidemiologic data, as well as from the
cancer bioassays for comparison purposes. Benchmark dose modeling and sample reference
concentrations (RfCs) are performed for various reproductive and developmental effects that are
candidates for the critical noncancer effect. Ultimately, a chronic RfC is presented based on the
critical endpoint of ovarian atrophy. Reference values for acute and subchronic exposures based
on the critical endpoint of reduced fetal weight are also presented in Chapter 10. Chapter 11
provides a dose-response characterization based on the information developed in Chapter 10, and
the references for the entire document are listed at the end.
1-2
-------
1.3. SUMMARY OF EPA'S CARCINOGEN ASSESSMENT (U.S. EPA, 1985)
Pertinent studies reported before 1986 were reviewed in Mutagenicity and
Carcinogenicity Assessment of 1,3-Butadiene (U.S. EPA, 1985). This document was peer-
reviewed by experts in the field, as well as in public sessions of the Environmental Health
Committee of EPA's Science Advisory Board. The studies presented in the 1985 document are
not reviewed in the present document but are briefly summarized below.
EPA reviewed six epidemiologic studies, which included four retrospective cohort
mortality studies, one nested case-control study, and an industrial hygiene and hematologic cross-
sectional survey. The first cohort study involved 6,678 hourly workers in a rubber tire
manufacturing plant in Akron, Ohio (McMichael et al., 1974). The standard mortality ratios
(SMRs) were calculated using the 1968 U.S. male population as the reference. Cause-specific
mortality was evaluated for 16 different occupational title groups (work areas) within the plant.
This study was followed up by a nested case-control study involving 455 of the 1,983 deaths
recorded between 1968 and 1973 (McMichael et al., 1976). The second cohort study was
conducted in 8,938 male workers in a rubber plant also located in Akron, Ohio (Andjelkovich et
al., 1976, 1977). The 1976 study used the U.S. male population as the reference for calculating
the SMRs, whereas for the 1977 study, the entire cohort was used to calculate the SMRs for 28
different work areas. The third cohort study included 2,756 workers at two styrene-butadiene
rubber facilities in eastern Texas (Meinhardt et al., 1982). The sex, age, race, and calendar-time
cause-specific rates of the U.S. population were used to calculate the SMRs. The last and most
comprehensive study was conducted in 13,920 workers at one Canadian and seven U.S. styrene-
butadiene rubber plants (Matanoski et al., 1982). The SMRs for black and white workers were
calculated separately. The cross-sectional survey was conducted on workers in the same styrene-
butadiene rubber plant studied by McMichael et al. (Checkoway and Williams, 1982). Blood
samples were obtained to evaluate hematology parameters. The survey was not designed to
evaluate mortality experience and did not contribute to cancer risk evaluation of 1,3-butadiene.
Of the five epidemiologic studies that evaluated cause-specific mortality, three cohort
studies demonstrated statistically significant excess mortality due to cancers of the lymphatic and
hematopoietic tissues (Andjelkovich et al., 1976; McMichael et al., 1976; Meinhardt et al.,
1982). The fourth cohort study by Matanoski et al. (1982) also showed increased leukemia, but
failed to achieve statistical significance. Lastly, the nested case-control study by McMichael et
al. (1976) showed statistically significant increased standardized risk ratios for cancers of the
lymphatic and hematopoietic tissues among workers with exposures of 5 years or more in one
area of the plant (synthetic rubber plant area), compared with either all the other workers or the
matched controls. Statistically significant excess cancer mortality was also observed for
gastrointestinal tract, respiratory tract, central nervous system, prostate, testicles, and urinary
1-3
-------
bladder in one or more studies. However, these excesses were not observed consistently across
all the studies.
Although excess mortality due to cancers of the lymphatic and hematopoietic tissues was
observed consistently in all the evaluated studies, the methodologic limitations prompted EPA to
conclude that the evidence was inadequate for determining a causal association between
exposure to 1,3-butadiene and cancer in humans. The methodologic limitations included too few
deaths from specific cancers to evaluate the causal association, exclusion of large portions of the
population due to lack of records, lack of adjustment for smoking, confounding by other
exposures such as benzene or styrene, and excess cancer mortality at other sites.
Two long-term animal studies presented strong evidence for the induction of cancers at
multiple anatomical sites in rats (HLE, 1981b) and mice (NTP, 1984). Sprague-Dawley rats
were exposed by inhalation to 1,3-butadiene at concentrations of 1,000 or 8,000 ppm 6
hours/day, 5 days/week, for 111 weeks and 105 weeks for males and females, respectively.
Statistically significant increased incidences in the following neoplasms were observed at one or
both concentrations: mammary gland tumors, thyroid follicular adenomas/carcinomas, and
Zymbal gland carcinomas in female rats and Leydig cell adenomas/carcinomas, pancreatic
exocrine adenomas, and Zymbal gland tumors in male rats. In addition, gliomas occurred in four
high-dose male rats. Non-neoplastic effects due to long-term exposure of rats to 1,3-butadiene
included clinical signs of toxicity, increase in liver weight in both sexes, marked to severe
nephropathy in 27% of the high-dose male rats compared with 9% or 10% of the controls, and
alveolar metaplasia in male rats.
Among B6C3FJ mice exposed by inhalation to 1,3-butadiene at 625 or 1,250 ppm for 6
hours/day, 5 days/week, neoplasms also developed at multiple anatomical sites; this study was
terminated at Week 60 to 61 because of high mortality in the treated groups, primarily due to
neoplasms. There was an overall increase in the number of animals with primary neoplasms and
animals with multiple neoplasms. Neoplasms showing statistically significant increased
incidences among both male and female mice were as follows: malignant lymphomas,
alveolar/bronchiolar adenomas/carcinomas, hemangiosarcomas of the heart, and forestomach
papillomas/carcinomas. In addition, mammary gland acinar cell carcinomas, ovarian granulosa
cell carcinomas, and hepatocellular adenomas/carcinomas occurred in female mice.
Nonneoplastic effects observed were testicular atrophy, chronic inflammation, fibrosis,
cartilaginous metaplasia, osseous metaplasia, and atrophy of the sensory epithelium of the nasal
cavity in male mice. Ovarian atrophy was observed in female mice. Some discrepancies were
noted for this study, but they were not considered to pose a significant impact on the overall
interpretation of the study.
1-4
-------
EPA also reviewed data from metabolism and mutagenicity studies, concluding that
inhaled 1,3-butadiene is metabolized to mutagenic epoxide intermediates.
In addition, EPA reviewed the carcinogenicity of related compounds (4-vinyl-l-
cyclohexene, epoxybutene, J/-l,2:3,4-diepoxybutane, and we50-l,2:3,4-diepoxybutane).
4-Vinyl-l-cyclohexene is carcinogenic in female mice (oral/gavage), based on increased
incidences of ovarian and adrenal gland neoplasms. Equivocal evidence was noted for malignant
lymphomas and alveolar/bronchiolar adenomas in male mice and clitoral gland neoplasms in
female rats (NTP, 1985). Skin painting of mice with we50-l,2:3,4-diepoxybutane induced
papillomas and squamous cell carcinomas (Van Duuren et al., 1963), and subcutaneous injection
with J/-l,2:3,4-diepoxybutane caused fibrosarcomas in mice and rats (Van Duuren et al., 1966).
Based on the studies in mice and rats, EPA concluded that there is sufficient evidence for
carcinogenicity of 1,3-butadiene in animals. EPA also concluded that evidence from
metabolism, mutagenicity, and carcinogenicity studies suggests that 1,3-butadiene presents a
genetic risk to humans.
Two developmental toxicity studies were reviewed. One study was conducted using
pregnant female Sprague-Dawley rats exposed to 200, 1,000, and 8,000 ppm 6 hours/day on
gestation days 6-15 (HLE, 198la). Developmental effects included slightly decreased fetal
weight and mean crown-rump length and increased skeletal variations and malformations. The
other study was inadequately reported (Carpenter et al., 1944).
Thus, in 1985, EPA presented the following conclusion regarding the qualitative
evaluation of the data for 1,3-butadiene: "On the basis of sufficient evidence from studies in two
species of rodents, and inadequate epidemiologic data, 1,3-butadiene can be classified as a
probable human carcinogen, Group B2." Using the classification scheme of the International
Agency for Research on Cancer (IARC), 1,3-butadiene would also be classified as a "probable"
human carcinogen, Group 2B.
The linearized multistage model was used to calculate the maximum likelihood estimate
for the incremental risk for 1,3-butadiene based upon the National Toxicology Program (NTP)
mouse data (NTP, 1984), the HLE (1981) rat data, and internal dosimetry derived from data on
mice and rats exposed to varying concentrations of 1,3-butadiene for 6 hours. The upper-limit
unit risk of 6.4 x 10'1 (ppm)"1 was a geometric mean of the values calculated for male and
female mice separately. This value was used to predict human responses in the epidemiologic
studies, which were then compared with the actual responses. According to EPA (U.S. EPA,
1985),
The comparisons were hampered by a scarcity of information in the epidemiologic
data concerning actual exposures, age distribution, and work histories. In
1-5
-------
addition, because there was no consistent cancer response across all of the studies,
the most predominant response, cancer of the lymphatic and hematopoietic
tissues, was chosen as being the target for 1,3-butadiene. Based on the
comparisons between the predicted and observed human response, the
extrapolated value from the mouse data was consistent with human response, but
in view of all the uncertainties and apparent inconsistencies in the epidemiologic
data, a fairly wide range of potency estimates and exposure scenarios would also
be satisfactory.
1-6
-------
2. OVERVIEW OF EXPOSURE TO 1,3-BUTADIENE
The purpose of this chapter is to present an overview of how human exposure to 1,3-
butadiene occurs. The chapter summarizes physical/chemical properties, production/use,
sources/emissions, and ambient air data. Pathways of exposure are briefly described, but no
quantitative estimates of exposure levels and numbers of people exposed are presented.
Therefore, this chapter should not be used as a source of data for enforcement purposes.
2.1. PHYSICAL/CHEMICAL PROPERTIES
1,3-Butadiene (CH2=CH-CH=CH2, CAS No. 106-99-0) is a colorless gas with mildly
aromatic odor (Sax and Lewis, 1987). It is noncorrosive and has a molecular weight of 54.09.
Its boiling point is -4.4°C (Weast, 1989) and its vapor pressure is 1,790 mm Hg (239 kPa) at
20°C (Santodonato, 1985). It is easily liquefied, with a density of 0.6211 g/mL at 20°C/liquefied
(Kirshenbaum, 1978; Verschueren, 1983). It is soluble in ethanol, diethyl ether, and organic
solvents (Verschueren, 1983; Sax and Lewis, 1987; Budavari, 1989) and very slightly soluble in
water, with a solubility of 735 mg/L at 20°C. 1,3-Butadiene has a flash point of -76°C (Sax and
Lewis, 1987), is slowly dimerized to 4-vinyl-l-cyclohexene (OSHA, 1990), and may form
peroxide upon exposure to air (Kirshenbaum, 1978). Because 1,3-butadiene is a highly volatile
gas, it is expected to partition predominantly to the atmosphere and then undergo rapid
destruction by photoinitiated reactions. The reaction with photochemically produced hydroxyl
radicals has a calculated half-life of approximately 6 hours and is expected to be the dominant
pathway for atmospheric removal (U.S. DHHS, 1992). Destruction of atmospheric 1,3-butadiene
by the gas-phase reaction with ozone and by the nighttime reaction with nitrate radicals in urban
areas is also expected to be significant (U.S. DHHS, 1992). The major photooxidation products
of 1,3-butadiene are acrolein and formaldehyde (Maldotti et al., 1980).
Data are limited on the fate of 1,3-butadiene in soil or water. Based on its physical
properties, rapid volatilization of 1,3-butadiene from either soil or water to atmosphere is
expected to dominate over all other potential environmental processes. Studies performed with
pure cultures indicate that 1,3-butadiene may be susceptible to microbial attack. On the basis of
estimated values, 1,3-butadiene is not expected to adsorb significantly to soil or sediment, nor is
it expected to bioconcentrate in fish or aquatic organisms (U.S. DHHS, 1992).
2.2. PRODUCTION AND USE
1,3-Butadiene was first produced in 1886 by the pyrolysis of petroleum hydrocarbons
(Kirshenbaum, 1978). Commercial production of 1,3-butadiene started in the 1930s (Kosaric et
al., 1987) and has involved three processes: catalytic dehydrogenation of n-butane and n-butene,
2-1
-------
oxidative dehydrogenation of n-butene, and recovery from the C4 coproduct (by-product) stream
from the steam-cracking process used to manufacture ethylene. The ethylene coproduct process
accounts for approximately 95% of U.S. and 85% of worldwide production (Morrow, 1990).
Approximately 12 billion pounds of this gas are produced annually worldwide and 3 billion
pounds in the United States (Morrow, 1990; USITC, 1990).
1,3-Butadiene is used as an intermediate in the production of polymers, elastomers, and
other chemicals. The major uses of this chemical are in the manufacture of styrene-butadiene
(SB) rubber (synthetic rubber) and of thermoplastic resins. In 1990, 1,3-butadiene was used in
the United States for SB rubber (30%), polybutadiene rubber (20%), adiponitrile/
hexamethylenediamine (15%), SB latex (10%), neoprene rubber (5%), acrylonitrile-butadiene-
styrene resins (5%), exports (4%), nitrile rubber (3%), and other (including specialty polymers)
(8%) (Anonymous, 1991).
2.2.1. SB Latex and Rubber Production
SB latex and rubber production is the major use for butadiene, accounting for 40% of
butadiene consumption. SB latex and rubber are used for a variety of products, including
automobile tires, textiles, paper, and adhesives.
The 1994 EPA report, Locating and Estimating Air Emissions From Sources of 1,3-
Butadiene, lists SB latex and rubber production as the major contributor to industrial butadiene
emissions (U.S. EPA, 1994a). About 74% of the industrial emissions are from SB latex and
rubber production. At least 26 facilities in the United States produce SB latex and rubber (SRI
International, 1993).
As stated previously, butadiene has very low water solubility and high vapor pressure;
thus, if it were released to an aqueous waste stream, it would immediately evaporate. It is then
logical to assume, and the data confirm, that the amount of butadiene found in secondary sources
such as wastewater and solid waste is minimal or nonexistent. The majority of the butadiene
releases during industrial production occur via process vents, so only emission factors for process
vents will be presented. The emission factors, as presented in the 1994 EPA report for process-
vent butadiene released during SB latex and rubber production, range from 0.00024 to 94.34 Ib
butadiene emitted per ton produced (mean of 7.10) measured at 18 facilities (U.S. EPA, 1994a).
2.2.2. Polybutadiene Production
The second largest use for butadiene is in production of polybutadiene, accounting for
more than 20% of butadiene consumption. Polybutadiene is used in tire manufacturing and in
the high-impact resin industry. Four companies at five locations in the United States currently
produce polybutadiene. The estimate for process-vent butadiene emissions from polybutadiene
2-2
-------
production, as stated in the 1994 EPA report, ranges from 0.00008 to 36.06 Ib butadiene emitted
per ton produced (mean of 6.14), measured at six facilities (U.S. EPA, 1994a).
2.2.3. Neoprene Rubber Production
Neoprene, or polychloroprene, rubber production accounts for 5% of butadiene
consumption. Neoprene rubber is primarily used in the automotive industry for belts, cables,
hoses, and wires. Three facilities currently produce neoprene, although only two use butadiene
as a raw material and the other starts with chloroprene. The two facilities identified in the 1994
EPA report that used butadiene as a raw material yield estimated process-vent butadiene
emissions from neoprene production ranging from 0.32 to 6.78 Ib butadiene emitted per ton
produced (mean of 4.04) (U.S. EPA, 1994a).
2.2.4. Acrylonitrile-Butadiene Resin Production
Acrylonitrile-butadiene (ABS) resins are used to make plastic components such as
automotive parts, pipes and fittings, appliances, telephones, and business machines, among many
other uses. ABS production accounts for 5% of butadiene consumption. Currently, 10 facilities
produce ABS resin, only 6 of which use butadiene as a raw material. The estimate for process-
vent butadiene emissions from ABS resin production ranges from 0.16 to 10.66 Ib butadiene
emitted per ton produced (mean of 4.22), measured at three facilities (U.S. EPA, 1994a).
2.2.5. Nitrile Elastomer Production
Nitrile elastomer or nitrile-butyl rubber is a specialty elastomer known for its resistance
to oil solvents and chemicals. Some uses include the manufacture of hoses, belts, cables, seals,
and gaskets. Nitrile elastomer is produced at nine facilities in the United States and accounts for
about 5% of total butadiene consumption. The estimate for process-vent butadiene emissions
from nitrile elastomer production ranges from 0.0004 to 17.80 Ib butadiene emitted per ton
produced, measured at six facilities identified in the 1994 EPA report (U.S. EPA, 1994a).
2.2.6. Adiponitrile Production
Adiponitrile (hexanedinitrile) is primarily an intermediate used in the production of nylon
6,6. Three facilities produce adiponitrile, but only two of these facilities use butadiene in
production. This accounts for 12% of butadiene consumption. Despite the large usage of
butadiene in adiponitrile production, emissions appear to be fairly small. The estimate for
process-vent butadiene emissions from adiponitrile production, based on actual emissions
reported at two facilities, is 0.12 Ib butadiene emitted per ton produced (U.S. EPA, 1994a).
-------
2.3. SOURCES AND EMISSION
1,3-Butadiene may be released to the environment as an intentional or fugitive emission
during its production, use, storage, transport, or disposal. Its sources and emission to the
environment can be classified as industrial production and use (1.6%), mobile sources (78.8%),
and other miscellaneous combustion sources (19.6%) (U.S. EPA, 1994a).
Industrial butadiene emissions arise from process vents, equipment leaks, and secondary
sources such as wastewater treatment. Because butadiene released to aqueous systems or
entering treatment plants is likely to evaporate completely, all emissions of butadiene can be
considered air emissions. Actual reported emissions of butadiene are available through the Toxic
Release Inventory, and the relative contribution of butadiene production to the national butadiene
emissions is 0.2% (U.S. EPA, 1994a). The EPA's 1996 Toxic Release Inventory report (U.S.
EPA, 2001) summarized the on-site and off-site releases of 1,3-butadiene for all industries for
1988 and 1994-1999. Significant reduction of release has been achieved, as shown in Table 2-1.
2.3.1. Mobile Sources
Butadiene is formed as a product of incomplete combustion of fossil fuels and has been
reported in the emissions from gasoline and diesel vehicles as well as aircraft. Emissions of
butadiene from combustion sources are commonly represented as a weight percentage of total
organic gas emissions. The relative contribution of mobile sources to the national butadiene
emissions is 78.8%, which includes both on-road and nonroad engines. Levels of butadiene in
gasoline and diesel fuel are expected to be insignificant, because butadiene tends to form a
varnish that can be harmful to engines; therefore, refiners try to minimize the butadiene content.
Because butadiene is not a component of gasoline, it is not present in mobile-source evaporative
or refueling emissions and is found only in exhaust emissions (U.S. EPA, 1992).
It should be noted that a recent reevaluation by Nordlinder et al. (1996) of the Concawe
report (1987) found that the concentrations of 1,3-butadiene in gasoline vapors were much lower
than had been reported. Two analyses also found negligible amounts of 1,3-butadiene in
gasoline vapors (Lofgren et al., 1991; Ramnas et al., 1994). When the investigators compared
the concentrations of benzene and butadiene in gasoline, they found concentrations to be 3% to
5% and <0.0005%, respectively. On the basis of these three reports, Nordlinder et al. (1996)
concluded that no significant amount of 1,3-butadiene is present in gasoline vapors.
2.3.1.1. On-Road Mobile Sources
On-road mobile sources include the following classes of vehicles: light-duty gasoline
vehicles (LDGVs), light-duty gasoline trucks, heavy-duty gasoline trucks, light-duty diesel
2-4
-------
Table 2-1. 1,3-Butadiene on-site and off-site releases (in pounds).
Year
1988
1994
1995
1996
1998
1999
On-site releases Off-site releases
Fugitive or On-site
nonpoint air Stack or point Surface water Underground land Total on-site Transfers off-site to
emissions air emissions discharges injection releases releases disposal
4,059,260
1,680,098
1,436,010
1,429,487
1,435,139
1,91 1,696 for
total air emission
2,945,362
1,260,880
1,613,728
1,296,809
1,303,975
522,504
7,118
5,393
11,001
8,834
1,900
1,500
0
0
1,000
732
720
7,817
396
277
263
7998
127
7,536,443
2,948,492
3,055,408
2,300,159
2,760,284
1,914,443
185,398
6,846
4,892
4,790
2,567
2,571
Total on- and
off-site
releases
7,721,841
2,955,338
3,060,300
2,743,350
2,762,851
1,917,014
to
-------
vehicles, light-duty diesel trucks, heavy-duty diesel trucks, and motorcycles. On-road mobile
sources account for 37.7% of national butadiene emissions.
Although data on the butadiene content of motor vehicle exhaust were lacking until the
late 1980s, butadiene emissions from LDGVs are now reasonably well understood. As
mentioned previously, butadiene is not a component of gasoline and is not present in evaporative
or refueling emissions; thus, only exhaust butadiene emissions are included. Butadiene has been
found to be removed effectively from motor vehicle exhaust by catalytic converters (McCabe et
al., 1992). Thus, nearly all on-road motor vehicle butadiene emissions come from older,
noncatalyst vehicles; new vehicles with nonfunctional catalysts; cold-start emissions from
catalyst vehicles; and diesel vehicles.
The emission factors calculated for all of the vehicles listed above range from 0.01 to
0.09 g/mile (U.S. EPA, 1994b). A composite emission factor of 0.0156 g/mile was calculated for
the calendar year 1990 by the EPA Office of Mobile Sources using the MOBILE model. The
composite emission factor represents all vehicles classes and is based on the percentage of total
vehicle miles traveled (VMT) attributable to each vehicle class (U.S. EPA, 1993a).
2.3.1.2. Nonroad Mobile Sources
Nonroad mobile sources include mobile gasoline- and diesel-powered equipment and
vehicles and other equipment types. Types of equipment included in this category range from
construction, industrial, and agricultural equipment to small engines used in lawnmowers, chain
saws, and other gasoline-powered equipment. Nonroad vehicles include motorcycles,
snowmobiles, golf carts, all-terrain vehicles (ATVs) used for off-road recreation, and recreational
and commercial marine vessels.
Generally, most nonroad engines are in use for many years and are noncatalyst engines.
The lack of a catalyst, in conjunction with the engine deterioration associated with increased
equipment age, may have profound effects on the amount of butadiene emitted. The EPA 1994
draft denotes that nonroad engines are expected to contribute 41% to the national butadiene
emissions (U.S. EPA, 1994a).
2.3.1.3. Aircraft
Human exposure to aircraft emissions is considered to be limited to the emissions that
occur during aircraft landing and takeoff (LTO). Airborne aircraft are assumed to fly at
sufficiently high altitudes that their emissions do not reach the surface. This assumption is likely
to be valid for butadiene because of its short atmospheric lifetime.
Butadiene has been reported in aircraft LTO emissions from military, commercial, and
general aviation. Based on the EPA SPECIATE database, the butadiene weight percentages for
2-6
-------
aircraft LTO hydrocarbon emissions range from 1.57% for general aviation (piston engines) to
1.89% from military aircraft (jet and piston engines). The 1994 EPA report estimates that 0.1%
of the national butadiene emissions is attributable to aircraft LTO (U.S. EPA, 1994a).
2.3.2. Miscellaneous Sources
The miscellaneous sources of butadiene emissions are grouped as miscellaneous chemical
production, secondary lead smelters, petroleum refining, and combustion sources (especially
biomass burning). Emissions from these sources account for 19.6% of the national butadiene
emissions.
2.3.2.1. Miscellaneous Chemical Production
The 1994 EPA report notes that butadiene is used to produce other elastomers and
plastics not mentioned previously, as well as pesticides and fungicides at 19 separate facilities in
the United States (U.S. EPA, 1994a). This process accounts for 8% of the butadiene use, but
contributes only 0.1% to the national average butadiene emissions. The emission factors for
process-vent butadiene released during miscellaneous chemical production range from 0.03 to
440 Ib butadiene emitted per ton produced (product varies), measured at only four facilities.
2.3.2.2. Secondary Lead Smelters
Secondary lead smelting involves reclamation of scrap automobile batteries to produce
elemental and lead alloys. The 23 such facilities in the United States are mostly located near
large population centers. The plastic and rubber components of the battery are the source of the
butadiene emissions, contributing 0.4% of the national butadiene emissions. The 1994 EPA
report lists uncontrolled butadiene emissions measured from a blast furnace yielding an average
emission factor of 0.79 Ib/ton (U.S. EPA, 1994a).
2.3.2.3. Petroleum Refining
The 1992 Toxic Release Inventory (TRI) contains the emission factor 270,389 Ib/year for
petroleum refining. Using this emission factor would make this source one of the largest
emitters of butadiene, contributing 0.2% to the national butadiene emissions. However, a recent
TRI report indicates that the release of 1,3-Butadiene from petroleum refining has decreased
significantly. The total air emission of this chemical for 1998 and 1999 was 150,705 Ibs/yr and
144,222 Ibs/yr respectively.
2-7
-------
2.3.2.4. Combustion Sources
Butadiene is, as mentioned previously, a product of incomplete combustion and has been
reported in the emissions from gasoline and diesel vehicles, as well as aircraft. Butadiene is also
released during the combustion of tobacco, biomass, and automobile tires, and may be released
from home heatings. However, only the combustion of biomass and automobile tires will be
discussed in this section because of the scarcity of data.
2.3.2.4.1. Tire burning. A 1992 report (Lemieux and DeMarini, 1992) showed that
approximately 240 million tires are discarded annually, of which only 25% are recycled. The
remaining tires are discarded in landfills, stockpiles, or illegal dumps. Tires are combusted
through accidental fires at stockpiles, illegal burning, tire-to-energy facilities, cement kilns, tire
manufacturing facilities, and as a supplemental fuel in boilers. Butadiene is a major constituent
of the tire manufacturing process and is present in emissions from uncontrolled tire burning.
Emission factors have been calculated for the open burning of tires (U.S. EPA, 1992; Lemieux
and DeMarini, 1992). These emission factors range from 234.28 Ib per 1,000 tons of chunk tires
to 277.95 Ib per 1,000 tons of shredded tires. No emission factor for butadiene from tire
incineration has been located. The Rubber Manufacturers Association (RMA) recently indicated
that annual scrap tire generation is now in the range of 260 to 270 million units per year (RMA,
1998). RMA claims that approximately 75% of annual scrap tire generation is now directed to
markets that make beneficial use of the scrap tire. Of the remaining 25%, most is directed to
legal landfilling or monofilling. Only a small number of tires are stockpiled or illegally dumped.
RMA also suggests, based on a 1994 EPA study (U.S. EPA, 1994a), that butadiene emissions
from the use of tires as a fuel in a well-controlled facility are negligible.
2.3.2.4.2. Biomass burning. Biomass burning includes residential wood combustion in both
fireplaces and wood stoves, open burning such as backyard burning of yard waste, slash burning,
land clearing/burning, agricultural burning, forest fires/prescribed burning, structural fires, and
other wildfires. Although these fires differ in many important characteristics, the fuels in all
cases are composed primarily of wood. The relative contribution of biomass burning to the
overall national butadiene emissions was calculated at 18.8% in the 1994 EPA report (U.S. EPA,
1994a).
Emission factor models based on field and laboratory data were developed by the U.S.
Forest Service (Peterson and Ward, 1989). These models incorporate variables such as fuel type
and combustion type (flaming or smoldering) and correlate butadiene emissions with carbon
monoxide emissions to develop emission factors for biomass burning (Campbell and Mangino,
2-8
-------
1994). The calculated emission factors range from 0.40 Ib/ton of yard waste burned to 0.90
Ib/ton for large wood burning in forest fires and prescribed burning.
Butadiene emissions have been reported from the combustion of wood (Sandberg et al.,
1975; Ward and Hao, 1992). The data of Ward and Hao (1992), in which both butadiene and
benzene were quantified from biomass burning, provide a butadiene:benzene ratio of 0.36 for
wood smoke.
2.4. AMBIENT CONCENTRATION OF 1,3-BUTADIENE
2.4.1. Air
The dominant sources for the release of 1,3-butadiene to the atmosphere are fugitive or
accidental emissions during its manufacture, use, transport, and storage. Low levels of 1,3-
butadiene are constantly emitted to the atmosphere from many sources, including exhaust from
motor vehicle engines using petroleum-based fuels (ATSDR, 1991).
2.4.1.1. Ambient Monitoring Data
Several EPA databases contain results of various air toxics monitoring programs. These
programs have set up monitoring devices that are used to collect air samples all over the United
States over a period of months or years. Three of these programs/databases contain data on 1,3-
butadiene. This section summarizes the three monitoring programs and presents annual average
concentrations of 1,3-butadiene derived from these programs.
One of these programs is the Aerometric Information Retrieval System (AIRS), which
became operational in 1987 and uses a network of monitoring stations called the State and Local
Air Monitoring System (SLAMS) (U.S. EPA, 1989a). This network consists of monitoring
stations set up by every State in accordance with regulations promulgated in response to
requirements of the Clean Air Act. EPA's Office of Air Quality Planning and Standards
(OAQPS) administers the AIRS program.
The AIRS program allows State and local agencies to submit local air pollution data and
also have access to national air pollution data (U.S. EPA, 1989a). EPA uses data from AIRS to
monitor the States' progress in attaining air quality standards for ozone, carbon monoxide,
nitrogen oxides, sulfur oxides, and lead through the use of State Implementation Plans (SIPs). In
addition to containing information about each monitoring site, including its geographic location
and operator, AIRS contains extensive information on ambient levels of many toxic compounds.
The AIRS catalogs ambient air pollution data from 18 to 55 monitors in 15 to 23 urban areas,
depending on the pollutant. These monitors collect a 24-hour sample every 12 days. However,
in some cases not every target compound was detected in every sample. Where this occurred,
half the minimum detection limit was used to average the data for this summary. The annual
2-9
-------
average ppb for each site was calculated using only those sites that provided four quarters of
monitoring data. The cities monitored and the average concentrations determined at each
monitoring site are listed in Table 2-2A for 1988 to 1994 and in Table 2-2B for 1992 to 1994.
Another air monitoring program is the Urban Air Toxic Monitoring Program (UATMP),
which EPA developed in 1987 to assist State and local agencies in determining the nature and
extent of urban air toxic pollution (McAlister et al., 1989, 1990, 1991; Wijnberg and Faoro,
1989). Data from the UATMP also are used in air toxic risk assessment models (U.S. EPA,
1989b,c; 1990a,b). In 1989, the UATMP had 14 monitors in 12 urban areas, and in 1990, the
UATMP had 12 monitors in 11 urban areas, of which 9 also participated in the 1989 monitoring
program.
In 1989 and 1990, the UATMP network simultaneously monitored 37 nonmethane
organic compounds, selected metals, benzo(a)pyrene (1989 only), formaldehyde, acetaldehyde,
and acetone for a 24-hour period once every 12 days. The UATMP database lists the data
collected from the monitoring network using two methods. In the first method, only the
concentrations above the detection limit of the compound are included in the data. In the second
method, if the concentration of a compound is below the detection limit, then one-half of the
compound's detection limit is incorporated into the data. The second method was used because
it seemed more reasonable and allowed a greater number of samples to be averaged. Data
collected in 1989 and 1990 were used in this summary. The cities monitored and the average
concentrations determined at each monotoring site can be found in Table 2-3.
Monitoring data for the UATMP collected from 1991 to 1994 have not yet been released
as separate reports. The data collected in those 4 years were entered into and reported as part of
the 1991-1994 AIRS database.
The National Ambient Volatile Organic Compounds (NAVOC) database contains
approximately 175,000 records on the concentrations of 320 volatile organic compounds (VOCs)
observed in 1-hour air samples taken every 24 hours between 1970 and 1987 (Shah and
Heyerdahl, 1988; Hunt et al., 1988). However, only the most current NAVOC data, taken during
1987, are used in this summary. In averaging data for this summary, samples that had nondetects
of 1,3-butadiene were included as one-half the detection limit. The NAVOC database includes
air samples collected using indoor and outdoor monitoring devices. Personal monitors were also
used. The types of locations of outdoor monitoring sites included remote, rural, suburban, and
urban areas, as well as near specific point sources of VOCs. Indoor monitoring sites consisted of
nonindustrial workplaces and residential environments. Personal monitors also are included in
the indoor category. This database was an interim precursor to the air toxics portion of AIRS.
For this summary, only the outdoor urban data were used. The cities monitored and the average
concentrations determined at each monitoring site can be found in Table 2-4.
2-10
-------
Table 2-2A. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information Retrieval System (AIRS)
for 1988 to 1991.
Sampling site
Washington, DC
Ft. Lauderdale, FL
Miami, FL
Miami, FL
Miami, FL
Miami, FL
Chicago, IL
Chicago, IL
Chicago, IL
St. Clair Co., IL
St. Clair Co., IL
St. Clair Co., IL
St. Clair Co., IL
Wichita, KS
Wichita, KS
Louisville, KY
Baton Rouge, LA
Baton Rouge, LA
Baton Rouge, LA
Land use of monitor location
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Residential/suburban
Mobile/urban & center city
Industrial/suburban
Industrial/Suburban
Industrial/suburban
Industrial/suburban
Residential/suburban
Residential/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Average concentration (ppb)
1988 1989 1990 1991
0.13
0.29
0.14
0.19
0.14
0.13
0.10
0.29
0.73
0.22
0.31
0.30
0.25
0.13
0.16
0.44
0.43
0.41
0.53
Number of samples
1988 1989 1990 1991
7
29
31
6
6
6
10
25
9
29
4
4
4
29
8
6
29
4
4
-------
Table 2-2A. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information Retrieval System (AIRS)
for 1988 to 1991 (continued).
Sampling sites
Baton Rouge, LA
Detroit, MI
St. Louis, MO
Camden, NJ
Camden, NJ
Camden, NJ
Camden, NJ
Newark, NJ
Plainfield, NJ
New York, NY
New York, NY
Dallas, TX
Houston, TX
Burlington, VT
Arlington Co., VA
Henrico Co., VA
Hampton, VA
Hopewell, VA
Roanoke, VA
Land use of monitor location
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Residential/suburban
Residential/suburban
Residential/suburban
Residential/suburban
Industrial/urban & center city
Residential/suburban
Residential/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Residential/suburban
Commercial/urban & center city
Commercial/urban & center city
Residential/suburban
Residential/suburban
Residential/suburban
Residential/suburban
Average concentration (ppb)
1988 1989 1990 1991
0.37
0.33 0.07
0.12
0.23
0.16
0.18
0.15
0.26
0.20
0.25
0.29
0.11
1.11 0.60 0.72
0.47
0.20 0.11
0.16 0.12
0.22 0.11
0.13 0.06
0.24 0.12
1,3 -Butadiene Average
0.67 0.26 0.29 0.10
Number of samples
1988 1989 1990
4
19
28
29
4
4
4
9
9
9
9
23
6 30 4
6
13
21
14
16
14
1991
28
18
12
22
15
22
to
to
-------
Table 2-2B. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information Retrieval System (AIRS)
for 1992 to 1994.
Sampling sites
Jefferson Co., AL
Jefferson Co., AL
Tarrant City, AL
Tarrant City, AL
Shelby Co., AL
Fresno, CA
Clovis, CA
Bakersfield, CA
Bakersfield, CA
Bakersfield, CA
Los Angeles, CA
Roseville, CA
Citris Heights, CA
Sacramento, CA
El Cajon, CA
Simi Valley, CA
Washington, DC
Washington, DC
Chicago, IL
Land use of monitor location
Residential/rural
Residential/rural
Residential/suburban
Residential/suburban
Agricultural/rral
Residential/suburban
Residential/urban & center city
Residential/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Residential/urban & center city
Mobile/suburban
Residential/suburban
Residential/suburban
Commercial/suburban
Residential/suburban
Commercial/urban & center city
Commercial/urban & center city
Industrial/urban & center city
Average concentration
1992 1993
0.12
4.32
0.07
0.30
0.28
0.65
0.16
0.41
0.34
0.18
0.36
0.32
0.09
(ppb)
1994
0.33
0.19
0.78
0.91
0.15
0.30
0.54
0.27
0.35
0.63
0.53
0.15
0.53
0.27
0.17
0.28
0.46
0.18
Number of samples
1992 1993
83
82
50
30
30
26
23
6
28
28
10
13
6
1994
79
9
81
10
78
31
111
9
23
105
30
31
84
28
29
16
15
14
to
oo
-------
Table 2-2B. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information Retrieval System (AIRS)
for 1992 to 1994 (continued).
Sampling sites
Lemont, IL
St. Clair Co., IL
Kansas City, KS
Baton Rouge, LA
Glen Burnie, MD
Essex, MD
Baltimore, MD
Baltimore, MD
Baltimore, MD
Baltimore, MD
Baltimore, MD
Alma, MI
Portage, MI
Midland, MI
Midland, MI
Midland, MI
Midland, MI
Midland, MI
Midland, MI
Midland, MI
Detroit, MI
Land use of monitor location
Residential/suburban
St. Louis metro area
Industrial/urban & center city
Commercial/urban & center city
Commercial/suburban
Residential/suburban
Residential/suburban
Industrial/suburban
Industrial/urban & center city
Residential/urban & center city
Industrial/urban & center city
Commercial/rural
Industrial/suburban
Commercial/rural
Commercial/rural
Industrial/rural
Agricultural/rural
Industrial/rural
Residential/suburban
Residential/rural
Commercial/urban & center city
Average concentration
1992 1993
0.18
0.09
0.10
0.26
0.09
0.45
0.06
0.08
0.07
0.07
0.07
0.07
0.07
0.07
0.72
0.06
0.10
0.17
0.42
0.15
0.25
0.17
0.12
0.25
0.10
0.05
0.08
0.09
0.07
0.07
0.07
0.08
0.07
0.07
0.08
(ppb)
1994
0.08
0.11
0.26
0.43
0.22
0.26
0.28
0.07
0.07
0.07
0.07
0.07
0.07
0.07
Number of samples
1992 1993 1994
56
21
50
58
39
22
25
56
55
55
55
55
28
28
24
15
10
11
14
60
39
59
50
57
48
14
4
61
61
61
61
60
31
30
15
13
12
13
8
54
58
48
60
22
61
61
61
11
11
to
-------
Table 2-2B. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information Retrieval System (AIRS)
for 1992 to 1994 (continued).
Sampling sites
Camden, NJ
Newark, NJ
Plainfield, NJ
Nassau Co., NY
Philadelphia, PA
San Antonio, TX
Clute, TX
Brownsville, TX
Brownsville, TX
Dallas, TX
Dallas, TX
Odessa, TX
Midlothian, TX
El Paso, TX
El Paso, TX
El Paso, TX
El Paso, TX
El Paso, TX
Texas City, TX
Texas City, TX
Harris Co., TX
Land use of monitor location
Residential/suburban
Industrial/urban & center city
Residential/suburban
Commercial/suburban
Residential/suburban
Residential/suburban
Residential/suburban
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Industrial/rural
Residential/suburban
Agricultural/rural
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/suburban
Residential/suburban
Residential/urban & center city
Residential/suburban
Agricultural/suburban
Average concentration
1992 1993
0.26
0.63
0.72
0.19
0.24
0.18
0.29
0.32
0.20
0.41
0.18
0.45
0.98
0.83
0.21
0.29 0.22
0.35
0.19 0.30
(ppb)
1994
0.49
0.73
0.79
0.70
0.08
0.37
0.21
0.37
0.37
1.68
0.37
0.43
1.16
1.49
1.29
0.38
1.54
0.29
0.56
Number of samples
1992 1993
14
8
8
38
29
7
19
82
34
39
39
14
79
8
22
5 35
12
5 28
1994
13
9
8
9
21
53
15
59
58
59
51
15
79
9
4
58
51
15
59
to
-------
Table 2-2B. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information Retrieval System (AIRS)
for 1992 to 1994 (continued).
Sampling sites
Houston, TX
Houston, TX
Houston, TX
Beaumont, TX
Beaumont, TX
Port Arthur, TX
Port Neches, TX
Port Neches, TX
Port Neches, TX
Corpus Christi, TX
West Orange, TX
Smith Co., TX
Fort Worth, TX
Fort Worth, TX
Fort Worth, TX
Grapevine, TX
Austin, TX
Burlington, VT
Rutland, VT
Waterbury, VT
Land use of monitor location
Industrial/suburban
Industrial/suburban
Industrial/suburban
Residential/suburban
Residential/suburban
Residential/suburban
Industrial/suburban
Residential/urban & center city
Residential/suburban
Commercial/suburban
Residential/suburban
Mobile/rural
Commercial/urban & center city
Commercial/urban & center city
Commercial/suburban
Residential/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/urban & center city
Commercial/suburban
Average concentration
1992 1993
0.44
0.24
0.05 0.39
0.56
0.05 0.25
0.10 0.23
0.32
5.22
0.18
0.20
0.31
0.15
0.08
(ppb) Number of samples
1994 1992 1993 1994
0.81
0.64
0.73 6
0.37 7
0.44 6
0.43
5.95
6.11
0.37
0.18
0.33
0.57
0.80
0.06
0.05
0.13
0.49
0.44
0.09
39
33
37
78
36
45
18
11
39
80
13
13
14
54
58
58
59
57
58
23
14
52
27
32
77
8
9
15
21
15
15
15
1,3-butadiene average concentrations
across sites by year
0.16 ppb 0.40 ppb
0.59 ppb
to
-------
Table 2-3. Summary of 1,3-butadiene ambient data from the Urban Air Toxics Monitoring Program (UATMP).
Sampling sites
Baton Rouge, LA
Chicago, IL
Camden, NJ
Dallas, TX
Fort Lauderdale,
Houston, TX
Miami, FL
Pensacola, FL
St. Louis, MO
Sauget, IL
Washington, DC-1
Washington, DC-2
Wichita, KS-1
Wichita, KS-2
Port Neches, TX
Orlando, FL
Toledo, OH
Land use of monitor location
Urban/center city -commercial
Suburban-residential
Suburban-residential
Urban/center city-commercial
Urban/center city -commercial
Suburban-residential
Urban/center city -commercial
Suburban-industrial
Urban/center city -commercial
Suburban-industrial
Urban/center city -commercial
Urban/center city -commercial
Urban/Center City -/Residential
Suburban-residential
Suburban-residential
Urban/center city -commercial
Suburban-residential
Sampling time/frequency
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
24-hour/every 12 days
Average concentration (ppb)a
1989 1990
0.39
0.24
0.20
0.08
0.20
0.60
0.11
0.09
0.20
0.11
0.29
0.16
0.09
1,3-butadiene
0.21
0.36
0.06
0.10
0.47
0.06
0.06
0.10
0.15
0.06
11.09
0.10
0.06
average
1.02
Detected/total
1989 1990
11/31
10/27
19/32
8/25
18/31
23/34
7/33
12/30
7/31
9/27
19/27
10/31
7/31
4/21
8/29
4/29
9/30
111/28
6/42
2/27
11/30
12/27
1/30
24/28
8/28
to
aThe arithmetic average concentration of all samples using half minimum detection limit (MDL) value for samples in which compound was not found.
bCalculated by averaging all 390 samples taken from 13 sites equally in 1989 and 349 samples from 12 sites in 1990.
-------
Table 2-4. Summary of outdoor urban data from the National Ambient Volatile Organic Compounds (NAVOC)
Database.
Sample years
9/22/87 to 10/4/87
9/16/87 to 9/28/87
10/4/87 to 10/4/87
10/4/87 to 10/4/87
9/9/87 to 10/7/87
9/29/87 to 9/27/87
Average concentration
used (ppb)a
0.30
0.35
0.60
0.40
0.25
0.30
Number of
samples
2
2
1
1
2
1
Sampling siteb
Bakersfield, CA
Concord, CA
Fremont, CA
Richmond, CA
San Jose, CA
Stockton, CA
Land
Urban
Urban
Urban
Urban
Urban
Urban
use of monitor
location
to
oo
"Overall 1,3-butadiene average concentration: 0.344 ppb. Calculated by averaging all nine samples taken from six cities equally.
bNumber of samples/sampling time equals 1/24 hours.
-------
Table 2-5 summarizes the average concentrations (in ppb) of 1,3-butadiene found at the
monitoring sites of each air monitoring program. The table also shows the total number of
observations for each average and the number of sites that monitored the compounds in each
program. For AIRS, the average concentrations of 1,3-butadiene are listed separately for 1987
through 1994. Some of the highest averages in the AIRS database were from suburban
residential sites in Houston and Port Neches, TX. Both of these cities have high point-source
emissions that could be affecting the monitor. The AIRS and UATMP data from Houston and
Port Neches were excluded to create alternate annual averages (ppb and Hg/m3) for the years
1988 through 1994 (where applicable) and are presented in Table 2-5. This alternate annual
average may be more representative of areas that are not near strong point sources.
Tables 2-6, 2-7, and 2-8 regroup and summarize Tables 2-2, 2-3, and 2-4 according to
the sampling locations, that is, rural, suburban, or urban settings. The data obtained from Port
Neches were not included in these averages because of the elevated levels caused by industrial
emissions.
It should be noted that methods of averaging the data are not consistent between the air
monitoring databases. Also, in the NAVOC monitoring network, samples were taken for 1 hour
in a 24-hour period, whereas the other monitoring networks collected a 24-hour air sample every
12 days.
It also should be noted that the ambient levels detected in these three databases are not
completely representative of an individual's actual exposure to 1,3-butadiene. Ideally, exposure
would be estimated on the concentrations in microenvironments and the times that people spend
in these microenvironments.
2.4.1.2. Ambient Source Apportionment
Measured concentrations in ambient air include contributions from a variety of source
categories. If sufficient information is available, such ambient monitoring data can be adjusted
to represent the amount attributed to a particular source using emissions inventory
apportionment. Three studies attempt to apportion sources according to their contribution to
ambient levels of 1,3-butadiene (U.S. EPA, 1994a; CARB, 1992; Ligocki, 1993). The studies
assume that all emissions to the atmosphere contribute proportionally to ambient concentration.
These three studies are summarized in Table 2-9. As observed in Table 2-9, the source
apportionment conducted by Systems Applications International for the American Automobile
Manufacturers Association (Ligocki, 1993) contains the category of biomass burning as a large
part of the inventory. The weight percentage of butadiene in total organic gases (TOG) for
emissions from residential wood combustion, open burning, forest fires, and other burning that
are used in this analysis are derived from a single estimate provided in EPA's SPECIATE
2-19
-------
Table 2-5. Summary of air monitoring program results for 1,3-butadiene.
AIRS
1988 level
Cone.
# Observation
# Sites
1989 Level
Cone.
# Observation
# Sites
1990 Level
Cone.
# Observation
# Sites
1991 Level
Cone.
# Observation
# Sites
1992 Level
Cone.
# Observation
# Sites
1993 Level
Cone.
# Observation
# Sites
1994 Level
Cone.
# Observation
# Sites
Annual average
DDb (U2/m3)
0.67 (1.48)
18
3
0.23 (0.57)
399
30
0.29 (0.64)
101
7
0.10(0.22)
117
6
0.16(0.40)
656
20
0.40 (0.88)
2069
64
0.59(1.30)
2666
70
Alternate annual average3
DDb (U2/m3)
0.46 (1.02)b
12
2
0.25 (0.55)b
369
29
0.21 (0.46)b
97
6
C
0.16(40)b
650
19
0.32 (0.71)d
1,931
59
0.42 (0.92)d
2,401
64
2-20
-------
Table 2-5. Summary of air monitoring program results for 1,3-butadiene
(continued).
UATMP
1989 Level
Cone.
# Observation
# Sites
1990 Level
Cone.
# Observation
# Sites
NAVOC
1987 Level
Cone.
# Observation
# Sites
"Alternate averages
sources.
Annual average Alternate annual average3
DDb (U2/m3) DDb (U2/m3)
0.21 (0.46)
390
13
1.02(2.25) 0.12(0.27)b
349 293
12 10
0.34 (0.75)f no data
9
6
do not include data from Houston and Port Neches, TX, due to impacts from strong point
°Houston, TX, was not monitored during this 4-quarter period.
dAverage ppb from all sites, excluding Houston and Port Neches, TX.
ePort Neches, TX, was not monitored during this 4-quarter period.
fAll urban California sites.
2-21
-------
Table 2-6. Summary of 1,3-butadiene ambient data from the EPA Aerometric Information Retrieval System (AIRS)
(Table 2-2A and 2-2B) based on sampling locations.
Year
1988
1989
1990
1991
1992
1993
1994
Rural area
Total samples/
number of
Range Average3 locations
0.07-0.45 0.13 271/6
0.05-0.20 0.10 494/10
0.07-0.37 0.18 522/11
Range
~
0.13-0.60
0.16-0.72
0.06-0.12
0.05-0.19
0.06-4.32
0.06-1.68
Suburban
Average
1.11
0.25
0.29
0.10
0.10
0.41
0.45
area
Total samples/
number of
locations
6/1
151/12
95/5
71/4
154/7
864/31
1135/32
Range
0.44-
0.10-
0.20-
0.07-
0.09-
0.08-
0.05-
Urban
Average
0.46
0.27
0.27
0.09
0.29
0.31
0.62
area
Total samples/
number of
locations
12/2
237/18
32/2
46/2
176/5
580/21
780/24
to
to
to
a 1,3-Butadiene average concentration in ppb.
Table 2-7. Summary of 1,3-butadiene data from Table 2-3 based on sampling locations.
Year
1989
1990
Range
0.09-0.60
0.06-0.47
Suburban
Total samples/ number of
Average a locations
0.27 155/5
0.14 177/6
Urban
Range
0.08-0.39
0.06-0.36
Average
0.18
0.15
Total samples/number of
locations
235/8
144/5
al,3-Butadiene average concentration in ppb.
-------
Table 2-8. Summary of 1,3-butadiene data from Table 2-4 based on sampling locations.
Urban
Year
Range
Averagea
Total samples/number of
locations
1987
0.25-0.60
0.37
9/6
al,3-Butadiene average concentration in ppb.
to
to
-------
Table 2-9. Summary of the relative contributions to ambient 1,3-butadiene
emissions given as percentage of total mg/year.
Study
EPA, 1994a
CARB, 1992
Ligocki, 1993
Mobile
sources"
78.7
96d
57
Stationary point
and area sources'5
2.4
4
5
Aircraft
0.1
Included under
mobile sources
3
Biomass
burning0
18.8
NDe
35
"Mobile sources include on-road and off-road vehicles and generally exclude trains and aircraft.
bArea and point sources generally include all manufacturing and industrial process, oil and gas production facilities,
commerce, residential fuel combustion, and other stationary fuel combustion.
°Biomass burning includes residential wood combustion, incineration, and other biomass burning.
dThe CARB off-road apportionment of mobile sources includes trains and aircraft.
eND=not determined.
database. An actual measurement of 1,3-butadiene TOG weight percentage for incineration of
wood was not found in the literature; therefore, the solid waste incineration TOG weight
percentage was used. A great deal of uncertainty is connected with the 1,3-butadiene emission
estimates that were developed for the biomass burning as well as for emissions from aircraft.
Many of the limitations revolve around the lack of real-world data on actual 1,3-butadiene
emissions and exposures for the scenarios mentioned above, as well as the allocation of these
scenarios nationwide. The emissions from residential wood combustion and forest fires vary by
season and region of the country. The mobile source and stationary source emissions would, for
the most part, remain constant throughout much of the year.
2.4.2. Indoor Exposure to 1,3-Butadiene
Information on 1,3-butadiene concentrations in homes or public buildings is limited at
this time. Indoor concentrations of 1,3-butadiene depend primarily on the presence of
environmental tobacco smoke (ETS) (CARB, 1992). Several studies indicate that on average
most individuals spend anywhere from about 60% to 70% of their time each day indoors at their
residence (Robinson et al., 1989; U.S. EPA, 1993b). In addition, individuals also spend a lot of
time at indoor workplaces. This makes indoor air a major route of exposure to 1,3-butadiene for
individuals who are exposed to tobacco smoke. It is also apparent that the potential for indoor
exposure can exceed outdoor exposure if ETS is taken into consideration. Lofroth et al. (1989)
and Brunnemann et al. (1990) measured 1,3-butadiene emissions in sidestream smoke ranging
from 200 to 400 |J,g/cigarette and 1,3-butadiene levels in smoke-filled bars ranging from 2.7 to
2-24
-------
19 |-ig/m3. As discussed in Section 2.4.5, heating of cooking oils can release butadiene to the
indoor air. Fugitive emissions from wood stoves or fireplaces could also release butadiene to the
indoor air. Further research and measurements are needed to quantify typical indoor 1,3-
butadiene exposures.
2.4.3. Inside Automobile
The 1993 Motor Vehicle Related Air Toxics Study (U.S. EPA, 1993a) reported that a
short-term microenvironment exposures study revealed that the in-vehicle exposure level of 1,3-
butadiene was determined to have a mean of 3.0 |-ig/m3 and a maximum measured level of 17.2
|-lg/m3. Similar levels were found in the air surrounding moving vehicles. This is based on
measurements made by a monitor mounted on top of a moving vehicle that showed a mean of 3.0
|-lg/m3 and a maximum of 6.9 |-lg/m3.
2.4.4. Water
The presence of 1,3-butadiene in drinking water in the United States has been noted in
the literature, but no concentrations or frequency of detection are available (U.S. EPA, 1978;
Kraybill, 1980).
2.4.5. Food
Certain cooking oils release butadiene to the air on heating. This probably occurs
primarily by volatilization, but could also involve formation due to low levels of combustion.
For example, 1,3-butadiene emissions are approximately 22-fold higher from unrefined Chinese
rapeseed oil than from heated peanut oil. Of three fatty acids tested, heated linolenic acid
produced the greatest amount of 1,3-butadiene. Although cooking oils in the United States are
refined for purity, U.S. rapeseed oil (canola) also emitted 1,3-butadiene (Shields et al., 1995).
Also, levels of <0.2 |-ig/kg 1,3-butadiene were found in retail soft margarine; the plastic tubs
containing the margarine contained from less than 5 |-ig/kg to less than 310 |-ig/kg (Startin and
Gilbert, 1984).
2.5. MODELED ON-ROAD MOBILE SOURCE BUTADIENE EXPOSURE ESTIMATES
A recently published study by the EPA Office of Mobile Sources discussed the
assessment of motor vehicle air toxics emissions and exposure for benzene, acetaldehyde,
formaldehyde, 1,3-butadiene, methyl tributyl ether (MTBE), and diesel particulate (U.S. EPA,
1999a). The analysis reflects emissions from only on-road motor vehicles (highway vehicles
only) located in nine urban areas: Chicago, Denver, Houston, Minneapolis, New York City,
Philadelphia, Phoenix, Spokane, and St. Louis. Modeling was performed for 1990, 1996, 2007,
2-25
-------
and 2020, and separate estimates were prepared for winter, spring, summer, and fall. The
forecast years include both base and control scenarios.
The results of the exposure analysis for 1,3-butadiene from on-road vehicle emissions are
presented in Table 2-10 for all nine urban areas. The fleet-average annual average toxics
exposures in the table show significant reductions between 1990 and 2020 with no further
vehicle or fuel controls. This is because of fleet turnover resulting in full implementation of the
Federal emission control regulations currently in place.
Exposure estimates also were prepared for three different demographic groups: total
population, outdoor workers, and children 0-17 years of age. (The estimates given in Table 2-10
are for the total population.) This study found that exposure to air toxics for outdoor workers is
generally about 20% higher than for the total population, whereas exposure for children is
typically slightly below the total population. This information appears in Table 2-11, which
shows the annual average 1,3-butadiene exposure for the three demographic groups analyzed in
this study for Chicago under the base control scenario. As seen in the table, 1,3-butadiene
exposure is highest for outdoor workers, whereas children and the total population show similar
results.
2.6. CONCLUSIONS ABOUT PATHWAYS OF EXPOSURE
Although 1,3-butadiene undergoes rapid destruction in the atmosphere, it is almost
always present at very low concentrations (generally in the range of 0.1 to 1.0 ppb) in urban and
suburban areas. This is primarily because vehicle exhaust provides a continuous source of 1,3-
butadiene release to the atmosphere. Although the data are limited, it appears that butadiene
levels in other environmental media and food are very low. On this basis it is generally assumed
that inhalation is the dominant route of human exposure.
Butadiene has been measured in cigarette smoke and therefore smokers are likely to have
elevated inhalation exposures over nonsmokers. Indoor inhalation exposure may be elevated
over outdoor levels because of butadiene emissions from cigarette smoke, wood burning, and
cooking operations involving some oils. Elevated inhalation exposures may occur to individuals
living near butadiene production facilities or facilities where it is made into polymeric materials.
Inhalation exposures are also likely to be higher inside vehicles and near roadways than in other
locations. Better data are needed on butadiene levels in these microenvironments (i.e., vehicles
and residences) in order to more accurately estimate total exposure for the general population.
2-26
-------
Table 2-10. Annual-average exposure results (|lg/m3) for 1,3-butadiene,
total population - all on-road vehicles, base control scenario.
Area
Chicago
Denver
Houston
Minneapolis
New York
Philadelphia
Phoenix
Spokane
St. Louis
Calendar year
1990
0.100
0.139
0.081
0.190
0.229
0.130
0.150
0.186
0.076
1996
0.057
0.104
0.060
0.122
0.123
0.073
0.112
0.126
0.071
2007
0.028
0.049
0.031
0.082
0.063
0.028
0.045
0.055
0.030
2020
0.025
0.045
0.027
0.089
0.048
0.023
0.044
0.046
0.028
Table 2-11. Annual average exposure results (|Ig/m3) for 1,3-butadiene in Chicago
by demographic group - all on-road vehicles, base control scenario.
Demographic
group
Total
population
Outdoor
workers
Children
0-17 years
Calendar year
1990
0.100
0.120
0.098
1996
0.057
0.069
0.056
2007
0.028
0.034
0.028
2020
0.025
0.030
0.024
2-27
-------
Occupational exposure to 1,3-butadiene is expected to occur at facilities that manufacture
1,3-butadiene or convert it into commercial polymers. Toll collectors, bus drivers, or other
workers who spend extended time periods near roadways are likely to have higher exposures
than the general population.
2-28
-------
3. PHARMACOKINETICS
Data from both in vitro and in vivo studies have established that metabolites, not the
parent compound, cause the toxic effects of exposure to 1,3-butadiene. Because differences have
been noted in the toxic responses to 1,3-butadiene among laboratory species, the
pharmacokinetics of 1,3-butadiene and its metabolites are important in assessing the risk of
cancer and reproductive and developmental effects associated with exposure to this chemical.
The U.S. Environmental Protection Agency (U.S. EPA, 1985) and the International Agency for
Research on Cancer (IARC, 1986) previously have reviewed the pharmacokinetics of 1,3-
butadiene. This chapter summarizes the recent research that has provided information on the
pathways and rates of metabolism of 1,3-butadiene, via both in vitro and in vivo studies, and the
in vivo disposition of 1,3-butadiene in several animal species, including molecular dosimetry and
urinary metabolites from occupationally exposed humans. This chapter is divided into sections
according to the various components of pharmacokinetics: absorption, biotransformation,
distribution of 1,3-butadiene and its metabolites, and excretion. In addition, a section is included
for the structurally similar compounds isoprene and chloroprene. The research summarized in
this chapter supports species differences in pharmacokinetics that may contribute to, but not be
solely responsible for, species differences in the toxic responses from exposure to 1,3-butadiene.
The chemical terminology used in the publications reviewed in this chapter has been
standardized for consistency. Epoxybutene (EB) is used for 1,3-butadiene monoepoxide, 1,3-
butadiene monoxide, l,2-epoxybutene-3, l,2-epoxy-3-butene, vinyl oxirane, and 3,4-epoxy-l-
butene; diepoxybutane (DEB) is used for l,2:3,4-diepoxybutane, butadiene diepoxide, and
butadiene bisoxide; butenediol (BDdiol) is used for l,2-dihydroxybut-3-ene and 3-butene-l,2-
diol; and epoxybutanediol (EBD) for 3,4-epoxybutanediol, 3,4-epoxybutane-l,2-diol, 3,4-epoxy-
1,2-butanediol, and l,2-dihydroxy-3,4-epoxybutane.
3.1. OVERVIEW OF PHARMACOKINETIC STUDIES
In brief, the primary route of exposure to 1,3-butadiene is via inhalation; therefore,
absorption occurs through the respiratory system. 1,3-Butadiene is then widely distributed
throughout the body. The first step of metabolism is primarily oxidation of 1,3-butadiene to EB.
EB can then be further oxidized to DEB, hydrolyzed to BDdiol, or conjugated with glutathione.
DEB can then also be hydrolyzed to EBD or conjugated with glutathione, while BDdiol can be
further oxidized to EBD. EBD can be hydrolyzed or conjugated with glutathione. These various
metabolites also appear to be well distributed in the body, and some of them have been measured
in a number of different tissues. 1,3-Butadiene can be excreted via the respiratory system as the
parent compound or as a metabolite. Metabolites of 1,3-butadiene are also excreted in urine and
3-1
-------
feces. This qualitative overview appears to apply to the various species that have been studied,
including humans; however, the enzymatic rates for specific metabolic routes differ among
species (and tissues).
Since the previous reviews by EPA in 1985 and IARC in 1986, considerable data have
been generated regarding the pharmacokinetics of 1,3-butadiene in various species. Most of the
studies have been of mice and rats; however, there are also some data on monkeys and humans.
Recent research has elucidated new metabolic pathways that had not been described previously
and has shown differences in tissue concentrations of reactive metabolites of 1,3-butadiene. In
vitro studies with liver, lung, and kidney cytosolic and microsomal fractions have elucidated
possible metabolic products and yielded estimates of kinetic constants for the various metabolic
pathways under controlled conditions. The in vitro studies indicated differences in the rates of
metabolism that may result in different concentrations of the reactive metabolites in vivo. In
vivo studies, which encompass all processes of pharmacokinetics—absorption, distribution,
biotransformation, and excretion—have provided an account of the disposition of 1,3-butadiene
during and following inhalation exposures in various animal species. The studies have
quantitated kinetic parameters governing disposition in rodents, identified tissue and urinary
metabolites, and identified DNA and hemoglobin adducts of reactive metabolites. The in vivo
pharmacokinetic studies validated the existence of certain reactive metabolites of 1,3-butadiene
and showed differences in tissue concentrations of the metabolites among species. The in vivo
studies verified that interspecies differences in rates of metabolism for the various
biotransformation pathways result in different tissue doses of reactive metabolites. An extensive
summary of the pharmacokinetics of 1,3-butadiene was published recently by Himmelstein et al.
(1997), and readers are referred to this reference for a more detailed description of the
pharmacokinetic research. This chapter summarizes the relevant research for evaluating the role
of 1,3-butadiene pharmacokinetics in the species differences in toxicity following inhalation
exposures.
3.2. ABSORPTION
Absorption refers to the rate at which a chemical leaves its site of administration and
enters the systemic circulation. The primary route of exposure to 1,3-butadiene is via inhalation;
therefore this section focuses on factors that affect absorption of 1,3-butadiene via inhalation.
Absorption via inhalation is dependent on the solubility of the chemical in blood versus
air. Several laboratories have reported blood:air partition coefficients, which are ratios of the
concentrations of chemical in blood versus air, for 1,3-butadiene and its metabolites in mice and
rats. Table 3-1 summarizes the results from various laboratories. 1,3-Butadiene is moderately
soluble in blood, with a partition coefficient of approximately 1, which means theoretically in a
O O
3-2
-------
Table 3-1. Tissuerblood partition coefficients".
Tissue
Liver
Lung
Muscle
Fat
l,3-Butadieneb
Rats Mice
0.80
0.62
0.99
15
1.0
1.1
3.0
14
3,4-Epoxvbutenec
Rats
1.4
1.1
0.39
2.7
Mice
1.1
1.5
0.64
2.5
l,2,3,4-DieDoxvbutaned
Rats
NRe
NR
NR
NR
Mice
1.4
NR
1.8
2.2
Tartition coefficients represent the thermodynamic equilibrium of the chemical in the tissue and blood.
bExperimentally determined partition coefficients were reported by Medinsky et al. (1994), who measured tissue:air and blood:air by the vial equilibration method of
Gargas et al. (1989). The blood:air partition coefficients for BD were 1.34 and 1.49 in mice and rats, respectively, and for EB were 36.6 and 50.4 in mice and rats,
respectively.
Experimentally determined partition coefficients were reported by Medinsky et al. (1994), who measured tissue:air and blood:air ratios by the vial equilibration method
of Gargas et al. (1989).
dData obtained from Sweeney et al. (1997). Sweeney et al. calculated tissue:blood partition coefficents from the ratio of tissue:hexane partition coefficients versus
blood:hexane partition coefficients. This method for measuring partition coefficients has not been previously validated or extensively used to date.
eValues for these partition coefficients were not reported.
-------
single breath 50% of the amount of 1,3-butadiene would enter the blood to equilibrate with the
1,3-butadiene present in the air. No significant species differences have been noted among the
reported blood:air partition coefficients.
Pharmacokinetic studies have been performed exposing rats, mice, and monkeys to 1,3-
[l-14C]butadiene to measure the percentage of the inhaled dose that was retained at the end of
exposure. Although the studies indicate qualitatively that 1,3-butadiene is absorbed into the
bloodstream from the air, the rate of absorption or amount absorbed cannot be estimated from the
collected data. Different protocols were used in the various studies and none completely
analyzed the total amount of radioactivity during and after exposures.
Bond et al. (1986) exposed mice and rats by inhalation to l,3-[l-14C]butadiene for 6 h at
exposure concentrations ranging from 0.08 to 1,000 ppm and 0.08 to 7,100 ppm for mice and
rats, respectively. Radioactivity measured at the end of exposure in the animals included the
amount in exhaled breath, urine, feces, and body. The measured radioactivity excluded the
amounts excreted via all routes during exposure. Except for the exposures to 0.8 and 7 ppm 1,3-
butadiene, the rats and mice appeared to retain similar percentages of the inhaled 1,3-butadiene,
4-20% and 1.5-17% for mice and rats, respectively. The amount retained at the end of 6 hours
was used to estimate an inhaled dose normalized to body weight for each species of animal. The
mice received approximately 2.5-11 times the inhaled dose of 1,3-butadiene per kg body weight
as rats in the experiments. The measured minute ventilation of mice normalized to the body
weight in this study was approximately threefold higher than rats, 96 versus 34 L/h/kg,
respectively.
Dahl et al. (1991) exposed monkeys for 2 h by inhalation to l,3-[l-14C]butadiene at
concentrations of 10, 300, or 8,000 ppm and quantitated 14C in exhaled breath, urine, and feces
during and up to approximately 96 h postexposure. The amount of radioactivity recovered was
1.5-2.9% of the inhaled amount of radiolabeled 1,3-butadiene; however, the data did not include
residual 14C still present in the monkeys at the end of 96 h or the amount of 14C exhaled as 1,3-
butadiene either during or following exposures.
The higher dose in mice would be due partly to higher ventilation per kg of body weight
of mice versus rats or monkeys, which would allow greater absorption of the chemical into the
blood. In addition, more extensive metabolism of 1,3-butadiene by the mice would enhance the
amount of 1,3-butadiene absorbed in the blood.
3.3. BIOTRANSFORMATION
This section highlights the major findings from several in vitro and in vivo studies that
have elucidated the pathways of 1,3-butadiene biotransformation, identified responsible enzymes,
and quantitated kinetic parameters that govern the rates of metabolism. The pathways of 1,3-
3-4
-------
butadiene biotransformation are illustrated in Figure 3-1 and identified in Table 3-2. Several of
the metabolites, which are indicated by brackets, are electrophilic and capable of reacting with
DNA and protein.
Table 3-3 summarizes the intermediate metabolites that have been identified either in
vitro or in vivo from the proposed biotransformation pathways and the corresponding animal
species in which the metabolite has been identified. The detection of the metabolite in the urine
only is indicated as such in the table. Some compounds have been detected in vivo only as
adducts of DNA or hemoglobin adduct, which is noted in the table.
3.3.1. Pathway Elucidation
The initial and predominate biotransformation step of 1,3-butadiene is oxidation to form
EB. EB has been detected in vitro during incubations of 1,3-butadiene with hepatic
postmitochondrial fractions from mouse, rat, monkey, or human (Schmidt and Loeser, 1985);
with hepatic microsomes from rats, mice, or humans (Cheng and Ruth, 1993; Csanady et al.,
1992; Duescher and Elfarra, 1994; Elfarra et al., 1991; Malvoisin et al., 1979a; Nieusma et al.,
1997); with lung postmitochondrial fractions from mouse and rat (Schmidt and Loeser, 1985);
with lung microsomes from mice, rats, and humans (Csanady et al., 1992); with lung tissue from
airways of mice and rats (Seaton et al., 1996); with kidney microsomes from mice (Sharer et al.,
1992); with bone marrow from humans and mice (Maniglier-Poulet et al., 1995); with
chloroperoxidase from Caldoriomycesfumago (Elfarra et al., 1991); with human
myeloperoxidase (Duescher and Elfarra, 1992; Maniglier-Poulet et al., 1995); and with cDNA-
expressed human cytochrome P450s (Duescher and Elfarra, 1994). EB was identified in vivo in
blood and tissues of mice, rats, and monkeys that were exposed by inhalation to 1,3-butadiene
(Bechtold et al., 1995; Bond et al., 1986; Himmelstein et al.,1994; Himmelstein et al., 1995;
Leavens et al., 1996; Leavens et al., 1997; Thornton-Manning et al., 1995a, 1997, 1998); and
exhaled EB has been measured in the atmosphere of closed chambers containing rats and mice
exposed by inhalation to 1,3-butadiene (Bolt et al., 1983; Filser and Bolt, 1984; Kreiling et al.,
1987).
EB can be oxidized further to form another reactive metabolite, DEB. DEB has been
detected in vitro during incubations of 1,3-butadiene with rat liver microsomes (Malvoisin and
Roberfroid, 1982; Cheng and Ruth, 1993) and during incubations of EB with mouse, rat, and
human liver microsomes (Krause and Elfarra, 1997; Malvoisin et al., 1979b; Nieusma et al.,
1997; Seaton et al., 1995). DEB has been detected in vivo in blood and tissues of mice and rats
exposed by inhalation to 1,3-butadiene (Bechtold et al., 1995; Bond et. al., 1986 ; Himmelstein et
al., 1994; Himmelstein et al., 1995; Thornton-Manning et al., 1995a,b; 1997, 1998).
5-5
-------
P450
1, 2 ,9, GST
6,15 11,12,13 ~<
GST
P45Q
EH
1,2-Epoxy-3 -butene
l,2:3,4-Diepoxybutane
CO2
EH
O
OH
OH
10
l,2-Dihydroxy-3,4-epoxybutane
EH / \ GST
OH
OH
OH
6,15
OH
Erithritol
Figure 3-1. Schematic of 1,3-butadiene metabolism".
P450 OH
3-Butene-1,
I
3,7,
8,14
CO2
O
OH
Acrylic Acid
aP450 stands for cytochrome P450, EH stands for epoxide hydrolase, GST stands for glutathione transferase, and GSH stands for glutathione. The reactive
metabolites are shown inside boxes. The urinary metabolites are numbered and listed in Table 3-1.
-------
Table 3-2. Identification of urinary metabolites in Figure 3-1.
Urinary metabolite
number
1
2
O
4
5
6
7
8
9
10
11
12
13
14
15
Urinary metabolite name
7V-acetyl-,S'-(2-hydroxy-3-butenyl)-L-cysteine,
,S'-(2-hydroxy-3 -buten- 1 -yl)-7V-acetyl-L-cysteine, or
1 -(TV- Acetyl-L-cystein-,S'-yl)-2-hydroxybut-3 -ene
N-acety\-S-( 1 -(hydroxymethyl)-2-propenyl)-L-cysteine,
or
-------
Table 3-3. Identified 1,3-butadiene metabolites".
oo
Species
Mice
Rats
Monkeys
Humans
Mice
Rats
Monkeys
Humans
Butadiene metabolite
3-Butenal Crotonaldehye EB DEB BDdiol
In vitro
+ + + +
+ + +
ND ND + ND ND
+b + + +
In vivo
ND ND + + +c
ND ND + + +c
ND ND + + ND
ND ND +e +f ND
EBD
ND
+
ND
ND
+d
+d
ND
+f
This table includes the intermediate metabolites that are shown in Figure 3-1 except for acrolein and acrylic acid, which have not been studied specifically in any
tissue in vitro or in vivo but have been suggested as intermediates for 3-carbon urinary metabolites identified by Nauhaus et al. (1996). In the table + indicates
the metabolite has been detected, - indicates the metabolite has not been detected, and ND indicates that the presence of the metabolite has not been examined.
bDetected in incubations with human myeloperoxidase but not human tissues.
°Detected as a urinary metabolite in animals following exposure to 1,3-butadiene and in blood or tissues of animals administered EB by intraperitoneal injection.
dHemoglobin and DNA adducts that may have been formed from EBD detected following exposure to 1,3-butadiene.
eHemoglobin and DNA adducts detected in occupationally exposed humans.
Hemoglobin and DNA adducts that may have been formed from either DEB or EBD detected in occupationally exposed humans.
-------
In addition, EB can be hydrolyzed chemically or by epoxide hydrolase and conjugated
with glutathione chemically or by glutathione transferases. EB is hydrolyzed to BDdiol, which
has been detected in vitro with incubations of 1,3-butadiene with rat liver microsomes (Cheng
and Ruth, 1993); with incubations of EB in the absence of NADPH with hepatic microsomes
from rats, mice, and humans (Krause et al., 1997; Malvoisin and Roberfroid, 1982); with
incubations of EB in the absence of NADPH with liver cytosols from mice and humans (Krause
et al., 1997); and with cDNA-expressed human microsomal epoxide hydrolase (Krause et al.,
1997). BDdiol has been detected in vivo in urine from mice and rats exposed by inhalation to
1,3-butadiene (Nauhaus et al., 1996) and from rats and mice administered EB by intraperitoneal
injection (Krause et al., 1997). Glutathione conjugates of EB have been detected in the bile from
rats administered EB intraperitoneally (Sharer and Elfarra, 1992). Mercapturic acids from
glutathione conjugation with EB have been identified in urine of mice and rats exposed by
inhalation to 1,3-butadiene (Bechtold et al., 1994; Nauhaus et al., 1996; Osterman-Golkar et al.,
1991; Sabourin et al., 1992); in urine of mice and rats administered EB intraperitoneally (Elfarra
et al., 1995; Richardson et al., 1998); and in workers exposed occupationally (Bechtold et al.,
1994; Ward et al., 1994).
Because of its electrophilic nature, EB also can form DNA and hemoglobin adducts in
both rats and mice. DNA adducts of EB have been detected in blood and tissues of mice and rats
exposed by inhalation to 1,3-butadiene (Jelitto et al., 1989; Koc et al., 1999; Koivisto et al.,
1997; Koivisto et al., 1998; Tretyakova et al., 1998). Hemoglobin adducts derived from EB have
been detected in blood of mice and rats exposed by inhalation to 1,3-butadiene (Albrecht et al.,
1993; Osterman-Golkar et al., 1991, 1998; Sun et al., 1989a; Swenberg et al., 1999; Tretyakova
et al., 1996) and in workers occupationally exposed to 1,3-butadiene (Osterman-Golkar et al.,
1996).
BDdiol, the hydrolysis product of EB, can be oxidized or conjugated with glutathione.
The oxidation of BDdiol has been studied in vitro in hepatic cytosols from rats, mice, and
humans (Kemper and Elfarra, 1996); in microsomes from mice, rats, and humans (Krause et al.,
1997); and in vivo in mice (Kemper et al., 1998). A mercapturic acid derived from glutathione
conjugation of BDdiol was detected in the urine of mice and rats exposed by inhalation to 1,3-
butadiene (Bechtold et al., 1994; Nauhaus et al., 1996; Sabourin et al., 1992).
DEB, the oxidation product of EB, can be hydrolyzed chemically or by epoxide hydrolase
and conjugated with glutathione chemically or by glutathione transferases. EBD, which can be
formed by the hydrolysis of DEB or the oxidation of BDdiol, has been identified from in vitro
incubations of 1,3-butadiene with rat liver microsomes (Cheng and Ruth, 1993; Malvoisin and
Roberfroid, 1982). Formation of EBD from 1,3-butadiene has not been studied in microsomal
fractions from other species. Other metabolites of hydrolysis that have been measured in vitro
3-9
-------
include erythritol in mouse, rat, and human microsomes incubated with DEB; and
anhydroerythritol in human microsomes incubated with DEB, although these products
represented only a small percentage of the DEB that was hydrolyzed (Boogaard and Bond, 1996).
Mercapturic acids that could have been derived from glutathione conjugation of DEB were
detected in the urine of mice and rats exposed by inhalation to 1,3-butadiene (Nauhaus et al.,
1996). Boogaard et al. (1996) identified two glutathione metabolites and the corresponding
trihydroxy metabolites that were formed from glutathione conjugation of DEB in liver cytosols of
mice and rats. Dihydropropanone, which could be formed through the pentose phosphate
pathway from DEB, was detected in the urine of rats exposed by inhalation to 1,3-butadiene
(Nauhaus et al., 1996).
Because of their electrophilic nature, both DEB and EBD form trihydroxybutyl DNA
adducts and hemoglobin adducts. Trihydroxybutyl adducts of DNA have been identified in vivo
in blood and tissues of mice and rats exposed to 1,3-butadiene by inhalation (Jelitto et al., 1989;
Koc et al., 1999; Tretyakova et al., 1998). Trihydroxybutyl hemoglobin adducts have been
identified in blood from rodents administered EB, EBD, or DEB intravenously (Perez et al.,
1997; Swenberg et al., 1999) and from humans exposed occupationally to 1,3-butadiene (Perez et
al., 1997; Swenberg et al., 1999).
An alternative but minor biotransformation pathway, which has only been demonstrated
definitively in vitro, is oxidation of 1,3-butadiene to 3-butenal. 3-Butenal has been detected in
incubations with chloroperoxidase from Gaidariomyces fumago at a pH of 6.0 (Duescher and
Elfarra, 1993). At the physiological pH of 7.4, 3-butenal is thought to be tautomerized rapidly to
crotonaldehyde, which has been detected in vitro in hepatic microsomal incubations from mice
only (Elfarra et al., 1991; Sharer et al., 1992); in lung and kidney microsomal incubations from
mice (Sharer et al., 1992); in chloroperoxidase incubations (Elfarra et al., 1991); and in
incubations with human myeloperoxidase (Duescher and Elfarra, 1992). In the mouse
microsomes, crotonaldehyde was formed at higher protein concentrations than those used in the
incubations in which only EB was detected, 1.7-2.0 mg protein/mL versus 0.3-0.6 mg
protein/mL, respectively. In addition, crotonaldehyde concentrations were approximately 30% of
EB concentrations in microsomal and chloroperoxidase incubations and approximately 6% of EB
concentrations in human myeloperoxidase incubations. Neither 3-butenal nor its product
crotonaldehyde were biotransformed from EB, as incubations with EB failed to produce either
compound. Although myeloperoxidase was shown to oxidize 1,3-butadiene to crotonaldehyde,
incubations with either human or mouse bone marrow failed to produce detectable quantities of
crotonaldehyde (Maniglier-Poulet et al., 1995). Experiments have not been conducted to
specifically identify 3-butenal or crotonaldehyde in blood or tissues in vivo. However, urinary
metabolites, which may result from 3-butenal or further biotransformation of crotonaldehyde to
3-10
-------
acrolein, have been detected in the urine of mice but not rats exposed by inhalation to 1,3-
butadiene, and represented 13.9% of total urinary metabolites that were quantitated (Nauhaus et
al., 1996). But, the metabolites attributed to acrolein biotransformation pathways (10.2% of
urinary metabolites) could also be attributed to biotransformation of EB.
Another potential metabolic pathway that was identified in vitro is biotransformation of
1,3-butadiene to l-chloro-2-hydroxy-3-butene (chlorhydrin). Duescher and Elfarra (1992)
detected chlorhydrin by gas chromatography mass spectrometry (GC-MS) in incubations of 1,3-
butadiene with physiological concentrations of KC1 and human myeloperoxidase purified from
human leukocytes. Chlorhydrin concentrations exceeded EB concentrations at KC1
concentrations greater than 50 mM. Chlorhydrin was not produced in the absence of KC1 or by
liver microsomes, chloroperoxidase, or horseradish peroxidase and was not a product of EB or
crotonaldeyde because incubation with these two chemicals did not produce chlorhydrin. This
particular product has not been specifically investigated in vivo, but as with 3-butenal, it may be
relevant to potential toxic effects in the lymphohematopoietic system.
Another chlorine metabolite that was identified in phosphate buffered saline in which EB
was incubated in the absence of tissues is l,2-dichloro-3,4-epoxybutane (Cheng et al., 1994).
Cheng and co-workers noted the disappearance of EB from the headspace of solutions in
phosphate-buffered saline and identified the product by comparing the fragmentation pattern
obtained from electron ionization GC-MS of the solvolysis product with synthetic standard.
Although the production of l,2-dichloro-3,4-epoxybutane was formed nonenzymatically at
physiological Cl concentrations, the relevancy of this pathway in vivo is not known.
3.3.2. Enzyme Identification
The major group of enzymes involved in the oxidation of 1,3-butadiene, EB, and BDdiol
are cytochromes P450. The generation of EB from 1,3-butadiene in a NADPH-generating
system was first demonstrated by Malvoisin et al. (1979a). The role of cytochromes P450 in the
biotransformation of 1,3-butadiene was demonstrated in vitro by induction of EB formation in
liver microsomes from rats pretreated with phenobarbital (Bolt et al., 1983; Malvoisin et al.,
1979a) and 20-methylcholanthrene (Bolt et al., 1983) and by inhibition of the reaction following
the addition of SKF525-A (Bolt et al., 1983; Malvoisin et al., 1979a) and dithiocarbamate to the
incubation with rat liver microsomes (Bolt et al., 1983). The role of cytochromes P450 in the
biotransformation of 1,3-butadiene was demonstrated in vivo by induction of the rate of
metabolism in rats pretreated with Arochlor (Bolt et al., 1984) and by reduced uptake of 1,3-
butadiene from the atmosphere in closed inhalation chambers by rats and mice pretreated with
pyrazole (Medinsky et al., 1994). Elfarra et al. (1991) suggested the role of cytochromes P450 in
the oxidation of 1,3-butadiene to both EB and crotonaldehyde in mouse liver microsomes by the
3-11
-------
dependence of the reactions on the addition of NADPH to the incubation and by inhibition of the
reactions with the addition of 1-benzylimidazole, a selective cytochrome P450 inhibitor, to the
incubations. The role of cytochrome P450 in the oxidation of BDdiol was suggested in vivo,
since pretreatment of male B6C3FJ mice with benzylimidazole, a cytochrome P450 inhibitor,
decreased the clearance of BDdiol administered intraperitoneally (25 mg/kg BW) by 44%.
Several isoforms of cytochrome P450 have been shown to be involved in the
biotransformation of 1,3-butadiene. The specific P450 isoforms that have been identified thus
far as important in the biotransformation of 1,3-butadiene to EB include P450 2E1 and P450
2A6. Csanady et al. (1992) suggested a significant role of cytochrome P4502E1 in the oxidation
of 1,3-butadiene through the use of correlation analysis with chlorzoxazone. Duescher and
Elfarra (1994) used microsomes from P-lymphoblastoid cells expressing human cDNA for
specific P450 isozymes to demonstrate that both P450 2E1 and P450 2A6 oxidize 1,3-butadiene.
Other isoforms oxidized 1,3-butadiene but were not major contributors to the overall activity
when normalized for the amount of enzyme present in the incubation. Cytochrome P450 2E1
appeared to be a high-affinity, low-capacity pathway, while cytochrome P450 2A6 was lower
affinity and higher capacity.
The major isoform of cytochrome P450 that oxidizes EB to DEB is cytochrome P450
2E1; however, 2A6, 2C9, and 3A4 have been shown to oxidize EB to DEB to some extent
(Krause and Elfarra, 1997; Seaton et al., 1995). Seaton et al. (1995) demonstrated from human
cDNA-expressed cytochromes P450 in human b-lymphoblastoid cells that P450 2E1 oxidized
EB to DEB in incubations with 80 |lM EB and that P450 2E1 and 3 A4 were the most active in
incubations with 5 |lM EB. Small amounts of DEB were formed in incubations with 5 |lM EB
and cytochromes P450 1 Al, 1A2, 2A6, 2D6, and 2F1; however, the rates were similar to the
rates of oxidation with control incubations of microsomes prepared from the parent cell line used
to express the cytochromes P450. Krause and Elfarra (1997) demonstrated, from human cDNA-
expressed cytochromes P450 in human P-lymphoblastoid cells incubated with 5 |lM EB, that the
predominate cytochrome that oxidized EB to DEB was P450 2E1. Although not as active as
cytochrome P450 2E1, cytochromes P450 2A6 and 2C9 also catalyzed the oxidation of EB to
DEB. Krause and Elfarra did not detect DEB in incubations with cytochrome P450 3 A4.
Lewis et al. (1997) attempted to provide explanations for species differences in 1,3-
butadiene metabolism among mice, rats, and humans through molecular modeling of the putative
active site region on P450 2E1. The molecular stucture for cytochrome P450 2E1 was
constructed on the basis of the CYP102 crystal structure. Differences were noted among species
for some of the amino acids in the binding pocket, which for mice and rats would facilitate
hydrogen bonding and increase the probability of a second epoxidation. The changes in residues
were thought to bring about a change in the size of the binding pocket. The mouse also has an
3-12
-------
additional methylene group at the active site to allow it to bind EB more strongly and increase
the probability of a second epoxidation.
Another enzyme capable of oxidizing BDdiol to EBD is alcohol dehydrogenase. Kemper
and Elfarra (1996) demonstrated that purified horse liver alcohol dehydrogenase oxidized BDdiol
in vitro. The involvement of alcohol dehydrogenase was indicated by inhibition of BDdiol
oxidation by methylpyrazole, as measured by NADH generation, in incubations of mouse, rat,
and human liver cytosols with BDdiol (Kemper and Elfarra, 1996). The role of alcohol
dehydrogenase in the oxidation of BDdiol was indicated in vivo, since pretreatment with 4-
methylpyrazole, a nonselective inhibitor of alcohol dehydrogenase, decreased clearance of
BDdiol administered to mice intraperitoneally by 82%.
The major detoxication enzymes for EB and DEB appear to be epoxide hydrolase and
glutathione transferase. The role of epoxide hydrolase was indicated by inhibition of EB
metabolism in rat liver microsomes incubated with styrene oxide and vinylcyclohexene oxide,
substrates for epoxide hydrolase (Malvoisin and Roberfroid, 1982), and by increased EB
formation in NADPH-fortified rat liver microsomes incubated with 1,1,1-trichloropropeneoxide,
another substrate for epoxide hydrolase (Bolt et al., 1983). Urinary metabolites that were formed
from 1,3-butadiene intermediates are predominately products of hydrolyis or glutathione
conjugation (Bechtold et al., 1994; Elfarra et al., 1995; Nauhaus et al., 1996; Osterman-Golkar et
al., 1991; Richardson et al., 1998; Sabourin et al. 1992). The involvement of glutathione
transferase in vivo in the biotransformation of 1,3-butadiene is evident by the depletion of
glutathione in the tissues of animals exposed by inhalation to 1,3-butadiene (Deutschmann and
Laib, 1989; Himmelstein et al., 1995; Kreiling et al., 1987; Laib et al., 1990).
As noted in Section 3.3.1 on metabolic pathways, other enzymes such as
myeloperoxidase and chloroperoxidase have been shown to be capable of oxidizing 1,3-
butadiene. The myeloperoxidase-mediated pathway is consistent with the ability of bone marrow
to oxidize 1,3-butadiene (Maniglier-Poulet et al., 1995), especially since Genter and Recio
(1994) did not detect cytochrome P450 2E1 in mouse bone marrow. Oxidation of 1,3-butadiene
by myeloperoxidase has not been studied extensively, particularly in vivo, to evaluate its
relevancy for determining a local tissue dose of reactive metabolites that could result in a toxic
effect on the lymphohematopoietic system.
3.3.3. Rates of Metabolism
Qualitatively the metabolic pathways are similar among species; however, quantitatively
the metabolism of 1,3-butadiene varies among species. Several investigators have noted
differences among species in enzymatic activities for several of the pathways of 1,3-butadiene
biotransformation shown in Figure 3-1. For example, differences among species exist in the
O 1O
3-13
-------
kinetic constants for oxidation of 1,3-butadiene to EB, oxidation of EB to DEB, and detoxication
of EB and DEB by epoxide hydrolase and glutathione transferase. The kinetic parameters for the
various biotransformation pathways have been estimated from several experiments, including in
vitro incubations with 1,3-butadiene, EB, DEB, and BDdiol in microsomal and cytosolic
fractions from tissues of animals including humans; in vivo inhalation exposures of rodents to
1,3-butadiene and the reactive metabolite EB; and in vivo exposures of rats administered EB and
DEB intravenously. Based on estimates of intrinsic clearance rates of activation versus
detoxication in the liver that were calculated from in vitro experiments and on kinetic parameters
that were estimated from in vivo experiments, the tissue and blood concentrations of the reactive
1,3-butadiene metabolites such as EB, EBD, and DEB would be expected to be greater in mice
than in either rats or humans exposed by inhalation to 1,3-butadiene.
3.3.3.1. In Vitro
Kinetic differences in the various pathways in the metabolism of 1,3-butadiene among
species have been demonstrated in vitro with tissue fractions from several organs. This section
summarizes the various studies that quantitated enzymatic activities as well as estimated kinetic
parameters (Tables 3-4 to 3-9) that governed the rates of reaction for the biotransformation
pathways shown in Figure 3-1. The kinetic parameters from the liver were used to estimate
intrinsic clearance rates (Table 3-10) to quantitatively compare the metabolism of 1,3-butadiene
among species. The clearance rates were compared because the metabolism of 1,3-butadiene and
its metabolites would be expected to be first order at the low 1,3-butadiene concentrations likely
to be encountered environmentally (see Section 3.3.3.3 on in vivo metabolism). In this section
the studies that measured enzymatic activities are discussed first, followed by a description of the
experiments that estimated in vitro kinetic constants for the various biotransformation pathways.
The enzymatic activity studies are divided by species, gender, and tissue differences; chronic
effects; and stereochemistry.
3.3.3.2. Enzymatic Activities
3.3.3.2.1. Species, gender, and tissue differences. The mouse consistently had the highest
enzyme activities compared with rat, human, or monkey. The tissue activities varied, but no
single tissue consistently had the greatest activity among species or genders.
One of the first in vitro studies to address species differences in the kinetics of 1,3-
butadiene metabolism was that of Schmidt and Loeser (1985). Enzyme activities calculated from
the rate of accumulation of EB, which in this study was a function of both formation from 1,3-
butadiene and removal of EB, were measured in liver and lung postmitochondrial fractions from
male and female mice (NMRI and E6C3Fl\ male and female rats (Wistar and Sprague-Dawley),
3-14
-------
Table 3-4. In vitro Michaelis-Menton kinetic parameters for mouse liver.
Reaction
1,3 -Butadiene to EB
EB to DEB
EB GSH conjugation
EB hydrolysis
BDdiol oxidation
DEB GSH conjugation
DEB hydrolysis
Source
Csanadyetal. (1992)
Duescher and Elfarra
(1994)
Csanadyetal. (1992)
Seatonetal. (1995)
Krause and Elfarra (1997)
Sharer etal. (1992)
Csanadyetal. (1992)
Kreuzeretal. (1991)
Csanadyetal. (1992)
Krause etal. (1997)
Kemper and Elfarra (1996)
Boogaard et al. (1996)
Boogaard and Bond (1996)
Vmax
(nmol/mg
prot/min)
2.6
9.2
0.2
1.4
4.3
107
500
19
5.79
4
10.4
162
32
Km (UM)
2
160
15.6
141
5,600
4,700
35,300
1,500
1,590
500
20,000
6,400
8,100
Table 3-5. In vitro Michaelis-Menton kinetic parameters for rat liver.
Reaction
1,3-ButadienetoEB
EB to DEB
EB GSH conjugation
EB hydrolysis
BDdiol oxidation
DEB GSH conjugation
DEB hydrolysis
Source
Csanadyetal. (1992)
Duescher and Elfarra (1994)
Seatonetal. (1995)
Krause and Elfarra (1997)
Sharer etal. (1992)
Csanadyetal. (1992)
Kreuzeretal. (1991)
Csanadyetal. (1992)
Krause etal. (1997)
Kemper and Elfarra (1996)
Boogaard et al. (1996)
Boogaard and Bond (1996)
Vmax
(nmol/mg
prot/min)
0.59
2
0.41
3.5
71
241
17
2.48
24.4
21.4
186
52.9
Km ( UM)
3.74
120
145
16,300
3,100
13,800
700
260
700
32,700
24,000
2,760
3-15
-------
Table 3-6. In vitro Michaelis-Menton kinetic parameters for human liver.
Reaction
1,3 -Butadiene to EB
EB to DEB
EB GSH conjugation
EB hydrolysis
BDdiol oxidation
DEB GSH conjugation
DEB hydrolysis
Source
Csanadyetal. (1992)
Duescher and Elfarra (1994)
Duescher and Elfarra (1994)
Seatonetal. (1995)
Seatonetal. (1995)
Seatonetal. (1995)
Krause and Elfarra (1997)
Krause and Elfarra (1997)
Csanadyetal. (1992)
Kreuzeretal. (1991)
Csanadyetal. (1992)
Csanadyetal. (1992)
Csanadyetal. (1992)
Krause etal. (1997)
Krause etal. (1997)
Kemper and Elfarra (1996)
Kemper and Elfarra (1996)
Kemper and Elfarra (1996)
Boogaard et al. (1996)
Boogaard et al. (1996)
Boogaard et al. (1996)
Boogaard et al. (1996)
Boogaard et al. (1996)
Boogaard et al. (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boosaard and Bond (1996)
Vmax
(nmol/mg
prot/min)
1.18
10.4
22.8
1.2
0.38
1.2
6.4
0.8
45.1
14
58.1
18.5
9.2
55.8
26.7
11.6
10.6
9.1
3.7
5.5
7.8
4.7
23.6
19
229
133
74
270
119
120
KmOllVI)
5.14
200
390
320
304
880
32,900
6,600
10,400
500
1,650
580
240
2,400
300
18,300
19,600
22,500
1,040
1,270
3,600
3,180
9,400
4,500
6,860
3,640
3,740
5,470
3,590
6,670
5-16
-------
Table 3-7. In vitro Michaelis-Menton kinetic parameters for mouse lung.
Reaction
1,3 -Butadiene to EB
EB GSH conjugation
DEB GSH conjugation
DEB hydrolysis
Source
Csanadyetal. (1992)
Sharer et al. (1992)
Csanadyetal. (1992)
Boogaard et al. (1996)
Boogaard and Bond (1996)
Vmax
(nmol/mg
prot/min)
2.31
12
273
38.5
49.8
KmOlM)
5,001
4,400
36,500
1,700
7,500
Table 3-8. In vitro Michaelis-Menton kinetic parameters for rat lung.
Reaction
Source
Vmax
(nmol/mg
prot/min)
Km(uM)
1,3-ButadienetoEB
EB GSH conjugation
DEB GSH conjugation
DEB hydrolysis
Csanadyetal. (1992)
Sharer et al. (1992)
Csanadyetal. (1992)
Boogaard et al. (1996)
Boogaard and Bond (1996)
0.16
3
44.2
17.1
19.3
7,750
4,700
17,400
4,200
7,100
Table 3-9. In vitro Michaelis-Menton kinetic parameters for human lung.
Reaction
1,3 -Butadiene to EB
DEB hydrolysis
Source
Csanadyetal. (1992)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Boogaard and Bond (1996)
Vmax
(nmol/mg
prot/min)
0.15
29.5
29.3
23
17.2
18.1
22.5
KmOlM)
2,000
2,830
1,800
5,000
2,440
1,910
3,120
3-17
-------
Table 3-10. Estimated intrinsic clearance for liver (Vmax/Km)a.
oo
oo
Metabolic reaction
Species
Mouse
Rat
Human
1,3-Butadiene
toBMO
2.5-56
0.84-8.0
1.4-6.2
BMO to BDE
0.43-0.55
0.011-0.14
0.034-0.037
BMO
hydrolysis
0.16-0.55
0.05-1.22
0.76-1.0
BMO GSH
conjugation
4.4-7.0
5.7-7.4
0.36
BDE
hydrolysis
0.17
0.97
0.48-1.34
BDE GSH
conjugation
7.8
2.5
0.13-0.36
BDdiol
oxidation
0.16
0.21
0.033-0.053
"Ratio was determined from Michaelis-Menton parameters determined from in vitro incubations with liver microsomal and cytosolic fractions (see Tables 3-2 to 3-4). The values cytosolic and microsomal
protein concentrations and liver fractions as a percentage of body weight were taken from Boogaard and Bond (1996). The cytosolic protein concentrations were 83, 108, and 45 mg protein/g liver for
mouse, rat, and human, respectively. The microsomal protein concentrations were 11.6, 16.8, and 14.5 mg protein/g liver for mouse, rat, and human, respectively. The liver was assumed to be 6.2%, 5%,
and 3.1% of the body weight in mice, rats, and humans, respectively. Units for clearance are L/h/kg body weight.
-------
male and female rhesus monkeys, and one human (gender not specified). EB was detected in
incubations with liver postmitochondrial fractions from all species, whereas only mouse and rat
lungs formed measurable amounts of EB. The sequence of liver activities (nmol/g tissue/45 min)
quantitated from headspace amounts of EB that accumulated during the incubation with 30,000
ppm 1,3-butadiene were mouse>rat=human>monkey. The mouse had approximately sixfold
higher activity in the lung compared with the rat. In addition, female mice had higher liver
activities and lower lung activities than the male mice, whereas male rats had higher activities in
both the liver and lung. The activity was greater in the lungs compared with the liver of male
mice, whereas they were approximately the same in female mice. In rats the lung activities of
both males and females were less than liver activities. Sharer et al. (1992) measured the
formation of EB by microsomes from liver, lungs, and kidneys of male B6C3FJ mice and
Sprague-Dawley rats. The microsomes were incubated in vials with headspace pressures of 1,3-
butadiene ranging from 48 to 52 mm Hg. The authors did not report the total pressure in the
headspace or to what molar concentration of 1,3-butadiene in the headspace or media that this
pressure was equivalent. Assuming that air was present in the headspace for the reaction and that
the headspace was at atmospheric pressure, the concentration of 1,3-butadiene in the headspace
was approximately 63,158 ppm. In all three tissues the concentration of EB (nmol/mg protein)
versus time was greater in mice than in rats. In mice the sequence of activities for the tissues was
kidney>liver=lung, 23.8 ± 2.9, 6.4 ± 1.2, and 6.1 ± 0.4 nmol EB/mg protein/min (mean ± sd),
respectively. In rats the sequence of activities for the tissues was liver>lung>kidney, 3 ± 0.6, 1.5
± 0.2, and 0.5 ± 0.02 nmol EB/mg protein/min (mean ± sd), respectively. Again, in these studies
the concentration of EB in the incubation would be a function of formation and removal;
however, this study compensated for removal of EB from the system by calculating percentage of
recovery.
Because bone marrow is a target tissue for 1,3-butadiene toxicity, Maniglier-Poulet et al.
(1995) quantitated the formation of EB from 1,3-butadiene by hydrogen-peroxide-fortified bone
marrow cells. EB was not detected in bone marrow cell lysates incubated with NADPH. Mouse
bone marrow cell lysates were incubated with 50,000 ppm 1,3-butadiene in the headspace and 1
mM hydrogen peroxide at 37°C and pH 7.4, and hepatic microsomes from male B6C3FJ mice
were incubated in the presence of 1 mM NADPH and 50,000 ppm 1,3-butadiene in the headspace
for up to 20 min. EB was quantitated in the headspace of each vial. The formation of EB by
mouse liver microsomes was 281 times greater than formation from the bone marrow cell lysates.
The relevancy of metabolism of 1,3-butadiene by bone marrow lysates to in vivo exposures
cannot be accessed since the concentration of hydrogen peroxide in bone marrow cells is not
known. Granulocytic respiratory burst is known to generate superoxide and hydrogen peroxide.
5-19
-------
Even though bone marrow oxidation of 1,3-butadiene to EB may not add significantly to
systemic concentrations of reactive metabolites, it may affect local tissue concentrations.
Seaton et al. (1996) measured the concentrations of EB in media with isolated airway
sections from male B6C3FJ mouse and Sprague-Dawley rat lungs incubated with 10,000 ppm
1,3-butadiene in the headspace of vials. The authors did not find dramatic differences in the
concentrations measured at 60 min of incubation between the mice and rats.
The role of epoxide hydrolase in the detoxication of EB and DEB also differs among
species. Nieusma et al. (1997) measured greater differences in the concentrations of DEB in
hepatic microsomes from rats pretreated with trichloropropenic oxide compared with
microsomes without trichloropropylene oxide. Nieusma and co-workers did not note the nearly
twofold increase in DEB concentrations with microsomes from B6C3FJ mice.
3.3.3.2.2. Chronic effects. The effect of repeated exposures on the metabolism of 1,3-butadiene
was quantitated in a study that measured enzyme activities in vitro in tissues from animals
exposed to 1, 3-butadine in vivo (Bond et al., 1988). Bond and co-workers exposed male
Sprague-Dawley rats and B6C3FJ mice to either 7,600 ppm (252 |lmol/L) or 760 ppm (252
|lmol/L) 1,3-butadiene, respectively, for 6 h/d for 5 days. Microsomes were prepared from the
lungs and livers of both species and from nasal tissue from rats at 18 h postexposure. The assay
measured the disappearance of 1,3-butadiene in the headspace of the reaction vial. In the lungs
of both species the rate of metabolism of 1,3-butadiene was reduced 50% in animals repeatedly
exposed by inhalation to 1,3-butadiene compared with animals exposed to air only. The liver
was unaffected in these experiments. The reported activities were not consistent with previous
work because the activities in the liver were less than the lung in both species, and the activities
in the lung and liver of the rat were greater than previously reported activities in the mouse.
However, no units were given for the activities, so it is unclear whether they were normalized,
and if they were normalized, whether it was per g of protein or g of tissue weight.
3.3.3.2.3. Stereochemistry. A few investigators have examined the role of stereochemistry in
species differences in metabolism and toxicity of 1,3-butadiene. There are stereochemical
differences in vitro among species, but the significance of the differences in vivo is not known.
Nieusma et al. (1997) incubated hepatic microsomes from male B6C3FJ mice and
Sprague-Dawley rats with headspace concentrations of 25,000 ppm 1,3-butadiene, (R)-EB, or
(S)-EB. Total EB from the 1,3-butadiene incubations were quantitated from the headspace by
gas chromatography-flame ionization detector (GC-FID), and after extraction from the media the
specific enantiomers were quantitated by GC-FID. Concentrations of the racemic and meso
forms of DEB were measured in the incubation media by GC-MS. Consistent with previous
3-20
-------
studies, microsomes from mice had higher enzyme activities, with approximately twice the
concentration of EB at 30 min of incubation. Mice formed approximately the same proportion of
(S)-EB to (R)-EB during the incubation, whereas rats initially formed three times more (S)-EB
than (R)-EB, but by the end of incubation the ratio was less than one. In hepatic microsomes
incubated with (S)-EB or (R)-EB, mice had higher DEB formation than rats. In addition, more
DEB was formed from (S)-EB than from (R)-EB in microsomes from mice. At 5 min of
incubation, the rat liver microsomes had significantly greater amounts of DEB formed from (R)-
EB than (S)-EB. The trend towards more DEB from (R)-EB was present at other timepoints, but
the variability in the data made the differences insignificant. There were also regiochemical
differences in hydrolysis, with microsomes from mice having epoxide ring opening occurring at
the C2 of (S)-EB but not (R)-EB, whereas microsomes from rats had no ring opening at C2 of
either enantiomer of EB.
Nieusma et al. (1998a) quantitated the formation of (R)-EB and (S)-EB from purified
reconstituted cytochrome P450 2E1 from rabbits, rats, and humans. Solutions containing the
enzymes were incubated with headspace concentrations of 250,000 ppm 1,3-butadiene, and EB
was measured in the headspace while the specific enantiomers of EB were measured in extracts
of the enzyme solutions. The oxidation rate of 1,3-butadiene to EB was greatest by P450 2E1
from rabbits, followed by humans, then rats: 7.79, 5.0, and 4.21 nmol EB/pmol P450 2El/min,
respectively. The ratio of (S)-EB to (R)-EB was not significantly different among the species.
The authors did not account for further metabolism of EB to DEB during the incubations.
Nieusma et al. (1998b) investigated the stereochemical aspects of conjugation of EB,
DEB, and EBD with glutathione in rat liver cytosols and freshly isolated rat hepatocytes. They
quantitated the percentage of glutathione depletion during incubations with 5 mM of the two
enantiomers of EB, the three stereoisomers of DEB, or the four stereoisomers of EBD. Based on
the amount of glutathione that was depleted, (S)-EB was more reactive than (R)-EB with
glutathione transferase, and meso-DEB was the least reactive of the three DEB stereoisomers.
Only (SR)-EBD and the (SS)-EBD appeared to be enzymatically conjugated with glutathione in
both cytosolic fractions and hepatocytes, since the percentage of glutathione depletion with the
other two stereoisomers was the same in incubations with or without hepatic cytosol and was not
significant in hepatocytes. (SS)-EBD appeared more reactive than (SR)-EBD in both cytosolic-
mediated reactions and in freshly isolated hepatocytes.
3.3.3.2.4. Kinetic constants. Several investigators have shown species differences in the kinetic
constants estimated for in vitro metabolism of 1,3-butadiene and its metabolites. There is
variability in the results from the studies, potentially because of differences in experimental
methods. Some studies quantitated only the racemic form of DEB while others quantitated the
3-21
-------
racemic and meso forms of DEB. Some studies did not control or account for loss of substrate or
product through volatilization and further metabolism. In addition, the analysis of the data varied
among studies. In general the results from all the studies indicated that mice have greater rates of
oxidation of 1,3-butadiene to EB and EB to DEB than do humans or rats. There was some
variability among the studies as to whether the rates were greater in humans or rats. The in vitro
studies of detoxication of the reactive 1,3-butadiene metabolites EB and DEB via glutathione
conjugation or hydrolysis have shown that the predominate pathway of detoxi cation depends on
the species. For mice and rats, the rate of metabolism of EB and DEB via glutathione
conjugation in the liver is greater than hydrolysis. In humans, hydrolysis dominated removal of
both EB and DEB. A comparison of the metabolism of 1,3-butadiene is illustrated in Table 3-10
with estimates of the ranges of intrinsic clearance for the various biotransformation pathways in
the liver for mice, rats, and humans. Overall, mice appear to have greater formation of the
reactive metabolites EB and DEB versus detoxication via epoxide hydrolase and glutathione
transferase compared with rats and humans.
Csanady et al. (1992) quantitated kinetic constants in liver and lung cytosolic and
microsomal fractions from mice, rats, and humans for the oxidation of 1,3-butadiene to EB, EB
to DEB, EB hydrolysis, and EB glutathione conjugation. The demographics were not given for
the human subjects (n=12 for liver; n=5 for lung) from which tissue samples were obtained.
Csanady measured loss of either 1,3-butadiene or EB from the headspace or formation of
glutathione conjugates in the media of vials containing the tissue fractions, and estimated the
parameters for each pathway with a two-compartment kinetic model that described distribution
of the compound between the vial headspace and media and elimination via metabolism. The
concentrations of the substrates used in the experiments were 600-25,000 ppm 1,3-butadiene and
20-200 ppm EB. For human liver, all 12 samples were used to estimate the kinetic constants for
the oxidation of butadiene, and for human lung, three of the five samples were pooled and used
to estimate kinetic constants. For both hydrolysis and glutathione conjugation, the human liver
samples were ranked according to enzymatic activity. Kinetic constants were determined for two
samples with highest and lowest activity for glutathione conjugation and for three samples with
high, median, and low activity for hydrolysis. The metabolic pathways in all species were
described by Michaelis-Menton kinetics except for hydrolysis of EB in lung microsomes from all
three species and glutathione conjugation of EB in human lung. Vmax and Km for the various
pathways are summarized in Tables 3-3 to 3-8. The first order rate constants for hydrolysis of
EB in the lung microsomal fractions were 0.0013 and 0.00186 min"1' mg protein"1 for rats and
mice, respectively. For the three human lung samples that were used, the first-order rate
constants for hydrolysis of EB were 0.004, 0.0032 and 0.0076 min"1' mg protein"1. For the human
lung the second-order rate constant for glutathione conjugation of EB in the cytosol was 0.00026
3-22
-------
mmol"1' minting protein"1. The values for glutathione conjugation in the microsomal fractions
of the liver and lung are not listed here because they were approximately equal to spontaneous
glutathione conjugation. Csanady and co-workers detected DEB in incubations of EB with
microsomal fractions from livers of mice but not from livers of rats or humans or lungs from all
three species. The strengths of this study are lower concentration ranges used for the reactions,
estimates of the kinetic constants that are not subject to error from loss of substrate during
exposure, and estimates that are not linear approximations. However, the estimates of the
Michaelis-Menton parameters according to the method used in the paper are interrelated. The
standard deviations reported in the paper are not presented here because they represent deviations
in estimates of the parameters but not intraspecies variability for a given reaction. The authors
used the data to estimate in vivo hepatic clearances for oxidation of 1,3-butadiene to EB and for
the clearance of EB via glutathione conjugation and hydrolysis. From these values (corrected in
erratum, Csanady et al., 1993) the activation-to-deactivation ratios for each species are 12, 4.5,
and 1.3 for mouse, human, and rat, respectively.
Duescher and Elfarra (1994) also estimated Michaelis-Menton kinetic constants for
oxidation of 1,3-butadiene to EB by liver microsomes from male Sprague-Dawley rats, male
B6C3FJ mice, and two male and four female humans. Duescher and Elfarra measured the
generation of EB in microsomes incubated with 1,3-butadiene at pressures up to 52 cm Hg,
which the authors estimated would correspond to concentrations ranging from approximately 167
to 4,400 |iM 1,3-butadiene in the incubation media. In these studies the accuracy of the
measurement of 1,3-butadiene concentrations both in the headspace and media is uncertain. To
account for loss of EB from the media via volatilization or further metabolism, the authors
corrected for recovery of EB, which was 40% to 60% for human samples and approximately 60%
for mice and rat samples. Double reciprocal plots of enzymatic activity versus 1,3-butadiene
concentration in the media were used to estimate apparent kinetic constants. These values are
listed in Tables 3-4 to 3-9. As seen in the tables, the estimated Vmax and Km for each species are
greater than the values estimated by Csanady et al. (1992). In this study, oxidation of 1,3-
butadiene to EB was similar in mice and humans and greater in mice or humans than in rats. The
discrepancies in data between the two studies may be due to different methods for calculating the
apparent kinetic constants, different incubation concentrations of 1,3-butadiene, and different
human samples. Because Duescher and Elfarra also found higher chlorxazone oxidation by the
human samples in their study compared with Csanady et al. (1992), they hypothesized that some
of the discrepancy between the two studies, at least in human tissues, was due to variability in
expression of P450 enzymes. However, this does not address the differences noted with the
rodent samples, which may be due to differences in experimental design between the two studies.
Unlike Csanady et al. (1992), this study did not estimate kinetic parameters for glutathione
o oo
3-23
-------
conjugation or hydrolysis in the same tissues, so activation-to-deactivation ratios cannot be
estimated from this study.
Glutathione conjugation of EB occurs primarily in the cytosolic fraction in vitro. Sharer
et al. (1992) measured activity of EB conjugation in rat liver and found that the activity measured
from cytosolic fraction was approximately 96% of the measured activity from the whole liver
homogenate (from reported data of one rat). In addition, Sharer et al. determined Vmax and Km for
glutathione conjugation of EB in liver, lung, kidney, and testes from both male B6C3FJ mice and
Sprague-Dawley rats. The authors did not account for loss of EB into the headspace during the
incubations, which could cause an error when calculating the Michaelis-Menton kinetic constants
from double reciprocal plots. Data for the liver and lung in both species is contained in Tables 3-
4 to 3-8. In addition the Vmax for kidney was 16 ± 2 and 7 ± 3 nmol conjugates/mg protein/min
(mean ± sd) in mice and rats, respectively, and for testes 30 ± 9 and 51 ± 14 nmol conjugates/mg
protein/min (mean ± sd) in mice and rats, respectively. The Km for kidney was approximately
6,600 and 5,800 |lM in mice and rats, respectively, and for testes 5,500 and 10,500 |lM in mice
and rats, respectively. Therefore estimated clearance (Vmax/Km) of EB via glutathione
conjugation would be approximately twofold higher in kidneys of mice compared with rats, and
there would be no difference in the testes between the species.
Kreuzer et al. (1991) studied the kinetics of EB in liver microsomal and cytosolic
fractions from male NMRI mice, Sprague-Dawley rats, and one male human. Kreuzer and co-
workers found no NADPH-dependent metabolism in the incubations of microsomes with EB.
First-order metabolism of EB was detected in the cytosol due to conjugation of EB with
glutathione. Epoxide hydrolase activity was found in the microsomes only and Michaelis-
Menton parameters (Tables 3-4 to 3-6) were estimated from a two compartment model that
described the distribution and metabolism of EB during the incubations.
Seaton et al. (1995) estimated Michaelis-Menton kinetic parameters for the oxidation of
EB to DEB in microsomal fractions from livers of male B6C3FJ mice, Sprague-Dawley rats,
male and female adult humans, and a male child (Tables 3-4 to 3-6). The microsomal fractions
were incubated with media concentrations of 5-1,000 |lM EB. DEB concentrations were
quantitated in the media by GC-MS and nonlinear regression was used to obtain values for Vmax
and Km. In addition Seaton et al. measured the enzymatic activity of conversion of EB to DEB in
liver microsomes from 10 human donors. The microsomes were incubated with 80 |lM EB in
the media. The activity varied by 60-fold, spanning the range from rats to mice. Some
limitations of the studies include failure to control for volatilization of EB and further
metabolism of EB and DEB by hydrolysis or glutathione conjugation. In addition, only the
racemic form of DEB was quantitated, not the meso form.
5-24
-------
Krause and Elfarra (1997) also estimated Michaelis-Menton parameters for oxidation of
EB to DEB (Tables 3-4 to 3-6). This study quantitated formation of both racemic and meso-DEB
in microsomes from male B6C3FJ mice and Sprague-Dawley rats and a male and female human.
DEB was quantitated by GC-FID in incubations of the microsomes with 0.25-40 |lM EB. This
study also did not control for volatilization of EB or loss of EB or DEB by hydrolysis. The
authors compensated for the loss by calculating percent recoveries of DEB, which were as low as
16% in some samples. The data were analyzed both by linear transformation and by nonlinear
curve fitting.
The values obtained in the two studies are variable, probably because of differences in
experimental methods and data analysis. Both studies show mice to have greater oxidation of EB
to DEB, but Seaton et al. (1995) found greater rates of oxidation in humans compared with rats,
while Krause and Elfarra (1997) found greater oxidation in rats compared with humans. The
differences in human samples may be due to differences in cytochrome P450 content.
Boogaard and Bond (1996) studied hydrolysis of DEB in liver and lung fractions from
mice, rats, and humans. In this study only racemic DEB was studied, not the meso form of DEB.
Microsomal and cytosolic fractions from livers and lungs of male B6C3FJ mice and Sprague-
Dawley rats and male and female humans were incubated with 0.18 to 15 mM (±)DEB. The
human liver samples were obtained from trauma victims, and lung samples were obtained from
cancer patients. Depletion of DEB from the incubation media was used to quantitate enzymatic
hydrolysis of DEB. The authors reported greater hydrolysis in microsomes compared with
cytosols, 51% versus 7.7% depletion of DEB in rats, and 29% versus 6.2% depletion in mice (not
determined in humans). The depletion of DEB in liver microsomes from all three species was
well described by Michaelis-Menten kinetics (Tables 3-4 to 3-6). The kinetics in the lung was
less well described by Michaelis-Menton equations, but the fits obtained were highly significant
(Tables 3-7 to 3-9). As with the data of Csanady, Vmax and Km estimated in this study are
interrelated from the parameter estimation. Standard deviations are not given in the tables
because the deviations listed by Bogaards and Bond supposedly were deviations in the parameter
estimates, not interanimal variability (although the authors stated for the liver that multiple
samples were done for each DEB concentration). Species differences were noted for hydrolysis
of DEB. Using the ratio of Vmax/Km for comparison, hydrolysis in the liver was greatest in the
human followed by rats then mice, whereas hydrolysis in the lung was greatest in humans
followed by mice then rats. Only the mouse had higher hydrolysis in the lung compared with the
liver. The variability of Vmax/Km in human liver microsomes was fourfold; however, the sample
size was not large enough to address either gender or age-related diffferences. To assess
nonenzymatic hydrolysis of DEB, depletion of DEB was measured in buffer, in heat-inactivated
liver microsomes from rats, and in liver microsomes and cytosols from mice and rats in the
3-25
-------
presence of 1,2-epoxycyclohexane, an inhibitor of microsomal epoxide hydrolase. The estimated
half-life for nonenzymatic hydrolysis of DEB in 0.1 M potassium phosphate buffer was 21.8 ±
0.4 h (n = 4). No depletion of DEB was detected at 60 min of incubation either with heat-
inactivated liver microsomes from rats or with liver microsomes from mice in the presence of
1,2-epoxycyclohexane. Depletion of DEB was 9.3% of the initial concentrations in incubations
with liver microsomes from rats in the presence of 1,2-epoxycyclohexane. Although the authors
reported that depletion of DEB in the cytosols of both species was not affected by 1,2-
epoxycyclohexane, the data given in the article do not support these conclusions. DEB was not
depleted during incubations with cytosols from mice and rats in the presence of 1,2-
epoxycyclohexane, versus 6.2% and 7.7% depletion with cytosols from mice and rats,
respectively, in the absence of 1,2-epoxycyclohexane. Some potential limitations of this study
include problems with the lung data, in that limited concentration ranges were used for the
experiments because of limited amounts of available samples and lower hydrolytic activity in the
lung. In addition, the authors only quantitated erythritol and anhydroerythritol, which is a
rearrangement product of EBD, but not EBD. The amount of erythritol detected in incubations
from all three species and anhydroerythritol detected from humans was not great enough to
balance the amount of DEB lost during the incubations; the authors did not give the percentage
of disappearance of DEB that was accounted for by the generation of erythritol and
anhydroerythritol.
Boogaard et al. (1996) studied glutathione conjugation of DEB in fractions from mice,
rats, and humans. Again, in this study only racemic DEB was studied, not the meso form of
DEB. Liver and lung cytosols from male B6C3FJ mice and Sprague-Dawley rats and liver
cytosols from male and female humans were incubated with 0.1 to 100 mM (±)DEB and 10 mM
glycine-[2-3H]glutathione. 13C nuclear magnetic resonance spectroscopy (NMR) and mass
spectroscopy were used to identify two isomeric conjugates, S-(l-(hydroxy-methyl)-2,3-
epoxypropyl)glutathione and S-(2-hydroxy-3,4-epoxy-butyl)glutathione, which rapidly
hydrolyzed to trihydroxy conjugates. The total conjugates were quantitated by high-pressure
liquid chromatography (HPLC). The glutathione conjugation of DEB displayed Michaelis-
Menton kinetics and the constants are given in Tables 3-4 to 3-8. The same limitations for data
analysis from Boogaard and Bond (1996) are also applicable to this study. Conjugation of DEB
with glutathione is more efficient in mice than in either rats or humans. DEB was shown to
chemically react with glutathione in buffer at rates that averaged from approximately 1.3% to
50% of the enzymatically measured rates in tissues, depending on the species and tissue.
A recent study has estimated Vmax and Km for the kinetics of BDdiol oxidation in vitro in
liver cytosols from male B6C3FJ mice and Sprague-Dawley rats and male and female humans
(Kemper and Elfarra, 1996) and found no significant species differences. The values of the
3-26
-------
Michelis-Menten parameters are given in Tables 3-4 to 3-6. This oxidation represents only
metabolism of BDdiol by alcohol dehydrogenase, although as shown in other studies, multiple
pathways are responsible for biotransformation of this intermediate.
3.3.3.3. In Vivo
This section describes kinetic parameters estimated from in vivo exposures of mice and
rats to 1,3-butadiene and its metabolites EB and DEB. Some of the parameters that will be
discussed also describe rates of uptake and excretion, but are included in this section for clarity.
The rates of metabolism that are discussed are for overall metabolism of 1,3-butadiene, EB,
DEB, or BDdiol. In vivo kinetic parameters for specific biotransformation pathways have not
been quantitated. This section excludes discussion of physiologically based pharmacokinetic
models and the parameter values used in the models, because both are discussed in detail in
Chapter 9. Consistent with the in vitro studies, mice have greater rates of metabolism of 1,3-
butadiene and EB in vivo, which are due to differences in physiological parameters as well as
biochemical parameters.
3.3.3.3.1. 1,3'-Butadiene. Bond et al. (1986) exposed mice and rats by inhalation to 1,3-[1-
14C]butadiene for 6 h at exposure concentrations ranging from 0.08 to 1,000 ppm and 0.08 to
7,100 ppm for mice and rats, respectively. Total radioactivity that was retained by the body was
measured at the end of exposure in the animals. The measured radioactivity excluded the
amounts excreted via all routes during exposure. The amount of radioactivity recovered from the
animals at the end of the 6-h exposure was 4% to 20% and 1.5% to 17% of the inhaled amount of
radioactivity for mice and rats, respectively. The authors defined the amount inhaled as a product
of measured alveolar ventilation and air concentration of 1,3-butadiene. The percentage
decreased with increasing exposure concentration, indicating saturation of metabolism.
Bolt et al. (1984) measured the concentrations of 1,3-butadiene in a closed inhalation
chamber containing male Sprague-Dawley rats exposed by inhalation to initial concentrations of
1,3-butadiene ranging from 1,000 to 4,000 ppm or by intraperitoneal injections of approximately
1,000 mL gas/g body weight (approximately 41 nmol/g body weight). The data were used to
estimate kinetic constants that described the disappearance of 1,3-butadiene from the chamber by
uptake and metabolism in the rat according to a classical two compartment model (Filser and
Bolt, 1979; Filser and Bolt, 1981; Bolt et al., 1981; Filser and Bolt, 1983). Bolt et al. showed
that metabolism of 1,3-butadiene by the rat appears saturated above approximately 1,000 ppm.
The inhalation uptake (k^Vj) of 1,3-butadiene is similar to the metabolic clearance (Cltot) by the
rats, 5,750 mL/h versus 4,490 mL/h, respectively, indicating that metabolism in the rat is
5-27
-------
practically limited by the uptake of 1,3-butadiene. The maximum metabolic elimination rate
(Vmax) was 220 |lmol/h/kg; Km was not estimated from the experiments.
Kreiling et al. (1986b) measured the concentrations of 1,3-butadiene in a closed system
that contained mice exposed by inhalation to initial concentrations of 1,3-butadiene ranging from
10 to 5,000 ppm, and analyzed the data as described by Bolt et al. (1984). In mice the
pharmacokinetics were also first order below approximately 1,000 ppm 1,3-butadiene. As with
rats, Cltot was similar in mice to k12Vl3 7,300 versus 10,280 mL/h, respectively. The rates in mice
were approximately twice the rates in rats. In addition, Vmax was 400 |lmol/h/kg, approximately
twice the rate of rats. The coefficient of dynamic distribution (Kst) was also twice as high in mice
compared with rats, 1.0 versus 0.5, respectively, which would mean that under exposure
conditions where metabolism is first order the steady-state concentrations of 1,3-butadiene in the
body of mice would be approximately twofold higher compared with rats. Overall, the results
from these studies suggest that under low exposure concentrations, higher metabolic clearance of
1,3-butadiene to other metabolites will be expected in mice compared with rats because of higher
respiration frequency and higher lung and liver perfusion.
Species differences have been noted in the extent of glutathione depletion in the tissues of
animals exposed to 1,3-butadiene in vivo. Kreiling et al. (1987) exposed male B6C3FJ mice and
Sprague-Dawley rats in a closed inhalation system to saturable concentrations of 1,3-butadiene
(2,000-4,000 ppm) for 15 h. The hepatic nonprotein sulfhydryl content in BD-exposed animals
versus air-exposed control animals was 4% and 72% of unexposed controls, respectively, for
mice and 76% and 88% of unexposed controls, respectively, for rats. Kreiling et al. (1988) and
Laib et al. (1990) reported depletion of nonprotein sulfhydryl content in male B6C3FJ mice,
Sprague-Dawley rats, and Wistar rats exposed to saturable concentrations of 1,3-butadiene
(>2,000 ppm) for 7 or 15 h in a closed inhalation system. The content in exposed animals was
78%, 66%, and 24% of unexposed controls at 7 h and 88%, 60%, and 4% of unexposed controls
at 15 h, for Sprague-Dawley rats, Wistar rats, and B6C3FJ mice, respectively. Deutschmann and
Laib (1989) measured the effects of 1,3-butadiene exposure concentration on nonprotein
sulfhydryl content in liver, lung, and heart tissue of male B6C3FJ mice and Sprague-Dawley rats.
The animals were exposed by inhalation for 7 h to 10, 50, 100, 250, 500, 1,000, or 2,000 ppm
1,3-butadiene, and nonprotein sulfhydryl content was determined in the tissues immediately
following exposure. Mice showed dose-dependent depletion in all tissues, with the lung showing
the greatest loss. In rats significant depletion was noted only in liver at the highest exposure
concentration. Himmelstein et al. (1995) measured glutathione depletion in the liver and lungs of
mice exposed to 62.5, 625, or 1,250 ppm 1,3-butadiene or in rats exposed to 62.5, 625, 1,250, or
8,000 ppm 1,3-butadiene. There was not a significant difference between rats and mice in
depletion of glutathione in the liver; glutathione concentrations in the liver of exposed animals
3-28
-------
were approximately 60% of controls in both species. In the lung there was a significant species
difference in the depletion of glutathione. The depletion was reduced significantly at all
exposure concentrations in mice, with the maximal depletion occuring at the 6 h timepoint in
mice exposed to 1,250 ppm 1,3-butadiene. The maximum depletion resulted in glutathione
concentrations that were approximately 26% of controls. In rats significant depletion of
glutatione in the lung occurred during exposures to 1,250 and 8,000 ppm 1,3-butadiene, with
maximal depletion occuring at the 6 h timepoint of both exposures. The maximum depletion
resulted in glutathione concentrations that were 74% of controls.
3.3.3.3.2. EB. Filser and Bolt (1984) measured EB in the atmosphere of a closed-chamber
inhalation system containing male Sprague-Dawley rats that were exposed either by inhalation
chamber to initial concentrations of EB ranging from approximately 500 to 1,500 ppm or by
intraperitoneal injection of 5.7 nmol EB/g body weight. In addition, the generation of EB was
measured in the atmosphere of a closed inhalation system while 1,3-butadiene was maintained at
a constant concentration of 1,500 ppm. A saturable level of metabolism of EB was not attained
in the exposed rats. As with 1,3-butadiene, the uptake of EB was not limited by metabolism
since k12Vj was similar to Cltot, 13,800 versus 13,400 mL/h, respectively. When the kinetic
parameters from uptake studies with 1,3-butadiene were used with kinetic parameters for EB to
predict concentrations of EB in the chamber atmosphere during exposures to 1,3-butadiene, the
concentrations of EB were overpredicted by approximately fivefold. Filser and Bolt
hypothesized that this may be due to an "intrahepatic first pass effect," where the EB formed
from 1,3-butadiene is preferentially oxidized, hydrolyzed, or conjugated with glutathione and is
not available systemically.
Kreiling et al. (1987) estimated in vivo pharmacokinetic parameters for EB in male
B6C3FJ mice that were exposed in closed chambers by inhalation to initial concentrations of
100-2,000 ppm EB or constant concentrations of 2,000-4,000 ppm 1,3-butadiene. The kinetic
parameters for the previously described two-compartment system, k12Vj and Cltot, were 33,500
and 24,900 mL/h, respectively, indicating that unlike in rats, metabolism of EB, not uptake,
limits clearance in mice. In addition, unlike in rats, metabolism of EB in mice appeared to be
saturable above atmospheric concentrations of 500 ppm EB. As with the kinetics of 1,3-
butadiene, mice had a higher Kst than rats, approximately 10-fold higher, because of higher
inhalation uptake and lower elimination than rats. Therefore mice would be expected to have a
higher body burden of EB than rats.
Valentine et al. (1997) quantitated classical pharmacokinetic parameters in male Sprague-
Dawley rats that were administered 71, 143, and 286 (Imol EB/kg body weight by intravenous
injection. The generation of DEB in the blood was below the limit of detection for the method of
3-29
-------
analysis. The AUC^T was increased significantly at the highest EB concentration indicating
saturation of metabolism; however, neither the systemic clearance (Cls) nor the volume of
distribution at steady state (Vss) were decreased significantly. Since the Vss and the volume of
central compartment (Vc) were similar, EB was not distributed significantly from the central
compartment. In addition Cls was equal to or greater than estimated liver perfusion in the rat at
all doses indicating that extrahepatic clearance such as exhalation contributed to elimination of
EB from the rat. The systemic clearance of EB in rats was less than the metabolic clearance
reported by Kreiling et al. (1987), 67-114 mL/min/kg versus 223 mL/min/kg, respectively. The
estimated terminal half life of EB ranged from 5.7 to 8.5 min. Species comparisons cannot be
done with the work of Valentine et al. (1997) because only rats were studied.
3.3.3.3.3. DEB. Only one study to date has evaluated the in vivo pharmacokinetics of DEB
(Valentine et al., 1997). Valentine and co-workers administered 523 |lmol/kg DEB to male
Sprague-Dawley rats intravenously and quantitated DEB in venous blood versus time. The data
were used to estimate classical two-compartment pharmacokinetic parameters. The kinetic
parameters for DEB were similar to EB in rats. DEB was not distributed significantly from the
central compartment because the Vss was similar to Vc. In addition Cls was approximately the
estimated liver perfusion in the rat, indicating that clearance of DEB in rat was not limited by
metabolism. The estimated terminal half-life of DEB was 13.8 min. Unfortunately, species
comparisons cannot be done with the work of Valentine et al. because only rats were studied.
3.3.3.3.4. BDdiol. One study to date has quantitated the kinetics of BDdiol in vivo. Kemper et
al. (1998) administered 10, 25, 50, 100, and 250 mg/kg body weight BDdiol to male E6C3Fl
mice intraperitoneally. Blood from the animals was collected at time points up to 4-8 h
following dosing and was analyzed for BDdiol; urine was collected for 24 h postexposure and
analyzed for BDdiol and metabolites. In addition mice were pretreated with either 4-
methylpyrazole, an alcohol dehydrogenase inhibitor and P450 inhibitor, or benzylimidazole, a
cytochrome P450 inhibitor, and administered 25 mg/kg BDdiol to study the effects of inhibition
on BDdiol clearance. Although results were difficult to interpret from intraperitoneal data,
Kemper et al. reported that plasma clearance of BDdiol appeared saturated at the higher
administered doses. Hepatic and renal reduced nonprotein thiol was significantly depleted in
mice exposed to 250 mg/kg BDdiol. Pretreatment with methylpyrazole provided some protection
from thiol depletion.
5-30
-------
3.4. DISTRIBUTION
To determine if differences in tissue concentrations of reactive metabolites correlate with
differences in responses among species and tissues, several studies have measured the blood and
tissue concentrations of 1,3-butadiene and its reactive metabolites EB and DEB during and
following inhalation exposures to 1,3-butadiene (Bechtold et al., 1995; Bond et al., 1987; Dahl et
al., 199l;Himmelstein etal., 1994, 1995; Thornton-Manning et al., 1995a,b; 1997, 1998). The
results from the studies indicate that higher concentrations of reactive metabolites in target
tissues will be achieved in mice versus rats because of differences in the rates of metabolism of
1,3-butadiene and its metabolites, which have been noted from in vitro and in vivo metabolism
studies. In addition, investigators also have quantitated hemoglobin and DNA adducts of
reactive metabolites in mice, rats, and humans exposed to 1,3-butadiene and found differences
among the species.
The following section summarizes the research that illustrates species differences in the
distribution of 1,3-butadiene and its reactive metabolites. None of the studies measuring blood
and tissue concentrations of DEB in vivo have quantitated meso-DEB, only the enantiomeric
forms (±)DEB. This section does not include the concentration reported in rodents as measured
by Bond et al. (1986) by the technique of cryogenic trapping because of inaccuracies of the
technique in comparison with GC-MS. Unfortunately, in monkeys cryogenic trapping was the
only technique employed for quantitation of 1,3-butadiene and metabolites following inhalation
exposures to 1,3-butadiene (Dahl et al., 1991). These results are included; however, the
concentrations should not be compared with concentrations from rodents measured by a different
technique. The analysis of data in tables from Dahl et al. (1990, 1991) is not included in this
summary because of possible inaccuracies in the data. The authors, when comparing distribution
of volatile and nonvolatile metabolites in blood, did not include the amount of CO2 as a
metabolite of 1,3-butadiene. In addition, as seen from Thornton-Manning et al. (1995a,b), the
blood concentrations of metabolites may not be at steady state at 2 h, and therefore the
comparison of blood concentrations at different exposure times may reflect differences in
distribution as well as metabolism.
3.4.1. Blood and Tissue Concentrations
3.4.1.1. Species Differences
To determine if differences in distribution between rats and mice existed following
inhalation exposures to 1,3-butadiene, Bond et al. (1987) measured the distribution of 14C at 1, 2,
4, 8, 18, 27, 43, 51, and 67 h and 6, 10, and 13 days postexposure in male Sprague-Dawley rats
following a 3.4-h nose-only inhalation exposure to 670 ppm l,3-[l-14C]butadiene and in male
B6C3FJ mice following a 3.4 h nose-only inhalation exposure to 65 ppm l,3-[l-14C]butadiene.
3-31
-------
The radioactivity that remained at the time of tissue sampling was distributed widely throughout
the body in both species. In both species at 1 h postexposure the highest concentration of
radioactivity (expressed as nmol 14C-butadiene equivalents/g tissue) was found in the bladder.
However, none of the tissues appeared to accumulate 1,3-butadiene or its metabolites. At all
timepoints following exposure, rats had higher concentrations of radioactivity in most tissues;
however, when the tissue concentrations were normalized to the total amount of radioactivity that
was inhaled, the normalized concentration in mice exceeded the rats. Except for the fat at 1,2,
and 4 h postexposure, the majority of the radioactivity that was measured in the tissues was due
to nonvolatile metabolites. The study does not reveal if there were distributional differences in
the concentrations of reactive metabolites because the radioactivity was not speciated. However,
the study does show that 1,3-butadiene and its metabolites do not accumulate in the body to a
significant extent, and that mice have a higher rate of metabolism than rats for conversion of 1,3-
butadiene to other metabolites because the tissue concentrations of total radioactivity per amount
of inhaled 1,3-butadiene were greater in mice than rats.
The blood and tissue concentrations of the reactive metabolites EB and DEB have been
quantitated in mice and rats by several investigators (Bechtold et al., 1995; Himmelstein et al.,
1994, 1995; Thornton-Manning et al., 1995a,b; 1997; 1998). None of the studies have
quantitated specific enantiomers of EB and have quantitated only the racemic form of DEB, not
the meso form. In all studies the blood and tissue concentrations of the reactive metabolites in
mice exceeded the concentrations measured in rats. Figures 3-2 and 3-3 summarize the blood
concentrations in male B6C3FJ mice and Sprague-Dawley rats at 3, 4, and 6 h of inhalation
exposure to concentrations of 1,3-butadiene ranging from 62.5 to 8,000 ppm. Tables 3-11 to
3-18 contain information related to blood and tissue concentrations of EB and DEB from the
experiments described in more detail below.
Himmelstein et al. (1994) measured blood concentrations of 1,3-butadiene and
metabolites during and up to 30 min following exposure in mice and rats exposed by nose-only
inhalation to 62.5, 625, and 1,250 ppm 1,3-butadiene for 6 h. Steady-state blood concentrations
of 1,3-butadiene were achieved in both species by 2 h. The blood concentrations in mice were
approximately twice the blood concentrations in rats because of higher alveolar ventilation per
gram of body weight of mice versus rats. In addition, steady-state blood concentrations of EB
were attained in both species by 2 h of exposure. The blood concentrations of EB in mice were
approximately 4-8 times higher than blood concentrations in rats. The blood concentrations of
EB (mean ± s.d.) ranged from 0.56 ± 0.04 to 8.6 ± 0.6 |lM and 0.07 ± 0.01 to 1.3 ± 0.09 |lM in
mice and rats, respectively. DEB was detected only in the blood of mice at concentrations
ranging from 0.65 ± 0.10 to 2.5 ± 0.4 |lM. The blood concentrations of DEB appeared to reach
steady state during the exposure for 1,250 pm 1,3-butadiene exposure only.
3-32
-------
CO
lOOOO-i
s °
.2 S
£ S
II
5 -o
•a o
o g<
o i
5 "S
1000-
100-
10-
o
(L)
3
O
I
•a
Mouse 4h timepoint
Mouse 6h timepoint
Rat 4h timepoint
I Rat 6 h timepoint
62.50
100.00
625.00
1250.00
8000.00
1,3-Butadiene exposure concentrations (ppm)
Figure 3-2. Comparison of blood concentrations of EB in male mice and ratsa.
The data were obtained from Bechold et al. (1995), Himmelstein et al. (1994), Himmelstein et al. (1995), Thornton-Manning et al. (1995a), and Thornton-Manning et
al. (1995b).
-------
lOOOO-i
^
s
K>
'S -S? 1000-
Blood concentration
l,2:3,4-diepoxybutane (pmol
H- 0
-00
1 ' '
CO
0
^
1 1
X
1
r
8 §
° 1
CO
O
1
I
1
I
1
I Mouse 3h timepoint
Mouse 4h timepoint
Mouse 6h timepoint
::Ell3:Rat 4h timepoint
^H Rat 6h timepoint
62.50 100.00 625.00 1250.00
1,3-Butadiene exposure concentration (ppm)
Figure 3-3. Comparison of blood concentrations of DEB in male mice and ratsa.
The data were obtained from Bechtold et al. (1995), Himmelstein et al. (1994), Himmelstein et al. (1995), Thornton-Manning et al. (1995a), and Thornton-Manning et
al. (1995b).
-------
Another study by Bechtold et al. (1995) measured blood concentrations of EB and DEB
in the blood of mice and rats exposed by nose-only inhalation to 100 ppm 1,3-butadiene. This
study supported higher concentration of EB and DEB in the blood of mice compared with rats,
consistent with the results of Himmelstein et al. (1994). However, this research used cryogenic
isolation and gas chromatography-gas chromatography mass spectrometry (GC-GCMS) to
achieve better sensitivity than Himmelstein (limits of detection twofold lower).
Himmelstein et al. (1995) also measured the concentrations of EB and DEB in lungs and
livers of male B6C3FJ mice exposed to 625 or 1,250 ppm 1,3-butadiene and in lungs and livers
of male Sprague-Dawley rats exposed to 625, 1,250, or 8,000 ppm 1,3-butadiene by nose-only
inhalation exposures. Limitations of this data include inaccuracy in reported concentrations as a
result of the time lag from euthanasia to collection of tissue, which will result in loss of some of
the parent and metabolites from the tissue. Since the rates of metabolism differ between mice
and rats, the amount removed during the period of collection will differ, and therefore the
difference in concentrations measured may not represent the true difference in concentrations
during exposure. Maximum concentrations (mean ± s.d.) that were measured in the lungs of
mice during the 625 ppm or 1,250 ppm exposures were 2,600 ± 200 and 3,700 ± 1,200 pmol/g
tissue for EB, respectively, and 710 ± 60 and 1,500 ± 100 pmol/g tissue for DEB, respectively.
EB, but not DEB, was detected in the lungs of rats, and the concentrations were approximately
11-16 times lower than in mice, 160 ±30, 310 ±70, and 1,300 ± 200 pmol/g tissue, for 625,
1,250, and 8,000 ppm 1,3-butadiene, respectively. In addition, EB, but not DEB, was detected in
the livers from both mice and rats. Maximum concentrations measured in the livers of mice
during the 625 ppm or 1,250 ppm exposures were 580 ± 120 and 930 ± 190 pmol /g tissue,
respectively. Concentrations of EB in the livers of rats were approximately 9-11 times lower
than mice, 60 ± 10, 160 ± 60, and 1,200 ± 100 pmol/g tissue, for 625, 1,250, and 8,000 ppm 1,3-
butadiene, respectively.
Thornton-Manning et al. (1995a) also tried to address differences among species by
measuring the concentrations of EB and DEB in blood and various tissues during and following
nose-only inhalation exposure of male Sprague-Dawley rats and male B6C3FJ mice to 62.5 ppm
1,3-butadiene for 4 h. The technique for blood collection was the same as that of Himmelstein et
al. (1994); however, the method for tissue collection was different in that Thornton-Manning
(1995a) froze the tissues in liquid nitrogen after collection and prepared them for analysis by GC-
GCMS with a modification of a vacuum line-cryogenic distillation method (Dahl et al., 1984). In
addition the method of analysis for 1,3-butadiene, EB, and DEB by GC-GCMS differed from
Himmelstein et al. (1994 and 1995), resulting in a lower limit of detection for DEB (reported
detection limits for tissues were 11-1,300 times lower than limits reported by Himmelstein and
co-workers). This lower limit of detection allowed for detection of DEB in the blood of rats
3-35
-------
exposed to 1,3-butadiene. However, the paper reported two different limits of detection for EB
and DEB and did not explain the difference. Some of the values (mouse lung EB concentrations,
and rat heart and lung DEB concentrations) reported by Thornton-Manning were below the
reported limits of detection conducted via the EPA-approved procedure. The same limitations
for interpretation of the data also apply with this research because of loss of compound from time
of euthanasia to tissue collection. Thornton-Manning et al. quantitated EB and DEB in blood,
heart, lung liver, fat, spleen, thymus, and bone marrow. Some tissues had undetectable amounts
of EB (rat lung and liver) and DEB (rat liver and bone marrow). Table 3-11 reports measured
concentrations. The concentrations were used to calculate tissue:blood concentration ratios to
illustrate relative to blood which tissues had the greatest accumulation of reactive metabolites
(Table 3-12). For EB the highest concentrations were in fat in both species, and for DEB the
highest concentrations were reported in blood in both species. Consistent with Himmelstein et
al. (1995), mice had higher tissue concentrations of reactive metabolites compared with rats. The
ratio of tissue concentrations of EB in mice versus rats ranged from 3 to 11.5, with the highest
concentration difference occurring in bone marrow. The ratio of tissue concentrations of DEB in
mice versus rats ranged from 4 to 163, with the highest concentration difference occurring in
lung (Table 3-13).
Although the higher concentrations of DEB in mouse blood compared with rat blood are
consistent with Himmelstein et al. (1994), the mouse blood DEB concentrations reported by
Thornton-Manning (1995a) appear to be approximately half the concentrations reported by
Himmelstein et al. (1994). Thornton-Manning (1995a) reported blood concentrations of DEB at
2 h and 4 h of exposure of 100 and 204 pmol/g versus reported concentrations at 3 h and 6 h by
Himmelstein (1994) of 200 and 695 pmol/g. If the increase of DEB versus time were linear, a
blood concentration of 365 pmol/g would be predicted for 4 h exposure from Himmelstein's
data. Part of the difference may result from differences in the actual exposure concentrations,
approximately 2.1 versus 2.8 |lmol/L 1,3-butadiene, for Thornton-Manning and Himmelstein,
respectively. These concentrations were calculated from actual exposure concentrations
expressed in ppm and the estimated molar volume for each laboratory.
The tissue:blood concentration ratios in mice (Table 3-12) were less than the
reported partition coefficients (Table 3-1) for EB in the liver and lung and for DEB in the liver
and fat. The tissue:blood concentration ratio of EB in fat was greater than the reported partition
coefficient. Other tissues could not be compared because the partition coefficients have not been
reported. Concentrations of EB have been shown to reach steady state in the body within 4 h;
therefore, the ratios in the liver and lung are lower than the partition coefficients, probably
5-36
-------
Table 3-11. Tissuerblood concentrations (pmol/g tissue)a.
Tissue
Blood
Heart
Lung
Liver
Fat
Spleen
Thymus
Bone marrow
3,4-Epoxybutene
Rats Mice
36 ±7
40 ± 16
NDb
ND
267 ± 14
7±6
12.5 ±3.2
0.2 ±0.1
295 ± 27
120 ± 15
33 ±9
8±4
1,302 ±213
40 ± 19
104 ±55
2.3 ±1.5
1,2,3,4-Diepoxybutane
Rats Mice
5± 1
3 ±0.4
0.7 ±0.2
ND
2.6 ±0.4
1.7 ±0.5
2.7 ±0.7
ND
204 ± 15
144 ± 16
114±37
20 ±4
98 ± 15
95 ± 12
109 ± 19
1.4 ±0.3
aThe concentrations of EB and DEB (mean ± SE; n=3 or 4) in each of the tissues of male Sprague-Dawley rats and male
B6C3F! mice following a 4 h exposure to 62.5 ppm 1,3-butadiene were taken from Thornton-Manning et al. (1995a).
bND is reported for tissues where the concentration of the analyte was below the limit of detection.
Table 3-12. Tissuerblood concentration ratiosa.
3,4-EDOxvbutene
Tissue
Heart
Lung
Liver
Fat
Spleen
Thymus
Bone Marrow
Rats
1.1
NDb
ND
7.4
0.19
0.35
0.006
Mice
0.41
0.11
0.027
4.4
0.14
0.35
0.008
1,2,3,4-Diepoxvbutane
Rats
0.6
0.14
ND
0.52
0.34
0.54
ND
Mice
0.71
0.56
0.098
0.48
0.47
0.54
0.007
aThe ratios were determined from the corresponding tissue and blood concentrations of EB and DEB in each of the
tissues of male Sprague-Dawley rats and male B6C3F! mice following a 4 h exposure to 62.5 ppm 1,3-butadiene
(Thornton-Manning etal, 1995a).
bND is reported for tissues where the concentration of the analyte was below the limit of detection.
5-37
-------
Table 3-13. Mouse:rat concentration ratiosa.
Tissue
Blood
Heart
Lung
Liver
Fat
Spleen
Thymus
Bone marrow
3,4-Epoxybutene
8.2
3
NDb
ND
4.9
5.7
8.3
11.5
1,2,3,4-Diepoxybutane
41
48
163
ND
38
56
4
ND
aThe ratios were determined from the corresponding tissue and blood concentrations of EB and DEB in each of the
tissues of male Sprague-Dawley rats and male B6C3F[ mice following a 4-h exposure to 62.5 ppm 1,3-butadiene
(Thornton-Manning etal, 1995a).
bND is reported for tissues where the concentration was not available for the rat.
because of metabolism of the compound in these tissues. For DEB, steady state was not
achieved during the 4-h exposure; therefore, the lower ratios could be caused by distribution of
the compound through the body or because of metabolism in these tissues.
The more accurate methodology of GC-MS has not been performed for the blood and
tissues of monkeys exposed to 1,3-butadiene by inhalation. The blood and tissue of monkeys
have been analyzed by a technique called cryogenic trapping to quantitate CO2, 1,3-butadiene,
EB, and DEB and BDdiol (Dahl et al., 1991). The technique has limitations in that the trapping
of compounds may not be specific; therefore, the concentrations cannot be compared with rodent
data quantitated by GC-MS. Dahl et al. (1991) exposed monkeys for 2 h by inhalation to 1,3-[1-
14C]butadiene at concentrations of 10.1, 310, or 7,760 ppm. The fraction of radioactivity
attributed to nonvolatile metabolites represented the majority of metabolites quantitated in blood,
which was 87% and 54% of total radioactivity in blood at 10 and 310 ppm 1,3-butadiene,
respectively. However, at 7,760 ppm 1,3-butadiene, the fraction attributed to 1,3-butadiene
represented the majority of material quantitated in blood, 68% of total radioactivity. The fraction
attributed to CO2 was a larger percentage of the total radioactivity in blood than either EB or
DEB. The fraction attributed to EB was 1.6 ± 0.5, 500 ± 280, and 1,100 ± 500 pmol 1,3-
[14C]butadiene equivalents/mL blood (mean ± SE, n = 3) for the exposures to 10.1, 310, or 7,760
ppm 1,3-butadiene, respectively. The fraction attributed to a combination of DEB and BDdiol
5-38
-------
was 1.9 ± 0.6, 290 ± 170, and 590 ± 240 pmol l,3-[14C]butadiene equivalents/mL blood (mean ±
SE, n = 3) for the exposures to 10.1, 310, or 7,760 ppm 1,3-butadiene, respectively. These two
fractions combined represented a total of 1%, 17.3%, and 4.7% of the radioactivity quantitated in
the blood for the 10.1, 310, and 7,760 ppm 1,3-butadiene, respectively.
3.4.1.2. Gender Differences
The potential gender differences in the kinetics of 1,3-butadiene has been researched to a
limited extent. Thornton-Manning et al. (1995b) examined gender differences in the distribution
of EB and DEB by determining tissue (blood, femur, lung, and liver) concentrations of EB and
DEB in male and female Sprague-Dawley rats exposed by nose-only inhalation to 62.5 ppm of
1,3-butadiene for 6 h. For EB only the lung concentration differed significantly between male
and female rats (Table 3-14). The concentration of EB in the lungs of male rats was five times
higher than in female rats. The highest concentrations of EB were observed in the adipose
tissues of both sexes. In contrast, all tissue concentrations of the DEB were greater in females
than in males. The concentrations of DEB in blood from female rats were 4.75-fold greater than
in blood from male rats. The greatest gender difference was seen in the concentration of DEB in
adipose tissue, with females having a sevenfold greater tissue concentration than males. The
mammary tissue of females also contained relatively high concentrations of DEB. However, the
data in this study are not consistent with results from previous studies. The concentrations of
DEB in blood, femur, and fat from male rats in this study following a 6-h exposure were
approximately 50% of the concentrations from a previously reported 4-h exposure (Thornton-
Manning et al., 1995a), and the lung concentrations of DEB were approximately twofold higher
for the 6-h exposure compared with the 4-h exposure.
Thornton-Manning et al. (1995b) suggest that the greater production of the highly
mutagenic DEB in females may play a role in the increased incidence of mammary tumors
observed in a chronic carcinogenicity study with rats (Owen et al., 1987). However, in contrast
with the rats, the blood concentrations of EB and DEB in female B6C3FJ mice are only 50% of
the concentrations in male mice following nose-only inhalation exposure for 6 h to 62.5 ppm 1,3-
butadiene (data obtained from Himmelstein et al., 1994, and Thornton-Manning et al., 1997).
Therefore, the gender differences noted from the rats cannot be generalized to all species.
3.4.1.3. Chronic Exposures
An important issue concerning distribution of 1,3-butadiene and its metabolites is
whether distribution is altered by chronic exposures. This is important because the cancer
bioassays are repeated daily exposures. Bond et al. (1988) showed reduced metabolism of
5-39
-------
Table 3-14. Femalermale concentration ratio".
Tissue
Blood
Femur
Lung
Fat
3,4-Epoxybutene
Rats" Micec
1.1
1.1
0.21
1.1
0.43
NDd
ND
ND
1,2,3,4-Diepoxybutane
Rats" Micec
4.8
6.4
3.4
7
0.49
ND
ND
ND
aRatios of the tissue concentrations of EB and DEB in female and male Sprague-Dawley rats and B6C3F! mice exposed
for 6 h by nose-only inhalation to 62.5 ppm 1,3-butadiene.
bThe tissue concentrations for both male and female rats were obtained from Thornton-Manning et al. (1995b).
The blood concentration was obtained from Himmelstein et al. (1994) for male mice and from Thornton-Manning et al.
(1995b) for female mice.
dND is reported for tissues where data were not available for one of the genders.
1,3-butadiene in vitro in lung microsomal incubations from rats and mice repeatedly exposed to
1,3-butadiene by inhalation.
To address the effects of chronic exposures, Thornton-Manning et al. (1997) exposed
female Sprague-Dawley rats and B6C3FJ mice by nose-only inhalation exposure to 62.5 ppm 1,3-
butadiene either for a single 6-h exposure or for 6 h/day for 10 days. Concentrations of EB and
DEB were analyzed in the blood, femur, mammary gland, fat, and ovary immediately following
exposure as described previously (Thornton-Manning et al., 1995a). Table 3-15 shows the
concentrations of EB and DEB in the blood of both species following either a single or repeated
exposure to 1,3-butadiene (In the table, the blood concentrations for rats following a single
exposure are the values reported in Thornton-Manning et al. [1995b] because the authors stated
that this was where the data were obtained. However, the values reported in Thornton-Manning
et al. [1997] for blood for the rat were greater than in the original citation, 44 versus 29.4 pmol/g
for EB and 14 versus 11.4 pmol/g for DEB. The authors gave no explanation for the
discrepancy.) Although blood concentrations of EB increased in both species following repeated
exposures, the increase was not statistically significant. Again, although not statistically
significant, blood concentrations of DEB in the rat increased following repeated exposures,
whereas blood concentrations of DEB in the mouse decreased. The authors found statistically
significant increases in concentration of EB in mammary tissue and fat in both mice and rats and
in femurs of rats. The authors found statistically significant increases in concentration of DEB in
5-40
-------
Table 3-15. Blood concentrations of reactive metabolites (pmol/g tissue)a.
3,4-Epoxybutene
Species
Rat
Mouse
Single
exposure
29.4 ± 2.0b
239 ± 24
Repeated
exposure
64 ±8
317± 19
1,2,3,4-Diepoxybutane
Single
exposure
11.4±2b
345 ±33
Repeated
exposure
17±2
247 ± 32
aBlood concentrations (mean ± SE, n = 3-6) of EB or DEB in female Sprague-Dawley rats and B6C3Fj mice following a
single 6-h nose-only inhalation exposure or following repeated 6-h nose-only inhalation exposures for 10 days to 62.5
ppm 1,3-butadiene (Thornton-Manning et al., 1997).
bOriginal value that was reported in Thornton-Manning et al. (1995b).
the mammary gland of rats and statistically significant decreases in concentration of DEB in the
lung, fat, and mammary glands of mice. Tissue:blood concentration ratios calculated from the
concentrations of EB and DEB illustrate that distribution of EB and DEB changes from single
and repeated exposures (Tables 3-16 and 3-17). However, the experiments conducted in
Thornton-Manning et al. (1997) cannot address whether these differences resulted from changes
in the rates of formation or degradation of EB and DEB following repeated exposures or
reflected the accumulation of the compounds in tissues as the analytes approached steady-state
following repeated exposures. In addition, as seen in Table 3-18, the female mouse:rat
concentration ratios of EB and DEB decrease with repeated inhalation exposure to 1,3-butadiene
Another study (Thornton-Manning et al., 1998) showed different results in female rats
exposed to higher concentrations of 1,3-butadiene. The study reported blood concentrations of
EB and DEB from female Sprague-Dawley rats exposed by nose-only inhalation to 8,000 ppm
1,3-butadiene for 8 h or repeatedly for 8 h/day for 10 days. In contrast to the 62.5 ppm 1,3-
butadiene exposures, the blood concentration of EB decreased, 4,030 ± 190 versus 3,310 ± 260
pmol/g tissue, for single and repeated exposures to 8,000 ppm 1,3-butadiene, respectively, but
the decrease was not statistically significant. Consistent with the 62.5 ppm 1,3-butadiene
exposures, blood concentration of DEB increased significantly, 11 ± 1 versus 17 ± 1 pmol/g
tissue, for single and repeated exposures to 8,000 ppm, respectively.
An interesting result from the two previously described chronic studies by Thornton-
Manning et al. (1997, 1998) is the comparison of blood and tissue concentrations of reactive
metabolites of 1,3-butadiene. In both studies, concentrations of EB and DEB were measured in
blood, femur, lung, mammary gland, fat, heart, uterus, and ovary. Tissue and blood
concentrations of EB were significantly greater in female rats exposed repeatedly to 8,000 ppm
5-41
-------
Table 3-16. 3,4-Epoxybutene tissuerblood concentration ratios".
Rats
Tissue
Femur
Lung
Mammary gland
Fat
Single
exposure
0.34
0.11
1.95
7.1
Repeated
exposure
0.23
NDC
4.8
7
Mice
Single
exposure
0.24
0.11
3.1
5
Repeated
exposure
NRb
0.47
3.8
5.3
aThe ratios were calculated from blood and tissue concentrations in female Sprague-Dawley rats and B6C3F[ mice
following a single 6-h nose-only inhalation exposure or following repeated 6-h nose-only inhalation exposures for 10 days
to 62.5 ppm 1,3-butadiene (Thornton-Manning et al, 1997). The blood concentrations used in the calculations are given
in Table 3-14.
bNR is reported for the tissues in which EB was not reported.
°ND is reported for the tissues in which EB was not detected.
Table 3-17. 1,2,3,4-Diepoxybutane tissuerblood concentration ratios".
Rats
Tissue
Femur
Lung
Mammary gland
Ovary
Fat
Single
exposure
0.64
0.43
0.95
0.56
0.7
Repeated
exposure
0.53
0.25
0.87
0.56
0.72
Mice
Single
exposure
0.62
0.63
0.77
0.49
0.59
Repeated
exposure
0.74
0.58
0.77
0.61
0.7
aThe ratios were calculated from blood and tissue concentrations in female Sprague-Dawley rats and B6C3F[ mice
following a single 6-h nose-only inhalation exposure or following repeated 6-h nose-only inhalation exposures for 10 days
to 62.5 ppm 1,3-butadiene (Thornton-Manning et al., 1997). The blood concentrations used in the calculations are given
in Table 3-14.
5-42
-------
Table 3-18. Mouserrat concentration ratios.
3,4-EDOxvbutenea
Tissue
Blood
Femur
Lung
Mammary
Ovary
Fat
Single
exposure
8.1b
5.7
8
13
5.6
Repeated
exposure
5
NRC
NDd
3.9
3.7
l,2,3,4-DieDoxvbutanea
Single
exposure
30b
31
43
24
28
25
Repeated
exposure
15
20
36
13
15
13
aMouse:rat ratios were determined from tissue concentrations of EB and DEB in female Sprague-Dawley rats and female
B6C3FJ mice following a single 6-h nose-only inhalation exposure or following repeated 6-h nose-only inhalation
exposures for 10 days to 62.5 ppm 1,3-butadiene (Thornton-Manning et al., 1997).
bAlthough Thornton-Manning et al. (1997) reported blood concentrations of DEB in rats as 14 pmol/g respectively
following a single 6-h exposure to 62.5 ppm 1,3-butadiene, the blood concentration (11.4 pmol/g) from the original work
(Thornton-Manning et al., 1995b) from which the concentration was originally reported is used to generate this table.
The reason for the discrepancy between the two papers is not clear.
°NR is reported for the tissues in which EB was not reported.
dND is reported for the tissues in which EB was not detected.
compared with female rats exposed repeatedly to 62.5 ppm 1,3-butadiene. However, the blood
and fat concentrations of DEB were not significantly different, the mammary gland and femur
concentrations of DEB decreased significantly, and the lung concentration increased significantly
in female rats repeatedly exposed to 8,000 ppm 1,3-butadiene compared with female rats exposed
repeatedly to 62.5 ppm 1,3-butadiene.
3.4.2. Molecular Dosimetry
Several investigators have examined binding of the metabolites of 1,3-butadiene to
hemoglobin, DNA, and nucleoproteins to correlate with the varying toxicity noted among animal
species and to develop an internal marker for use as a dosimeter to measure 1,3-butadiene
exposure in humans. Early work focused mainly on quantitation of total radioactivity associated
with hemoglobin, DNA, or nucleoproteins in tissues following in vivo exposures to 1,3-
butadiene, and more recent work has focused on identification and quantification of specific
adducts of hemoglobin and DNA. The studies note species differences in concentration of the
> A O
5-43
-------
molecular markers, with the differences most pronounced at higher concentrations and more
similarity at lower concentrations. The differences are not as significant as the species
differences in toxic responses. A marker that is specific for the internal dose of DEB has not
been identified. This section includes only in vivo research that studied the association between
the exposure concentration of 1,3-butadiene and specific biomarker concentrations as well as
species differences in concentrations of the biomarkers for a given 1,3-butadiene exposure
concentration.
3.4.2.1. Hemoglobin Adducts
Several investigators have studied the use of hemoglobin adducts as biomarkers of
exposure to 1,3-butadiene. Most of the studies to date have quantitated adducts formed by EB,
although some recent studies have quantitated hemoglobin adducts resulting from either EBD or
DEB conjugation with hemoglobin. Several investigators have noted species differences in
hemoglobin adduct formation from metabolites of 1,3-butadiene.
Sun et al. (1989b) injected male B6C3FJ mice and male Sprague-Dawley rats
intraperitoneally with 1,3-[14C]-BD at doses of 1, 10, 100, or 1,000 |lmol/kg, and hemoglobin
adduct formation was quantitated at 24 h after exposure. Hb adduct concentrations were
quantitated by measuring total 14C associated with globin isolated from the collected blood
samples and was not metabolite specific. Hb adduct formation was linearly related to doses up to
100 |imol/kg for both species. The Hb adducts accumulated linearly after repeated injections of
100 |imol/kg for 3 days. The BD-derived Hb adducts showed lifetimes of approximately 24 and
65 days in mice and rats, respectively, which correlates with the lifetimes of red blood cells. In
this study rats were four to five times more efficient in forming hemoglobin adducts compared
with mice; however, this may be because of the route of administration and may not reflect the
dosimetry that would be expected following inhalation exposures.
Osterman-Golkar et al. (1991) investigated binding of EB to hemoglobin as a potential
indicator of 1,3-butadiene exposure. Osterman-Golkar and co-workers exposed male Wistar rats
by inhalation to 1,3-butadiene at concentrations of 250, 500, or 1,000 ppm for 6 h/day, 5
days/week for 2 weeks. Adducts of EB at the N-terminal valine in hemoglobin were analyzed by
the N-alkyl Edman procedure followed by quantitation with GC-MS of the thiohydantoin
derivatives. The investigators only quantitated adducts formed from binding at carbon 1 of the
enantiomers of EB. In this study, the hemoglobin adduct formation was stable and increased
linearly with exposure concentration, proving to be a useful dosimeter for long-term exposure to
1,3-butadiene. However, the authors noted that the quantitation method used in the study lacks
the sensitivity to detect the low levels of adducts that would be expected in humans from
occupational exposure to 1,3-butadiene.
3-44
-------
Albrecht et al. (1993) also studied the formation of adducts from EB in female Wistar
rats and CB6F1 mice exposed to 0, 50, 200, 500 or 1,300 ppm 1,3-butadiene by inhalation for 6
h/day for 5 days. Hemoglobin was isolated 18 h following the last exposure, and N-terminal
valine adducts were isolated by a modified Edman degradation procedure. The two regioisomers
formed from binding of EB at either carbon 1 or 2 to valine were quantitated, but the
diastereomers for each regioisomer were not resolved by the method used. The hemoglobin
adducts quantitated in mice were approximately three to six times greater than in rats. The rats
exhibited a linear dose-dependent increase in adduct concentration up to exposure concentrations
of 1,300 ppm 1,3-butadiene, whereas the mice exhibited a nonlinear dose-dependent increase in
adduct concentration.
Osterman-Golkar et al. (1998) measured N-terminal valine adducts of EB in male
B6C3FJ mice and Sprague-Dawley rats exposed by inhalation to 0, 2, 10, or 100 ppm 1,3-
butadiene 6 h/day, 5 days/week for 1, 2, 3, or 4 weeks. In the study, the diastereomers formed
from binding at carbon 1 of EB with valine, N-(2-hydroxy-3-butenyl)valine, were quantitated,
and the amount of adducts from carbon 2 was based on the ratio of adducts that had been
measured from other in vitro and in vivo experiments. In contrast to Albrecht et al. (1993), the
concentrations of adducts increased linearly with 1,3-butadiene exposure concentration in mice
but not in rats. The adduct levels were approximately four times greater in mice compared with
rats at 100 ppm 1,3-butadiene, but at lower 1,3-butadiene exposure concentrations the adducts
were similar. The authors reported that blood concentrations of EB estimated from the
hemoglobin adduct levels agreed well with reported concentrations of EB in BD-exposed mice
and rats (Himmelstein et al., 1994; Thornton-Manning et al., 1995a).
Swenberg et al. (2000) reported hemoglobin adducts derived from EB. The investigators
measured diastereomers formed from conjugation at carbon 1 and carbon 2 of EB that were
approximately two times greater in female and male B6C3FJ mice compared with Crl:CDBR
rats exposed by inhalation to 1,000 ppm 1,3-butadiene 6 h/day, 5 days/week for 13 weeks.
Recent work has focused on developing methods to measure hemoglobin adducts from
DEB and EBD. Perez et al. (1997) developed a method to measure N-(2,3,4-
trihydroxybutyl)valine adducts, which can be formed from reaction of either DEB or EBD with
N-terminal valine. In male Wistar rats exposed to 0, 50, 200, or 500 ppm 1,3-butadiene by
inhalation 6 h/day for 5 days, the concentrations of two of the stereoisomeric forms of N-(2,3,4-
trihydroxybutyl)valine were approximately 3-32 times greater than the hemoglobin adducts
formed by EB at carbon 1. The ratio was greatest at the lowest 1,3-butadiene exposure
concentration. This suggests that the majority of the trihydroxyvaline adducts were formed from
EBD, because EB blood concentrations in rodents exceed DEB concentrations.
5-45
-------
Gender differences have not been noted in studies with hemoglobin adducts from EB or
trihydroxybutyl adducts from EBD or DEB. In the previously described study of Albrecht et al.
(1993), the investigators also compared adduct formation in male and female C3HX101/EL mice
exposed by inhalation to 0-500 ppm 1,3-butadiene (specifics of exposures were not given) and
did not find a gender difference in adduct formation. The results for female Wistar rats reported
by Albrecht et al. (1993) cannot be compared with those of Osterman-Golkar et al. (1991) for
male Wistar rats to examine gender differences in rats for several reasons: (a) only one exposure
concentration was comparable, (b) Albrecht quantitated both regioisomers of EB whereas
Osteman-Golkar quantitated only one, and (c) the duration of the exposures differed between
studies. Tretyakova et al. (1996) reported greater concentrations of N-terminal valine adducts
from EB in female B6C3FJ mice compared with male mice exposed to 1,000 ppm 1,3-butadiene
6 h/day, 5 days/week for 13 weeks; however, the authors did not state whether the difference was
statistically significant and the variation in the data was as high as 64%. In a Health Effects
Institute (HEI) report from Swenberg et al. (2000), male and female B6C3FJ mice and Sprague-
Dawley rats were exposed by inhalation to 1,000 ppm 1,3-butadiene 6 h/day, 5 days/week for 13
weeks. There was not a statistically significant difference in the concentration of hemoglobin
adducts from EB (total included both regioisomers) or of trihydroxybutyl adducts formed from
either DEB or EBD between genders of either species.
Studies of molecular dosimetry in exposed workers are limited. Osterman-Golkar et al.
(1996) studied hemoglobin adducts in 17 workers exposed to 1,3-butadiene in a petrochemical
plant and nine referents employed at the same factory but not exposed to 1,3-butadiene. With
stationary and personal monitoring devices, the ambient 1,3-butadiene concentrations for workers
handling 1,3-butadiene containers was 11.2 ± 18.6 mg/m3 (approximately 5 ppm) versus < 1.2
mg/m3 (approximately 0.5 ppm) for maintenance and laboratory workers. The Hb adduct
measured was 2-hydroxy-3-butylvaline, formed by reaction of N-terminal valine with carbon 1 in
EB. Higher concentrations of Hb adducts (0.16 ± 0.099 pmol/g) were recorded in the workers
handling 1,3-butadiene containers compared with maintenance workers, laboratory workers, and
nine unexposed controls (approximately 0.05 pmol/g).
Perez et al. (1997) detected the trihydroxybutylvaline adducts in occupationally exposed
humans. Concentrations of trihydroxyvaline adducts in two humans occupationally exposed to
approximately 1 ppm 1,3-butadiene were 10 and 14 pmol/g globin compared with two human
control values of 1.8 and 3.3 pmol/g globin. An adduct of EB, N-(2-hydroxy-3-butenyl)valine,
was measured in the same individuals; 0.15 and 0.2 pmol/g for exposed versus 0.05 and 0.02
pmol/g for controls. Consistent with rodents, trihydroxybutylvaline adducts were greater than
EB adducts by 70-fold.
5-46
-------
In an HEI report by Swenberg et al. (2000), N-(2,3,4-trihydroxybutyl)valine adducts were
measured in exposed Chinese workers and compared with concentrations in nonexposed U.S.
laboratory workers. The mean total amount of trihydroxybutylvaline adducts in the U.S.
laboratory workers was 36 ± 23 pmol/g globin for nonsmokers (n=7) and 40 ± 9 for smokers
(n=4) compared with 39 ± 13 in a control set of Chinese workers (n=25). Swenberg et al.
reported that the control values were overestimations because 15/51 controls had concentrations
below the method's limit of detection. The BD-exposed Chinese workers had concentrations of
trihydroxybutylvaline adducts of 88 ± 59 pmol/g globin. The difference between control and
exposed Chinese workers was highly significant (p<0.001).
3.4.2.2. DNA Adducts
Research has also focused on the use of DNA adducts as biomarkers of 1,3-butadiene
exposure. Investigators have attempted to quantitate adducts from EB, EBD, and DEB. More
recent studies have shown that the guanine adducts formed from either EBD or DEB
predominate over the guanine adducts formed from EB, indicating that EBD may be a major
metabolite in vivo. As with the hemoglobin adducts, an adduct specific for DEB has not been
quantitated in vivo in animals exposed by inhalation to 1,3-butadiene.
Kreiling et al. (1986a) reported the radioactivity associated with liver nucleoproteins and
DNA after exposure of male B6C3FJ mice and Wistar rats in a closed inhalation system to 1,3-
[l,4-14C]butadiene until the animals took up 0.21 mmol of 1,3-butadiene (4 h and 6 h, for mice
and rats, respectively). The livers were collected 30 h after exposure for analysis of radioactivity
associated with hepatic nucleoproteins and DNA. The mice had approximately twice as much
radioactivity associated with nucleoproteins as rats, but the amount of radioactivity associated
with DNA was similar between species. The nature of the chemical species associated with the
two fractions was not characterized.
Jelitto et al. (1989) reported species-dependent differences in the in vivo formation of
DNA adducts by male B6C3FJ mice and male Wistar rats exposed in a closed inhalation system
to initial concentrations of 500 ppm l,3-[l,4-14C]butadiene until 98% of total radioactivity was
eliminated from the headspace. Liver DNA was isolated and qualitatively analyzed for N7
guanine adducts of EB and DEB (although the adduct monitored could be formed from EBD
also) by alkaline elution and comparison of HPLC profiles with synthesized adduct standards.
Jelitto and co-workers identified 7-N-(2-hydroxy-3-l-yl)guanine (adduct of EB at carbon 1) and
7-N-(2,3,4-trihydroxybutyl)guanine (adduct of either EBD or DEB) in livers from mice but not
rats. Jelitto also exposed male Sprague-Dawley rats and B6C3FJ mice for 7 h to 250, 500, and
1,000 ppm 1,3-butadiene (although not stated, the exposure is assumed to be a constant
concentration) and immediately after exposure isolated cell nuclei from liver and lung tissue to
3-47
-------
analyze by alkaline elution. The tissue from mice but not rats showed protein-DNA and DNA-
DNA cross-links for all exposures.
Two studies have investigated the enantioselective formation of EB-derived DNA
adducts. Koivisto et al. (1997) exposed male Sprague-Dawley rats to 200 ppm 1,3-butadiene by
inhalation 6 h/day for 5 days and isolated DNA from liver to quantitate the specific enantio- and
regioisomeric adduct formation at the N-7 position of guanine. The adducts were quantitated by
means of a 32P-postlabeling/HPLC assay. The major adduct formed was the carbon 2 isomer
derived from the R enantiomer of EB (47% of all N-7 guanine adducts). The other three adducts
were the carbon 1 isomer from the R enantiomer, the carbon 2 isomer from the S enantiomer, and
the carbon 1 isomer from the S enantiomer; the contributions to the total N-7 guanine adducts
were 22%, 18%, and 14%, respectively.
Koivisto et al. (1998) extended the work to study differences in adduct concentrations in
the lung and testis of male mice exposed by inhalation to 0, 50, 200, or 500 ppm 1,3-butadiene 6
h/day for 5 days. DNA was isolated from the lung and testis and analyzed for EB-derived, N-7
guanine adducts. Both organs had a dose-dependent increase in N-7 adducts, and at the highest
concentration the adduct concentration in the testis was approximately twice that in the lung. In
addition, the isomeric distribution of the adducts differed between tissues. In the lung, the
predominant form of the adduct was the carbon-1 isomer from the S enantiomer of EB (49% of
the total N-7 guanine adducts), whereas only two of the four isomeric forms were detected in the
testis. The carbon-2 isomer from the S enantiomer was the predominant form of the N-7 guanine
adducts (71% of total) in the testis. The overall S/R ratio of the adducts was similar between
tissues, 77/23 for the lung and 71/29 for the testis. Koivisto et al. (1998) hypothesize that the
tissue differences in enantiomeric adduct distribution may be due to differences in DNA repair.
More recent studies have quantitated guanine adducts formed from either DEB or EBD in
rats. Tretyakova et al. (1998) developed a method for detection of DNA adducts with liquid
chromatography electrospray ionization tandem mass spectrometry. Using the developed
technique, Tretyakova et al. reported the concentrations of hydroxybutenyl guanine adducts of
EB and trihydroxybutyl guanine adduct from either EBD or DEB in the liver of B6C3FJ mice and
F344 rats exposed by inhalation to 1,250 ppm 1,3-butadiene 6 h/day for 10 days. The guanine
adducts were approximately twice as high in mice compared with rats. The trihydroxybutyl
guanine adduct was approximately 1.5-fold higher than the hydroxybutenyl guanine adducts from
EB in both species, consistent with results for hemoglobin adducts.
A recent adduct study by Koc et al. (1999) addressed the relationship between exposure
dose and molecular dosimetry of guanine adducts in mice and rats. The results of the research
indicated that EBD may be a major metabolite in the blood and tissues of animals exposed to 1,3-
butadiene and a continued need for a unique marker for DEB binding to tissue DNA. Koc et al.
3-48
-------
exposed female B6C3FJ mice and F344 rats to 0, 20, 62.5, or 625 ppm 1,3-butadiene by
inhalation for 6 h/day, 5 days/week for 4 weeks and quantitated the molecular dosimetry of N-7
guanine adduct formation by EB, DEB, and EBD in the liver, lung, and kidney. The adducts
measured included the racemic and meso forms of N-7-(2,3,4-trihydroxybut-l-yl)-guaninie
(THB-Gua), -7-(2-hydroxy-3-buten-l-yl)-guanine (EB-Gua I), andN-7-(l-hydroxy-3-buten-2-
yl)-guanine (EB-Gua II). THB-Gua can be formed from the various stereoisomers of either DEB
or EBD, and EB-Gua I and EB-Gua II are formed from EB. The adducts were quantitated by
liquid chromatography/positive ion electrospray ionization/mass spectrometry/mass
spectrometry. The number of adducts was similar in the three tissues of both species. EB-Gua
displayed a linear increase with 1,3-butadiene exposure concentration in both species, whereas
THB-Gua plateaued after 62.5 ppm in rats but continued to increase in mice, although not
linearly. This correlates with the results of Thornton-Manning et al. (1998), which reported no
increase in blood concentrations in rats exposed to 8,000 ppm 1,3-butadiene compared with rats
exposed to 62.5 ppm 1,3-butadiene by inhalation. In this study, interspecies differences in the
number of N-7 adducts were lower than interspecies differences in concentrations of metabolites
from previous pharmacokinetic research. In the various tissues, the number of adducts per 106
normal guanine bases was higher in mice than in rats at 625 ppm 1,3-butadiene, but similar at
lower concentrations. THB-Gua adducts were more abundant than EB-Gua adducts (6-27 times
higher); the difference was greatest at higher exposure concentrations of 1,3-butadiene.
Assuming that DNA repair is the same for the various adducts, THB-Gua would be expected to
be lower than EB-Gua if formed solely from DEB. Based on previously published blood
concentrations of EB and DEB at similar exposure concentrations and exposure conditions, Koc
and co-workers hypothesize that the majority of THB-Gua adducts arise from EBD and that mice
have more efficient repair of DNA than rats.
3.5. EXCRETION
1,3-Butadiene may be excreted via the respiratory tract as the parent compound or as a
metabolite, or in the urine and feces as a metabolite. Several studies have quantitated the amount
of radioactivity from inhaled 1,3-butadiene eliminated by rats, mice, and monkeys by various
routes of excretion following inhalation studies (Bond et al., 1986; Dahl et al., 1991).
Unfortunately, none of the inhalation studies have done a mass balance to completely quantitate
the amount of inhaled 1,3-butadiene that was exhaled as parent or metabolite, excreted via urine
or feces, or retained in the body. The studies estimated a total inhaled dose based on measured
respiratory rates and the exposure concentrations used in the experiments. However, the exhaled
concentrations of 1,3-butadiene during or following exposure were not reported in any of the
studies, the retained dose was not measured in monkeys, and the exhaled CO2 concentration
3-49
-------
during exposure was not measured for rodents. Results from the studies were reported as
percentages of total inhaled dose or as a percentage of the total retained dose, which was
measured at euthanasia following exposure for mice and rats or as the total amount recovered in
96 h after exposure by all routes of excretion in monkeys. Because only one terminal carbon was
labeled, the specific activities of the recovered metabolites are not known for CO2 or the three-
carbon metabolites. Therefore, the total amount of metabolite formed cannot be quantitated, and
assessment of the percentage dose eliminated by each pathway is uncertain because of lack of
knowledge of the amount of unlabeled metabolite that may be present in the sample. Because of
the different protocols used for the studies, comparisons among the species with the data that
have been gathered and reported are not adequate to quantitatively assess species differences in
pharmacokinetics of excretion. The information reported here summarizes the pathways that are
responsible for excreting 1,3-butadiene and the identification of the metabolites excreted via each
route.
3.5.1. Pathways of Excretion
The major pathways for excretion of inhaled 1,3-butadiene by rodents are exhaled breath
and urine (Bond et al., 1986), and although the absolute amounts of l,3-[14C]butadiene
equivalents differ, the percentage of retained dose excreted by each pathway is similar between
species. The exhaled air after exposure (note that the amount of 1,3-butadiene that was exhaled
as parent material during exposure was not quantified) accounted for 27-77% of the amount of
14C retained at the end of a 6-h inhalation exposure for rats to 80-7,100 ppm l,3-[l-14C]butadiene
and 23-48% of the amount of 14C retained in mice exposed to 8-1,000 ppm l,3-[l-14C]butadiene.
At the same exposure concentrations of 1,3-butadiene, the total percentage of retained dose that
was exhaled was similar between rats and mice; however, the rats had slightly higher percentages
of retained l,3-[14C]butadiene equivalents exhaled as CO2 and slightly lower percentages exhaled
as 1,3-butadiene and metabolites when compared with mice. Mice and rats excreted similar
percentages (approximately 30-50%) of the retained 14C in urine following exposure to the same
concentrations of 1,3-butadiene.
The rate of 1,3-butadiene excretion by rats and mice was not affected by exposure
concentration (0.08-7,100 ppm 1,3-butadiene). Half-lives for urinary excretion of radioactivity
were similar for both rats and mice (5.6 and 4.6 h, respectively), and although fecal excretion was
somewhat greater in rats (22 h) than in mice (8.6 h), the error was large for rats so the difference
was not significant (Bond et al., 1986). Elimination of radioactivity from the blood and tissues
of rats and mice after inhalation exposure to 1,3-[14C]-BD was biphasic; half-lives for initial
removal were 2-10 h and for slower elimination were 5-60 days.
5-50
-------
Similar to rodents, in monkeys the important routes of excretion appeared to be
exhalation of 1,3-butadiene, metabolites of 1,3-butadiene, and CO2, and urinary excretion of
metabolites (Dahl et al., 1991). Dahl and co-workers quantitated the excretion of 14C by various
pathways from monkeys exposed for 2 h to l,3-[l-14C]butadiene at concentrations ranging from
10 to 8,000 ppm. Dahl and co-workers measured 14C during inhalation and up to 96 h after
exposure in exhaled breath, urine, and feces. Although exhaled breath was collected, the
investigators only report 14C exhaled as CO2 or metabolites but not as parent compound. The 14C
recovered via the various routes of excretion accounted for only 1.40-2.88% of the estimated
total amount of inhaled 1,3-butadiene. The majority of the amount inhaled would be assumed to
be exhaled 1,3-butadiene; however, neither exhaled 1,3-butadiene nor the amount of 14C retained
in the monkeys was quantitated in the experiments, therefore the percentage of 1,3-butadiene
exhaled is not known. Of the excretion routes quantitated, CO2 represented the major route of
excretion at 10.1 ppm 1,3-butadiene; at 310 ppm, urinary excretion represented the major route
of excretion; and at 7,760 ppm, exhalation of other volatile metabolites represented the major
route of excretion of the pathways that were quantitated. The percentage of the inhaled amount
of 1,3-butadiene that was excreted as CO2 or in urine and feces decreased with increasing
exposure concentration of 1,3-butadiene, indicating saturation of metabolism of 1,3-butadiene.
The volatile metabolites excluding 1,3-butadiene and CO2 that were exhaled increased as a
percentage of the inhaled amount of 1,3-butadiene at the highest exposure concentration.
Biliary excretion has been shown to be a minor pathway for elimination of 1,3-butadiene
and its metabolites in vivo. Following ip injection of 14.3 or 143 |lmol/kg of EB, two
glutathione conjugates, ,S'-(2-hydroxy-3-buten-l-yl)glutathione (I) and-S'-(l-hydroxy-3-buten-l-
yl)glutathione (II), were detected in the bile of rats (Sharer and Elfarra, 1992). Total conjugates
excreted in 60 min averaged 7.6% ± 4.2% of the administered dose with approximately a 3:1
ratio of conjugates 1:11. At either dose, the amount of conjugates excreted in 30 min was at least
85% of that excreted in 120 min. When the dose of EB was varied between 14.3 and 286
|lmol/kg and the combined amounts of conjugates I and n excreted in 60 min were determined,
an apparent linear dose-relationship was obtained. Saturation was not observed at these dose
levels. Although the study showed that glutathione conjugates of EB are formed in vivo after
administration of EB, biliary excretion of glutathione conjugates account for only a small portion
of the administered dose.
Urinary excretion was a major route of elimination in Sprague-Dawley rats and B6C3FJ
mice administered [4-14C]EB by intraperitoneal injection (Richardson et al., 1998). Richardson
and co-workers administered 1, 5, 20, and 50 mg EB/kg body weight to rats and mice and at 48 h
after exposure recovered 53-58% and 46-67% of the administered radioactivity in urine, feces
and cagewash from rats and mice, respectively. The unrecovered percentage could have been
3-51
-------
exhaled as parent or metabolite or still retained in the body. Urine contained the greatest amount
of radioactivity with 47-52% and 39-62% of the administered radioactivity in rats and mice,
respectively. There were no significant differences among doses or between species.
3.5.2. Identification of Metabolites
Studies on urinary excretion of 1,3-butadiene metabolites in mice, rats, hamsters,
monkeys, and humans have shown that all these species produce predominantly urinary
metabolites resulting from hydrolysis and glutathione conjugation of the reactive 1,3-butadiene
metabolites. The metabolites have been characterized in urine following inhalation exposures to
1,3-butadiene and ip injections of EB. Clear species differences exist as to which metabolite is
formed predominantly as a result of differences in metabolic pathways responsible for
detoxication of 1,3-butadiene metabolites. These differences correlate well with species
differences in enzyme activity measured in vitro for detoxication of EB. In mice, rats, and
hamsters, mercapturic acid metabolites resulting from glutathione conjugation at carbon 1 or
carbon 2 of EB represent the largest percentage of the urinary metabolites that were formed. In
humans and monkeys the largest percentage of metabolites result from initial hydrolysis of EB.
Several inhalation studies have identified and quantitated metabolites of 1,3-butadiene in
urine of animals exposed to 1,3-butadiene (Nauhaus et al., 1996; Osterman-Golkar et al., 1991;
Sabourin et al., 1992).
One of the first studies to report mercapturic acids derived from EB conjugation with
glutathione was Osterman-Golkar et al. (1991), who exposed male Wistar rats to 0, 250, 500, and
1,000 ppm 1,3-butadiene by inhalation 6 h/day, 5 days/week, for 2 weeks. Urine from exposed
animals was collected in the morning before exposure and in the evening after exposure. The
urinary metabolite S-(2-hydroxy-3-butenyl-l-yl)-N-acetyl-L-cysteine was quantitated in urine
following deacetylation, derivatization, and separation by HPLC. The work is limited in that
only one urinary metabolite was quantitated; however, the work did show linear increase in the
metabolite up to 1,000 ppm 1,3-butadiene and no accumulation of metabolites over time.
Sabourin et al. (1992) analyzed urinary metabolites from male F344/N rats administered
15 (imol l,3-[l-14C]butadiene intraperitoneally four times at 2-h intervals. Of the eight different
peaks that were separated and identified by HPLC, the two predominant peaks were
characterized further by GC-MS, JH nuclear magnetic resonance (NMR) spectroscopy, and gas
chromatography-chemical ionization mass spectroscopy (GC-CEVIS) following derivatization.
One peak was identified as l,2-dihydroxy-4-(N-acetylcysteinyl-S)-butane (MI) and the other as
the -acetylcysteine conjugate of EB (Mil). Sabourin et al. used HPLC to quantitate metabolites
in urine from F344/N rats, Sprague-Dawley rats, B6C3FJ mice, and Syrian hamsters (gender not
specified) exposed by inhalation for 2 h to 7,600 ppm l,3[l-14C]butadiene and male cynomolous
3-52
-------
monkeys exposed for 2 h to 10, 300, and 8,000 ppm l,3[l-14C]butadiene. At approximately
8,000 ppm 1,3-butadiene, the summation of MI and ME in all species constituted the largest
percentage of the urinary metabolites that were quantitated. However, the ratio of MI to MI+MII
varied among species and correlated (^=0.82) with epoxide hydrolase activities determined in
vitro (the substrate used for the activities was benzo[a]pyrene-4,5-(K-region)-oxide).
Concentration-dependent differences in the urinary profile of metabolites were not seen in
monkeys at 300 and 8,000 ppm 1,3-butadiene. The investigators were unable to determine if the
lowest concentration differed from the higher exposure concentrations because all the peaks
except for MI were below the limit of detection.
In another study characterizing urinary metabolites in rodents (Nauhaus et al., 1996),
male B6C3FJ mice and Sprague-Dawley rats were exposed by nose-only inhalation to 800 ppm
l,3-[l,2,3,4-13C]butadiene for 2 h. The urine was collected up to 20 h after exposure and
qualitatively analyzed by NMR spectroscopy. The amount of metabolites was quantitated in the
urine from one exposed rat and mouse. The total amount of urinary metabolites collected for the
rat and mouse was 13 and 7 (Imol, respectively, or 80 and 270 |lmol/kg, respectively, when
normalized to body weight. Nauhaus identified a total of 10 different metabolites, 9 of which
were present in mouse urine and 5 in rat urine. Consistent with Sabourin et al. (1992), Nauhaus
and co-workers found the majority of urinary metabolites in rodents following exposure to 1,3-
butadiene resulted from glutathione conjugation of EB, 70.4% and 60.8% of total metabolites for
the mouse and rat, respectively. The predominant metabolite in the mouse was N-acetyl-S-(2-
hydroxy-3-butenyl)-L-cysteine (43.9% of total metabolites), which results from glutathione
conjugation at carbon 1 of EB, and the rat formed predominantly N-acetyl-S-(l-(hydroxymethyl)-
2-propenyl)-L-cysteine (52.8% of total metabolites), which results from glutathione conjugation
at carbon 2 of EB. The rat also had a higher percentage of urinary metabolites resulting from
hydrolysis of EB compared with the mouse, 26.5% versus 7.1%, which is consistent with the
reported epoxide hydrolase activities in each species (Sabourin et al., 1992). Metabolites that
may have resulted from biotransformation of DEB differed between the mouse and rat. N-
Acetyl-S-(l-(hydroxymethyl)-2,3-dihydroxypropyl)-L-cysteine, which could be formed from
glutathione conjugation at carbon 2 of DEB, represented 4.6% of the total metabolites in the
urine from the mouse but was not identified in urine from the rat. 1,2-Dihydroxypropanone,
which the authors hypothesized could be derived from phosphorylation of DEB or the hydrolysis
product erythritol, represented 5.5% of total metabolites in rat urine but was not detected in
mouse urine. The amount of urinary metabolites that were hypothesized to be derived from DEB
was four times greater in the mouse than in the rat when expressed as |lmol/kg body weight.
However, the hypothesized DEB-derived metabolites also could have been formed following
biotransformation of BDdiol, which was detected in urine of both species, 2.9% and 5% of total
O CO
JOJ
-------
urinary metabolites in the mouse and rat, respectively. A metabolite of 3-butenal, N-acetyl-S-(l-
hydroxy-3-butenyl)-L-cysteine, was detected in the urine of the mouse but not of the rat and
represented 4% of the urinary metabolites. Nauhaus and co-workers also detected three-carbon
metabolites that could not be assigned definitively to any one intermediate; these metabolites
accounted for 10% and 5% of urinary metabolites in the mouse and rat, respectively.
Two studies have investigated the metabolites of EB in urine of rodents exposed ip to EB
(Elfarra et al., 1995; Richardson et al., 1998). N-Acetylcysteine derivatives of the two
glutathione conjugates of epoxybutene identified in the bile of rats by Sharer and Elfarra (1992)
were detected in the urine of rats and mice administered EB ip (Elfarra et al., 1995). When rats
were injected with EB at doses ranging from 71.5 to 285 |lmol/kg, urinary excretion of S-(2-
hydroxy-3-buten-l-yl)-N-acetyl-L-cysteine (I) and S-(l-hydroxy-3-buten-2-yl)-N-acetyl-L-
cysteine (II) within 8 h after exposure increased linearly with dose; the total amount of the two
mercapturic acids combined averaged 17 ± 4% of the administered dose at all exposure
concentrations. However, in mice the percentage of dose excreted as mercapturic acids I and II
was dose dependent. Mice excreted 9 ± 3%, 7 ± 3%, and 26 ± 13% of the administered dose at
the 71.5, 143, and 285 |lmol/kg dose of EB, respectively. Consistent with the research of
Nauhaus et al. (1996), rats preferentially excreted mercapturic acid n over I (approximate ratio
3:1), whereas mice preferentially excreted mercapturic acid I over n (approximate ratio 1.85:1).
Richardson et al. (1998) administered 1, 5, 20, or 50 mg/kg [4-14C]EB to male E6C3Fl
mice and Sprague-Dawley rats and collected urine up to 48 h after exposure. Richardson et al.
identified nine metabolites, including diastereomeric pairs from two regioisomers, in urine of rats
administered EB ip and nine metabolites in urine of mice, two of which were different from
metabolites in urine from rats. Consistent with other studies that have investigated urinary
metabolites resulting from 1,3-butadiene exposure, Richardson and co-workers found that the
majority of urinary metabolites resulting from EB were glutathione conjugates at carbon 1 or 2,
with mice forming slightly more mercapturic acids at carbon 1 than carbon 2 at all doses except
1 mg/kg. Rats formed approximately twice the mercapturic acids from glutathione conjugation at
carbon 2 compared with carbon 1. In both species, the R stereoisomer predominated over the S
stereoisomer for glutathione conjugation at both carbon 1 and carbon 2. Several novel
metabolites were identified by Richardson et al. (1998), including two mercaptoacetic acids,
which were most probably produced from deamination of EB-cysteine conjugates followed with
catabolism by decarboxylases, in urine from mice, and (4-[N-acetyl-L-cystein-S-yl]-2-
hydroxybutanoic acid, which was probably an oxidation product of l,2-dihydroxy-4-[N-acetyl-L-
cysteinyl]butane), in urine from rats. Notable was the absence of urinary metabolites derived
from DEB biotransformation or from 3-butenal in the urine of both species. However, unlike
Nauhaus et al. (1996) who detected the three-carbon metabolites 3-(N-acetyl-L-cystein-S-
3-54
-------
yl)propan-l-ol and 3-(N-acetyl-L-cystein-S-yl)-3-propanoic acid in both species, Richardson et
al. detected these three carbon metabolites only in the urine from rats administered EB ip. The
difference in the profile of urinary metabolites following exposure of rodents to EB from ip
injection versus inhalation to 1,3-butadiene could be due to differences in metabolism because of
the kinetics by the different routes of administration or possibly because some the metabolites
identified from 1,3-butadiene exposure do not proceed through the EB pathway.
Richardson et al. (1998) also found that the stereochemistry of EB appears to affect
biotransformation of the compound. Selective stereoisomers were formed in both species
administered racemic EB ip. In contrast to the results from Nieusma et al. (1997), both rats and
mice had more urinary metabolites originating from the R-stereoisomer than the S-stereoisomer
of EB. However, there were no stereoisomeric differences between the species.
In addition to identification of urinary metabolites in laboratory animals, two in vivo
studies have reported data on the urinary excretion of 1,3-butadiene metabolites by humans. The
studies have not identified all metabolites present in urine from humans exposed to 1,3-
butadiene, but have focused on quantitation of metabolites that make up the largest percentage of
metabolites in urine from laboratory animals exposed by inhalation to 1,3-butadiene.
In the first study, Bechtold et al. (1994) attempted to measure two metabolites of 1,3-
butadiene, l,2-dihydroxy-4-(N-acetylcysteinyl-S-)-butane (MI) and l-hydroxy-2-(N-
acetylcysteinyl-S-)-3-butene (Mil), in urine of workers employed at the Texaco Chemical Co. in
Port Neches, Texas, a 1,3-butadiene extraction plant. MI and Mil were selected because they
were the predominant metabolites identified in urine from laboratory animals exposed by
inhalation to 1,3-butadiene. The study population included (a) exposed employees who worked
in two areas (described as low- and high-exposure areas) with time-weighted average
concentrations of 3 to 4 ppm 1,3-butadiene over the previous 6 months, (b) intermediate-exposed
employees who worked variable time periods in low- and high-exposure areas, (c) nonexposed
employees who worked in areas with historical time-weighted average concentrations of less than
0.1 ppm 1,3-butadiene, and (d) outside controls who had no known exposure to 1,3-butadiene.
Urine samples were analyzed from 7 exposed employees, 3 intermediate-exposed employees, 10
nonexposed employees, and 9 outside controls. MI, but not Mil, was detected in the urine
samples. The average values of MI for exposed employees, intermediate-exposed employees,
nonexposed employees, and outside controls were 3,200 ± 1,600, 1,390 ± 550, 630 ± 190, and
320 ± 70 ng/mL, respectively. Although the actual exposure to 1,3-butadiene for each individual
was not known, urinary concentrations of MI for the exposed groups (combined) were
significantly higher (p<0.05) compared with the outside control group.
In the second study, Ward et al. (1996b) reported increased concentrations of the urinary
metabolite l,2-dihydroxy-4-(N-acetyl-cysteinyl)-butane in workers at a styrene-butadiene rubber
3-55
-------
plant. Exposure was assessed in workers from areas of higher exposure (reactor, recovery, tank
farm, laboratory) and lower exposure (blend, coagulation, bailers, shipping, utilities, shops) with
badge dosimeters. Concentration of the metabolite was measured in urine. The detection limit
(0.25 ppm) was exceeded in 20/40 dosimeter readings in the high-exposure group and in 0/20
readings in the low-exposure group. Urine samples were analyzed from 16 high- and 9 low-
exposure subjects. Expressed as ng/mg creatinine, metabolite concentrations were 1,690 ± 201.3
and 355 ± 250 in Study I (conducted in 10/91), and 761 ± 245 and 684 ± 176 in Study H
(conducted in 6/92), respectively, for the high- and low-exposure groups.
3.6. PHARMACOKINETICS OF STRUCTURALLY RELATED COMPOUNDS
Studies with isoprene (2-methyl-butadiene) and limited studies with chloroprene (2-
chloro-butadiene) have shown some similarities with the pharmacokinetics of 1,3-butadiene.
Similar to 1,3-butadiene, the initial step in biotransformation of isoprene and chloroprene
is oxidation to monoepoxides (Bartsch et al., 1979; Bogaards et al., 1996; Del Monte et al., 1985;
Gervasi and Longo, 1990; Wistuba et al., 1994). Isoprene can also form diepoxides (Bogaards et
al., 1996; Bond et al., 1991; Dahl et al., 1987; Del Monte et al., 1985; Gervasi and Longo, 1990;
Wistuba et al., 1994). The biotransformation of chloroprene to a diepoxide has not been studied.
Epoxide hydrolase is involved in the biotransformation of isoprene (Bogaards et al., 1996; Bond
et al., 1991; Dahl et al., 1987; Del Monte et al., 1985; Gervasi and Longo, 1990; Wistuba et al.,
1994); however, glutathione transferase does not appear to be a major pathway for
biotransformation of isoprene (Buckley et al., 1999). The role of glutathione transferase in the
metabolism of chloroprene has been implicated because of depletion of nonprotein sulfhydryl in
tissues from mice and rats exposed by inhalation to chloroprene (Melnick et al., 1996a; Plugge
and Jaeger, 1979).
Del Monte et al. (1985) showed that mouse hepatic microsomal monooxygenases
converted isoprene to two monoepoxides and a diepoxide. Specifically, 3,4-epoxy-3-methyl-1-
butene and 3,4-epoxy-2-methyl-l-butene were major and minor metabolites, respectively, with
the latter representing about 20% of the former. Of the two monoepoxides, only 3,4-epoxy-2-
methyl-1-butene metabolite was metabolized further in microsomal incubations to a diepoxide.
Addition of l,2-epoxy-3,3,3-trichloropropene to microsomal incubations with isoprene increased
the concentrations of 3,4-epoxy-3-methyl-1-butene in incubation media, indicating the role of
epoxide hydrolase in the biotransformation of isoprene. Because of the rapid hydrolysis of the
two monoepoxides in water, Del Monte et al. (1985) measured the diols of the epoxides to
quantitate enzymatic activity. Biotransformation of isoprene to the diol of 3,4-epoxy-3-methyl-1-
butene was O2 and NADPH dependent and inhibited by cytochrome P450 inhibitors such as CO,
SKF 525-A, and metyrapone. For production of the diol of 3,4-epoxy-3-methyl-1-butene from
3-56
-------
isoprene, the Km (mM) and Vmax (nmol diol/mg protein/min) values were 0.09 and 1.8, 0.08 and
2.9, and 0.07 and 2.1, respectively, in hepatic microsomes from control, phenobarbital-induced,
and 3-methylcholanthrene-induced mice. In addition, for the production of the diepoxide from
3,4-epoxy-2-methyl-l-butene, the Km (mM) and Vmax (nmol diepoxide/mg protein/min) values
were 0.24 and 1.7, 0.29 and 5.1, and 0.22 and 2.0, respectively, in microsomes from control,
phenobarbital-induced, and 3-methylcholanthrene-induced mice. The increase in Vmax for the
formation of the diepoxide was significant (p<0.01) in incubations with hepatic microsomes from
phenobarbital-treated mice.
Gervasi and Longo (1990) provided species comparisons for the in vitro metabolism of
isoprene by hepatic microsomal preparations from rats, mice, rabbits, and hamsters. Hepatic
microsomal preparations from all these species metabolized isoprene to 3,4-epoxy-3-methyl-1-
butene and 3,4-epoxy-2-methyl-l-butene. The former was the major metabolite in all species.
For the production of the diol of 3,4-epoxy-3-methyl-1-butene from isoprene, the Km (mM) and
Vmax (nmol diol/mg protein/h) were 0.08 and 0.24, 0.09 and 1.79, 0.2 and 0.66, and 0.06 and
1.20, respectively, in incubations with hepatic microsomes from rats, mice, rabbits, and hamsters.
The Vmax was almost an order of magnitude larger in the mouse compared with the rat.
Microsomal preparations from all species further metabolized the 3,4-epoxy-2-methyl-1-butene
isoprene dioxide (2-methyl-l,2,3,4-diepoxybutane).
In contrast to Del Monte et al. (1985) and Gervasi and Longo (1990), Wistuba et al.
(1994) and Bogaards et al. (1996) found production of diepoxides of isoprene from both 3,4-
epoxy-3-methyl-1-butene and 3,4-epoxy-2-methyl-l-butene.
Wistuba et al. (1994) studied the stereoselective metabolism of isoprene in liver
microsomes from rats and mice. In addition to the two monoepoxides, the investigators also
detected diepoxide formation from both monoepoxides. There was regioselectivity with the liver
microsomes from mice, but not from rats. Wistuba et al. (1994) also found stereoselectivity in
the hydrolysis by epoxide hydrolase and glutathione conjugation by glutathione transferase.
Bogaards et al. (1996) compared the metabolism of isoprene among liver microsomes
from mice, rats, and humans. In all species, 3,4-epoxy-3-methyl-1-butene was the predominate
monoepoxide of isoprene, and the enzymatic activities for the formation of both monoepoxides
did not differ among species in microsomal incubations with the epoxide hydrolase inhibitor
cyclohexene oxide. However, significantly less 3,4-epoxy-3-methyl-1-butene and 3,4-epoxy-2-
methyl- 1-butene were detected in rat and human microsomes, but not mouse microsomes,
incubated without cyclohexene oxide. For microsomes from rats, the concentrations of 3,4-
epoxy-3-methyl-1-butene and 3,4-epoxy-2-methyl-1-butene were reduced to 50% and 70% of
concentrations detected in incubations with cyclohexene oxide. For microsomes from humans,
concentrations of 3,4-epoxy-3-methyl-1-butene and 3,4-epoxy-2-methyl-1-butene were reduced
3-57
-------
to 4% and 25% of concentrations detected in incubations with cyclohexene oxide. The
enzymatic activities for the formation of the diepoxide from 3,4-epoxy-3-methyl-1-butene in
hepatic microsomes ranged from 780 to 1,210 pmol diepoxide/min/nmol P450, with microsomes
from humans exhibiting the lowest activity and microsomes from CD-I mice exhibiting the
highest activity. However, activity in microsomes from B6C3FJ mice was approximately the
same as humans, 806 vs. 780 pmol diepoxide/min/nmol P450. The enzymatic activities for the
formation of the diepoxide from 3,4-epoxy-2-methyl-1-butene in hepatic microsomes ranged
from 666 to 1,360 pmol diepoxide/min/nmol P450, with microsomes from humans exhibiting the
lowest activity and microsomes from Wistar rats exhibiting the highest activity.
Bogaards et al. (1996) also determined the human cytochromes P450 responsible for
oxidation of isoprene to the two monoepoxides and diepoxides. Of the cytochromes P450
examined, the highest activity for the formation of 3,4-epoxy-3-methyl-1-butene from isoprene
was measured in microsomes prepared from cell lines expressing cytochrome P450 2E1 followed
by cytochromes P450 2B6 > P450 2D6 > P450 2A6. 3,4-Epoxy-2-methyl-1-butene and the
diepoxides were detected only in incubations with microsomes from cell lines expressing
cytochrome P4502E1.
Bogaards et al. (1999) studied the enzymatic conjugation of the two monoepoxides of
isoprene, 3,4-epoxy-3-methyl-1-butene and 3,4-epoxy-2-methyl-1-butene, with glutathione. The
studies used purified glutathione-S-transferases of the alpha, mu, pi, and theta class from rats and
humans. 3,4-Epoxy-2-methyl-1-butene was conjugated more efficiently than 3,4-epoxy-3-
methyl- 1-butene. For both rat and human, the mu and theta class were more efficient than the
alpha and pi class at conjugating the monoepoxides. The rat glutathione transferases were more
efficient than the human transferases at metabolizing both monoepoxides; the theta class in rats
had 2.1- to 6.5-fold higher activities than in humans, and the mu class had 5.2- to 14-fold higher
activities than in humans. There were also differences in the site of conjugation between the two
monoepoxides. 3,4-Epoxy-3-methyl-1-butene was almost exclusively conjugated with glutatione
at carbon 3, while no selectivity was noted with 3,4-epoxy-2-methyl-1-butene. In S9 fractions
from livers of mice, rats, and humans, mice and rats had similar rates of conjugation, while the
rates in humans were 25- to 50-fold lower.
Similar to 1,3-butadiene, the pharmacokinetics of isoprene in vivo appear to differ among
species.
In a study by Dahl et al. (1987), groups of 30 male F344 rats were exposed by nose-only
inhalation to [14C]isoprene at concentrations of 8.0, 266, 1,480, or 8,200 ppm for 6 h (5.5 h for
the 8,200 ppm), and urine, feces, and exhalants were collected over a 66 h postexposure period.
The retained 14C (total amount remaining in animal at the end of exposure including isoprene)
decreased as a percentage of inhaled 14C with increasing exposure concentrations, suggesting
3-58
-------
saturation of metabolism. The investigators defined metabolite (nonisoprene)-associated
radioactivity as the total amount of 14C measured in exhaled CO2, urine, feces, and carcass
during the 66 h postexposure period. The authors did not account for the metabolites including
CO2 exhaled during exposure or lost during sacrifice. The major pathway for excretion of
isoprene in rats appeared to be the urine for all exposure concentrations. The authors noted,
however, that this finding was tentative and may be the result of a labeling artifact because the
terminal carbons of isoprene were not labeled. During the postexposure period, >75% of the
metabolite-associated radioactivity was excreted in the urine. The radioactivity in the feces and
exhaled CO2 accounted for 1.7%-13.4% and 1.9%-5.3% of the metabolite-associated
radioactivity, respectively. The half-life (mean ± SE) for urinary excretion of 14C was 10.2 ± 1.0
h (range of 8.8 to 11.1 h). The investigators used cryogenic trapping to tentatively identify
metabolites in the blood and tissues. Generally, the concentration of metabolites in the blood
increased with exposure concentration and duration of exposure. The authors noted that 85% of
the radioactivity in the blood was associated with material of low volatility, which probably
represented covalently bound metabolites, conjugates of isoprene metabolites, or tetrols. The
radioactivity trapped at -45°C, which was attributed to the diols and diepoxides of isoprene,
accounted for the majority of the volatile metabolites at all exposure concentrations. The
percentage of inhaled, isoprene-derived 14C present as diepoxides or diols in the blood remained
fairly constant with time but decreased with exposure concentration. Only at the two highest
exposure concentrations were materials detected that possessed volatilities matching those of
isoprene and isoprene monoepoxides. Assessing the distribution of isoprene and its metabolites
in some animals of the 1,480 ppm exposure group revealed that fat, followed by liver and blood,
contained the largest amounts of the radioactivity (concentrations were not reported).
Bond et al. (1991) conducted similar experiments as Dahl et al. (1987), but with male
B6C3FJ mice at exposure concentrations of 20, 200, and 2,000 ppm isoprene. Isoprene
significantly reduced minute ventilation in mice exposed to 2,000 ppm isoprene compared with
controls exposed to air only. This effect was not seen in rats (Dahl et al., 1987). Except for the
lowest exposure concentrations, mice retained similar percentages of inhaled 14C compared with
rats. However, mice had lower percentages of metabolite-associated radioactivity excreted in the
urine (52.4%-73.4%) but higher percentages excreted as CO2 (2.1%-18.4%) and in feces
(5.5%-35.5%) compared with rats exposed to similar concentrations of isoprene.
Peter et al. (1987) also noted differences in the pharmacokinetics of isoprene in male
Wistar rats and male B6C3FJ mice. Animals were exposed in closed inhalation systems to
concentrations of isoprene as high as 4,000 ppm for up to 10 h. At concentrations <300 ppm, the
rate of metabolism was found to be directly proportional to the isoprene concentration, but
saturation of metabolism was detected at higher concentrations. Pharmacokinetic parameters
3-59
-------
were estimated from a two-compartment model as described for similar experiments with 1,3-
butadiene (Bolt et al., 1984; Kreiling et al., 1986b). The Vmax for the metabolism of isoprene in
rats and mice was 130 and 400 |imol/h/kg, respectively. The estimated half-lives were 4.4 and
6.6 min for mice and rats, respectively. Estimated metabolic clearance was approximately twice
as high in mice compared with rats, 12,000 vs. 6,200 mL/h, respectively. The estimated
clearance of uptake from the chamber was more than twice as high in mice compared with rats,
16,000 vs. 7,300 mL/h, respectively, and approximately equal to the estimated metabolic
clearance, indicating that metabolism of isoprene was limited by uptake. The thermodynamic
partition coefficient was approximately equivalent in mice and rats, 7.0 vs. 7.8, respectively, and
the steady-state, whole body:air concentration ratio was also approximately the same in mice and
rats, 1.7 vs. 1.2, respectively. The endogenous rate of isoprene production was approximately 4
times greater in rats than mice, 1.9 vs. 0.4 |lmol/h/kg, respectively.
Buckley et al. (1999) compared the routes of excretion and the urinary metabolites in
male B6C3FJ mice and Fischer 344 rats after ip administration of 64 mg [14C]isoprene/kg body
weight. The disposition and time course of 14C did not differ among species; there were no
statistically significant differences in the percentage of administered 14C excreted among the
various routes of excretion, which included volatiles in breath, CO2 in breath, urine, feces, and
carcass and tissues. Exhaled volatiles accounted for the largest percentage (approximately 50%)
of administered radioactivity in mice and rats, followed by urine, which accounted for
approximately 30% of the administered radioactivity in mice and rats. CO2 in breath and the
carcass and tissues accounted for approximately 2% and 2-3% of administered radioactivity,
respectively. Feces accounted for 0.2-7% of the administered radioactivity; however, feces from
mice were thought to be contaminated by urine. The profile of urinary metabolites differed
among species. In the rat, four major metabolite peaks were identified in the urine. The
biotransformation of isoprene in mice appeared to be more complex because the urine from
exposed mice contained several minor metabolite peaks in addition to four metabolite peaks that
were detected in the urine from rats. The major metabolite peaks in urine from rats were
identified as 2-hydroxy-2-methyl-3-butenoic acid (53% of the total urinary radioactivity), 2-
methyl-3-buten-l,2-diol (23%), and the glucuronide conjugate of 2-methyl-3-buten-l,2-diol
(23%). The other major metabolite peak in rat urine was not uniquely identified. In mouse urine,
2-hydroxy-2-methyl-3-butenoic acid, 2-methyl-3-buten-l,2-diol, and 2-methyl-3-buten-l,2-diol
accounted for 15%, 3.5%, and 2.5% of the total urinary radioactivity, respectively. The
unidentified metabolite peak in mouse urine, which was thought to be directly correlated with the
peak in rat urine, accounted for 25% of the total urinary radioactivity. None of the metabolites in
either species were derivatives of glutathione conjugates; therefore, in contrast to 1,3-butadiene,
5-60
-------
glutathione conjugation does not appear to be a major pathway in the biotransformation of
isoprene.
Chloroprene (2-chloro-butadiene) is also structurally similar to 1,3-butadiene, but its
pharmacokinetics have not been studied extensively. Studies have shown that the
biotransformation of chloroprene results in the formation of peroxides that may interact with
tissue thiols (Haley, 1978). Furthermore, cytochrome P450 may form an epoxide intermediate
similar to that formed during 1,3-butadiene metabolism (Bartsch et al., 1979). In contrast to
isoprene, glutathione transferase may play a role in the biotransformation of chloroprene. Plugge
and Jaeger (1979) found significant depletion of nonprotein sulfhydryl at 24 h postexposure in
the lungs of male Sprague-Dawley rats that were fasted and exposed by inhalation to 100 or 300
ppm chloroprene for 4 h compared with rats that were fasted and exposed to air only. Melnick et
al. (1996a) found significant depletion of nonprotein sulfhydryl in the livers from male and
female B6C3FJ mice exposed to 80 ppm chloroprene by inhalation for 12 weeks, and in livers
from male and female Sprague-Dawley rats exposed by inhalation to 200 ppm for 12 weeks.
Female but not male rats had significant depletion of nonprotein sulfhydryl following 1 day of
exposure to 200 ppm chloroprene.
In summary, studies have shown that the structurally similar isoprene and chloroprene are
metabolized to monoepoxide intermediates, and the isoprene monoepoxides are further
metabolized to diepoxides. The monoepoxides and diepoxides from isoprene were detected with
in vitro incubations with microsomes from all species investigated, including humans. Although
the enzymatic activities for the formation of the monoepoxides and diepoxides did not differ
among species, enzymatic hydrolysis of the monoepoxides is significantly greater in humans
compared with rats and mice. In vivo inhalation studies with rats and mice exposed to isoprene
showed some differences in the disposition of isoprene compared with 1,3-butadiene. As with
1,3-butadiene, the in vivo rate of isoprene metabolism was greater in mice compared with rats.
However, the contributions of urinary excretion and CO2 exhalation toward the elimination of
1,3-butadiene and isoprene differ between the two chemicals. In addition, glutathione transferase
plays an important role in the biotransformation of 1,3-butadiene. Although metabolism of the
isoprene monoepoxides by glutathione transferase has been measured in vitro, the lack of urinary
metabolites of glutathione conjugates indicate it is not involved in the biotransformation of
isoprene in vivo. Glutathione transferase may be involved in the biotransformation of
chloroprene.
3.7. DISCUSSION
Current data indicate that toxicity of 1,3-butadiene depends on the metabolic activation to
reactive intermediates. Because differences among species have been noted in toxic responses to
3-61
-------
1,3-butadiene exposure, the pharmacokinetics are important in assessing the risk of adverse
health effects from exposure to this compound. This chapter summarized research related to
absorption, distribution, biotransformation, and excretion of 1,3-butadiene with emphasis on
identifying species differences in the pharmacokinetics of 1,3-butadiene. The research supports
the idea that species differences in the pharmacokinetics may contribute but not be solely
responsible for differences in toxicity between mice and rats. However, data gaps still exist in
the knowledge about 1,3-butadiene metabolism that limit the ability to accurately assess the risk
to humans from exposure to 1,3-butadiene based on experimental animal data.
Qualitatively, metabolism of 1,3-butadiene is similar among species that have been
studied, including humans. Oxidation of 1,3-butadiene and intermediate metabolites of 1,3-
butadiene and detoxication of intermediate metabolites epoxide hydrolase and glutathione
transferase are the major metabolic pathways in biotransformation of 1,3-butadiene. Several
reactive intermediates have been identified in either in vivo or in vitro studies as shown in Figure
3-1, including 3-butenal, crotonaldehyde, EB, DEB, and EBD. However, the major metabolites
that have been detected directly in vivo are EB and DEB. EBD has been detected indirectly in
vivo by DNA and hemoglobin adducts and also may be a major metabolite.
However, although the pathways of biotransformation are qualitatively similar among
species, there are quantitative differences in predominant pathways for detoxication and in the
rates of metabolism by the various pathways. Although hepatic microsomes from mice, rats, and
humans all oxidize 1,3-butadiene to EB and EB to DEB, based on the ratio of Vmax/Km in liver
(estimated intrinsic clearance), which would be relevant at lower exposure concentrations of 1,3-
butadiene, the rate of oxidation is greatest in mice with humans and rats approximately
equivalent. For the detoxication pathways, mice and rats predominantly remove EB and DEB
through conjugation by glutathione transferase, whereas epoxide hydrolase is more important in
humans. Using the estimated intrinsic clearance (Vmax/Km) (Table 3-9) as a means of comparison
among species, we find that 5.4-11% of the intrinsic clearance of EB is through oxidation to
DEB in mice, compared with 0.13-2.4% and 2-3% for rats and humans, respectively.
The differences in the rates of metabolism noted in vitro have also been noted in vivo in
rats and mice, and have been reflected in blood and tissue concentrations of reactive metabolites,
in the profile of urinary metabolites, and in biomarkers such as hemoglobin adducts and DNA
adducts. Inhalation studies have demonstrated that 1,3-butadiene is readily absorbed into the
blood, distributed as parent compound and metabolites throughout the body, and eliminated in
exhaled breath as parent and metabolites or in urine as metabolites. The inhalation uptake of 1,3-
butadiene by mice is approximately twice the uptake of rats, partly because of higher ventilation
per kg of body weight, higher tissue perfusion, and higher rates of metabolism. EB and DEB are
the only two metabolites of 1,3-butadiene that have been measured directly in vivo in rats and
3-62
-------
mice, with the blood and tissue concentrations in mice significantly greater than those reported in
rats. Reported blood concentrations of EB are 4-8 times higher in mouse blood than rat blood,
whereas reported tissue concentrations are 3-16 times greater in mice compared with rats. For
DEB, the species differences are greater. The reported blood concentrations of DEB in mice are
approximately 40-fold higher than in rats, whereas tissue concentrations of DEB are 4-163-fold
higher in mice compared with rats. These differences are greater at higher exposure
concentrations because blood concentrations of DEB were similar in female rats exposed to
8,000 ppm 1,3-butadiene compared with female rats exposed to 62.5 ppm 1,3-butadiene for 6 h.
This plateau in DEB blood concentrations resulting from saturation of the oxidation of EB to
DEB may explain in part the difference of carcinogenic response between mice and rats.
Profiles of urinary metabolites that have been detected among species following
inhalation exposures to 1,3-butadiene show clear species differences. The profiles reflect the
predominant clearance of EB in mice through initial conjugation with glutathione versus the
predominant clearance of EB in humans through initial hydrolysis. The greater body burden of
1,3-butadiene and its metabolites in mice versus rats is also reflected in the recovery of total
urinary metabolites normalized to the body weight of the animals, which was approximately
three times greater in mice compared with rats exposed by inhalation to equal concentrations of
1,3-butadiene.
The effect of repeated exposures on 1,3-butadiene pharmacokinetics has been studied to a
limited extent. One in vitro study showed reduced activities from microsomes prepared from the
liver and lungs of mice and rats exposed repeatedly to 1,3-butadiene. In vivo, EB and DEB
concentrations have been quantitated in mice and rats exposed repeatedly to 1,3-butadiene and
compared with animals with a single exposure to 1,3-butadiene. Although the disposition of EB
and DEB appeared to change in both species from single and repeated exposure, the design of the
studies could not address whether the changes were from metabolic alterations or from continued
accumulation as the metabolites approached steady-state during repeated exposures.
Molecular dosimetry studies also have shown species-related differences in the formation
of various adducts. Hemoglobin adducts and DNA adducts formed from EB are approximately
two- to sixfold higher in mice than in rats at 1,3-butadiene exposure concentrations exceeding
approximately 100 ppm, but similar between mice and rats at lower exposure concentrations, and
therefore do not reflect species differences in susceptibility to toxic response from exposure to
1,3-butadiene. Recently, trihydroxybutyl adducts of hemoglobin and DNA have been identified.
These adducts are formed from either DEB or EBD. In rats, the trihydroxybutylvaline adducts of
hemoglobin exceeds the concentration of EB-derived hemoglobin adducts by 3- to 32-fold,
although blood concentrations of EB are higher than DEB in rats. In addition, the
trihydroxybutylguanine adduct exceed the EB-derived guanine adducts in both species. Based on
3-63
-------
blood concentrations of DEB and EB, the majority of the trihydroxybutyl adducts are thought to
be derived from EBD, indicating that EBD may be a major metabolite of 1,3-butadiene in vivo.
Currently, no data exist on species differences in concentrations of EBD in vivo, representing a
major data gap in the research on 1,3-butadiene pharmacokinetics.
Other data gaps in the research on species differences in pharmacokinetics of 1,3-
butadiene include lack of in vivo measurements of other potentially reactive metabolites such as
3-butenal and crotonaldehyde, although these metabolites would be expected to be minor in vivo
based on in vitro data and urinary data. In addition, the stereochemistry of 1,3-butadiene
pharmacokinetics may be important. Limited research has recently shown some species
differences in stereometabolism as well as species and tissue differences in adduct formation.
Data on interindividual variability in humans related to the pharmacokinetics of 1,3-butadiene is
limited. The greatest variability reported was in enzyme activities for oxidation of EB to DEB
measured in hepatic microsomes, which varied approximately 60-fold. However, variability for
Vmax/Km reported for conjugation of DEB by glutathione transferase and hydrolysis by epoxide
hydrolase varied by threefold. Although not specifically studied for 1,3-butadiene or its
metabolites, polymorphisms in cytochromes P450, epoxide hydrolase, or glutathione transferases
could be expected to affect the pharmacokinetics of 1,3-butadiene in vivo. For example,
Bogaards et al. (1999) reported differing activities for the alpha, mu, pi, and theta classes of
glutathione transferases for conjugation of isoprene monoepoxides, structurally similar
compounds to EB. The mu and theta classes were more efficient at conjugating the epoxides,
and since these classes are expressed in 50% and 40-90% of the population, respectively,
individuals lacking these enzymes may have altered detoxification of the epoxides. In addition,
limited data exist to address gender and age differences in the pharmacokinetics of 1,3-butadiene.
Lack of knowledge in these areas limits the ability to make precise assessments of the risk to
humans from exposure to 1,3-butadiene.
5-64
-------
4. MUTAGENICITY
The mutagenic effects of 1,3-butadiene have been reviewed extensively (Rosenthal, 1985;
de Meester, 1988; Arce et al., 1990; Norppa and Sorsa, 1993; Jacobson-Kram and Rosenthal,
1995; Pacchierotti et al., 1998a; IARC, 1999). There is extensive evidence that 1,3-butadiene
induces genotoxic effects in a variety of in vitro and in vivo test systems through its metabolism
to several DNA-reactive intermediates (primarily the epoxide metabolites EB, DEB, and EBD).
Most of the in vivo studies discussed in the cited reviews were assays in mice and rats using
cytogenetic endpoints, and the results generally support the dichotomy in carcinogenic response
in which mice are more responsive than rats. Table 4-1 summarizes much of the earlier data on
1,3-butadiene and its metabolites. This review will focus on recently published studies
performed in vivo (both somatic and germ cell effects), with an emphasis on those studies
providing information relative to the mode of action of 1,3-butadiene metabolites. Tables 4-2
through 4-5 present a tabular display of the studies discussed in the following sections.
4.1. 1,3-BUTADIENE
4.1.1. Gene Mutations
Most of the earlier in vivo genotoxicity studies used cytogenetic endpoints (aberrations,
micronuclei [MN], or sister chromatid exchange [SCE]). This reflected the dearth of available in
vivo assays for measuring gene mutations. The ability to detect mutations at the hprt locus
obtained from T lymphocytes from exposed mammals including mice, rats, monkeys, and
humans provides an important step in developing an understanding of chemically induced
mutational processes. Cochrane and Skopek (1994a) used B6C3FJ mice to evaluate the
mutagenic potential of 1,3-butadiene. Mice were exposed for 6 h/day, 5 days/week for 2 weeks
to 1,3-butadiene at 625 ppm. The induced hprt mutant frequency was 6.2 x 10"6 compared with
1.2 x 10"6 from unexposed controls. They reported that about half of the mutations were
frameshift, and of the base-pair mutations induced by 1,3-butadiene mutations at AT base pairs
predominated.
Tates et al. (1994, 1998) reported mixed results in two strains of mice exposed to 1,3-
butadiene. In the first experiments male (102/E1 x C3H/E1)F1 mice were exposed to 0, 200,
500, or 1,300 ppm of 1,3-butadiene 6 h/day for 5 days. 1,3-Butadiene treatment resulted in a
linear dose-related increase in hprt mutant frequency; only the high dose increase was
significantly different from controls (Tates et al., 1994). In a second experiment using the same
strain, mice were exposed at 500 and 1,300 ppm (Tates et al., 1998). Because of technical
problems there were no data from the 1,300 ppm exposed group and the authors reported a ten-
fold increase in hprt mutant frequency in the 500 ppm group. However, when four control and
4-1
-------
Table 4-1. Summary of earlier mutagenicity testing of 1,3-butadiene, 1,3-butadiene monoepoxide and
l,2:3,4-diepoxybutane (derived from tables in IARC, 1999).
Test system
Results
Dose
(LEC or HIC)a
Comments
1,3-butadiene
S. typhimurium TA100, reverse mutation
S. typhimurium TA1535, reverse mutation
S. typhimurium TA1537, reverse mutation
S. typhimurium TA98, reverse mutation
S. typhimurium TA97, reverse mutation
E. coli WP2 , reverse mutation
Drosophila melanogaster, sex-linked recessive lethal
Mouse lymphoma L5178Y, gene mutation
Chinese hamster ovary cells in vitro, SCE
Human lymphocytes, in vitro, SCE
Human lymphocytes, in vitro, SCE
B6C3F! mouse bone marrow in vivo, SCE
Sprague-Dawley rat bone marrow in vivo, SCE
B6C3F! mouse bone marrow in vivo micronucleus
B6C3F! mouse peripheral blood in vivo, micronucleus
Sprague-Dawley rat bone marrow in vivo, micronucleus
positive
positive
negative
negative
negative
negative
negative
negative
positive
negative
positive
positive
negative
positive
positive
negative
1,080 ppm
86 ppm
1,080 ppm
1,800 ppm
1,800 ppm
1,080 ppm
500 ppm
650 ppm
1.35
2, 160 ppm
108
116 ppm
4,000 ppm
116 ppm
7 ppm
70 ppm
500 ppm
requires metabolic activation
requires metabolic activation
inhalation exposure
requires activation
6 h inhalation
6 h inhalation
6 h inhalation
13 wk inhalation 6 h/day, 5d/wk
2 wk inhalation 6 h/day, 5d/wk
1 wk inhalation 6 h/day, 5d/wk
to
-------
Table 4-1. Summary of earlier mutagenicity testing of 1,3-butadiene, 1,3-butadiene monoepoxide and
l,2:3,4-diepoxybutane (derived from tables in IARC, 1999) (continued).
Test system
Results
Dose
(LEC or HIC)a
Comments
1 ,2-epoxy-3-butene
S. typhimurium TA100, reverse mutation
S. typhimurium TA1535 reverse mutation
S. typhimurium TA1537, TA 1538, TA 98, reverse mutation
E.coliWP2, reverse mutation
human TK6 cells in vitro, gene mutations tk locus
human TK6 cells in vitro, gene mutations hprt locus
Chinese hamster cells in vitro, SCE
positive
positive
negative
positive
positive
positive
positive
26
1,750
8,750
not reported
17.5
10.5
0.07
without metabolic activation
without metabolic activation
not tested with metabolic activation
without metabolic activation
without metabolic activation
with and without metabolic
activation
l,2:3,4-diepoxybutane
S. typhimurium TA100, TA 1535, reverse mutation
S. typhimurium TA1537, TA 1538, TA 98, reverse mutation
E. coli WP2, reverse mutation
Saccharomyces cerevisiae D4, gene conversion
Saccharomyces cerevisiae D3, mitotic recombination
Saccharomyces cerevisiae D7, reverse mutation
Neurospora crassa, reverse mutation
Drosophila melanogaster, somatic mutation
Drosophila melanogaster, sex-linked recessive lethal
positive
negative
positive
positive
positive
positive
positive
positive
positive
20,5
167
167
430
400
130
4,300
430
100
without metabolic activation
with or without metabolic activation
without metabolic activation
without metabolic activation
without metabolic activation
without metabolic activation
in feed
injection
-------
Table 4-1. Summary of earlier mutagenicity testing of 1,3-butadiene, 1,3-butadiene monoepoxide and
l,2:3,4-diepoxybutane (derived from tables in IARC, 1999) (continued).
Test system
Results
Dose
(LEC or HIC)a
Comments
l,2:3,4-diepoxybutane (continued)
Drosophila melanogaster, sex-linked recessive lethal
Chinese hamster ovary cells in vitro, gene mutations, hprt locus
Mouse lymphoma L5 178Y cells in vitro, gene mutations, tk locus
Chinese hamster ovary cells in vitro, SCE
Swiss- Webster mouse bone marrow in vivo, SCE
NMRI mouse bone marrow in vivo, SCE
Chinese hamster bone marrow in vivo, SCE
positive
positive
positive
positive
positive
positive
positive
175
2.15
0.3
0.025
1
22
34
in feed
without metabolic activation
without metabolic activation
without metabolic activation
single ip injection
2 h inhalation
2 h inhalation
aLEC, lowest effective concentration; HIC, highest ineffective concentration; in vitro tests, |J,g/ml; in vivo tests, mg/kg bw.
-------
Table 4-2. Summary table of genetic effects of 1,3-butadiene (details in text).
Test system
Results
Dose
(LEC or HIC)a
Reference
Gene mutations
B6C3F! mouse T lymphocytes, hprt locus, in vivo
B6C3F! mouse T lymphocytes, hprt locus, in vivo
(102 x C3H)F1 mouse T lymphocytes, hprt locus, in vivo
GDI mouse T lymphocytes , hprt locus, in vivo
B6C3F! mouse T lymphocytes, hprt locus, in vivo
B6C3F! mouse T lymphocytes, hprt locus, in vivo
B6C3F! mouse, bone marrow, lad locus, in vivo
T-stock female mice, spot test, in vivo
F344 rat T lymphocytes, hprt locus, in vivo
F344 rat T lymphocytes, hprt locus, in vivo
Human T lymphocytes, hprt locus, in vivo (autoradiographic assay)
Human T lymphocytes, hprt locus, in vivo (autoradiographic assay)
Human T lymphocytes, hprt locus, in vivo (cloning assay)
Human T lymphocytes, hprt locus, in vivo (cloning assay)
positive
positive
positive
negative
positive
positive
positive
positive
positive
positive
positive
positive
negative
negative
490
1,000
390
1,000
970
50
490
390
580
290
1.2 (avg)
0.1 (avg)
1.2 (avg)
0.3 (avg)
Cochrane & Skopek, 1994a
Tales etal., 1994
Tatesetal., 1998
Tatesetal., 1998
Mengetal., 1998
Mengetal., 1999a
Recio & Goldsworthy, 1995
Adleretal., 1994
Mengetal., 1998
Mengetal., 1999a
Wardetal., 1994
Wardetal., 1996b
Hayes etal, 1996
Tatesetal., 1996
Cytogenetic effects-Somatic cells
(102 x C3H)F1 mouse bone marrow & peripheral blood MN, in vivo
(102 x C3H)F1 mouse spleen & peripheral blood MN, in vivo
Human peripheral blood, chromosome aberrations, in vivo
positive
positive
negative
40
100
0.1 (avg)
Adleretal., 1994
Stephanouetal., 1998
Auetal., 1995
-------
Table 4-2. Summary table of genetic effects of 1,3-butadiene (details in text) (continued).
Test system
Results
Dose
(LEC or HIC)a
Reference
Cytogenetic effects-Somatic cells (continued)
Human peripheral blood, chromosome aberrations, in vivo
Human peripheral blood, chromosome aberrations & SCE, in vivo
Human peripheral blood, chromosome aberrations, MN, & SCE, in vivo
negative
negative
negative
0.8 (avg)
0.3 (avg)
0.6 (avg)
Hallberg et al., 1997
Srametal., 1998
Sorsaetal., 1994
Cytogenetic effects-Germ cells
CD-I male mice, dominant lethal, in vivo
CD-I male mice, dominant lethal, in vivo
CD-I male mice, dominant lethal, in vivo
CD-I male mice, dominant lethal, in vivo
CD-I male mice, dominant lethal, in vivo
C3H male mice, heritable translocation, in vivo
C3H male mice, heritable translocation, in vivo
Sprague-Dawley rats, dominant lethal, in vivo
positive
positive
positive
positive
positive
positive
positive
negative
155
980 (acute)
10 (subchronic)
98 (subchronic)
50 (subchronic)
390 (5days)
1,000 (5days)
580 (subchronic)
Hackett et al., 1988a
Anderson etal., 1993
Anderson etal., 1993
Brinkworth et al., 1998
Anderson etal., 1998
Adleretal., 1995
Adleretal., 1998
Anderson etal., 1998
LEC, lowest effective concentration; HIC, highest ineffective concentration; in vitro tests, |_lg/ml; in vivo tests, mg/kg bw.
-------
Table 4-3. Summary table of genetic effects of l,2-epoxy-3-butene (EB) (details in text).
Test system
Results
Dose
(LEC or HIC)a
Reference
Gene mutations
B6C3F! mouse T lymphocytes, hprt locus, in vivo, ip
(102 x C3H)F1 mouse T lymphocytes, hprt locus, in vivo, ip
(102 x C3H)F1 mouse T lymphocytes, hprt locus, in vivo,
drinking water
B6C3F! mouse T lymphocytes, hprt locus, in vivo, inh
Lewis rat T lymphocytes, hprt locus, in vivo, ip
Lewis rat T lymphocytes, hprt locus, in vivo, drinking water
Fisher 344 rat T lymphocytes, hprt locus, in vivo, inh
Human TK6 cells, hprt locus, in vitro
positive
positive
negative
positive
negative
negative
negative
positive
100
3 x33
17
9
40
7
8
10.5
Cochrane & Skopek, 1994a
Tatesetal., 1998
Tales etal., 1998
Mengetal., 1999b
Tatesetal., 1998
Tatesetal., 1998
Mengetal., 1999b
Cochrane & Skopek, 1994b
Cytogenetic effects-Somatic cells
(102 x C3H)F1 mouse spleen MN, in vivo, ip
(102 x C3H)F1 mouse bone marrow MN, in vivo, ip
CD-I mouse bone marrow MN, in vivo, ip
B ALB/c mouse spleen MN & SCE, in vivo, ip
B ALB/c mouse peripheral blood MN, in vivo, ip
CD-I mouse splenocyte SCE, in vitro
CD rat splenocyte SCE, in vitro
Lewis rat spleen MN, in vivo, ip
Sprague-Dawley rat bone marrow MN, in vivo, ip
Human peripheral blood SCE, in vitro
positive
positive
positive
positive
positive
negative
negative
positive
positive
positive
40
20
40
48.8
22
70
70
80
120
3.5
Xiao & Tates, 1995
Adleretal., 1997
Anderson etal., 1997
Stephanouetal., 1997
Russoetal., 1997
Kligerman et al., 1999
Kligerman et al., 1999
Xiao & Tates, 1995
Anderson etal., 1997
Uuskula et al., 1995
-------
Table 4-3. Summary table of genetic effects of l,2-epoxy-3-butene (EB) (details in text) (continued).
Test system
Results
Dose
(LEC or HIC)a
Reference
Cytogenetic effects-Somatic cells (continued)
Human peripheral blood SCE, in vitro
Human peripheral blood SCE, in vitro
positive
negative
8.75
70
Bernardinietal., 1998
Kligerman et al., 1999
Cytogenetic effects-Germ cells
(102 x C3H)F1 mouse sperm. MN, in vivo, ip
B ALB/c mouse sperm. MN, in vivo, ip
(102 x C3H)Flmale mouse dominant lethal, in vivo, ip
Sprague-Dawley rat testes, MN, in vitro
Sprague-Dawley rat spermatids, MN, in vivo
positive
positive
negative
negative
positive
40
55
120
70
13
Xiao & Tales, 1995
Russoetal., 1997
Adleretal., 1995
Sjoblom & Lahdetie, 1996
Lahdetieetal., 1997
oo
LEC, lowest effective concentration; HIC, highest ineffective concentration; in vitro tests, |J,g/ml; in vivo tests, mg/kg bw.
-------
Table 4-4. Summary of genetic effects of l,2:3,4-diepoxybutane (DEB) (details in text).
Test system
Results
Dose
(LEC or HIC)a
Reference
Gene mutations
B6C3F! mouse T lymphocytes, hprt locus, in vivo, ip
(102 x C3H)F1 mouse T lymphocytes, hprt locus, in vivo, ip
(102 x C3H)F1 mouse T lymphocytes, hprt locus, in vivo, drinking
water
B6C3F! mouse T lymphocytes, hprt locus, in vivo, inh
Lewis rat T lymphocytes, hprt locus, in vivo, ip
Lewis rat T lymphocytes, hprt locus, in vivo, drinking water
Fisher 344 rat T lymphocytes, hprt locus, in vivo, inh
Human TK6 cells, hprt locus, in vitro
positive
negative
negative
positive
negative
negative
positive
positive
7
40
21
9
100
8.5
2
0.2
Cochrane & Skopek, 1994a
Tatesetal., 1998
Tales etal., 1998
Mengetal., 1999b
Tatesetal., 1998
Tatesetal., 1998
Mengetal., 1999b
Cochrane & Skopek, 1994b
Cytogenetic effects-Somatic cells
(102 x C3H)F1 mouse spleen MN, in vivo, ip
(102 x C3H)F1 mouse bone marrow MN, in vivo, ip
CD-I mouse bone marrow MN, in vivo, ip
B ALB/c mouse peripheral blood MN, in vivo, ip
CD-I mouse splenocyte SCE, in vitro
CD rat splenocyte SCE, in vitro
Lewis rat spleen MN, in vivo, ip
Sprague-Dawley rat bone marrow MN, in vivo, ip
Human peripheral blood SCE, in vitro
Human peripheral blood SCE, in vitro
positive
positive
positive
positive
positive
positive
positive
positive
positive
positive
15
9
30
9
1.7
1.7
15
25
0.17
1.7
Xiao & Tales, 1995
Adleretal., 1995
Anderson el al., 1997
Russoelal., 1997
Kligerman el al., 1999
Kligerman el al., 1999
Xiao & Tales, 1995
Anderson el al., 1997
Norppaelal., 1995
Kligerman el al., 1999
-------
Table 4-4. Summary of genetic effects of l,2:3,4-diepoxybutane (DEB) (details in text) (continued).
Test system
Results
Dose
(LEC or HIC)a
Reference
Cytogenetic effects-Germ cells
(102 x C3H)F1 mouse sperm. MN, in vivo, ip
B ALB/c mouse sperm. MN, in vivo, ip
(102 x C3H)Flmale mouse dominant lethal, in vivo, ip
Sprague-Dawley rat testes, MN, in vitro
Sprague-Dawley rat spermatids, MN, in vivo
positive
positive
positive
positive
positive
30
9
18
0.4
16.7
Xiao & Tales, 1995
Russoetal., 1997
Adleretal., 1995
Sjoblom & Lahdetie, 1996
Lahdetie etal., 1997
a LEC, lowest effective concentration; HIC, highest ineffective concentration; in vitro tests, Hg/ml; in vivo tests, mg/kg bw.
-------
Table 4-5. Summary table of genetic effects of 3,4-epoxy-l,2-butanediol(EBD) (details in text).
Test system
Results
Dose
(LEC or HIC)a
Reference
Gene mutations
Human TK6 cells, hprt locus, in vitro
positive
47
Cochrane & Skopek, 1994b
Cytogenetic effects-Somatic cells
(102 x C3H)F1 mouse bone marrow MN, in vivo, ip
(102 x C3H)Flmale mouse dominant lethal, in vivo, ip
positive
positive
120
26
Adleretal., 1997
Adleretal., 1995
Cytogenetic effects-Germ cells
Sprague-Dawley rat testes, MN, in vitro
Sprague-Dawley rat spermatids, MN, in vivo
negative
positive
102
30
Sjoblom & Lahdetie, 1996
Lahdetieetal., 1997
LEC, lowest effective concentration; HIC, highest ineffective concentration; in vitro tests, jjg/ml; in vivo tests, mg/kg bw
-------
two treated mice with zero mutants were excluded from the analysis, the difference was not
significant (initial group size was five animals). In the third experiment with this strain using
500 and 1,300 ppm 1,3-butadiene, an additional shorter expression time was incorporated into
the design. The hprt mutant frequencies at 1,300 ppm were higher than the 500 ppm, but neither
were significantly different from controls. There were no differences in mutant frequencies
between the 35- and 70-day expression times in the 1,300 ppm group, but the mutant frequency
at 70 days was twice that at 35 days expression time. The authors could offer no apparent reason
for the failure to detect a mutagenic response in this third experiment. The mice for all of the
above experiments were bred and treated in Dr. Adler's laboratory in Neuherberg, Germany.
A single experiment used CD-I male mice exposed to 1,3-butadiene concentrations of
12.5, 65, or 130 ppm in Dr. Anderson's laboratory at BIBRA. The authors stated that statistical
analysis indicated no trace of a mutagenic effect (Tates et al., 1998), as would be expected with
the concentrations used.
Meng et al. (1998) reported on a study in which female B6C3FJ mice and F344 rats were
exposed by inhalation to 1,250 ppm 1,3-butadiene for 1 or 2 weeks (6 h/day, 5 days/week).
Groups of animals were necropsied before exposure (controls) and weekly up to 10 weeks after
the last exposure. The researchers measured hprt mutants in both spleen and thymus using the T-
cell cloning assay. Mutant frequencies (mf) in both tissues of both species increased for several
weeks and then declined. Maximal frequencies were: in thymus, 11.3 x 10"6 in mice (2 weeks)
and 4.9 x 10"6 in rats (3 weeks); in spleen, 19.7 x 10"6 in mice (5 weeks) and 8.4 x 10"6 in rats (4
weeks). They determined a relative mutagenic potency (RMP) as the ratio of cumulative increase
in mutant frequency in treated versus controls. For the spleen the RMP was about fivefold
greater in mice than rats, which is similar to differences in metabolism. In a more recent report
(Meng et al., 1999a), female mice and rats were exposed to 1,3-butadiene at 0, 20, 62.5, and 625
ppm, 6 h/day, 5 days/wk for 2 or 4 weeks. Animals were sacrificed at 2, 4, 5, 6, and 8 weeks
after exposure for hprt mutant analysis in spleen T-cells. Mf increased linearly in both species at
625 ppm. Mf in mice was about fivefold and tenfold above controls after 2 and 4 weeks'
exposure, and in rats, though linear, was only twofold above controls after 4 weeks' exposure. In
the time-course experiments, the patterns were very similar to those observed in the previous
study with 1,250 ppm 1,3-butadiene. Calculated RMPs were as follows:
62.5 ppm 625 ppm
11 61
F344 rats not significant 7
4-12
-------
Molecular analysis of mutants from control and 1,3-butadiene-treated mice indicated a significant
decrease in base substitution mutations. Of base substitutions, approximately 60% were at AT
sites in controls, whereas about 55% were at GC sites in the treated groups (most in the
nontranscribed strand). The decrease in base substitutions was reflected by a significant increase
in the frequency of large deletions in 1,3-butadiene-derived mutants.
Several recent studies have measured in vivo mutations using the phage lad or lacZ
genes incorporated into a rodent genome. Recio and Goldsworthy (1995) summarized
experiments in which male B6C3FJ /ac/transgenic mice were exposed to 62.5, 625, and 1,250
ppm 1,3-butadiene (6 h/day, 5 days/week) for 4 weeks. Two weeks after the last exposure,
animals were euthanized and DNA was extracted from bone marrow to be examined for lad
mutagenesis. Mutant frequencies increased in a dose-response manner, reaching an apparent
plateau at 625 ppm (about a fourfold increase above controls). Sequence analysis of lad mutant
colonies from the 625 and 1,250 ppm groups indicated an increased frequency of point mutations
at A:T base pairs. A second group of animals was exposed to 1,3-butadiene at 625 ppm for only
5 days (Recio et al., 1996a). The observed lad mutant frequency was 12.6 x 10"5 compared to
15.3 x 10"5 after the 4-week exposure at 625 ppm. Subsequently, lad mutagenesis was examined
in spleen samples from the same animals (Recio et al., 1998). Although mutant frequencies were
significantly increased at all concentrations, the maximum response (also fourfold above
controls) was reached at 62.5 ppm. Mutants from the air controls and the 1,250 ppm exposed
groups were amplified and sequenced to identify mutational sites. As in the bone marrow
mutants, the frequency of mutations at A:T sites was increased. But more striking was an
increase in G:C ->• A:T at non-CpG sites (21% of mutations in the 1,250 ppm group compared
with 5% in controls). The authors stated that this increase in G:C bases was not observed in bone
marrow. These findings are consistent with those observed in 1,3-butadiene-induced hprt mutant
T lymphocytes from B6C3FJ mice (Cochrane and Skopek, 1994b).
The mouse spot test, although seldom used, was developed to detect mutations in seven
recessive genes (expressed as visible spot changes in coat color) in exposures of pregnant female
mice. Adler et al. (1994) exposed pregnant females to 0 and 500 ppm 1,3-butadiene on days 8-
12 of pregnancy. The incidence of visible spots in the exposed group was significantly above
controls. Other endpoints indicative of maternal or fetal toxicity, litter size, number weaned, and
malformations were not different from controls.
Several studies of genetic effects in exposed workers have recently been reported. Ward
et al. (1994) measured the frequency of hprt mutations in lymphocytes of workers in a 1,3-
butadiene production plant (two studies) and in a styrene-butadiene rubber (SBR) plant. In the
first study exposure estimates were based on 8-h samples in two production areas and in a central
control area. Mean 1,3-butadiene concentration in the production areas was 3.5 ppm, but the
4-13
-------
majority of samples showed concentrations below 1 ppm; mean 1,3-butadiene concentration in
the control area was 0.03 ppm. Variant frequencies at the hprt locus in PHA-stimulated
peripheral blood T-cells of a high-exposure group were increased more than threefold compared
with the low-exposure and nonexposed groups. The eight individuals in the high-exposure group
had hprt variant frequencies varying from 0.94 x 10"6 to 8.98 x 10"6 and the variant frequency
generally correlated with the level of the metabolite dihydroxybutane in the urine. Whether the
difference was due to differences in exposure or genetic differences in metabolism cannot be
ascertained from the data. A second study was conducted in the same plant about 1 year later
(Ward et al., 1996). Measured 1,3-butadiene concentrations in personal samplers were markedly
lower, 0.30 ± 0.59, 0.21 ± 0.21, and 0.12 ± 0.27 ppm in areas defined as high, medium, and low
exposure (no controls were reported for the second study). The corresponding hprt variant
frequencies were 5.33 ± 3.76, 2.27 ± 0.99, and 2.14 ± 0.97 x 10'6, respectively. Individual data
were not reported for this study, but again there is a high standard deviation in the highly exposed
group. The Ward et al. (1996) paper also reported preliminary results from workers in a styrene-
butadiene rubber plant. Workers were assigned to high (20 of 40 personal samplers exceeded the
0.25 ppm detection limit and 11 had a concentration over 1 ppm) and low (none of 26 exceeded
the detection limit) exposure groups. In nonsmokers, the hprt variant frequencies were 7.47 ±
5.69 and 1.68 ± 0.85 x 10"6 for the high and low groups, respectively. Although the variant
frequency for smokers in the high-exposure group (6.24 ± 4.37) was not different from
nonsmokers, the frequency for smokers in the low-exposure group was about twice that of the
nonsmoker group (3.42 ± 1.57). These preliminary findings with small sample sizes and no
detail about smoking history or other confounding factors raise several unanswerable questions
regarding the contribution of 1,3-butadiene exposure to the observed differences in hprt variants.
Hayes et al. (1996) employed the T-cell cloning assay to detect mutant frequencies in
lymphocytes of workers in a rubber production factory. 1,3-Butadiene levels were measured
using personal samplers during the 6-h work shift and expressed as a 6-h time-weighted average
(TWA). These were supplemented with several grab samples. Three different work areas were
identified: initial distillation and recovery from dimethyl formamide, polymerization, and
recovery, with median air levels of 3.5, 1.0, and 1.1 ppm, respectively. The T-cell cloning assay
was performed from postshift blood samples. Unexposed subjects were age- and gender-
matched and a brief questionnaire was administered. Tabular hprt mutant frequencies were
presented grouped only by gender and exposed versus unexposed. Mean mutant frequencies
were somewhat higher in females than males. Smoking (in males only) was not different in
either group, but mutant frequency did significantly increase with age. Mean mutant frequencies,
raw and adjusted for age, sex, cloning efficiency, and exposure status were similar in exposed
4-14
-------
and nonexposed workers. Adjusted mean frequency for total exposed workers was 18.0 x 10"6
compared with 13.6 x 10"6 for nonexposed workers.
In a third study, Tates et al. (1996) used the T-cell cloning assay on blood samples
collected from workers in a 1,3-butadiene plant in the Czech Republic. Workers were sampled
in 1993 and 1994, but most of the blood samples from 1993 were lost to technical errors. A
detailed analysis was conducted on the later group of 19 exposed and 19 nonexposed workers
from other parts of the same plant. Personal samplers indicated a mean 1,3-butadiene
concentration of 1.76 ppm, with individual samples ranging from 0.012 ppm to 19.77 ppm. The
geometric mean hprt mutant frequencies (adjusted for age, smoking, and cloning efficiency) were
7.10 x 10"6 for exposed and 10.59 x 10"6 for the controls. The range of mutant frequencies among
individuals was similar for both groups and individual mutant frequencies in the exposed group
were not correlated with concentrations of 1,3-butadiene detected in the personal samplers.
The results in both the T-cell cloning assay groups are in apparent conflict with the Ward
et al. (1994, 1996) findings both for exposed versus nonexposed and for smokers versus
nonsmokers. There are a number of differences in the protocols of the two procedures that could
account for the differences reported. Also, the study populations differed. Ward et al. (1994,
1996) conducted studies at a monomer production facility and an SBR production facility in
Texas, Tates et al. (1996) conducted the study in a monomer facility in the Czech Republic, and
Hayes et al. (1996) conducted the study in a polybutadiene rubber production plant in China.
Hence, these differences as well as the lack of precision of exposure determinations preclude any
direct comparison of these reports.
4.1.2. Cytogenetic Effects
4.1.2.1. Somatic Cells
Most of the rodent in vivo cytogenetic studies on 1,3-butadiene—especially in somatic
cells—have been thoroughly treated in the reviews cited in the introduction of this chapter. In
those studies, positive results were reported for all cytogenetic endpoints studied in mice and
negative results were consistently reported in rats.
Adler et al. (1994) exposed both male and female (102 x C3H)F1 mice to 1,3-butadiene
at 50, 200, 500, and 1,300 ppm for 6 h/day for 5 days and measured micronuclei in bone marrow
and peripheral blood sampled 18-24 h after exposure. Significant increases were seen at all dose
levels in both tissues, with the slope of the response decreasing after 200 ppm and essentially
leveling at 500 ppm. At 50 ppm the responses were similar in both sexes, but at higher
concentrations the incidence of MN was significantly greater in males (19.2 vs. 14.1 at 500 ppm).
In a later study Stephanou et al. (1998) exposed (102 x C3H)F1 males to 0, 130, 250, or 500 ppm
1,3-butadiene 6 h/day for 5 days. Animals were sacrificed 2, 5, and 13-14 days after the end of
4-15
-------
exposure, spleens were dissected, and splenocytes were collected for micronucleus analysis.
Blood samples were obtained from the 13-14 day group for peripheral blood micronucleus
analysis. Although MN frequencies were elevated in all groups analyzed, again the slope
decreased with increased concentration. Toxicity in the spleen precluded obtaining sufficient
cells for measurement on day 2 in the 500 ppm group. Significantly fewer MN were recorded at
the latest sampling at all concentrations. CREST antibody analysis used with splenocytes
showed only a weak aneugenic response in day 2 samples from the 130 and 250 ppm groups.
Anderson et al. (1997) reported negative results using the Comet assay in liver and bone
marrow in CD-I male mice after exposure to 1,3-butadiene at 0, 12.5, 65, or 130 ppm, 6 h/day, 5
days/wk for 4 weeks. (This assay detects DNA migrating out of the cell nucleus during
electrophoresis. Damaged DNA, with strand breaks migrates further than intact DNA and
measurement of length, area, and density of the "tail" provides an index of DNA damage.)
There have been several studies evaluating cytogenetic effects of exposed workers. Au
et al. (1995) measured chromosome aberration frequencies in blood samples of 10 exposed
workers and 10 matched controls from the same population used in the Ward et al. (1996b) study
cited above. They reported measurable, but not significant (p>0.1), increases in chromosome
aberrations and chromatid breaks. Also, cells were exposed to gamma rays in Gl and aberrations
were measured in the subsequent metaphase. With this indirect measure of DNA repair,
chromatid breaks, deletions, and dicentrics were all significantly higher in cells from 1,3-
butadiene-exposed workers. A second study by the same group exhibited similar results
(Hallberg et al., 1997). Male workers in a 1,3-butadiene production plant wore personal film
badges during a work shift, at the end of which they donated a blood and urine sample. Mean
exposure was 2.4 ppm (TWA) among 24 exposed workers and 0.3 ppm among 19 control
workers. Chromosome aberration frequencies in standard assays and in gamma ray challenge
assays were higher in exposed workers, but in neither case was the increase statistically
significant. Recently, Sram et al. (1998) examined cytogenetic damage in peripheral
lymphocytes of workers in a 1,3-butadiene monomer production plant in the Czech Republic.
Nineteen exposed workers and 19 matched controls from other parts of the same plant wore
personal monitors for one work shift, after which they contributed blood samples. Both
chromosome aberrations and SCE were significantly higher in exposed workers, even when
corrected for smoking. Micronuclei were higher in exposed than in controls, but the difference
was not significant. This negative finding is questionable because of the surprisingly high
incidence of micronuclei in the nonsmoking control group (17.45 vs. 10.75 for control smokers).
DNA damage measured by the Comet assay also was not elevated in the exposed workers.
4-16
-------
Sorsa et al. (1994) investigated chromosomal damage in blood lymphocytes sampled in
1993 from workers in the factories described by Tates et al. (1996) above. Chromosome
aberrations, MN, and SCE frequencies were not elevated above samples from unexposed
persons. They did note that smoking had a slight effect in MN formation and SCE but not in
chromosome aberrations. Preliminary data measuring chromosome aberrations and micronuclei
in blood samples from the 1994 group of workers were reported by Tates et al. (1996). The
percentage of aberrant cells was significantly increased (p<0.01) in exposed subjects; however,
the frequency of micronuclei in lymphocytes was similar in exposed and unexposed subjects.
Evaluation of data for each subject would be required to determine the basis for the apparent
discrepancy of the results between the two years.
4.1.2.2. Germ Cells
The question of whether 1,3-butadiene has the capacity to induce stable genetic damage
has been addressed in dominant lethal and heritable translocation experiments. Hackett et al.
(1988a) exposed CD-I mice by whole-body inhalation to 200, 1,000, and 5,000 ppm 1,3-
butadiene for 6 h/day on 5 consecutive days. A significant increase in the number of early deaths
of implants per pregnant female was observed during the first 2 weeks after exposure, but the
results were not dose related. Males exposed to 200 ppm and 1,000 ppm sired litters with
increased numbers of early deaths. This effect was not seen in the group exposed to 5,000 ppm.
In addition, in the first week post-exposure the number of dams showing two or more dead
implants per pregnancy was increased for all dose groups. Because these effects were noted in
the first week or two following exposure, the authors suggested that spermatozoa and spermatids
were the susceptible germinal stage.
In a second dominant lethal study, Anderson et al. (1993) treated male CD-I mice with
acute and subchronic inhalation exposures to 1,3-butadiene and mated the males to untreated
females. In the acute study, the male mice were exposed to 0, 1,250, or 6,250 ppm for 6 h. The
females were examined for live fetuses, postimplantation deaths, and gross malformations. The
only statistically significant result was a small decrease in implantations at 1,250 ppm. In the
subchronic study, the males were exposed to 0, 12.5, or 1,250 ppm for 6 h/day, 5 days/week for
10 weeks. Statistically significant effects were observed at both exposure levels. An increase in
late deaths and abnormal fetuses was observed at 12.5 ppm. A decrease in the number of
implantations and an increase in the number of dominant lethal mutations were found at 1,250
ppm, as well as an increase in fetal malformations. Further evaluation of the fetuses as well as
evaluation of animals allowed to deliver their young was published in a report and a paper by
Anderson et al. (1995, 1996). In fetuses recovered on gestation day 17, skeletal changes were
evaluated in the small number of malformed fetuses and their normal litter mates and controls,
4-17
-------
and karyotypes were examined from the livers of these fetuses. A few minor skeletal changes
were seen in the skeleton, particularly in skull bones of fetuses with exencephaly. Pups from the
other half of the litters were raised to 37 weeks of age for control and high-exposure animals in
the acute study, and for 75 weeks for all animals in the subchronic study. There was no obvious
increase in tumor incidence above control values, but a histologic examination had not been
completed.
Adler and Anderson (1994) used male (102/E1 x C3H/E1)F1 mice to assess the stage at
which male germ cells are affected by 1,3-butadiene. Adult males were exposed by inhalation to
0 or 1,300 ppm, 6 h/day for 5 consecutive days. Four hours after the end of exposure, each male
was mated at a ratio of 1:2 to untreated virgin females. Females judged bred by the presence of a
vaginal plug were replaced with new females, and mating continued for 4 consecutive weeks.
Females were killed on gd 14 to 16 and examined for numbers of live and dead implants.
Exposure of male mice to 1,300 ppm resulted in an increase in dead implants during the first to
the third weeks of mating; however, statistical significance (p<0.01) was reached only in the
second week. When expressed as a percentage of dominant lethals, a significant increase was
seen in the second (12.4%,/?<0.01) and third (5.5%,/?<0.05) weeks. Because of the time course
for dominant lethal mutations to manifest as dead implantations, 1,3-butadiene again appears to
induce dominant lethality in spermatozoa and late spermatids.
Brinkworth et al. (1998) exposed CD-I mice, aged 8-10 weeks to either 0, 12.5, or 125
ppm 1,3-butadiene. The purpose of this study was to undertake a dominant lethal study in mice
at sub-chronic low dose (6 h/day, 5 days/wk, 10 weeks) exposure to 1,3-butadiene. CD-I mice,
aged 8-10 weeks were assigned to 3 groups and exposed to either 0, 12.5, or 125 ppm 1,3-
butadiene and then mated to untreated virgin females. Incidence of early deaths were
significantly elevated only in the high (125 ppm) dose group relative to the control group
(p<0.01). The incidence of late deaths and of dead fetuses was clearly higher in both dose groups
than in the control, but the difference was not statistically significant.
Anderson et al. (1998) reported on a comparative study testing the hypothesis that genetic
effects could be transmitted by males to their offspring following subchronic inhalation exposure
to 1,3-butadiene in mice and rats. Because it had been previously shown that mice require only 4
weeks of exposure (Adler and Anderson, 1994), male CD-I mice were exposed to 0 ppm, 12.5
ppm, 65 ppm, or 130 ppm for 6 h/day, 5 days/wk, for 4 weeks. In the rat study, Sprague-Dawley
males were exposed to 0 ppm, 65 ppm, 400 ppm, or 1,250 ppm for 6 h/day, 5 days/wk, for 10
weeks. Each exposed male was mated with two untreated virgin females. Females were killed
the day before parturition (day 17 after finding a vaginal plug in mice and on day 19 after finding
a sperm-positive vaginal smear in rats). Females were examined for the number of live fetuses,
the number of fetuses with gross malformations, and the number of postimplantation early and
4-18
-------
late deaths. In mice there were no significant increases in implants between the three dose
groups and controls. However, there was a significant increase in the mean number of early
embryonic deaths in the 65 and 130 ppm dose groups, with a NOAEL of 12.5 ppm. There was
no significant increase in mean number of late deaths or dead or abnormal fetuses from control.
In rats, there was a statistically significant (p<0.05) reduction in implants in the 65 ppm exposure
group, but not in the 400 and 1,250 ppm dose groups. There was no significant increase in the
mean number of early or late deaths or congenital malformations between treated and control rat
groups. In both rodent studies neither mating frequency nor pregnancy rate was significantly
reduced as a result of exposure.
Pacchierotti et al. (1998a) conducted a cytogenetic analysis of chromosome aberrations in
first-cleavage embryos derived from 1,3-butadiene exposed male mice. Male (102/E1 x
C3H/E1) mice were exposed for 5 consecutive days (6 h/day) to 0, 130, 500, or 1,300 ppm of
1,3-butadiene in air. On the last day of inhalation exposure, mice were mated during a 3-week
period with untreated females for evaluation of transmissible chromosome damage. First-
cleavage fertilized oocytes following paired matings were analyzed for chromosome and
chromatid break and exchange aberrations for each dose group. Statistically significant increases
over controls were observed in the first mating week of mice exposed to 500 and 1,300 ppm
doses and in the second mating week of animals treated at the 1,300 ppm dose. There was no
evidence of increased frequency of zygote aberrations obtained from animals in third week
matings. Aberration induction was 2.5 times higher in the first than in the second mating week,
suggesting that late developmental stages of spermatozoa were again more sensitive to butadiene
than earlier stages. Of 72 aberrations observed, 96% of them were chromosomal breaks or
exchanges.
To determine whether the cytogenetic damage induced by 1,3-butadiene was
transmissible, heritable translocation studies were conducted (Adler et al., 1995, 1998). Male
C3H/ El mice were exposed by inhalation to 1,3-butadiene at 500 ppm (Adler et al., 1998) or
1,300 ppm (Adler et al., 1995) for 5 days for 6 h/day and mated to 102/E1 virgin females for 7
days (days 8-14 post-exposure). Offspring were tested for translocations as revealed by reduced
litter size and confirmed by cytogenetic analysis of meiotic and somatic cells. The translocation
frequency from treated males was 1.15% and 2.7% for the 500 and 1,300 ppm groups compared
with 0.05% for historical controls. The authors stated that the data best fit a linear equation: y =
0.05 + 6.9 x I0"5x. Using this equation, the authors calculated the doubling dose (that dose which
induces as many translocation mutations as occur spontaneously per generation) of 725 ppnrh, or
a concentration of approximately 24 ppm over the 5 days of 6 h/day exposure.
The results of the dominant lethal studies suggest that only postmeiotic stages are affected
by 1,3-butadiene exposure. Xiao and Tates (1995) measured micronuclei in germ cells of male
4-19
-------
(102 x C3H)F1 mice after inhalation exposure to 1,3-butadiene at 500 or 1,300 ppm 6 h/day for 5
days. Mice were sacrificed at 2, 5, 11, and 15 days after exposure to sample various stages of
spermatogenesis. Significant increases were observed only at the 15-day sampling, indicative of
damage to early spermatocytes. In a second experiment, males were exposed at 200, 500, and
1,300 ppm under the same conditions and only early spermatocytes were analyzed. The
incidence of MN was 2.80, 3.40, and 2.60, respectively, for the three concentrations (control
incidence was 0.33). Both body weight and testis weight were noticeably reduced at the highest
concentration. These results were confirmed by Tommasi et al. (1998) using the same strain of
mice and 1,3-butadiene concentrations of 130, 250, and 500 ppm. The frequency of MN again
was increased only in early spermatocytes, with values significantly higher than controls only at
250 ppm 1,3-butadiene.
Anderson et al. (1997) also used the Comet assay on haploid cells from the testis of the
males in the study cited earlier, which used a 4-wk exposure and observed no indication of a
response. Later, Brinkworth et al. (1998) exposed CD-I males to 1,3-butadiene at 12.5 and 1,250
ppm, 6 h/day for 1 day or for 10 wk (5 day/wk). After exposure one testis was taken for the
Comet assay and the other was injected with 3H-thymidine for later determination of unscheduled
DNA synthesis (UDS). A weak but significant positive response was seen with the Comet assay
after 1,250 ppm 1,3-butadiene in the 1-day but not the 10-wk exposure. For the UDS assay mice
were sacrificed 17 days after 3H-thymidine injection so that cells that were early spermatids
could mature to spermatozoa. Although the mean counts were higher than controls in both
groups for the 1-day exposure, high variability between animals meant that the differences were
not significant. The authors also reported that the recovery of sperm from the cauda epididymis
was at least 100-fold lower than expected, indicating toxicity.
4.2. METABOLITES
A search of the mutagenicity literature databases in Toxnet revealed a surprising number
of more than 400 citations to DEB; about three-fourths of these appeared prior to 1991. These
papers account for two-thirds of the total number of papers citing 1,3-butadiene or either of the
two more common epoxide metabolites (EB and DEB). DEB was one of the mutagens of choice
of investigators exploring various aspects of chemically induced mutations in a wide variety of
biological systems. It is noteworthy that the stereoisomers of DEB were recognized to exhibit
differential mutagenic activity as early as 1963 (Moutschen et al., 1963). The focus of the
following sections is on those systems and endpoints that might provide information relevant to
an understanding of the mode of action of 1,3-butadiene.
4-20
-------
4.2.1. Gene Mutations
In the same paper cited earlier, Cochrane and Skopek (1994a) reported on the mutagenic
activity of EB and DEB in the mouse hprt assay. Mice received three daily intraperitoneal (ip)
injections of 60, 80, or 100 mg/kg of EB or 7, 14, or 21 mg/kg of DEB. Mutant frequencies in
hprt from splenic T-cells were dose related for both metabolites, with maximal responses of 8.6
x 10"6 and 13 x 10"6 for EB and DEB, respectively. In a study using human TK6 cells, Cochrane
and Skopek (1994b) showed that all three epoxide-forming metabolites were effective at
inducing mutations at both the hprt and tk loci. Concentrations inducing approximately twofold
increases in mutant frequency at the hprt locus were 3.5 |J,M, 150 |J,M, and 450 |J,M for DEB,
EB, and 3,4-epoxy-l,2-butanediol, respectively. Later, Steen et al. (1997a,b) characterized the
mutants induced by EB and DEB in TK6 cells. Although both metabolites induced significant
increases in A:T to T:A transversions, only DEB resulted in a higher incidence of deletions.
In a series of experiments, Tates et al. (1998) measured hprt mutations in both mice and
rats after exposure to either EB or DEB. A clear dose related increase was seen only in male
(102 x C3H)F1 mice that had received three ip injections of 33 or 100 mg/kg EB (48 h between
injections). Mice injected with single ip injections of 20 or 40 mg/kg of DEB or 3 injections of 7
or 14 mg/kg DEB showed no evidence of a mutational response. Likewise negative responses
were reported in Lewis rats injected three times with 33 or 100 mg/kg EB or singly with 20 or 40
mg/kg DEB. Neither EB nor DEB was mutagenic in rats exposed to concentrations of 0.1, 0.3,
or 1.0 mM of the chemicals in drinking water for 30 days.
In addition to the 1,3-butadiene experiments with mice and rats, Meng et al. (1999b)
exposed females of both species to both EB and DEB. Using published data and PBPK models,
they calculated concentrations of the metabolites that, using inhalation exposures, would give
blood-level concentrations of the metabolites similar to those seen after exposure to 62.5 or 625
ppm of 1,3-butadiene. Animals were exposed to EB at 2.5 and 25 ppm, 6 h/day, 5 days/wk for 4
weeks. A preliminary 6-h exposure of female mice at 25 ppm to measure blood levels confirmed
the similarity to 625 ppm 1,3-butadiene. For DEB, researchers selected concentrations of 2 and 4
ppm. Calculations indicated that 4 ppm would result in blood concentrations of DEB similar to
62.5 ppm of 1,3-butadiene, but because of concerns of nasal toxicity to the highly reactive DEB,
researchers chose to use 4 ppm as the high concentration and used 2 ppm as a backup level. As
with the 1,3-butadiene experiments, animals were sacrificed at various times after exposure (up
to 9 weeks) for measuring hprt mutants in spleen T-cells. With EB, hprt mutant frequencies
were significantly elevated only at 7 weeks post-exposure in mice. In rats, neither concentration
caused a significant increase, but there was a significant upward trend also at 7 weeks post-
exposure. Results with DEB were strikingly different. Mutant frequencies were significantly
above controls even at 1 week after exposure in both species. In mice, mutant frequencies
4-21
-------
increased only marginally at 2, 4, and 6 weeks post-exposure, reaching a maximum of 2.8-fold
above controls, before declining to control level by 9 weeks. In rats, the increase was greater,
with a maximum of 3.3-fold above controls by week 4 followed by a steeper decline, again
reaching control levels by 9 weeks post-exposure.
4.2.2. Cytogenetic Effects
4.2.2.1. Somatic Cells
Adler et al. (1995) injected (102 x C3H)F1 males and females with DEB at 0, 4.5, 9, 18,
and 36 mg/kg and sampled bone marrow 24 h later. A clear linear dose-response of MN was
observed, with no differences between sexes, unlike the greater response in males seen
previously with 1,3-butadiene exposure (Adler et al., 1994). In a later study (Adler et al., 1997),
mice were injected with 20, 40, and 80 mg/kg of EB (both sexes) or 60, 120, and 240 mg/kg of
EBD (males only). Bone marrow MN increased with dose after EB injection, with all doses
significantly above untreated controls. Even though the MN response for EBD was also dose
related, only the higher two doses were significantly higher than controls, and the response at 240
mg/kg was about the same as the response after 40 mg/kg of EB.
Xiao and Tates (1995) evaluated the cytogenetic effects of EB and DEB in both (102 x
C3H)F1 mice and Lewis rats. Male animals of both species received single ip injections of 40 or
80 mg/kg of EB. Animals were sacrificed at various time intervals after treatment and spleens
were processed for scoring of micronuclei. EB was almost four times more effective in the
mouse as in the rat. For DEB, mice were injected with 15 or 30 mg/kg and rats received single ip
injections of 20, 30, or 40 mg/kg. In a separate experiment, rats received 3 daily injections of 10
mg/kg. The response in splenocytes was similar in both mice and rats at 30 mg/kg. Using
selected slides from this study, the investigators employed centromeric probes and fluorescent in
situ hybridization (FISH) to investigate the origin of the induced micronuclei. They reported
(Xiao et al., 1996) that in both mice and rats a clear clastogenic response was observed for the
metabolites, and that only mice exhibited a weak anuegenic response. A positive response was
also reported by Anderson et al. (1997) for both metabolites in CD-I mice and in Sprague-
Dawley rats. In that study, there seemed to be little if any difference between the species in
response to either metabolite, though differences in control levels complicate the comparison.
Stephanou et al. (1997) injected BALB/C mice with EB at 24.4, 48.8, and 73.2 mg/kg and
measured micronuclei and SCE at 2 and 16 days after treatment. Although induction of
micronuclei reached a maximum after 48.8 mg/kg, SCE induction continued to increase up to the
highest dose used. That the two endpoints reflect different genetic lesions was further indicated
in that the residual level of SCE at 16 days of the two higher concentrations was only marginally
higher than the response of the lowest dose, whereas for micronuclei the response at 16 days was
4-22
-------
only slightly lower than after 2 days and the dose-response curves were essentially identical in
pattern.
Russo et al. (1997) also detected increased incidence of MN in peripheral blood from
BALB/C mice injected with EB. They also found DEB to be more than twice as effective an MN
inducer as EB.
Kligerman et al. (1999) recently reported on a series of experiments providing additional
evidence that the species differences in response to 1,3-butadiene exposure reflect differences in
formation of active metabolites rather than differences in processing genetic lesions to observed
toxic responses. They examined SCE and aberrations in mouse, rat, and human lymphocytes
treated in G0 with EB or DEB. One-hour treatments with EB at concentrations up to 1,000 |J,M
were without effect in all three species. Longer treatments of mouse lymphocytes resulted in
slight increase in SCE (2 h) and toxicity with few second-division cells (4 h). Conversely, DEB
was an effective inducer of both SCE and aberrations in all three species, exhibiting linear
responses up to 40 |J,M (the highest concentration tested). Responses from whole-blood or
splenic lymphocytes were consistently lower than from isolated lymphocytes in both rats and
mice, reflecting the absence of glutathione transferase activity in lymphocytes.
The role of glutathione S-transferase (GST) genes GSTM1 and GSTT1 enzymes in the
detoxification of butadiene metabolites has been evaluated by measuring the induction of SCE in
cultured human lymphocytes. Uuskula et al. (1995) found that SCE induction in lymphocytes
from GSTMl-null individuals was 31% higher than in lymphocytes from GSTM1-positive
individuals treated with 50 or 250 mM EB. The same group (Norppa et al., 1995) reported no
difference in SCE induction among GSTM1 nulls and GSTM1-positive lymphocytes when
treated in vitro with DEB; however, they observed a 60% increase in SCE in lymphocytes from
GSTTl-null individuals treated with 2 or 5 |J,M DEB. Subsequently, these investigators
(Bernardini et al., 1998) observed increased SCE response to EB in human lymphocyte cultures
from GSTTl-null donors when compared with cultures from GSTT1-positive donors. They also
reexamined the GSTT1 phenotypes in the earlier GSTMl-null study and found that only two of
the six GSTMl-null donors were also GSTTl-nulls and that the difference in SCE response was
still 1.4 times higher in the GSTMl-null/GSTTl-positives than the GSTM1-positive donors.
Neither GSTM1 nor GSTT1 deficiency affected the induction of SCE by 250 or 500 |J,M of EBD
(Bernardini et al., 1996). In a separate study, Kelsey et al. (1995) found that GSTT1 deficiency
significantly increased the frequency of SCE induced by DEB in lymphocyte cultures of workers
exposed to 1,3-butadiene.
In the study cited above, Kligerman et al. (1999) reported that differences in response
between whole blood and isolated lymphocytes were seen only between GSTT1 null and GSTT1+
groups. Differences in phenotype of GSTM1 were not seen within the two GSTT1 groups. The
4-23
-------
authors did note that variability in response within the groups or within individuals sampled at
different times suggested that other enzyme polymorphisms or dietary factors might also be
involved.
4.2.2.2. Germ Cells
Xiao and Tates (1995) measured micronuclei in germ cells from the mice and rats in the
study cited above. In mouse germ cells, the incidence of micronuclei was similar to controls on
days 1 and 3 after exposure, but was significantly increased on day 14. In rats, EB was equally
effective on days 1 to 3 (late spermatids) and day 20 (early spermatocytes), and the frequency of
micronuclei at 80 mg/kg was slightly higher than that observed on day 14 in the mouse. DEB
increased the frequency of micronuclei only on day 3 after treatment in mouse germ cells,
whereas significant increases of micronuclei were observed in rat germ cells at all doses and all
time periods.
Russo et al. (1997) observed similar results in spermatids and spermatocytes in male
BALB/C mice treated with EB and DEB; however, they reported significant increases in MN in
spermatocytes as well as in spermatids. As with peripheral blood MN, DEB was the more
effective inducer of MN.
Sjoblom and Lahdetie (1996) used an in vitro meiotic micronucleus assay to examine the
effects of EB, DEB, and EBD in seminiferous tubule sections of male Sprague-Dawley rats.
Tissue sections were cultured for 4 days with EB at 100, 500, or 1,000 |imol/L; DEB at 5, 10, or
20 |J,mol/L; or EBD at 10, 50, or 100 |J,mol/L. The frequency of micronuclei was increased only
by DEB and the increase was clearly dose-related. The authors suggest that EB requires further
metabolism by P450 enzymes, which they indicate do not occur in rat testes microsomes. That
this was an in vitro "artefact" was confirmed in their subsequent report, in which the three
metabolites were injected ip into male Sprague-Dawley rats and sampled for micronucleus
formation at several time periods after injection (Lahdetie et al., 1997). All three metabolites
were active in the induction of micronuclei measured in spermatids and representing meiotic
stages from Gl onward. Although on face value the three metabolites show peak responses in
different stages of meiotic prophase, detailed analysis indicates that the differences are not
statistically significant. Also, because of metabolism there are certainly different concentrations
of the various metabolites in the different stages of spermatogenesis.
Dominant lethal studies have also been conducted with all three epoxide metabolites.
Adler et al. (1995) injected male E6C3Fl mice ip with 18, 36, or 54 mg/kg DEB. Toxicity of the
high dose resulted in reduced fertility, and the implantation rate was too low to measure
dominant lethality in the first two mating periods. Increased incidence of dead implants was
observed in the next mating period, corresponding to treatment of late spermatids. A dominant
4-24
-------
lethal effect with the two lower doses was seen only in the early matings representing mature
spermatozoa. Researchers also injected a second group of males with 17, 26, 34, 43, or 52 mg/kg
EBD to be subsequently mated with virgin females 7, 14, 21, and 28 days post-treatment. In
cytogenetic analysis of first-cleavage-division embryos from the 26 mg/kg group, they recorded
increased chromosome aberrations in embryos from only the first mating, again representing
effects on spermatozoa. Embryos derived from early spermatids (day 14 and 21 matings) and
spermatocytes (day 28 mating) had aberration levels not significantly above controls. The three
higher doses exhibited extreme toxicity, precluding cytogenetic analysis.
In a companion study, Tiveron et al. (1997) injected gonadotropin-treated B6C3HF1
females with 26, 34, 43, or 54 mg/kg of DEB. Chemical injections were 24 h after injection of
PMS and 24 h before the injection of HCG. Matings to untreated males were 4 h after HCG and
cytogenetic analysis conducted on one-cell zygotes. Although matings were reduced in all
treatment groups, the effect was not dose related. Further, among the successful matings there
was no reduction in the percent of oocytes fertilized or in the percentage of zygotes reaching
first-cleavage metaphase. Significant increases in aberrations were observed at all dose levels,
with the response best fitting to a linear quadratic curve. Comparing the response in this study
with the treated male study cited above (Adler et al., 1995), they reported that oocytes transmit 4
to 8 times fewer aberrations than treated sperm. The authors cited the known presence of DNA
repair capacity in preovulated oocytes and its absence in late spermatocytes and mature sperm as
well as the fact that oocytes have significantly more cytoplasmic protein available for binding the
DEB, thus reducing the effective nuclear dose of DEB.
In a later study, Adler et al. (1997) injected male (102 x C3H)F1 mice ip with either 120
mg/kg of EB or 240 mg/kg of DEB. In this study, males were mated to both (102 x C3H)F1 and
NMRI females. Surprisingly, neither EB nor DEB showed a dominant lethal effect, despite the
dose-related effect of both in the bone marrow micronucleus test discussed above.
Experiments detecting cytogenetic damage in male germ cells indicate that the dominant
lethal effects detected only from late spermatids and spermatogonia are not due to differential
sensitivity to damage of those cell types, but rather are a reflection of the biology of
spermatogenesis. Genetic damage in spermatogonia or premeiotic spermatocytes, especially the
type of chromosome damage resulting in micronuclei formation, is most likely to be cytotoxic.
Also, DNA repair activity ceases by the mid-spermatid stage; hence damage (DNA adducts)
would remain in the sperm and be first "processed" in the first cleavage division of the zygote.
4.3. DISCUSSION
The current database of more than 600 publications provides ample evidence of the
breadth of mutagenic responses that result from exposure to 1,3-butadiene of a variety of
4-25
-------
biological systems ranging from bacteria to human beings. It is also clear that the mutagenic and
genotoxic responses require the metabolic activation of 1,3-butadiene to several DNA-reactive
intermediates, especially DEB, EB, and possibly EBD (data on this metabolite are scant).
Consistent with (more accurately, as a consequence of) the differences in metabolism described
in Chapter 3, mutagenic responses in mice are consistently greater than the responses in rats
when animals are exposed to 1,3-butadiene. That the species difference in hprt response is
greater after 4 weeks exposure than after 2 weeks (Meng et al., 1999a) indicates that the
differences continue to diverge with the length of exposure. That lad mutation frequencies do
not increase from 1 week to 4 weeks exposure (Recio et al., 1996b) is not a direct contradiction
of the Meng finding because several differences in experimental design preclude a direct
comparison of the two studies.
A range of studies, both in vitro and in vivo, on the two major metabolites indicate that
EB primarily induces point mutations and small deletions, whereas DEB, as a bifunctional
alkylating agent, results in larger deletions. Additionally, DEB is mutagenically active in in vitro
systems at concentrations similar to those found in the blood of mice exposed to 1,3-butadiene by
inhalation. EB requires concentrations in vitro 10- to 100-fold greater than those found in the
blood of 1,3-butadiene-exposed mice. Studies with EB and DEB measuring hprt mutations in
female mice and rats have revealed, at least initially, some unexpected results (Meng et al.,
1999b). Although, as expected, EB was positive in the mouse and negative in the rat, at
equivalent exposure concentrations DEB was about threefold more mutagenic in Fischer 344 rats
than in C3B6F1 mice. Similarly, Xiao and Tates (1995) reported greater response in MN in bone
marrow and male germ cells of Lewis rats than in (102 x C3H)F1 mice after ip exposure to
DEB. Beyond differences in general metabolic rates, the rate of hepatic clearance in the rat is
about two times slower than in the mouse, with possibly higher blood levels in the rat. Also,
DEB seems to be less cytotoxic in rats, and as mutant yield is clearly a function of surviving
cells, higher survival would result in higher observed mutant frequencies.
For alkylating agents, DNA adducts are the molecular precursors of the mutants observed
in the various assay systems discussed above. Measurement of DNA adducts in target tissues can
provide a direct and more accurate determination of the dose of carcinogen bound to DNA. As
discussed in Chapter 3, a number of DNA adducts of 1,3-butadiene metabolites have been
identified, mostly with guanine and adenine bases. The recent studies measuring specific DNA
adducts have resolved some questions but have raised additional questions regarding the
contribution of the different metabolites in both carcinogenic and mutagenic responses resulting
from exposure to 1,3-butadiene. The report that EB-DNA adducts are not markedly different in
rats and mice suggests that EB does not play a insignificant role in carcinogenesis of the two
4-26
-------
species (Koc, et al., 1999). However, they also found that the THB-Gua adducts were
predominantly EBD-derived adducts, rather than from DEB.
Many of the in vivo mutagenicity studies described above were part of a collaborative
effort of several European laboratories in a project, "Multi-endpoint analysis of genetic damage
induced by 1,3-butadiene and its major metabolites in somatic and germ cells of mice, rats and
man: Genetic risk estimation by the parallelogram method." Mutations occurring in germ cells
may be inherited by future generations and may contribute to genetic disease. In the weight-of-
evidence scheme in the Guidelines for Mutagenicity Risk Assessment (U.S. EPA, 1986), 1,3-
butadiene is in the highest category described based on animal data. As described in the
Guidelines, there are a number of reasons that make it unlikely that there could ever be sufficient
data relative to chemically induced genetic damage in humans.
The parallelogram approach was devised by Sobels (1977) as a way to estimate heritable
genetic effects using data from somatic cells from both rodents and humans and data from rodent
germ cell assays (measuring heritable effects) as three corners of a parallelogram from which the
fourth (heritable effects in humans) could derived. Although the concept is simple in form, there
are obviously many variables, both within and across species, that necessarily affect the
confidence of any derived estimate. In the output of the European effort (Pacchierotti et al.,
1998a), the results of the studies were summarized and genetic risk estimates were derived using
the parallelogram concept and the more classical doubling dose concept. (The doubling dose can
be traced back to the efforts to estimate the effects of ionizing radiation in the mid-1950s.) The
doubling dose (DD) is simply that dose (concentration x time) that induces as many mutations
per generation as occur spontaneously. When the dose-response is a linear function (y = a + bx),
DD is simply the spontaneous incidence divided by the slope (a/b). Because the only heritable
genetic endpoint data available are from the mouse heritable translocation test, the researchers
used cytogenetic data in their calculations. Somatic cell data were from MN in mouse bone
marrow (Adler et al., 1994) and chromosome aberrations in human lymphocytes (Sram et al.,
1998). Using the linear portion of the curve of the mouse data researchers derived an estimated
DD of 2,411 ppnrh. From the human data only an estimated concentration of 1.58 ppm for
exposed workers was available. They used an exposure time of three years (lymphocyte lifespan)
and assumed a work year of 46 weeks and 40 h/week. Combining these assumed figures with the
observed frequencies of aberrant cells of 3.11% in exposed workers and 2.08% in controls, they
calculated a DD of 16,394 ppnrh. In the second heritable translocation paper, Adler et al. (1998)
calculated a DD of 725 ppnrh. Using simple proportions, a DD for heritable translocations in
humans is calculated to be 4,930 ppnrh.
Alternatively, the parallelogram can be applied using the induced damage per unit dose.
Using the same studies as above with slopes of 1.12 x 10"6 and 0.69 x 10"6 for MN and
4-27
-------
translocations and the point estimate of 1.24 x 10"6 per ppm for aberrations in human
lymphocytes, they calculated an estimated rate of heritable translocations in human germ cells of
0.76 x 10"6 per gamete per ppm of 1,3-butadiene.
Of course, application of the parallelogram method is overly simplistic and encompasses
many assumptions, some a function of the chemical and some totally independent of any
chemical exposure. Also, it must be recognized that the events measured in the cited studies are
not truly toxic events. Although it is certainly true that translocations and other chromosome
aberrations are critical and causative factors in many genetic diseases, their mere existence is not
predictive of disease. This is because the impact of a mutation (in this case chromosome
aberration) is a function of the physiological function of the gene (or genes) involved rather than
the molecular nature of the genetic alteration.
The studies cited here, along with the many earlier genotoxicity studies discussed in the
cited reviews, provide clear evidence that 1,3-butadiene is both mutagenic and clastogenic
through its metabolism, primarily because of the mono- and diepoxide. For most endpoints, the
difunctional DEB is clearly more effective than the monofunctional EB. Where both mice and
rats have been studied, mice are more responsive than rats, except for the recent germ cell
studies. Whether this exception is strain specific (among or between species) can only be
answered with future work.
The role of glutathione S-transferase is also clearly established for the genotoxic effects
of butadiene in human lymphocytes. That the induction of SCEs in human lymphocytes by the
three metabolites is affected differently by GSTM1 and GSTT1 complicates risk estimation for
human beings, assuming that SCE is a reliable biomarker for exposure (or effect).
4-28
-------
5. REPRODUCTIVE AND DEVELOPMENTAL EFFECTS
5.1. REPRODUCTIVE EFFECTS
Several toxicity studies for 1,3-butadiene have examined reproductive effects in multiple
species: one in rats, guinea pigs, rabbits, and dogs (Carpenter et al., 1944); three in rats (Owen et
al., 1987; Owen and Glaister, 1990; Bevan et al., 1996); three in mice (NTP, 1984, 1993; Bevan
et al., 1996); and an acute "sperm head morphology" study in B6C3FJ mice (Hackett et al.,
1988a). These studies are summarized in Table 5-1. In addition, several dominant lethal studies
have been published that are related to reproductive toxicity; these are reviewed in Chapter 4.
There are no data on reproductive or developmental effects in humans from exposure to 1,3-
butadiene.
5.1.1. Carpenter et al. (1944)
Four groups, each consisting of 24 albino rats, 12 guinea pigs, 4 rabbits, and 1 dog, were
exposed to 0, 600, 2,300, or 6,700 ppm 1,3-butadiene 7.5 h/day, 6 days/week for 8 months in
546-L chambers. Except for the dogs, which were all female (only one in each group), the
animals were divided equally between the two sexes. Terminal body weights in rats were
reduced to 90.5%, 86.3%, and 81.2% in the 600, 2,300, and 6,700 ppm groups, respectively,
relative to controls. A similar trend was noted for male guinea pigs, while body weights for dogs
and rabbits fluctuated. Microscopic lesions were observed solely in the liver, in which a mild,
cloudy swelling was noted in 68% of the animals exposed to 6,700 ppm. Few results regarding
fertility of rats, guinea pigs, and rabbits exposed to 1,3-butadiene were given in the report.
Fertility, defined as the number of litters produced within a given time, was reduced in rats, with
3.3, 2.7, 2.5, and 2.6 litters being produced by animals exposed to 0, 600, 2,300, or 6,700 ppm,
respectively. Because the results were not analyzed statistically and other details regarding the
duration of the mating periods were not presented, it is not possible to conclude that 1,3-
butadiene either did or did not have an effect on fertility in rats. Data presented on other species
also were inadequate to conclude whether 1,3-butadiene had any effect on fertility.
5.1.2. Owen et al. (1987); Owen and Glaister (1990)
This 2-year chronic bioassay study is the same as the Hazleton Laboratories Europe, Ltd.
(HLE, 1981b), study discussed previously by EPA (U.S. EPA, 1985). Male and female CD strain
(Sprague-Dawley derived) rats (110 of each sex per group) were exposed by inhalation to
1,3-butadiene (99.2% purity) at target concentrations of 0, 1,000, or 8,000 ppm, 6 h/day, 5
days/week for 105 (females) or 111 (males) weeks. The average weekly concentration of the 1,3-
butadiene dimer impurity (4-vinyl -1-cyclohexene) was 413 ± 219 ppm (v/v). Ten males and 10
5-1
-------
Table 5-1. A summary of the reproductive and developmental effects of 1,3-butadiene.
Species
Dose, route, and time of
exposure
Effects
LOAEL"
Reference
Reproductive Effects
Male and female rats, guinea
pigs, rabbits; female dogs
0, 600, 2,300, 6,700 ppm,
7.5 hr/day, 6 days/wk, 8 mos
I body wt in rats, guinea
pigs; mild cloudy swelling in
liver at 6,700 ppm in all
species; I fertility in rats
no statistical analysis
Carpenter etal., 1944
Male and female rats
0, 1,000, 8,000 ppm, 6h/day,
5 days/wk, 105 wks
(females), 111 wks (males)
I mortality with exposure;
benign and malignant
mammary tumors, uterine
sarcomas, thyroid follicular
cell tumors (females); Leydig
cell tumors, pancreatic
exocrine adenomas (males)
1,000 ppm for 105 or 111
wks
>8,000 ppm for non-
neoplastic lesions
Owen et al., 1987; Owen and
Glaister, 1990
Male and female B6C3FJ
mice
0, 625, 1,250 ppm, 6 hr/day,
5 days/wk, 60 or 61 wks
to
I mortality; numerous
neoplasms including
mammary gland; ovarian
atrophy, uterine atrophy,
(females); testicular atrophy
(males)
625 ppm for 60 or 61 wks
NTP, 1984
Male and female B6C3FJ
mice
0, 6.25, 20, 62.5, 200, 625
ppm, 6 hr/day, 5 days/wk,
103 wks (some killed at 9
and 15 mos)
I mortality (> 20 ppm);
ovarian atrophy, germinal
epithelial hyperplasia,
angiectasis, granulosa cell
tumors, uterine atrophy,
mammary tumors (females);
testicular atrophy (males)
6.25 ppm for 103 wks
62.5 ppm for 15 mos
200 ppm for 9 mos
NTP, 1993
Male and female rats and
mice
0, 1,000 ppm, 6 hr/day, 5
days/wk, 13 wks
no effects on reproductive
system (rats); ovarian
atrophy (female mice);
testicular atrophy (male
mice)
I,000ppmbforl3 wks in
mice
Bevanetal., 1996
-------
Table 5-1. A summary of the reproductive and developmental effects of 1,3-butadiene (continued).
Species
Dose, route, and time of
exposure
Effects
LOAEL"
Reference
Reproductive Effects (continued)
Male B6C3FJ mice
(sperm head
morphology study)
Male 102/E1XC3H/E1 mice
(flow cytometric analysis of
spermatogonial cells)
0, 200, 1,000, 5,000
ppm, 6 hr/day, 5 days;
killed at 5 wks after
exposure
0, 130, 500, 1,300 ppm, 6
hr/day, 5 days; killed 21 days
after exposure and later
1 % epididymal sperm,
t sperm head
abnormalities
I testis weight, I round and
elongated spermatids
1,000 ppm for 5 days
130 ppm for 5 days
Hackett et al., 1988a
Pacchierotti et al., 1998b
Developmental Effects
Pregnant CD rats
Pregnant CD rats
Pregnant CD-I mice
0, 200, 1,000, 8,000 ppm, 6
hr/day, GD 6-15; killed GD
20
0, 40, 200, 1,000 ppm, 6
hr/day, GD 6-15
0, 40, 200, 1,000 ppm, 6
hr/day, GD 6-15
t maternal toxicity, I fetal
body wt and CPX, t skeletal
and eye defects
t maternal toxicity, no fetal
effects
t maternal toxicity, I fetal
wt, t skeletal variants
1,000 ppm for 10 days
1,000 ppm for 10 days
40 ppm for 10 days
IISRP, 1982
Hackett etal., 1987a
Hackett etal., 1987b
'Lowest-observed-adverse-effect-level
' Frank effects on ovarian and testicular atrophy at this exposure level, not a LOAEL.
-------
females were killed at 52 weeks. A comprehensive postmortem examination, including necropsy
and histopathologic examination, was conducted on all gross lesions, all tissues from control and
high-exposure groups, and selected tissues from low-exposure groups. Nonneoplastic lesions
were not induced in reproductive organs in either male or female rats, although benign and
malignant mammary tumors, uterine sarcomas, and thyroid follicular cell tumors were seen in
females, while Leydig cell tumors and pancreatic exocrine adenomas were observed in males.
5.1.3. NTP(1984)
The first inhalation chronic bioassay study conducted by the National Toxicology
Program (NTP, 1984) showed that, in addition to the numerous neoplasms induced by high
concentrations of 1,3-butadiene in male and female B6C3FJ mice, nonneoplastic lesions also
were induced in reproductive organs. Male and female mice were exposed to 0, 625, or 1,250
ppm 1,3-butadiene 6 h/day, 5 days/week and then killed after 60 or 61 weeks of exposure.
Among female mice, ovarian atrophy was seen in 40/45 (89%) mice exposed to 625 ppm and in
40/48 (83%) mice exposed to 1,250 ppm, compared with an incidence of only 2/49 (4%) in
control mice. Involution of the uterus, which was considered a manifestation of ovarian atrophy,
was seen in 7/46 (15%) and 14/49 (29%) mice exposed to 625 and 1,250 ppm, respectively,
compared with 0/49 control mice. Uterine involution was characterized by fewer and less
prominent endometrial glands. A low incidence of mammary gland neoplasms (acinar cell and
adenosquamous carcinomas) was induced by 1,3-butadiene; nonneoplastic mammary lesions
were not induced. Testicular atrophy was observed in 19/47 (40%) mice exposed to 625 ppm
and in 11/48 (23%) mice exposed to 1,250 ppm compared with 0/50 control mice. Statistical
analysis showed that the increased incidences of the lesions in male and female mice were
significant (p<0.05) for all groups compared with their respective controls.
5.1.4. NTP (1993)
NTP (1993) conducted a second inhalation chronic bioassay study in male and female
B6C3FJ mice exposed to lower concentrations of 1,3-butadiene. Concentrations were 0, 6.25,
20, 62.5, 200, or 625 ppm 1,3-butadiene for 6 h/day, 5 days/week for 103 weeks, with interim
evaluations at 9 and 15 months. Additional male mice were exposed to 200 ppm of 1,3-
butadiene for 40 weeks or 625 ppm for 13 weeks (8,000 ppnrweeks), or 312 ppm for 52 weeks
or 625 ppm for 26 weeks (16,000 ppnrweeks), followed by observation for the remainder of the
2 years (stop-exposure protocol). It should be emphasized that this study was designed to study
neoplastic and general lexicological, rather than reproductive, endpoints. Further details are
presented in Chapter 6.
5-4
-------
The effects of 1,3-butadiene on reproductive organs in female mice are presented in Table
5-2. Ovarian atrophy was seen in the 200 ppm and 625 ppm exposure groups sacrificed for the
9-month interim evaluation. The atrophic ovaries were characterized by the absence of oocytes,
follicles, and corpora lutea. No occurrences of this lesion were noted in the lower exposure
groups. Hyperplasia of the germinal epithelium was observed in one animal exposed to 625 ppm
for 9 months. Germinal epithelial hyperplasia was described as prominent down-growth of the
mesothelial surface into the parenchyma of the ovary, forming tubular and glandlike structures.
At the 15-month interim evaluation, ovarian atrophy was observed in mice exposed to 20 ppm or
higher, with a significant increase in the incidence at 62.5 ppm or higher compared with
concurrent controls. Hyperplasia of the germinal epithelium was seen at 200 and 625 ppm but
was not significant. Angiectasis (dilation of blood vessels) was seen in one mouse in the control
group, one exposed to 6.25 ppm, and two exposed to 200 ppm. The ovary, which was evaluated
at 15 months in only two female mice exposed to 625 ppm, was atrophic in both. Among female
mice exposed to 1,3-butadiene for 2 years, ovarian atrophy was observed in all exposure groups
at incidences that were significantly elevated compared with controls. Therefore, using ovarian
atrophy as an endpoint of reproductive toxicity, a no-observed-adverse-effect level (NOAEL)
could not be defined in this mouse study. Although the functional integrity of the female
reproductive system was not assessed, it can be assumed that animals without oocytes or follicles
would be infertile and would express reduced estrogenic and progestin secretory capacities.
Uterine atrophy was seen at the two highest concentrations at 9 months, but was seen only
at the highest concentration at the 15-month evaluation. After 2 years, the incidence of uterine
atrophy among mice exposed to 200 and 625 ppm did not increase relative to that observed at 9
months.
Data regarding the effect of 1,3-butadiene on the reproductive organs of male B6C3FJ
mice are summarized in Table 5-3. The testes of a majority of males exposed to the highest
concentration of 1,3-butadiene (625 ppm) were atrophic at the 9- and 15-month interim
evaluations and at termination of the 2-year study. In the stop-exposure study, testicular atrophy
was observed after 2 years in 5/50 mice exposed to 200 ppm 1,3-butadiene for 40 weeks and 3/50
exposed to 625 ppm for 13 weeks (8,000 ppnrweeks), and in 3/50 mice exposed to 312 ppm for
52 weeks and 5/50 exposed to 625 ppm for 26 weeks (16,000 ppnrweeks). The lack of a more
prominent response in mice exposed to 625 ppm for 26 weeks may have been due to insufficient
time for induction of testicular atrophy or, if atrophy was induced during exposure, the possibility
of lesion repair following termination of the exposure and before the examination at 2 years.
5-5
-------
Table 5-2. Reproductive tract lesions in female B6C3Ft mice exposed to 1,3-butadiene by inhalation.
Lesion
Concentration (ppm)
0
6.25
20
62.5
200
625
9-month interim evaluation
Ovary3
Atrophy
Germinal
epithelial
hyperplasia
(NOS)
Uterus3
Atrophyd
10
0 (0%)b
0 (0%)
10
0 (0%)
—
—
—
—
—
—
—
—
10
0 (0%)
0 (0%)
10
0 (0%)
10
9 (90%)c
0 (0%)
10
3 (30%)
8
8 (100%)c
1 (13%)
8
6 (75%)
15-month interim evaluation
Ovary3
Atrophy
Germinal
epithelial
hyperplasia
Angiectasis
Uterus3
Atrophyd
10
0 (0%)
0 (0%)
1 (10%)
10
0 (0%)
10
0 (0%)
0 (0%)
1 (10%)
1
0 (0%)
10
1 (10%)
0 (0%)
0 (0%)
10
0 (0%)
10
9 (90%)c
0 (0%)
0 (0%)
10
0 (0%)
10
7 (70%)c
3 (30%)
2 (20%)
10
0 (0%)
2
2 (100%)e
1 (50%)
0 (0%)
2
2 (100%)
2-year studyf
Ovary3
Atrophy
Germinal
epithelial
hyperplasia
49
4 (8%)
/K0.001
2 (4%)
/K0.001
49
19 (39%)
/K0.001
3 (6%)
p=OA60
48
32 (67%)
/K0.001
8 (17%)
p=Q.on
50
42 (84%)
/K0.001
15 (30%)
/K0.001
50
43 (86%)
/K0.001
15 (30%)
/7=0.010
79
69 (87%)
/K0.001
18 (23%)
/K0.001
-------
Table 5-2. Reproductive tract lesions in female B6C3Ft mice exposed to 1,3-butadiene by inhalation (continued).
Lesion
Concentration (ppm)
0
6.25
20
62.5
200
625
2-year studyf (continued)
Angiectasis
Uterus3
Atrophyd
4 (8%)
p=0.259
50
1 (2%)
6 (12%)
p=0366
49
0 (0%)
3 (6%)
^=0.606
50
1 (2%)
13 (26%)
^=0.017
49
1 (2%)
14 (28%)
p=Q.02l
50
8 (16%)
17 (22%)
p=OA25
78
41 (53%)
"Number of animals for which this site was examined microscopically.
bNumber of observations; percentage in parentheses.
°p<0.01, pairwise comparison with controls by Fisher's exact test.
dStatistical tests were not conducted for these lesions.
ep<0.05, pairwise comparison with controls by Fisher's exact test.
{p values for the statistical analysis (logistic regression test) for the 2-year study are presented; the value for the trend test is in the column for the control group,
and the value for pairwise comparisons of individual exposed group with the corresponding control group is in the column for the exposed groups.
Source: NTP, 1993.
-------
Table 5-3. Reproductive tract lesions in male B6C3Ft mice exposed to 1,3-butadiene by inhalation.
Lesion
Concentration (ppm)
0
6.25
20
62.5
200
625
9-month interim evaluation
Testes3
Absolute
weight (g)
Relative
weight (mg/g
BW)C
Atrophyd
10
0.117±
0.002
2.89 ±0.06
0 (0%)
10
0.117±
0.003
2.92 ±0.09
e
10
0.114±
0.003
2.76 ±0.09
—
10
0.103±0.004b
2.87 ±0.12
—
10
0.102±0.002b
2.54±0.05b
0 (0%)
10
0.059 ±0.003b
1.57±0.03b
6 (60%)
15-month interim evaluation
Testes3
Absolute
weight (g)
Relative
weight (mg/g
BW)
Atrophyd
10
0.116±
0.003
2.62 ±0.07
0 (0%)
10
0.113±
0.003
2.79 ±0.08
—
10
0.104 ±
0.004
2.48 ± 0.04
0 (0%)
10
0.112 ±0.003
2.66 ±0.07
—
10
0.100±0.003b
2.39±0.05f
0 (0%)
7
0.071 ± 0.004b
1.80±0.05b
4 (57%)
oo
-------
Table 5-3. Reproductive tract lesions in male B6C3Ft mice exposed to 1,3-butadiene by inhalation (continued).
Lesion
Concentration (ppm)
0
6.25
20
62.5
200
625
2-year study
Testesa
Atrophyd
50
1 (2%)
50
3 (6%)
50
4 (8%)
48
2 (4%)
49
6 (12%)
72
53 (74%)
"Number of animals for which this site was examined.
b/><0.01, pairwise comparison with controls by Williams' or Dunnett's test.
CBW = body weight.
dStatistical tests were not conducted for these lesions.
eTestes were not examined microscopically at this concentration.
fp<0.05, pairwise comparison with controls by Williams' or Dunnett's test.
Source: NTP, 1993.
-------
5.1.5. Bevan et al. (1996)
In a study on the subchronic effects of 4-vinylcyclohexene in rats and mice, groups of male
and female rats and mice were exposed by inhalation to 1,000 ppm 1,3-butadiene for 6 hr/day, 5
days/wk for 13 weeks. Exposed male and female rats showed no effects on body weight or
weight gain, but there was an increase in absolute and relative liver weights, and male rats had
increased relative kidney weights with hyaline droplet accumulation in renal proximal tubules.
No other microscopic effects were seen, and there were no clinical chemistry or hematologic
effects in rats. In male and female mice exposed to 1,000 ppm 1,3-butadiene, there were no
effects on body weight or weight gain. There was a significant decrease in circulating
erythrocyte mass and an increase in reticulocyte counts in both males and females; male mice had
mild macrocytic anemia. The most remarkable finding was ovarian atrophy in 6/10 female mice.
Females also showed an increase in absolute and relative liver weight and a decrease in relative
spleen weight and splenic atrophy in 3/10 mice. Males showed reduced testicular weights and
testicular atrophy in 10/10 mice exposed to 1,000 ppm 1,3-butadiene. These data demonstrate
that ovarian and testicular atrophy can develop as early as 13 weeks after exposure to a high
concentration of 1,3-butadiene.
5.1.6. Hackett et al. (1988a)
This sperm-head morphology study was conducted in B6C3FJ mice at Pacific Northwest
Laboratories for NTP as part of a series of studies to investigate the effects of 1,3-butadiene on
reproductive function. Adult male B6C3FJ mice were exposed to 1,3-butadiene (99.88% purity;
174 ± 13 ppm mean headspace dimer [4-vinyl-l-cyclohexene] concentration) at concentrations of
0 (filtered air), 200, 1,000, or 5,000 ppm 6 h/day for 5 successive days. Measured concentrations
(mean ± standard deviation [SD]) were 199 ± 6.12, 999 ± 22.6, and 4,980 ± 130 ppm. Positive
controls received intraperitoneal injections of 167 mg/kg of ethyl methane sulfonate daily for 5
consecutive days. The mice were killed 5 weeks after exposure, weighed, and examined for
gross lesions, with particular emphasis on the reproductive tract. Sperm collected from the right
epididymis were examined for abnormal heads (blunt hook, banana, amorphous, pinhead, two
heads/two tails, short) and other abnormalities (primarily midpiece abnormalities).
Piloerection and dyspnea were observed within the first 20 to 30 min after exposure in
mice receiving 5,000 ppm; no clinical signs of toxicity were noted for the other groups. The
percentages of epididymal sperm with normal morphology were 98.08%, 97.23% (p<0.05), and
96.34% (p<0.05) at 200, 1,000, and 5,000 ppm, respectively, compared with 98.40% for controls;
these values also showed a significant exposure-related trend (p<0.05). The percentages of the
following abnormalities were significantly elevated compared with controls (/X0.05): blunt
hooks at 5,000 ppm, bananas at 1,000 and 5,000 ppm, and pinheads at 1,000 ppm. Amorphous,
5-10
-------
two heads/two tails, and shorts were not significantly elevated at any dose. The predominant
types of abnormalities were the banana followed by blunt hook and amorphous. The authors
speculated that late spermatogonia or early primary spermatocytes were sensitive to
1,3-butadiene. The authors also stated that examining the sperm at only one time point following
termination of exposure precluded a determination of the stage of spermatogenesis affected by
the chemical.
5.1.7. Pacchierotti et al. (1998b)
Reproductive effects of 1,3-butadiene were studied in male mice by flow cytometric
analysis of spermatogonial cells. Male (102/E1XC3H/E1) mice were exposed for 5 consecutive
days (6 h/day) to 0, 130, 500 or 1,300 ppm of 1,3-butadiene in air. Results of the flow cytometric
analysis of testis cell fractions at different butadiene concentrations and times after exposure
revealed that testis weight was significantly decreased at all doses tested 21 days after exposure
termination. At longer time intervals, progressive recovery was seen, which was dose-dependent.
Testis weight loss paralleled the sequential depletion of round and elongated spermatid
populations. These data, together with the dominant lethal studies (discussed in Chapter 4)
confirm the toxicity of 1,3-butadiene on late developmental stages of spermatozoa in mice.
5.2. DEVELOPMENTAL EFFECTS
The developmental toxicity study sponsored by the International Institute of Synthetic
Rubber Producers (IISRP, 1982) is the same as the Hazleton study discussed briefly in the 1985
EPA document (U.S. EPA, 1985). Hackett and co-workers also conducted two developmental
toxicity studies, one using rats (Hackett et al., 1987a) and one using mice (Hackett et al., 1987b).
The study using rats was conducted to confirm and extend the findings of the IISRP (1982) study
in rats, and the mouse study was conducted for comparison of a rodent species more sensitive
than the rat to the toxic effects of 1,3-butadiene. The acute and subchronic effects of 1,3-
butadiene inhalation exposure in male mice on fetal abnormalities have already been alluded to
in the dominant lethal studies of Anderson et al. (1993, 1998), Adler and Anderson (1994),
Pacchierotti et al. (1998a), and Brinkworth et al. (1998).
5.2.1. IISRP (1982)
Pregnant Sprague-Dawley CD rats were exposed by inhalation to 1,3-butadiene at target
concentrations of 0, 200, 1,000, or 8,000 ppm 6 h/day on gestation day (gd) 6 to 15 and killed on
gd 20. Measured concentrations (mean ± SD) were 2.8 ± 1.2, 202 ± 14, 990 ± 24, and 7,647 ±
375 ppm for 0, 200, 1,000, and 8,000 ppm, respectively. The purity of the 1,3-butadiene was not
5-11
-------
reported; the mean concentration of the dimer impurity, 4-vinyl-l-cyclohexene, was 108.6 ±
53.59 ppm, well below the target of 300 ppm.
Maternal body weight gain was markedly depressed in dams exposed to 1,000 and 8,000
ppm, especially during gd 6 to 9; a significant decrease was also noted during gd 9 to 12 in rats
exposed to 8,000 ppm. During the later stages (gd 12 to 15 and 16 to 20), body weight gain was
similar to controls. The gravid uterus and extragestational weights were similar to controls, but
extragestational weight gain was significantly depressed by 17% (p<0.05) in dams exposed to
1,000 ppm and by 24% in dams exposed to 8,000 ppm (p<0.05). No effects were observed on
other measures of maternal toxicity.
Fetal body weight and crown/rump length were significantly reduced at 8,000 ppm
(p<0.05). The percentage of fetuses with major skeletal defects was significantly elevated at
1,000 and 8,000 ppm and minor skeletal defects were significantly elevated only at the lowest
concentration. The percentage of fetuses showing minor external/visceral defects, predominantly
subcutaneous hematomas, was significantly elevated only at 1,000 ppm, but the percentage was
similar in all three experimental groups. The incidence of bilateral lens opacity was elevated at
all concentrations but was significantly elevated only at 8,000 ppm. The incidence of marked-to-
severe wavy ribs and the total number of abnormal ossifications and irregular ossification of the
ribs were elevated at 8,000 ppm. The incidence of thoracic bipartite centers was elevated in all
exposed groups; a dose-response relationship was not observed. Other malformations and
variations occurred at incidences similar to those of controls or were not significantly elevated
compared with controls.
5.2.2. Hackett et al. (1987a)
Pregnant rats were exposed to 0, 40, 200, or 1,000 ppm 1,3-butadiene (99.84% purity; 197
± 6 ppm mean headspace dimer concentration) via inhalation on gd 6 to 15 (6 h/day). The
measured concentrations (mean ± SD) were 40.1 ± 0.62 (mean ± SD), 199.8 ± 2.61, and 1,005 ±
11.9 ppm, respectively. No clinical signs of toxicity were observed. Final body weights were
similar to those of controls; body weight gain, however, was reduced by about 30% (p<0.05) in
the 1,000 ppm group during the first 5 days of exposure (gd 6 to 11). From gd 11 to 20, body
weight gain was not significantly different from that of controls. The gravid uterine weights and
extragestational weights (whole-body weight minus gravid uterine weight) were similar to those
of controls, but extragestational weight gain was significantly lower (16%;/?<0.05) in dams
exposed to 1,000 ppm than in control dams.
The overall pregnancy rates were similar among all groups, ranging from 80% among dams
exposed to 40 ppm to 93% among controls and dams exposed to 1,000 ppm (Table 5-4). Fetal
measures, including the numbers of implantations/dam, resorptions/litter, dead
5-12
-------
Table 5-4. Maternal toxicity in Sprague-Dawley CD rats exposed to 1,3-butadiene by
inhalation.
Parameter
No. dams assigned
No. of deaths
No. pregnant (%)
Whole-body weight
(g)
DayO
Day 20
Body weight gain
(g)
Days 0-6
Days 6- 11
Days 11-16
Days 16-20
Gravid uterine
weight (g)
Extragestational
weight0 (g)
Extragestational
weight gaind(g)
Significant clinical
signs
Concentration (ppm)
0
30
0
28 (93)
242±3.7a
362 ±7.1
21.4 ± 1.6
25. 5 ± 1.3
29.2 ± 1.4
44.5 ± 1.8
73.0 ±2.9
289 ±5.7
47.6 ±2.8
None
reported
40
30
0
24 (80)
239 ±3.2
351 ±5. 9
21. 1± 1.6
23.6 ± 1.3
30.9 ± 1.7
36.7 ±2.5
69.5 ±3. 5
282 ±3. 9
42.7 ±2.2
None reported
200
30
0
26 (87)
244 ±3.0
369 ±6.6
22.9 ± 1.3
26.6 ± 1.5
31.7 ± 1.9
43.6 ±2.3
73.9 ±2.8
295 ±5. 8
50.9 ±3.0
None reported
1,000
30
0
28 (93)
242 ± 4.0
354 ±6.1
20.1 ± 1.5
17.5 ± 1.9b
31.2±2.1
43.2 ±2.9
71.2±4.1
283 ±3. 5
39.9 ± 3. 5b
None reported
aMean ± standard error.
b/><0.05, compared with corresponding control.
°Body weight on gd 20 minus gravid uterine weight.
dExtragestational weight minus body weight on gd 0.
Source: Hackett et al., 1987a.
5-13
-------
fetuses/litter, fetal body weights, sex ratios, malformations, and variations, were not affected by
exposure to 1,3-butadiene. Overall, no developmental toxicity was observed in rats exposed to
40 to 1,000 ppm during gd 6 to 15.
5.2.3. Hackett et al. (1987b)
Pregnant CD-I mice were exposed to 0, 40, 200, or 1,000 ppm 1,3-butadiene (99.88%
purity; 338 ± 72 ppm mean headspace dimer concentration), 6 h/day on gd 6 to 15. Measured
concentrations were 39.9 ± 0.06, 199.8 ± 3.0, and 1,000 ±13.1 ppm (mean ± SD).
The effects of 1,3-butadiene on maternal toxicity in CD-I mice are summarized in Table
5-5. Three animals exposed to 1,000 ppm showed signs of dehydration: two died on gd 15, and
early parturition occurred in the third. No other clinical signs of toxicity were observed.
Exposure-related decreases in whole-body weights on gd 18, body weight gain during gd 11 to
16, gravid uterine weight, extragestational weight, and extragestational weight gain were
significantly reduced in the 1,000 ppm exposure group compared with controls. Whole-body
weight gain during gd 11 to 16 and extragestational weight gain were also reduced in the 200
ppm exposure group. None of these parameters were significantly affected in dams exposed to
40 ppm. The pregnancy rates in mice were uniformly low in all groups and unaffected by
exposure to 1,3-butadiene.
The effects of 1,3-butadiene on various parameters of developmental toxicity in CD-I
mice are summarized in Tables 5-6 and 5-7. More resorptions per litter were observed among
control dams than among exposed dams. Fetal body weights were reduced in all exposed groups
compared with controls, and the reduction showed a significant exposure-related trend. The
overall fetal body weights (males and females combined) were reduced by 4.5% at 40 ppm (not
significant), 15.7% at 200 ppm (p<0.05), and 22.4% at 1,000 ppm (p<0.05) (Table 5-6).
Significant differences from controls were seen at all treatment concentrations for fetal males and
at the two higher concentrations for fetal females. Placental weights showed an effect similar to
that of fetal body weights. Malformations occurred sporadically and at low frequencies in all
exposure groups; increases were seen in several skeletal variations (Table 5-7). The frequency of
supernumerary ribs was greatly elevated at 200 and 1,000 ppm; 6% of the fetuses/litter were
affected at 200 ppm (p<0.05) and 9.9% at 1,000 ppm (p<0.05) compared with 1.7% in controls
and 1.6% in the 40 ppm exposure group (not significant). There also was a marked increase in
the total number of fetuses with supernumerary ribs at the 200 and 1,000 ppm exposure levels. A
number of these were of normal length as opposed to those considered rudimentary ribs or
ossification sites at lumbar 1. The frequency of reduced ossification of the sternebrae was
elevated at 200 (p<0.05) and 1,000 ppm (/X0.001) (Fisher exact test); the litter incidence was
elevated but not significantly. The percentages of reduced ossifications at all sites and the
5-14
-------
Table 5-5. Maternal toxicity in pregnant CD-I mice exposed to 1,3-butadiene by
inhalation.
Parameter
No. dams assigned
No. of deaths
No. pregnant (%)
Whole-body weight (g)
DayO
Day 18
Body weight gain (g)
Days 0-6
Days 6- 11
Days 11-16
Days 16-18
Gravid uterine weight (g)
Extragestational weightd
(g)
Extragestational weight
gain6
(g)
Significant clinical signs
Concentration (ppm)
0
32
0
18(56)
28.4±0.25a
54.9 ± 1.21b
2.7 ±0.3
5. 5 ±0.4
13.3±0.6b
5.5±0.3b
19.3 ± 1.00b
35.5±0.48b
7.60±0.48b
None
40
33
0
19 (57)
28.3 ±0.32
55.4 ± 1.09
3.0 ±0.3
5. 8 ±0.3
12.7 ±0.4
5.7 ±0.3
20.3 ±0.80
35.1 ±0.44
6.99 ±0.38
None
200
31
0
21 (68)
28.2 ±0.32
52.5 ± 1.01
2.5 ±0.2
5.6 ±0.3
11.4±0.5C
4.7 ±0.4
18.0 ±0.87
34.5 ±0.46
6.20±0.38C
None
1,000
33
O
22 (67)
28.4 ±0.32
50.8±0.86C
2.3 ±0.2
4.8 ±0.3
10.6±0.4C
4.8 ±0.3
16.8±0.67C
34.1±0.36C
5.91±0.28C
Dehydration
aMean ± standard error.
bp<0.05, significant linear trend.
c»<0.05, pairwise comparison with corresponding control parameter.
Body weight on gd 18 minus gravid uterine weight.
eExtragestational weight minus body weight on gd 0.
Source: Hackett et al, 1987b.
5-15
-------
Table 5-6. Developmental toxicity in CD-I mice exposed to 1,3-butadiene by
inhalation.
Parameter
No. pregnant (%)
No. litters with live
fetuses
No. implantations/dam
No. resorptions/litter
Early resorptions
Dead fetuses/litter
No. fetuses/no, litters
examined
Fetal body weight (g)
Females
Males
Placental weight (mg)
Females
Males
Sex ratio (% males)
Concentration (ppm)
0
18(56)
18
12.7 ±0.52
1.06 ±0.22
1.00 ±0.23
0
11.7 ±0.66
1.34±0.03b
1.30±0.03b
1.38±0.03b
86.8±2.99b
83.1±3.03b
89.3±3.03b
51.6±3.91
40
19 (57)
19
13.3 ±0.44
0.84 ±0.22
0.58 ±0.21
0
12.5 ±0.52
1.28 ±0.01
1.25 ±0.01
1.31±0.02a
85.4 ±2.29
80.9 ±2.46
89.5 ±2.27
49.8 ±3. 06
200
21 (68)
21
13.0 ±0.64
0.67 ±0.20
0.43±0.13a
0
12.3 ±0.62
1.13±0.02a
1.10±0.02a
1.13±0.02a
78.6±3.24a
74.7±3.52a
80.1±2.35a
51. 5 ±3.68
1,000
22 (67)
20
13.1 ±0.43
0.90 ±0.19
0.75 ±0.16
0
12.2±0.51
1.04±0.03a
1.06±0.02a
1.06±0.02a
72.6 ± 1.88a
70.1±2.33a
74.5 ± 1.81"
51. 8 ±3.29
a/><0.05, pairwise comparison with corresponding control.
b/><0.05, significant linear trend.
Source: Hackett et al, 1987b.
5-16
-------
Table 5-7. Malformations and variations in CD-I mice exposed to 1,3-butadiene by
inhalation.
Parameter
No. fetuses/no, litters examined
No. fetal heads examined
Malformations2
Exencephalus
Open eye
Limb flexure
Fused sternebrae
Fused ribs
Variations
Pale
Hydroureter
Abnormal stemebraec>d
Misaligned sternebrae
Ossification site between
sternebrae 5 and 6
Supernumerary ribsc-d
Supernumerary ribs
(total number)
Normal length
Rudimentary
Ossification site at lumbar 1
Reduced ossification (all sites)0
Skull
Sternebrae
Vertebrae (centra)
Phalanges
Concentration (ppm)
0
211/18
105
1/1
1/1
2/1
—
—
2/2
2/2
0.6 ±0.9
10/6
—
1.7 ±2.3
30/11
6/5
13/6
11/5
1.7 ±1.7
—
31/13
—
—
40
237/19
120
b
—
—
—
2/2
—
6/3
0.4 ±0.7
3/3
1/1
1.6 ±2.1
30/9
5/1
19/8
6/4
1.2±1.5
—
20/9
1/1
—
200
259/21
130
—
—
—
—
—
—
—
0.4 ±0.8
9/8
1/1
6.0±3.6e
127/20
29/9
81/20
17/10
2.7 ±2.7
2/2
57/1 6f
—
—
1,000
244/20
120
2/2
1/1
—
2/2
—
—
—
0.8±1.3e
10/8
3/3
9.9±3.0e
198/20
68/10
120/16
10/7
3.9±2.6e
3/1
76/1 9f
1/1
2/16
Expressed as number of fetuses/number of litters; includes only those findings occurring in more than one fetus
or at more than one concentration.
b—, no malformations or variations noted.
°Mean percentage per litter (mean ± SD).
dp<0.05, linear trend, orthogonal contrast test.
ep<0.05, Tukey's test.
fp<0.05, Fisher exact test (fetal incidence).
Source: Hackett et al., 1987b.
5-17
-------
percentages of abnormal sternebrae (misaligned, scrambled, or cleft) per litter were also
significantly elevated at 1,000 ppm (p<0.05). The percentages of supernumerary ribs and
abnormal sternebrae also showed significant linear trends.
These studies show that inhalation exposure to 1,3-butadiene causes maternal toxicity,
manifested as reduced body weight gain, in the mouse at 200 and 1,000 ppm; therefore, the
NOAEL for maternal toxicity is 40 ppm. 1,3-Butadiene also caused developmental effects,
manifested by reduced fetal body weight and increased frequency of skeletal variations at 200
and 1,000 ppm. In addition, inhalation exposure to 1,3-butadiene during gestation caused a
significant reduction in body weight of male fetuses at 40 ppm. Therefore, a NOAEL for
developmental toxicity in CD-I mice could not be obtained. Although 1,3-butadiene did not
induce a detectable increase in major malformations in the mouse fetus, the dose-related
increases in supernumerary ribs and reduced ossifications, particularly of the sternebrae, indicate
altered development and are a cause for concern.
5.3. STRUCTURE-ACTIVITY RELATIONSHIPS
A number of studies using 1,3-butadiene metabolites or chemically related agents have
contributed to our understanding of the species susceptibility and reproductive organ toxicity of
1,3-butadiene. Data on structure-activity relationships and structural analogs of 1,3-butadiene are
summarized in Table 5-8.
Studies using the mono- and diepoxide metabolites of 1,3-butadiene in mice and rats have
contributed to a better understanding of species differences and of the responsible agent in
ovarian toxicity (Doerr et al., 1995; 1996). Doerr et al. (1995) administered l,2-epoxy-3-butene
(EB) (1.43 mM/kg), l,2:3,4-diepoxybutane (DEB) (0.14 mM/kg), or vehicle (sesame oil) to
immature (28-day-old) female B6C3FJ mice ip for 30 days. Follicle counts were reduced from
controls as follows: EB caused a 98% reduction of small and 87% reduction of growing follicles,
while, at a 10-fold lower dose, DEB caused an 85% reduction of small and 65% reduction of
growing follicles. In a second study by Doerr et al. (1996), EB (0, 0.005, 0.02, 0.09, 0.36, or
1.43 mM/kg), DEB (0, 0.002, 0.009, 0.036, 0.14, or 0.29 mM/kg), or vehicle (sesame oil) was
administered intraperitoneally once daily to immature female B6C3FJ mice and Sprague-Dawley
rats for 30 days. At the high dose, EB resulted in reduced ovarian and uterine weights (p<0.05)
and decreased follicular counts in mice; rats, however, were unaffected. DEB caused decreased
ovarian weights (p<0.05) in mice and rats at >0.14 mM/kg and decreased uterine weights
(p<0.05) in mice at >0.14 mM/kg and in rats at 0.29 mM/kg. In mice, ED50 values for EB were
0.29 and 0.40 mM/kg and for DEB were 0.1 and 0.14 mM/kg for small and growing follicles,
respectively. EB did not deplete the ovary of small or growing follicles in the rat, and significant
depletion with DEB occurred only at 0.14 mM/kg (30%-40% of the follicular population
5-18
-------
Table 5-8. A summary of reproductive and developmental toxicity effects of chemicals structurally similar to 1,3-
butadiene.
Chemical
l,2-Epoxy-3-
butene (EB)
1,2:3,4-
Diepoxybutane
(DEB)
l,2-Epoxy-3-
butene (EB)
1,2:3,4-
Diepoxybutane
(DEB)
Species
Female
B6C3F; mice
(28-day)
Female
B6C3F; mice
(28-day)
or
Female
Sprague-
Dawley rats
(28-day)
Female
B6C3F; mice
(28-day)
or
Female
Sprague-
Dawley rats
(28-day)
Dose, route, and time
of exposure
1.43 mM/kg intra-
peritoneally for 30 days
0.14 mM/kg intra-
peritoneally for 30 days
0.005, 0.02, 0.09, 0.36,
1.43 mM/kg
intraperitoneally once
daily for 30 days
0.002, 0.009, 0.036,
0.14, 0.29 mM/kg
intraperitoneally once
daily for 30 days
Effects
Significant I in small (98%) and growing
(87%) follicles
Significant I in small (85%) and growing
(63%) follicles
>0.36 mM/kg (31.0 mg/kg): decreased follicle
counts (ED50 = 0.29 and 0.40 mM/kg for
small and growing follicles)
1.43 mM/kg (100.1 mg/kg): decreased
ovarian and uterine weights
None
>0. 14 mM/kg (12. 05 mg/kg): decreased
ovarian and uterine weights; decreased
follicular counts (ED50 = 0.1 and 0.14 mM/kg
for small and growing follicles)
>0. 14 mM/kg (12. 05 mg/kg): decreased
ovarian weight, decreased follicular counts;
0.29 mM/kg (25 mg/kg): decreased uterine
weight, severe general toxicity precluded
complete evaluation
LOAEL
N/A
0.36 mM/kg
>1.43 mM/kg
0.14 mM/kg
0.14 mM/kg
Reference
Doerr et al.,
1995
Doerr et al.,
1996
VO
-------
Table 5-8. A summary of reproductive and developmental toxicity effects of chemicals structurally similar to 1,3-
butadiene (continued).
Chemical
4-Vinyl-
cyclohexene
4-Vinyl-
cyclohexene
Species
Male and
female F344
rats
Male and
female
B6C3F; mice
Male and
female
B6C3F; mice
Male and
female
Sprague-
Dawley rats
Dose, route, and time
of exposure
0, 50, 100, 200, 400,
800 mg/kg by gavage, 5
days/wk, 13 wks
0, 200, 400 mg/kg by
gavage, 5 days/wk, 105
wks
0, 75, 150, 300, 600,
1,200 mg/kg by gavage,
5 days/wk, 13 wks
0, 200, 400 mg/kg by
gavage, 5 days/wk, 105
wks
0, 50, 250, 1,000 ppm
by inhalation, 5
days/wk for 13 wks
0, 250, 1,000, 1,500
ppm by inhalation, 5
days/wk for 13 wks
Effects
Male and female rats and male mice
13 weeks: no reproductive effects
105 weeks: no reproductive effects
Female mice
13 weeks: 1,200 mg/kg, all had I primary and
mature Graafian follicles; ovaries in other
groups not examined
105 weeks: 200 and 400 mg/kg, mixed benign
tumors, hyperplasia, granulosa cell tumors,
and granulosa cell carcinomas in ovaries;
tubular cell hyperplasia
1,000 ppm: 5/10 females showed ovarian
atrophy
1,500 ppm: 2/10 females showed ovarian
atrophy
LOAEL
>800 mg/kg for 13 wks
>400 mg/kg for 105 wks
N/A
200 mg/kg for 105 wks
1,000 ppm for 13 wks
1,500 ppm for 13 wks
Reference
NTP, 1986
Bevanetal.,
1996
to
o
-------
Table 5-8. A summary of reproductive and developmental toxicity effects of chemicals structurally similar to 1,3-
butadiene (continued).
Chemical
4-Vinylcyclo-
hexene
4-Vinylcyclo-
hexene
Species
Female (28-
day) B6C3FJ
mice
Swiss (CD-I)
male and
female mice,
continuous
breeding
protocol
Dose, route, and time
of exposure
800 mg/kg/day intra-
peritoneally for 30
days; ovaries collected
at 30, 60, 120, 240, and
360 days from
beginning of treatment
100, 250, 500
mg/kg/day by gavage
for approximately 18
weeks
Effects
>30 days: reduced ovarian weight, reduced
number of small (11% of controls) and
growing (22% of controls) follicles that were
progressively decreased over time
>240 days: increased serum FSH levels
360 days: complete ovarian atrophy,
preneoplastic lesions, increased plasma
androstenedione concentrations, no estrous
cyclicity
Fl females: 500 ms/ks~reduced primordial
(33% decrease), growing (55% decrease), and
antral follicles (33% decrease); no effect on
other reproductive measures
Fl males: 500 mg/ks~reduced testicular
spermatid count (17% decrease); no effect on
other reproductive measures
LOAEL
800 mg/kg/day ip for 30
days
500 mg/kg/day
500 mg/kg/day
Reference
Hooser et al.,
1994
Grizzle et al.,
1994
to
-------
Table 5-8. A summary of reproductive and developmental toxicity effects of chemicals structurally similar to 1,3-
butadiene (continued).
Chemical
4-Vinyl-
cyclohexene
Isoprene
Ethylcyclo-
hexene
Ethylcyclo-
hexene mono-
epoxide
Vinylcyclo-
hexane
Vinylcyclo-
hexane mono-
epoxide
Cyclohexene
Cyclohexene
monoepoxide
Epoxy butane
Species
Female
B6C3F; mice
(28-day)
Dose, route, and time
of exposure
1.5 mM/kg intra-
peritoneally for 30 days
7.34 mM/kg intra-
peritoneally for 30 days
1.5 mM/kg intra-
peritoneally for 30 days
1.43 mM/kg intra-
peritoneally for 30 days
7.5 mM/kg intra-
peritoneally for 30 days
1.43 mM/kg intra-
peritoneally for 30 days
7.5 mM/kg intra-
peritoneally for 30 days
1.43 mM/kg intra-
peritoneally for 30 days
1.43 mM/kg intra-
peritoneally for 30 days
Effects
Significant I in small (86%-88%) and
growing (72-83%) follicles
Significant I in small (76%) and growing
(46%) follicles
Significant I in growing (37%) follicles
No effect
No effect
No effect
No effect
No effect
No effect
LOAEL
N/A
Reference
Doerr et al.,
1995
to
to
-------
Table 5-8. A summary of reproductive and developmental toxicity effects of chemicals structurally similar to 1,3-
butadiene (continued).
Chemical
4-Vinylcyclo-
hexene
4-Phenylcyclo-
hexene
Benzo[a]-pyrene
(+ control)
4-Vinylcyclo-
hexene
diepoxide
4-Vinyl-l-
cyclohexene
diepoxide
Species
Female
B6C3F; mice
(28-day)
Adult (58-day)
and immature
(28-day)
female Fischer
344 rats
Male and
female
B6C3F; mice
Male and
female Fischer
344 rats
Dose, route, and time
of exposure
6 mM/kg (650 mg/kg)
intraperitoneally for 30
days
3 or 6 mM/kg (475 or
950 mg/kg) intra-
peritoneally for 30 days
0.32 mM/kg (80
mg/kg) on 1st and 7th
days of dosing
80 mg/kg (0.57
mM/kg) intraperi-
toneally for 30 days
0, 2.5, 5, 10 mg/animal
topically 5 days/week
for 13 or 105 weeks
0.3.75,7.5, 15,30,60
mg/animal topically 5
days/week for 13
weeks; 0, 15, or 30
mg/animal for 105
weeks
Effects
Significant reduction in number of small and
growing follicles
None
Significant reduction in number of small and
growing follicles
Adults
Reduced uterine weight, reduced number of
regular estrous cycles; reduced number of
primordial (33% of control) and primary
(38%) follicles
Immature
Reduced number of primordial (19%),
primary (45%), and growing (54%) follicles;
Females
13 weeks: >5 mg/animal~ovarian atrophy; 10
mg/animal~uterine atrophy
105 weeks: >2.5 mg/animal~ovarian atrophy
Males
105 weeks: > 5 mg/animal~inflammation of
epididymis
13 weeks: None
105 weeks: None
LOAEL
N/A
N/A
5 mg/animal for 13 wks
2.5 mg/animal for 105
wks
5 mg/animal for 105 wks
None
Reference
Hooser et al.,
1993
Flaws etal.,
1994a
NTP, 1989
to
oo
-------
Table 5-8. A summary of reproductive and developmental toxicity effects of chemicals structurally similar to 1,3-
butadiene (continued).
Chemical
4-Vinylcyclo-
hexene
diepoxide
4-Vinylcyclo-
hexene
diepoxide
Species
Female (28-
day) B6C3FJ
mice
Female (28-
day) Fischer
344 rats
Female (28-
day) Fischer
344 rats
Dose, route, and time
of exposure
80 mg/kg intraperi-
toneally for 6, 8, 10, or
12 days
80 mg/kg intraperi-
toneally for 1 or 15
days
Effects
8 days: increased atretic primordial follicles;
12 days: increased atretic primary follicles,
significant loss of follicles (64%)
10 days: increased atretic primordial follicles;
12 days: increased atretic primary follicles,
significant loss of follicles (35%)
1 day: increased number of primary follicles,
decreased box expression
15 days: reduced number of primordial and
primary follicles, increased bax expression
LOAEL
80 mg/kg ip for 8 days
80 mg/kg ip for 10 days
N/A
Reference
Kaoetal., 1999
Borman et al.,
1999
fj\
to
-------
Table 5-8. A summary of reproductive and developmental toxicity effects of chemicals structurally similar to 1,3-
butadiene (continued).
Chemical
4-Vinylcyclo-
hexene
diepoxide
4-Vinylcyclo-
hexene
Vinylcyclo-
hexene 1,2-
monoepoxide
Isoprene
Species
Male (28-day)
B6C3F; mice
Male B6C3FJ
mice
Male F344
rats
Dose, route, and time
of exposure
320 mg/kg/day intra-
peritoneally for 5, 10,
15, 20, 25, or 30 days;
two additional groups
dosed for 30 days and
allowed to recover for
30 or 60 days
40, 80, 160, 320 mg/kg
intraperitoneally for 30
days
800 mg/kg intra-
peritoneally for 30 days
200 mg/kg intra-
peritoneally for 30 days
0, 70, 220, 700, 2,200,
7,000 ppm inhalation 6
h/day, 5 days/week, 13
or 26 weeks
Effects
>5 days: decreased testis weight
5 days: loss of Type I and B spermatogonia in
Stages II to VI and of preleptotene
spermatocyte in Stages VI to VIII
30 days: no germ cells except spermatogonial
stem cells in testis, reduced seminal vesicle
weight
>30 days recovery: 100% recovery in
seminiferous tubule
60 days recovery: spermatozoa in epididymis
>80 mg/kg: reduced testis weight
> 160 mg/kg: greatly reduced or absent
epididymal sperm, scattered empty
seminiferous tubules, some stages of
spermatids and spermatocytes absent
320 mg/kg: reduced seminal vesicle weight
None
None
13 and 26 weeks: 7,000 ppm-- testicular
atrophy
13 weeks: none
26 weeks: 7,000 ppm-- hyperplasia of
interstitial cells
LOAEL
320 mg/kg ip for 5 days
80 mg/kg ip for 30 days
7,000 ppm for 13 weeks
7,000 ppm for 26 weeks
Reference
Hooser et al.,
1995
Melnicketal.,
1994
to
-------
depleted). Severe toxicity occurred in the highest DEB dose group in rats, which precluded
complete evaluation. These results clearly show that mice are more sensitive than rats to the
ovotoxic effects of the mono- and diepoxides of 1,3-butadiene, and that DEB is the more potent
ovotoxicant in both species.
The 1,3-butadiene dimer, 4-vinylcyclohexene, has been tested in short- and long-term
toxicological and carcinogenicity studies in rats and mice. Male and female F344 rats and male
B6C3FJ mice dosed by gavage with 0, 200, or 400 mg/kg of 4-vinylcyclohexene in corn oil 5
days/week for 105 weeks failed to develop neoplastic or nonneoplastic lesions in the
reproductive organs (NTP, 1986). However, in female mice, the incidences of granulosa cell
neoplasms, mixed benign tumors, granulosa cell hyperplasia, and tubular cell hyperplasia were
increased, but no increase in the incidence of nonneoplastic lesions was reported. In the 13-week
study, none of the male or female rats (0, 50, 100, 200, 400, or 800 mg/kg) or the male mice (0,
75, 150, 300, 600, or 1,200 mg/kg) showed reproductive effects, but all female mice treated with
1,200 mg/kg had reduced numbers of primary and mature Graafian follicles whether they
survived until termination (5/10) or died before termination (5/10). Ovaries in other dose groups
were not examined for numbers of follicles.
Bevan et al. (1996) reported on a study in which mice and rats were exposed by
inhalation to 4-vinylcyclohexene for 6 h/day, 5 days/week for 13 weeks. Male and female
B6C3FJ mice were exposed to 0, 50, 250, or 1,000 ppm while rats were exposed to 0, 250, 1,000,
or 1,500 ppm. In mice, 5/10 females exposed to 1,000 ppm showed ovarian atrophy, and in rats
exposed to 1,500 ppm 2/10 females showed ovarian atrophy.
In a study to more carefully evaluate the ovarian toxicity of 4-vinylcyclohexene, Hooser
et al. (1994) reported on mice exposed by intraperitoneal injection to 800 mg/kg/day for 30 days.
Animals were killed at 30, 120, 240, and 360 days from the start of exposure and ovarian follicle
counts were performed. At 30 days, there was a significant decrease in follicle counts (30% of
control numbers), but not until 240 days were follicle counts reduced to <1% of control values.
At the latter time point, there was also an increase in serum follicle stimulating hormone (FSH),
indicating a lack of feedback inhibition by estradiol on the hypothalamus and pituitary. At 360
days, complete ovarian atrophy was reported, as well as a sustained increase in serum FSH
levels, and preneoplastic changes in the ovary (blood-filled cystic structures lined with flattened
cells in which there were small, irregularly shaped foci of hypertrophic cells).
The functional consequences of the germ-cell effects of 4-vinylcyclohexene were studied
by Grizzle et al. (1994) in mice exposed to 100, 250, or 500 mg/kg/day in a continuous breeding
study protocol. CD-I male and female mice were dosed with 4-vinylcyclohexene in corn oil by
gavage at doses of 0, 100, 250, or 500 mg/kg/day for 1 week, then cohabited in breeding pairs for
14 weeks and dosed throughout the mating period. Litters born were evaluated, then euthanized.
5-26
-------
Litters born after week 15 were reared until postnatal day (PND) day 21 with continuous
exposure to the dams, then control and high-dose Fl weanlings were raised and bred within dose
groups. Testicular spermatid count was decreased in Fl males by 17%, whereas treated females
had significantly reduced numbers of primordial (33% decrease), growing (55% decrease), and
antral follicles (33% decrease) after treatment with 500 mg/kg compared to controls. However,
no changes in testis weight, ovarian weight, or estrous cyclicity were seen, and no effect on
reproductive competence was detected.
Doerr et al. (1995) studied structure-activity relationships for the ovarian toxicity of 4-
vinylcyclohexene and several related olefms and their monoepoxides. 4-Vinylcyclohexene was
administered intraperitoneally to female B6C3FJ mice at 7.5 mmol/kg for 30 days as a positive
control. The analogues of vinylcyclohexene (ethylcyclohexene, vinylcyclohexane, cyclohexene:
all analogues with only a single unsaturated site) were similarly given at 7.5 mmol/kg, while their
monoepoxides were given at 1.43 mmol/kg (maximum tolerated doses). Following day 30, mice
were killed and the ovaries were removed and sectioned for histologic examination. Besides
vinylcyclohexene, only the analogue ethylcyclohexene caused a significant reduction in growing
(37%) but not small follicles, and not to the same extent as vinylcyclohexene (86%-88% for
small and 72%-83% for growing follicles). None of the monoepoxides of the vinylcyclohexene
analogues affected small or growing follicle counts. Additional groups of animals were given
isoprene at 7.34 mmol/kg and epoxybutane (EB analogue) at 1.43 mmol/kg, while EB was given
at 1.43 mmol/kg, and DEB at 0.14 mmol/kg. At a dose similar to that for vinylcyclohexene,
isoprene reduced small follicles by 76% and growing follicles by 46%. EB almost completely
depleted the population of both small and growing follicles at a dose more than fivefold below
that for vinylcyclohexene, whereas epoxybutane (a monoepoxide analogue of EB that cannot
form a diepoxide) did not affect follicle counts. DEB, at a dose tenfold lower than that for EB,
depleted 85% and 63% of small and growing follicles, respectively. These structure-activity data
indicate that the tested analogues of vinylcyclohexene that had only a single unsaturated site (and
their monoepoxide metabolites) are not ovotoxic, and that only those related compounds that can
be metabolized to a diepoxide or were a diepoxide were ovotoxic. In addition, the study showed
a relationship between chemical reactivity, as measured by nicotinamide alkylation, and
ovotoxicity; the diepoxide metabolites of vinylcyclohexene and DEB were approximately 3.5 to
10-fold more reactive than their monoepoxide precursors and other structurally related
monoepoxides.
Similarly, Hooser et al. (1993) studied the effect of substituting a phenyl group for the
vinyl group in 4-vinylcyclohexene. Mice were treated with either 4-vinylcyclohexene (6
mM/kg/day ip for 30 days) or 4-phenylcyclohexene (3 or 6 mM/kg/day ip for 30 days). A
positive control group was treated with benzo[a]-pyrene (0.32 mM/kg) on the first day of dosing
5-27
-------
and again 7 days later. Animals were killed shortly after exposure ended, and both 4-
vinylcyclohexene and benzo[a]-pyrene caused dramatic reductions in small and growing
follicles, but no treatment-related effects were seen with 4-phenylcyclohexene, possibly related to
its inability to form a mono- or diepoxide.
The reproductive toxicity of 4-vinylcyclohexene diepoxide has also been reported in
studies from several laboratories. When male and female F344 rats were treated dermally with 0,
3.75, 7.5, 15, 30, or 60 mg/animal or B6C3FJ male and female mice were treated dermally with
0, 2.5, 5, or 10 mg/animal of 4-vinyl-l-cyclohexene diepoxide 5 days/week for 13 weeks, the
only reproductive effects noted were diffuse ovarian atrophy in 10/10 female mice that received
10 mg/mouse and in 4/10 receiving 5 mg/mouse, and uterine atrophy in 2/10 female mice that
received 10 mg/mouse (NTP, 1989). In a 2-year study, male and female rats were exposed
dermally to 0, 15, or 30 mg/animal, and male and female mice were exposed to 0, 2.5, 5, or 10
mg/animal. No reproductive lesions were seen in rats, but ovarian atrophy occurred in almost all
mice treated with 2.5, 5, and 10 mg/mouse (43/49, 42/49, and 47/50, respectively, compared with
12/50 for controls). Ovarian atrophy was similar to that observed in animals exposed to 1,3-
butadiene and was characterized by a complete absence of follicles and corpora lutea. Tubular
hyperplasia also occurred at a high incidence in all dose groups (35/49, 38/49, and 34/50,
respectively, compared with 5/50 for controls). In addition, the combined incidence of luteomas,
granulosa cell tumors, benign mixed tumors, and malignant granulosa cell tumors was increased
in the mid- and high-dose groups (5 or 10 mg/animal, respectively). In male mice, subacute
inflammation of the epididymis occurred in 0/50, 6/50, and 13/49, respectively, compared with
0/50 for controls. These data clearly show that mice are more susceptible to 4-vinylcyclohexene-
induced reproductive toxicity than are rats regardless of the route of exposure (i.e., oral versus
dermal).
Several studies have investigated the effects of 4-vinylcyclohexene diepoxide on the
destruction of ovarian follicles. Flaws et al. (1994a) reported the effects of age at treatment of
female rats on the ovarian toxicity of 4-vinylcyclohexene diepoxide. Adult (58-day) and
immature (28-day) female Fischer 344 rats were treated with 80 mg/kg ip for 30 days. Adult rats
showed significantly reduced uterine weight, number of regular estrous cycles, and number of
primordial (33% of control) and primary (38% of control) follicles. In immature rats, the effects
were reduced primordial (19%), primary (45%), and growing (54%) follicles, with no other signs
of toxicity. These data indicate that 4-vinylcyclohexene causes ovarian toxicity in both adult and
immature animals, and that the effects on uterine weight may be indirect as a consequence of the
reduction in number of regular estrous cycles and estrogen production. Kao et al. (1999) reported
on the differential response to sensitive indicators of ovarian toxicity in immature female mice
and rats exposed to 4-vinylcyclohexene diepoxide (80 mg/kg ip for 6, 8, 10, or 12 days). An
5-28
-------
increase in atretic follicles was detected in mice after 8 days, and in rats after 10 days, whereas a
significant loss in number of follicles was detected in both species after 12 days. However, the
reduction in number of follicles in mice (64%) was greater than that in rats (35%). Finally,
Borman et al. (1999) reported that immature female Fischer 344 rats given a single ip dose of 80
mg/kg 4-vinylcyclohexene diepoxide showed an increase in the number of primary follicles along
with decreased expression of the cell death gene bax, in contrast to a group of animals treated
with the same dose for 15 days, which showed a reduced number of primordial and primary
follicles and increased bax expression. These data suggest that a single dose of 4-
vinylcyclohexene diepoxide may cause a reduction in apoptosis in favor of DNA repair, resulting
in an apparent "protective" effect.
Studies to look more carefully at the effects of 4-vinylcyclohexene and its epoxides on
male reproduction and fertility were reported by Hooser et al. (1995). Male (28 day) B6C3Fj
mice were treated with 320 mg/kg/day 4-vinylcyclohexene diepoxide for 5, 10, 15, 20, 25, or 30
days. Two additional groups were dosed for 30-days and allowed to recover for 30 or 60 days.
Reductions in testis weight were noted after 5 days of exposure, as well as loss of Type I and B
spermatogonia in Stages II to VI and of preleptotene spermatocytes in Stages VI to VIII. After 30
days there were no germ cells except spermatogonial stem cells in the testis, and reduced seminal
vesicle weight. However, after a 30-day recovery period, there was 100% recovery of germ cells
in the seminiferous tubules, and after 60 days, spermatozoa populated the epididymis. A dose-
response study (40, 80, 160, 320 mg/kg/day ip for 30 days) showed that reduced testis weight
could be detected at 80 mg/kg; at >160 mg/kg, there were greatly reduced or absent epididymal
sperm and scattered empty seminiferous tubules, with some stages of spermatids and
spermatocytes absent. At 320 mg/kg, seminal vesicle weight was also reduced. Treatment with
4-vinylcyclohexene (800 mg/kg ip for 30 days) or vinylcyclohexene-l,2-monoepoxide (200
mg/kg ip for 30 days) showed no effects on male reproductive organs.
Finally, differences in the susceptibility and reproductive effects in male rats and mice
were seen in inhalation studies with isoprene, the 2-methyl analogue of 1,3-butadiene (Melnick et
al., 1994). Male F344 rats and B6C3Fj mice were exposed to 0, 70, 220, 700, 2,200, and 7,000
ppm isoprene, 6 h/day, 5 days/week for either 13 or 26 weeks followed by a 26-week recovery
period. After 13 weeks of exposure, no effects were observed in rats at any concentration, but
testicular atrophy occurred in mice at 7,000 ppm. Following 26 weeks of exposure, all treated
rats in the 7,000 ppm group had hyperplasia of the interstitial cells of the testis (p<0.01; 10/10
vs. 1/10 controls); however, following the 26-week recovery, there was only a marginal increase
(not significant) in benign testicular tumors: 9/30 compared with 3/30 for controls. Mice had an
increase in the incidence of testicular atrophy following 26 weeks of exposure to 7,000 ppm
5-29
-------
(p<0.05; 5/10 vs. 0/10 controls). After 26 weeks of recovery, mice had a slight increase (not
significant) in testicular atrophy at 7,000 ppm (3/29 compared with 0/29 for controls).
5.4. PLAUSIBLE MECHANISMS OF REPRODUCTIVE TOXICITY
Differences in some aspects of reproduction between test species and humans contribute
to uncertainty in assigning human health reproductive risk for 1,3-butadiene. One of the
limitations in using laboratory rodent models for correlating age-related changes in human
reproductive physiology is the difference in the period of lifetime estrous/menstrual cycling,
being much shorter in rodents (6 to 8 months duration) than in humans (30 years). This fact
makes applicability of long-term exposures to presumptive toxicants between rodents and
humans difficult (May and Finch, 1987-1988), but data on several agents in test animals and
humans showing ovotoxicity and reproductive failure (Hoyer and Sipes, 1996) suggest that
findings of estrous/menstrual irregularities, reductions in follicle numbers, and ovarian failure
should be considered adverse reproductive effects and assumed to indicate the potential for
reproductive toxicity in humans (Heindel, 1999). Thus, the questions raised by Christian (1996)
as to the rationale for concern in the observation of ovarian atrophy in aged animals have been
addressed.
In mice exposed to 1,3-butadiene for 2 years, there was dose-related atrophy in both ovary
and testis (NTP, 1993), with the ovary being the more sensitive target organ. In the case of
ovarian atrophy, there was a significant effect even at the lowest dose tested (6.25 ppm).
Because follicle counts were not conducted in this study, there is the possibility that significant
follicular loss might have been detected at even lower exposure levels and at much earlier times
in the study. In support of this possibility, Hooser et al. (1994) showed that mice exposed to 4-
vinylcyclohexene for 30 days (800 mg/kg/day, ip) and examined at 30, 120, 240, and 360 days
from the start of exposure had a significant decrease in follicle counts (30% of control numbers)
at the end of the 30-day treatment period. At 240 days after the beginning of the 30-day exposure
period, follicle counts were reduced to <1% of control values with a concomitant increase in
serum FSH. At 360 days, complete ovarian atrophy and increased serum FSH levels were found,
at which time preneoplastic changes were detected in the ovary. Interestingly, animals continued
to cycle until the 360-day time point. Thus, one of the first signs of ovarian atrophy may be the
reduction in follicle counts, followed much later by ovarian atrophy and tumor formation.
Further support of this series of events comes from a study by Kao et al. (1999) in which
immature mice and rats were exposed to 80 mg/kg ip of 4-vinylcyclohexene diepoxide. Increases
in atretric primordial follicles were seen in the ovaries as early as 8 days of exposure in mice and
10 days in rats, with reductions in primary follicles in both species after 12 days exposure. This
greater effect on primordial follicles appears to be related to the fact that 4-vinylcyclohexene
5-30
-------
diepoxide is less efficiently detoxified in primordial and preantral follicles than in large preantral
follices (Flaws et al., 1994b). In addition, 4-vinylcyclohexene diepoxide causes atresia by
promoting the process of apoptosis; because the normal process of follicular atresia occurs
primarily in the primordial and small growing follicles (Springer et al., 1996a and b), 4-
vinylcyclohexene (and likely 1,3-butadiene) primarily affects the primordial and small growing
follicles. Springer et al. (1996c) also showed that follicular viability may be affected by 4-
vinylcyclohexene diepoxide via an effect on granulosa cells. Flaws et al. (1994b) showed that rat
preantral follicles versus larger follicles had a reduced ability to detoxify 4-vinylcyclohexene
diepoxide, another possible reason for a greater effect on primordial and primary follicles than on
larger growing follicles. A recent study by Borman et al. (1999) reported a so-called protective
effect of a single day of exposure to vinylcyclohexene diepoxide (80 mg/kg ip) on immature rat
primary ovarian follicles. It is unclear what the mechanism underlying this effect might be,
except that expression of the cell death gene bax was decreased, suggesting a suppression of
apoptosis in favor of DNA damage repair.
There are multiple pacemakers of reproductive senescence, including the increasing de-
synchronization of the temporal order of neuroendocrine signals of the central nervous system
and the exhaustion of ovarian follicles (Nelson et al., 1995; Wise et al., 1997). Menopause
marks the permanent end of fertility in women. If 1,3-butadiene acts on the human ovary as it
does in mice, then it is possible that the exocrine function (ovulation of oocytes) could be
adversely affected, leading to infertility and premature menopause. As a result the endocrine
function would also be affected. There is evidence that the postmenopausal ovary retains some
important endocrine functions despite follicular loss (Chakravarti et al., 1976; Judd et al., 1976;
Vermeulen, 1976; Schachter et al., 1979; Aiman et al., 1986; Nagamani et al., 1986; Hughes et
al., 1991). Thus, if 1,3-butadiene acts on the endocrine function of the ovary, causing a
premature attenuation of steroidogenesis, then the beneficial effects of estradiol are lost, resulting
in atrophy of the sensitive tissues, decreased calcium absorption, increased bone resorption,
accelerated bone loss and osteoporosis, a rise in serum triglycerides, increased VLDL and LDL
lipoproteins, and an increased LDL/HDL cholesterol ratio, a profile that favors atherosclerosis
(Vagenakis, 1989). Clearly, there is a research need to determine human ovarian sensitivity to
1,3-butadiene and its metabolites.
Another potential concern worth discussing here is the relationship between antecedent
ovarian hypoplasia and atrophy, followed by hyperplasia and subsequent ovarian neoplasia.
Capen and co-workers (1995) have shown that agents that either destroy or diminish the numbers
of Graafian follicles in the ovary result in decreased sex hormone secretion (especially estradiol)
and lead to compensatory overproduction of pituitary gonadotropins FSH and luteinizing
hormone (LH), which may place the mouse ovary at an increase risk to develop tumors via a
5-31
-------
hormone-mediated mechanism. This hypothesized link is further discussed by Hoyer and Sipes
(1996) and supported by the work of Hooser et al. (1994) showing the time course of effects on
follicular atrophy, loss of feedback control of FSH secretion via the hypothalamo-pituitary axis,
and subsequent observation of preneoplastic lesions in the ovary. Ovarian toxicity and
carcinogenicity has been documented for at least eight chemicals recently tested in National
Toxicity Program prechronic and chronic rodent studies. The chemicals that yielded
treatment-related ovarian lesions, pertinent here, were 1,3-butadiene, 4-vinylcyclohexene, and
vinylcyclohexene diepoxide (Maronpot, 1987). Whether these chemicals and conditions have
similar effects in women is unknown at this time.
The active metabolite for inducing ovarian atrophy has been clearly shown to be the
diepoxide metabolite of 1,3-butadiene (Doerr et al., 1995, 1996). This requirement for
epoxidation and activation to the diepoxide form is supported by these and other studies on 1,3-
butadiene and its structural analogues as well as their mono- and diepoxide metabolites (Melnick
et al., 1994; Doerr et al., 1995). Flaws et al. (1994a) and Kao et al. (1999) have shown that rats,
which are resistant to the ovotoxicity of 1,3-butadiene and 4-vinylcyclohexene, are affected by
treatment with the diepoxide metabolite of 4-vinylcyclohexene. Thus, while there are definite
species differences in effects seen between rats and mice with respect to ovarian toxicity from the
parent compound, there are much less striking differences in sensitivity to the diepoxide
metabolite.
The uterine atrophy observed in some of the toxicology studies (e.g., NTP, 1984, 1993) is
presumed to be due to reduced estrogen stimulation of the uterus by the atrophic ovary, rather
than a direct effect of the toxicant. Since the early work of Astwood (1938), it has been well
known that ovarian estrogens have a trophic effect on the uterus. Recently, it has been shown
that estrogen agonists promote a three- to fourfold increase in uterine wet weight and vaginal
cornification in wild-type females, whereas estrogen receptor knockout mice are totally
insensitive (Korach et al., 1996). Such studies clearly provide evidence for the importance of
ovarian estrogens in maintaining reproductive tract tissues and organs. Although this theory is
biologically plausible and the likely explanation for the uterine atrophy observed in 1,3-
butadiene-exposed mice with depressed ovarian activity, additional research is needed to verify
this hypothesis.
Testicular atrophy, as reflected by reduced testis weight following 1,3-butadiene exposure
for 9, 15, or 24 months in male mice (NTP, 1993), indicates gonadal sensitivity in the male as
well as in the female. However, the ovary is more sensitive than the testis because ovarian
atrophy occurs at very low concentrations (6.25 ppm) of 1,3-butadiene compared with that seen
in males after 2 years of exposure (LOAEL = 200 ppm). The sperm-head morphology study
5-32
-------
showed that male mice are affected at concentrations > 1,000 ppm (Hackett et al., 1988a) and the
dominant lethal test showed that male mice may be affected at 200 and 1,000 ppm (Hackett et al.,
1988b), again indicating that higher exposure concentrations are necessary to induce toxic effects
in male mice than in female mice. Hooser et al. (1995) studied the effects of the diepoxide
metabolite of vinylcyclohexene (320 mg/kg/day ip) on male mice killed after 5, 10, 15, 25, or 30
days of exposure. Decreases in testis weight began at 5 days and there was loss of Type I and B
spermatogonia in Stages II to VI and of preleptotene spermatocytes in Stages VI to VIII.
Following a 30-day recovery period, the seminiferous tubules were 100% repopulated, and after
60 days epididymal spermatozoa were present. Thus, the effects on the male germ cells appear to
be reversible whereas those in the female are not. Chabbra et al. (1990) had earlier shown that
male rats are resistant to vinylcyclohexene diepoxide, whereas females in both rats and mice are
affected. The reason for these gender and species differences in sensitivity needs further
investigation.
The differential intercalation of toxic chemicals with chromosomal macromolecules of
oocytes and spermatocytes may be a possible reason for gender differences. The diffuse state of
chromosomes of oocytes, allowing a more efficient intercalation between base pairs and potential
alkylation, is different from the dense chromosomal compaction of maturing late spermatocytes.
This explanation has been suggested and hypothesized as a mechanism for differential
susceptibility in mice for a number of alkylating agents (Sudman et al., 1992). Because both
male and female mice are adversely affected by 1,3-butadiene, albeit females more so, the
exposure effect may be attributed to a dose gradient. That is, less dose is needed for intercalation
of diffuse chromosomal macromolecules in oocytes than in late spermatocytes. It is of interest
that a diepoxybutane-adenine adduct has been detected in Chinese hamster ovarian cells (Leuratti
et al., 1994) and several guanosine adducts have been detected following butadiene monoxide
exposure (Selzer and Elfarra, 1996). The reader is referred to the chapter on mutagenicity
(Chapter 4) for further discussion of these issues. Whether this possible condition/mechanism
holds for women exposed to 1,3-butadiene as it might for rodent species is not known. Several
studies indicate that 1,3-butadiene affects spermatozoa and spermatids via a mutagenic effect, as
determined by postimplantation deaths during the first 3 weeks after exposure (see Chapter 4).
The relationship between germ-cell toxicity and functional capacity has been evaluated by
Grizzle et al. (1994), who exposed mice by gavage to 4-vinylcyclohexene (100, 250, or 500
mg/kg/day) in a continuous breeding protocol. A marginal reduction in testicular effects was
seen at 500 mg/kg, with a highly significant reduction in ovarian follicles at the same dose level,
but no effect on other measures of reproductive toxicity, fertility, or fecundity. Given the data
from Hooser et al. (1994) discussed above, this is not surprising because a >90% reduction in
ovarian follicle count was not associated with a change in FSH levels or estrous cycling at 120
5-33
-------
days after the beginning of a 30-day treatment. Complete ovarian atrophy did not occur until 240
days from the beginning of treatment, with a concomitant increase in FSH levels. These data
also support the gender difference in sensitivity between males and females, as testicular
morphology was only marginally affected while ovarian toxicity was significantly increased. In
addition, Hooser et al. (1995) showed reduced testicular weight and loss of germ cells in male
mice treated with 4-vinylcyclohexene diepoxide (320 mg/kg ip) for 5 days or 80 mg/kg for 30
days, an effect that is fully reversible; in contrast, a dose of 80 mg/kg ip to female mice and rats
for 12 days can lead to a permanent and irreversible loss of ovarian follicles (Kao et al., 1999).
5.5. SUMMARY AND CONCLUSIONS
Evidence has been presented showing that 1,3-butadiene induces reproductive and
developmental effects in rodents. The three long-term toxicity studies conducted in Sprague-
Dawley CD rats (Owen et al., 1987; Owen and Glaister, 1990) and E6C3Fl mice (NTP, 1984,
1993) as well as the subchronic toxicity studies (NTP, 1993; Bevan et al., 1996) suggest that
mice are much more sensitive than rats to the reproductive effects of 1,3-butadiene.
Nonneoplastic reproductive toxicity was not observed in either male or female rats exposed by
inhalation to 1,3-butadiene at concentrations up to 8,000 ppm for 2 years. Ovarian atrophy was
seen in mice at all exposure levels after 2 years of exposure (NTP, 1993), a time at which ovarian
atrophy is expected to appear in aged animals because of normal senescence mechanisms.
However, ovarian atrophy following 1,3-butadiene exposure occurred in a dose-related fashion,
and at significantly greater incidence in exposed mice versus controls. For example, ovarian
atrophy was observed as early as 13 weeks in female mice exposed to 1,000 ppm (Bevan et al.,
1996), after 9 months exposure to 200 and 625 ppm, after 15 months exposure to 62.5 ppm and
above, and after 2 years exposure to 6.25 ppm and above (NTP, 1993). Therefore, the dose-
response relationship observed for ovarian atrophy and the significant increase at the lowest dose
relative to that seen in control animals of a similar age is evidence for a causal relationship
between ovarian atrophy and exposure to 1,3-butadiene at 6.25 ppm.
Only a small amount of data is available regarding the reproductive or developmental
effects of the metabolites of 1,3-butadiene, EB, and DEB (Doerr et al., 1995, 1996). These
studies showed that mice are more sensitive than rats to the ovotoxic effects (follicular loss) of
EB and DEB, and that DEB is the more potent ovotoxicant in both species. Ovarian atrophy was
also observed in mice after exposure to the 1,3-butadiene dimer 4-vinylcyclohexene, adminis-
tered by inhalation for 6 h/day, 5 days/week for 13 weeks at 1,000 ppm (Bevan et al., 1996), or
by gavage for 5 days/week for 13 weeks at 1,200 mg/kg (NTP, 1986), or to its diepoxide, 4-
vinyl-1-cyclohexene diepoxide, administered by topical application (10 mg/animal) for 13 weeks
or 2 years (NTP, 1989). Ovarian atrophy induced by 1,3-butadiene, 4-vinylcyclohexene, or 4-
5-34
-------
vinyl-1-cyclohexene diepoxide is characterized by the absence of oocytes, follicles, and corpora
lutea. The functional integrity of the reproductive system in animals exposed to 1,3-butadiene
has not been tested, but the severity of the ovarian lesion is indicative of reproductive
dysfunction. Several studies provide a body of evidence showing that the earliest ovarian effect
that can be detected is DNA degradation indicative of apoptosis in small follicles (Springer et al.,
1996a), followed by atresia of follicles (Kao et al., 1999), with subsequent loss of follicles and
eventual ovarian atrophy, leading to early reproductive senescence and possibly increased risk for
ovarian tumor formation (Hooser et al., 1994; Hoyer and Sipes, 1996).
As for the functional consequences of the ovarian effects, Grizzle et al. (1994) showed
that exposure of mice to 4-vinylcyclohexene in a continuous breeding study resulted in a
significant reduction in ovarian follicle counts and a slight decrease in spermatid count, but did
not show alterations of reproductive competence in Fl animals. This is not surprising because
reproduction in mice is extremely robust, so that cohabitation and mating for longer periods of
time to the point of complete ovarian atrophy may be necessary to see the impact of such
significant follicular reductions. Rats administered 4-vinylcyclohexene orally or 4-vinyl-l-
cyclohexene diepoxide dermally did not develop ovarian atrophy, although 2/10 developed
ovarian atrophy when exposed by inhalation to 1,500 ppm (Bevan et al., 1996). These data
demonstrate a species-specific response similar to that after exposure to 1,3-butadiene. Several
other studies in which the vinylcyclohexene diepoxide metabolite was administered to rats
clearly show the sensitivity of the rat ovary to the diepoxide metabolite of 4-vinylcyclohexene
(Flaws et al., 1994a; Springer et al., 1996a; Kao et al., 1999) when administered
intraperitoneally.
The gender differences in response to 1,3-butadiene are also supported by studies on
structural analogues. For example, Melnick et al. (1994) reported testicular atrophy in male mice
exposed by inhalation to 7,000 ppm isoprene for 13 or 26 weeks, and no effect was seen in rats
exposed to the same concentration at 13 weeks, with some hyperplasia of the interstitial cells
evident at 26 weeks. Hooser et al. (1995) explored the sensitivity of male mice to 4-
vinylcyclohexene diepoxide and showed reduced testicular weight and a loss of germ cells after 5
days exposure to 320 mg/kg ip, with complete loss of germ cells except for spermatogonial stem
cells after dosing for 30 days. This effect was completely reversible after a 30-60 day recovery
period, whereas the ovarian toxicity reported in the female at 80 mg/kg ip for 12 days was
permanent and irreversible.
Developmental effects observed after exposure to 1,3-butadiene consisted primarily of
reduced fetal body weight and minor skeletal defects such as abnormal ossifications, abnormal
sternebrae, and supernumerary ribs. The species-specific response for developmental effects
induced by 1,3-butadiene was similar to that for reproductive effects, with mice showing greater
5-35
-------
sensitivity than rats. This difference was also seen for maternal toxicity as manifested by
decreased weight gain (whole-body and extragestational) in both rats and mice. The NOAEL for
maternal effects is 200 ppm for rats (IISRP, 1982; Hackett et al., 1987a) and 40 ppm for mice
(Hackett et al., 1987b). Even though the HSR? (1982) study showed increased frequencies for
bipartite thoracic centers and minor skeletal defects combined at 200 ppm in rats, the response
was not clearly dose-related. Several developmental effects occurred at significantly increased
frequencies at 8,000 ppm and showed a dose-response relationship: major skeletal defects
combined, wavy ribs, and abnormal ossification of the ribs (IISRP, 1982). The results from the
IISRP (1982) study were not confirmed in the study by Hackett et al. (1987a), which showed no
developmental toxicity in the same rat strain similarly exposed to 1,3-butadiene at concentrations
up to 1,000 ppm. Therefore, the NOAEL for developmental effects in rats is 1,000 ppm (IISRP,
1982; Hackett et al., 1987a). A NOAEL for developmental effects could not be defined for the
mouse, because male fetal body weight was decreased at 40 ppm in the Hackett et al. (1987b)
study. The mechanism for the developmental effects of 1,3-butadiene or its congeners has not
been explored.
Reproductive and developmental toxicity studies, therefore, show species specificity for
exposure to 1,3-butadiene, with mice being much more sensitive than rats. Pharmacokinetic
studies show that the uptake of 1,3-butadiene and its activation to epoxide metabolites are greater
in mice than in rats (see Chapter 3). These data also indicate that the availability of 1,3-
butadiene, EB, and DEB is greater in mice than in rats at comparable exposure concentrations.
Although the pharmacokinetic data do not fully explain the species differences, they suggest that
the basis of the difference in sensitivity may be related to the greater availability of toxic
intermediates of 1,3-butadiene metabolism in mice.
A few studies provide pharmacokinetic data specifically for reproductive organs. Nose-
only exposure of mice and rats to l,3-[14C]-butadiene resulted in greater or similar concentrations
of radioactivity (expressed as nM/g of tissue) in tissues of rats than those of mice under condi-
tions in which the rats were exposed to a tenfold higher concentration of 1,3-butadiene (Bond et
al., 1987). If tissue uptake was expressed as 1,3-butadiene equivalents/|_iM inhaled 1,3-buta-
diene, however, radioactivity levels were 15 to 100 times higher in mice. Mammary tissue,
which had 4.6-fold higher concentration in rats than in mice, was the only tissue analyzed that
was relevant to evaluating reproductive effects of 1,3-butadiene. Thornton-Manning et al. (1997)
exposed female Sprague-Dawley rats and female B6C3FJ mice to 62.5 ppm 1,3-butadiene by
inhalation for 6 hours (single exposure) or 6 hours/day for 10 days (repeated exposure) and
measured EB and DEB in several tissues, including mammary (EB and DEB) and ovary (DEB
only). In mammary tissue, EB levels were about 12 times greater in the mouse than in the rat
following the single exposure and about 4 times greater following the repeated exposures,
5-36
-------
whereas DEB levels were about 24 times greater following the single exposure and about 13
times greater following the repeated exposures. In addition, EB levels in mammary tissue were
about 5 times greater in the rat and about 1.7 times greater in the mouse following repeated
exposures than following the single exposure, whereas DEB levels in mammary tissue were
about 1.4 times greater in the rat and about 1.4 times lower in the mouse following repeated
exposures than following the single exposure. In the ovary, DEB levels were about 28 times
greater in the mouse than in the rat following the single exposure and about 15 times greater
following the repeated exposures. DEB levels in the ovary were about 1.7 times greater in the rat
and about 1.1 times lower in the mouse following repeated exposures than following the single
exposure. In an in vitro study, Sharer et al. (1992) showed that microsomes from the testes of
rats and mice are ineffective in forming EB, but the cytosol fraction was very effective in
forming glutathione conjugates. Keller et al. (1997) showed that ovarian microsomes from rats
and mice were unable to form mono- or diepoxides of vinylcyclohexene. Therefore, it is unlikely
that toxic effects on the testes or ovaries are due to metabolites formed within those tissues;
rather, they are due to metabolites formed elsewhere.
The link among ovarian atrophy, increased FSH levels, and formation of tumors in the
ovary is intriguing. Hoyer and Sipes (1996) reviewed the data related to this sequence of events
and postulated that preneoplastic and neoplastic changes may result from the loss of follicles in
the ovary and disruption of the feedback loop for endocrine regulation at the level of the
hypothalamus and/or pituitary, resulting in increased serum FSH levels which act as a promotor
of a genotoxic effect on the ovary. The data of Hooser et al. (1994) support this hypothesis,
showing that the long-term sequence of events can occur after as little as 30 days of exposure of
mice to the vinylcyclohexene diepoxide, which initiates accelerated follicular loss, eventual
ovarian atrophy, altered hormonal balance, and preneoplastic changes.
In conclusion, the animal data suggest that there is a potential reproductive hazard for
permanent and irreversible ovarian toxicity in women exposed by inhalation to 1,3-butadiene.
There is also a potential risk for reproductive hazards to the testes and germ cells in men, but the
effects appear to be reversible, because spermatogonial stem cells are not lost. Studies of 1,3-
butadiene, vinylcyclohexene, their epoxide metabolites, and structurally related compounds
provide strong evidence that the ovarian toxicity in mice is attributable specifically to the
diepoxide metabolites. Thus, although there is a potential hazard for ovarian toxicity in women
exposed to 1,3-butadiene, the risk would be expected to be quantitatively less than for mice
because pharmacokinetic data suggest that humans form less DEB than do mice (see Chapters 3
and 10). The animal data also show that there is a potential for developmental effects in humans
upon in utero exposure to 1,3-butadiene, and that these effects may occur at concentrations below
those causing maternal effects.
5-37
-------
6. TOXICITY IN ANIMALS
This chapter updates the evaluation of toxicity observed in experimental animals,
reviewing subchronic and chronic studies published from 1985 through December 1998, with the
exception of reproductive and developmental effects, which are reviewed in Chapter 5. The
available data from acute exposure studies suggest that 1,3-butadiene has low acute toxicity.
Data on acute toxicity mainly predate 1985 and are summarized by Himmelstein et al. (1997);
they are not reviewed in this document. On the other hand, a review of the chronic rat bioassay
conducted by Hazleton Laboratories (1981b) is included, even though the study was already
reviewed by U.S. EPA (1985), because it is the major rat cancer study.
6.1. SUBCHRONIC TOXICITY
Irons et al. (1986a,b) conducted studies to assess the potential of 1,3-butadiene to induce
myelotoxicity by exposing male B6C3FJ mice and male NIH Swiss mice to 1250 ppm 1,3-
butadiene, 6 h/day, 5 days/week for 6 weeks. Treatment-related hematological changes included
decreases in red blood cell counts, total hemoglobin, and hematocrit, and increases in mean cell
volume and circulating micronuclei in both strains of mice. The observed anemia was not
accompanied by significant alterations in mean corpuscular hemoglobin concentration, increases
in reticulocyte counts, or increases in the frequency of nucleated erythrocytes in peripheral blood.
These hematologic changes were considered to represent a macrocytic-megaloblastic anemia,
because they were accompanied by mild megaloblastic changes in bone marrow cells.
Exposure of male B6C3FJ mice to 1,250 ppm 1,3-butadiene for 6 h/day, 5 days/week, for
6 or 12 weeks did not produce any persistent effects on humoral or cell-mediated immunity
(Thurmond et al., 1986). Relative thymus weights were unaffected, but relative spleen weights
were decreased 20% and spleen cellularity was decreased 29% in exposed mice. Extramedullary
hematopoiesis and erythroid hyperplasia in exposed mice correlated with a twofold increase in
thymidine incorporation into spontaneously proliferating splenocytes. Although the number of
IgM antibody plaque-forming cells (PFC) per 106 splenocytes was unchanged, a 30% decrease in
PFC/spleen was observed. Proliferation of alloantigens was similar for 1,3-butadiene-exposed
splenocytes and controls. The mitogenic response of mature T lymphocytes to
phytohemagglutinin was significantly suppressed after exposure to 1,3-butadiene for 6 or 12
weeks.
6-1
-------
6.2. CHRONIC TOXICITY
6.2.1. Rat Study
In a long-term inhalation bioassay in rats (HLE, 1981b; Owen et al., 1987; Owen and
Glaister, 1990), groups of 110 male and 110 female Sprague-Dawley rats were exposed to 0,
1,000, and 8,000 ppm of 1,3-butadiene 6 h/day, 5 days/week, for 105 weeks (females) and 111
weeks (males). The low exposure level (1,000 ppm) was based on the threshold limit value
(TLV) for occupationally exposed personnel, while the high exposure (8,000 ppm) was limited
by safety requirements to be below the lower explosive limit of 1,3-butadiene in air. Ten males
and 10 females from all groups were killed after 52 weeks for interim clinical and histopathology
investigations.
Statistically significant reductions in survival occurred in both sexes at the highest
exposure concentration (p<0.05 in males, p<0.01 in females). Increased mortality in female rats
resulted from the sacrifice of animals with large subcutaneous masses, whereas in male rats,
renal lesions were the likely major cause of the increased death rate. Following this exposure
there were no effects on hematology, blood chemistry, urine analyses, or neuromuscular function
that definitely could be associated with the treatment. The only treatment-associated
changes were in clinical conditions, suppression of body weight, and increases in the weight of
certain organs. Table 6-1 lists the relative weights of those organs that showed a treatment-
related increase in either sex at 1 or 2 years. The relative liver weights of both sexes at both
exposure levels were statistically significantly elevated with the exception of high-exposure
females at 1 year. There were no associated changes in the liver even at the ultrastructure level.
The absolute and relative kidney weights were statistically significantly increased in male rats
exposed to 8,000 ppm for 2 years. This was associated with severe nephrosis compared to the
controls (Table 6-2). The relative heart weights were increased in the same group, which may
have been attributable to blood pressure changes resulting from the developing kidney changes.
Statistically significant increases in lung and spleen weights also were noted in high-exposure
males at 2 years with increased focal metaplasia in the lungs.
6.2.2. Mouse Study
A 2-year chronic inhalation toxicity and carcinogenicity study on the effects of 1,3-
butadiene on B6C3FJ mice was conducted by NTP (1993). In this study, groups of 70 male and
70 female mice were exposed by inhalation 6 h/day, 5 days/week to 0, 6.25, 20, 62.5, or 200 ppm
1,3-butadiene for periods up to 103 weeks; groups of 90 male and 90 female mice were similarly
exposed to 625 ppm 1,3-butadiene, which was the lowest exposure level in the previous NTP
(1984) study. The additional animals in the 625-ppm exposure group were included because
6-2
-------
Table 6-1. Relative organ weights (g/100 g body weight) in Sprague-Dawley rats.
Organ
1 Year
Male
Organ
Kidneys
Liver
Heart
Lung
Spleen
Number
examined
1,000
ppm
104
105a
96
97
92
10
8,000
ppm
102
124b
100
93
92
10
weight (% Current control)
2 Year
Female
1,000
ppm
104
116a
103
105
100
10
8,000
ppm
102
110
103
108
93
10
Male
1,000
ppm
104
lllb
104
104
93
50
8,000
ppm
124b
125b
116a
114a
121a
32
Female
1,000
ppm
104
118b
97
106
107
32
8,000
ppm
102
121b
97
103
133
24
a Statistically significantp<0.05.
b Statistically significant/X0.01.
Source: Adapted from Owen etal., 1987.
Table 6-2. Incidence of nephropathy in male Sprague-Dawley rats .
Finding
Number examined
Nephropathy
None
Minimal
Slight
Moderate
Marked
Severe
0 ppm
100 (45)
13(4)
29 (12)
38(19)
10(7)
3(2)
7(1)
No. of animals affected
1,000 ppm
100 (50)
25(7)
32 (20)
27(16)
7(4)
3(2)
6(1)
8,000 ppm
100(31)
9(0)
11(1)
42 (17)
11(2)
14(9)
13(2)
a The figures in parentheses are the values for animals killed at the end of the study.
b A statistically significant exposure-related trend (p<0.05) for fatal nephropathy was established. "Fatal" was
defined for the statistical treatment by the Peto method of analysis.
Source: Adapted from Owen et al., 1987.
-------
high mortality rates, observed previously at this exposure concentration, might interfere with the
scheduled interim evaluations. Interim evaluations were conducted at 9 and 15 months.
Mean body weight gains of male and female mice exposed to 6.25-625 ppm 1,3-
butadiene for 103 weeks were similar to those of controls. However, concentration-related
decreases in survival were seen in male and female mice exposed to concentrations >20 ppm 1,3-
butadiene (Table 6-3, Figures 6-1 and 6-2), primarily because of the development of malignant
neoplasms. No female mice exposed to 200 or 625 ppm or male mice exposed to 625 ppm
survived to the end of the study. Statistical analysis for the probability of survival was estimated
using the Kaplan and Meyer (1958) procedure; Cox (1972) and Tarone's (1975) life table test
was used to identify concentration-related trends.
At the 9- and 15-month interim evaluations, no clinical findings other than those
associated with lesion development and moribundity were observed. Some statistically
significant organ weight changes were observed at interim evaluations in male and female mice
exposed to 1,3-butadiene concentrations >62.5 ppm. Effects related to toxicity to reproductive
organs are discussed in Chapter 5.
Hematological indices measured at the interim evaluations showed significant (p<0.05)
decreases in erythrocyte counts, hemoglobin concentration, and packed cell volume in male mice
exposed to >62.5 ppm and in female mice exposed to >200 ppm at 9 months. Mean cell volume
was significantly increased in male mice exposed to 625 ppm and in females exposed to >200
ppm at 9 months. A similar profile of hematological changes was observed in male and female
mice exposed to 625 ppm for 15 months. Increases in the percentage of erythrocytes with
Howell-Jolly body inclusions and mean cell hemoglobin were seen at 9 and 15 months. At the
15-month interim evaluation, males exposed to 625 ppm 1,3-butadiene had a significantly
increased mean platelet value, a finding that correlated with the development of neoplasms.
Because these hematological changes were not associated with increases in reticulocyte counts or
in frequency of polychromatic erythrocytes in peripheral blood, they were attributed to a partial
or poorly regenerative bone marrow response to decreased levels of circulating erythrocytes.
There were no significant changes in total serum enzyme activity of lactate dehydrogenase
(LDH) or creatine kinase in mice evaluated at 9 months. LDH values at the 15-month evaluation
were increased in males and females exposed to >200 ppm. At 625 ppm, LDH-1 and LDH-2
were decreased and LDH-5, the principal enzyme in skeletal muscle and liver, was increased.
Histopathological effects observed at the 9-month evaluations included bone marrow
atrophy (depletion of cells) in 50% of males and in 13% of females at the highest concentration
(625 ppm). The atrophy increased in severity from mild depletion of hematopoietic cells at 9
months to marked depletion in mice that died or were killed at or before 15 months. An
increased incidence of bone marrow hyperplasia and an increased incidence or severity of
6-4
-------
Table 6-3. Survival of male and female B6C3Ft mice exposed to 1,3-butadiene by
inhalation for 103 weeks.
Concentration (ppm)
0
6.25
20
62.5
200
625
Male
Animals initially in study
9-Month interim evaluation"
15 -Mo nth interim evaluation3
Natural deaths
Moribund kills
Accidental deaths"
Missing2
Animals surviving until study termination
Percent survival at end of study0
Mean survival(days)d
Survival analysis6
70
10
10
6
9
0
0
35
70
597
/X0.001
70
10
10
5
6
0
0
39
78
611
^=0.430N
70
10
10
11
15
0
0
24
49
575
^=0.044
70
10
10
12
15
0
1
22
46
558
p=0.02l
70
10
10
23
23
0
0
4b
8
502
/X0.001
90
10
7
39
33
1
0
0
0
280
pO.OOl
Female
Animals initially in study
9-Month interim evaluation"
15 -Mo nth interim evaluation"
Natural deaths
Moribund kills
Accidental deaths"
Animals surviving until study termination
Percent survival at end of study0
Mean survival (days)d
Survival analysis6
70
10
10
3
10
0
37
74
608
/X0.001
70
10
10
7
10
0
33
66
597
p=0.510
70
10
10
11
14
1
24
50
573
/?=0.013
70
10
10
8
31
0
11
23
548
/X0.001
70
10
10
12
37
1
0
0
441
/X0.001
90
8
2
33
46
1
0
0
320
/X0.001
a Censored from survival analyses.
b Includes one animal that died during the last week of the study.
c Kaplan-Meier determinations. Survival rates adjusted for interim evaluations, accidental deaths, and missing animals.
d Mean of all deaths (uncensored, censored, terminal sacrifice).
e The result of the life table trend test (Tarone, 1975) is in the control column, and the results of the life table pairwise
comparisons (Cox, 1972) with the controls are in the dosed columns. A negative trend or lower mortality in a dose group is
indicated by N.
Source: NTP, 1993.
6-5
-------
1.0
0.0
WEEKS ON STUDY
Figure 6-1. Kaplan-Meier survival curves for male B6C3Ft mice exposed to 1,3-butadiene
by inhalation for 103 weeks.
Source: NTP, 1993.
6-6
-------
>
>
a:
u
C/1
en
<
en
O
ce
Q.
FEMALE MICE
• COMTROL
O 8.25
20 PPM
D 62.5 PPM
• 200 PPM
O 625 PPM
0.0
SO 75
WEEKS ON STUDY
Figure 6-2. Kaplan-Meier survival curves for female B6C3Ft mice exposed to 1,3-
butadiene by inhalation for 103 weeks.
Source: NTP, 1993.
6-7
-------
hematopoiesis of the spleen, liver, and lung occurred in females exposed to the three highest
concentrations (>62.5 ppm). Thymic necrosis (atrophy) and decreased thymus weights were
seen at the 9-month evaluation in males and females exposed to 625 ppm. Thymic necrosis also
occurred in females exposed to 62.5 or 200 ppm.
In the 2-year study, bone marrow atrophy was recorded in 50% of males and 14% of
females exposed to 625 ppm.
The incidence of liver necrosis was increased at the higher exposure concentrations in
males and females, occurring in 8%, 10%, 16%, 27%, 29%, and 26% of males and in 4%, 4%,
14%, 10%, 38%, and 21% of females exposed to 0 (controls), 6.25, 20, 62.5, 200, and 625 ppm,
respectively. Centrilobular hepatocellular necrosis of the liver was seen in 4% and 8% of males
exposed to 62.5 and 625 ppm, respectively, and in 2%, 8%, and 9% of females exposed to 62.5,
200, and 625 ppm, respectively. Hepatocellular necrosis was not seen in any of the concurrent
controls. Liver necrosis with no particular lobular distribution was found primarily in animals
with malignant lymphoma and hemangiosarcoma; centrilobular hepatocellular necrosis was often
found in animals described as anemic and in animals with atrial thrombi.
Myocardial mineralization, a lesion of unknown pathogenesis, occurred with increased
frequency in both sexes at 625 ppm (males, 27%; females, 14%), but was not observed in
controls. A low incidence was observed at the lower concentrations. Myocardial mineralization
was also observed in a separate stop-exposure study in which male mice were exposed to 312
ppm 1,3-butadiene for 52 weeks or 625 ppm for 13 or 26 weeks, and observed for periods up to
103 weeks. The incidence of myocardial mineralization for these three exposure groups was
12%, 18%, and 28%, respectively. Details of the stop-exposure study are presented in Section
o o
J.J.
Minimal to mild olfactory epithelial atrophy occurred in females exposed to 625 ppm and
in males exposed to concentrations >20 ppm. However, the incidence in males exposed to 625
ppm was lower than that seen in females. The olfactory epithelial lesions were unilateral at the
lower concentrations and bilateral at the higher concentrations. The lesions were similar to those
seen in the NTP (1984) study, but osseous or cartilaginous metaplasia was not observed. The
investigators considered olfactory nasal atrophy a possibly compound-related lesion.
Compared with controls, mice exposed to 1,3-butadiene exhibited increased incidences of
proliferative lesions (hyperplasia) in several organs, including the heart, lungs, forestomach,
ovaries, mammary gland, and Harderian gland (Table 6-4). Hyperplasia of the endothelium
(cardiac blood vessels), alveolar epithelium, forestomach epithelium (focal), germinal epithelium
and granulosa cells of the ovaries, mammary gland, and Harderian gland were all considered
preneoplastic lesions. Other preneoplastic lesions identified in the 2-year study were
hepatocellular foci (basophilic, clear cell, mixed cell, and eosinophilic) in female mice exposed
6-8
-------
Table 6-4. Incidence of hyperplasia in male and female B6C3Ft mice exposed to 1,3-butadiene by inhalation for
103 weeks.
Organ/tissue
Heart, endothelium
Lung, alveolar epithelium
Forestomach, epithelium
Ovary, germinal epithelium
Ovary, granulosa cells
Mammary gland
Harderian gland
Sex
M
F
M
F
M
F
F
F
F
M
F
Concentration (p
0
0/50 (0%)
0/50 (0%)
2/50 (4%)
5/50 (10%)
4/50 (8%)
4/50 (8%)
2/49 (4%)
1/49 (2%)
2/50 (4%)
1/50 (2%)
1/50 (2%)
6.25
1/49 (2%)
2/50 (4%)
9/50(18%)
5/50 (10%)
3/50 (6%)
5/49 (10%)
3/49 (6%)
0/49 (0%)
0/50 (0%)
3/49 (6%)
5/49 (10%)
20
0/50 (0%)
1/50 (2%)
6/50 (12%)
3/50 (6%)
3/50 (6%)
4/47 (9%)
8/48 (17%)
2/48 (4%)
2/50 (4%)
4/50 (8%)
9/48(19%)
62.5
2/48 (4%)
4/49 (8%)
13/49(27%)
9/50(18%)
5/48 (10%)
7/48(15%)
15/50(30%)
3/50 (6%)
4/50 (8%)
6/47(13%)
4/49 (8%)
pm)
200
4/48 (8%)
5/50 (10%)
17/50 (34%)
1 1/50 (22%)
4/48(18%)
14/50 (28%)
15/50(30%)
3/50 (6%)
7/50 (14%)
8/47 (17%)
4/49 (8%)
625
5/73 (7%)
8/80 (10%)
12/73 (16%)
11/78(14%)
40/72 (56%)
47/79 (59%)
19/79(23%)
2/79 (3%)
2/80 (3%)
5/40(13%)
7/66(11%)
Source: NTP, 1993.
-------
to 1,3-butadiene. Hepatocellular foci were observed in 16% of controls and in 29%, 38%, 24%,
10%, and 5% of females exposed to 6.25, 20, 62.5, 200, and 625 ppm, respectively. Hyperplastic
lesions were also observed in separate studies with male B6C3FJ mice using variable exposure
and durations (stop-exposure experiments). Hyperplasia in these studies occurred primarily in
the endothelium (cardiac blood vessels), alveolar epithelium, forestomach epithelium, and
Harderian gland (Table 6-5).
Table 6-5. Incidence of hyperplasia in male B6C3Ft mice exposed to
1,3-butadiene by inhalation in the stop-exposure study.
Organ/tissue
Heart, endothelium
Lung, alveolar epithelium
Forestomach, epithelium
Harderian gland
Concentration (duration of exposure)
0 ppm
0/50 (0%)
2/50 (4%)
4/50 (8%)
1/50 (2%)
200 ppm
(40 weeks)
6/50 (12%)
18/50 (36%)
10/48 (21%)
4/48 (8%)
312 ppm
(52 weeks)
3/50 (6%)
14/50 (28%)
20/48 (42%)
6/48 (13%)
625 ppm
(13 weeks)
7/50 (14%)
10/50 (20%)
8/50 (16%)
3/42 (7%)
625 ppm
(26 weeks)
7/50 (14%)
11/50(22%)
15/50 (30%)
7/36 (19%)
Source: NTP, 1993.
6.3. CARCINOGENICITY
6.3.1. Rat Study
The details of the study design and toxicity data have been described in Section 6.2.1.
Under the conditions of this study, 1,3-butadiene was carcinogenic at multiple organ sites in both
male and female Sprague-Dawley rats by inhalation (Table 6-6). Rats were exposed to 0, 1,000,
and 8,000 ppm 1,3-butadiene, 6 h/day, 5 days/week, for 105 weeks (females) and 111 weeks
(males). There were increased incidences and exposure-response trends (p<0.05) in several
tumor types. In males, there were increased incidences of pancreatic exocrine adenomas and
testicular Leydig cell tumors. Four tumor sites in females showed treatment-related effects
including total mammary gland adenomas and carcinomas, thyroid follicular cell adenomas,
Zymbal gland carcinomas (significant trend), and uterine sarcomas (significant trend).
Furthermore, the average number of mammary gland fibroadenomas per female rat bearing a
mammary gland tumor was 1.38 in controls, 3.70 in the 1,000 ppm exposure group, and 3.33 in
the 8,000 ppm exposure group (Melnick and Huff, 1993). The occurrence of glial cell tumors in
exposed male rats may also be related to exposure because these neurological tumors are
uncommon in untreated Sprague-Dawley rats. The metabolism of 1,3-butadiene in Sprague-
6-10
-------
Table 6-6. Incidences of tumors in Sprague-Dawley rats exposed to
1,3-butadiene by inhalation.
Tissue and tumor
Pancreas
Exocrine adenoma
Uterus
Sarcoma
Zymbal gland
Adenoma
Carcinoma
Mammary gland
Benign
Malignant
Total
Thyroid
Follicular cell adenoma
Follicular cell
carcinoma
Brain glial cell
Testis
Leydig cell tumor
Total number of tumor-
bearing rats
Subtotal tumor-bearers
No tumors
Single tumors
Multiple tumors
Number of rats examined
0 ppm
3
-
1
0
0
1
1
3
1
1
0
84
16
26
58
100
Male
1,000 ppm
1
-
1
0
2
0
2
5
0
4
o
J
70
30
32
38
100
8,000 ppm
1Qa,b
-
1
1
0
0
0
1
0
5a
8a,b
87
13
30
57
100
0 ppm
2
1
0
0
32
18
50
0
0
-
-
97
3
41
56
100
Female
1,000 ppm
0
4
0
0
64b
15
79b
2
2
-
-
98
2
24
74
100
8,000 ppm
0
5a
0
4a
55a'b
26
8p,b
10a,b
1
-
-
94
6
17
77
100
a Increasing trend, /><0.05.
b Increased in comparison with chamber controls (0 ppm), p<0.05.
Source: Adapted from Owen et al., 1987, and Melnick and Huff, 1993.
6-11
-------
Dawley rats is saturated at concentrations between 1,000 and 2,000 ppm (Laib et al., 1990),
which may explain why the incidences of tumors of the brain, uterus, and mammary gland of rats
exposed to 8,000 ppm were not higher than those exposed to 1,000 ppm (Melnick and Huff,
1993).
6.3.2. Mouse Studies
The first NTP mouse inhalation study of 1,3-butadiene was terminated early because of
induction of fatal neoplasms (NTP, 1984); therefore, a second study (NTP, 1993) was conducted
to better characterize the exposure-response relationship for neoplasms and nonneoplastic lesions
induced in mice by exposure to 1,3-butadiene for 2 years. The concentrations ranged from
100-fold lower (6.25 ppm) up to the lowest concentration (625 ppm) used in the first study. In
addition, stop-exposure studies were conducted to assess the relationship between concentration
nd duration of exposure on the induction of neoplasms by 1,3-butadiene. Results of this study
have also been published by Melnick et al. (1990a-c) and Melnick and Huff (1992). Miller and
Boorman (1990) provided morphological descriptions of the neoplastic lesions induced in
B6C3FJ mice by 1,3-butadiene. The results are presented here in two parts: 2-year study and
stop-exposure study.
6.3.2.1. 2-Year Study (NTP, 1993)
The details of the study design are described in Section 6.2.2. For neoplasms that were
considered to be lethal tumors, the tumor incidence was analyzed using the life table test, a
survival-adjusted procedure appropriate for rapidly lethal tumors (Cox, 1972; Tarone, 1975). For
incidental tumors (tumors discovered as a result of death from an unrelated cause), the primary
statistical method used was the logistic regression test. Alternative statistical methods included
the Fisher exact test and the Cochran-Armitage trend test (Armitage, 1971; Gart et al., 1979),
analyses based on the overall proportion of tumor-bearing animals. Tests of significance
included pairwise comparisons of each dose group and a test for an overall concentration-
response trend.
Exposure of male and female mice to 1,3-butadiene induced a variety of common and
uncommon tumors at multiple sites. The incidences of primary neoplasms associated with
exposure to 1,3-butadiene (for the 2-year study) are presented in Tables 6-7 and 6-8. The
percentage of animals bearing malignant tumors increased from about 30% in the controls to
nearly 90% in the highest exposure group, 625 ppm. The results of interim evaluations for 9
months and 15 months are presented in Tables 6-9 and 6-10.
6-12
-------
Table 6-7. Incidence of primary neoplasms in male B6C3Ft mice exposed to 1,3-butadiene by inhalation for
103 weeks.
Target organ
All organs
Heart
Lungs
Forestomach
Liver
Neoplastic lesion
Malignant lymphoma (histiocytic,
lymphocytic, mixed, NOS, or
undifferentiated)
Histiocytic sarcoma
Malignant lymphoma or histiocytic
sarcoma
Hemangiosarcoma
Alveolar/bronchiolar adenoma
Alveolar/bronchiolar carcinoma or
adenocarcinoma
Alveolar/bronchiolar adenoma,
adenocarcinoma, or carcinoma
Squamous cell papilloma
Squamous cell papilloma or
carcinoma
Hepatocellular adenoma
Concentration (ppm)
0
4/50 (8%)a
9.8%b
^<0.001C
0/50 (0%)
0.0%
^<0.001C
4/50 (8%)
9.8%
^<0.001C
0/50 (0%)
0.0%
^<0.001C
18/50(36%)
46.9%
p=0.200d
5/50 (10%)
14.3%
^<0.001C
21/50 (42%)
54.9%
^<0.001C
1/50 (2%)
2.5%
^<0.001d
1/50 (2%)
2.5%
^<0.001C
13/50(26%)
32.1%
p=0.042d
6.25
2/50 (4%)
5.1%
^=0.302N
0/50 (0%)
0.0%
2/50 (4%)
5.1%
^=0.302N
0/49 (0%)
0.0%
NA
20/50 (40%)
47.3%
^=0.517
6/50 (12%)
15.4%
^=0.577
23/50 (46%)
54.5%
^=0.552N
0/50 (0%)
0.0%
^=0.535N
0/50 (0%)
0.0%
^=0.481N
13/50(26%)
31.3%
^=0.552
20
4/50 (8%)
12.2%
^=0.528
4/50 (8%)
10.6%
^=0.051
8/50 (16%)
21.5%
^=0.118
1/50 (2%)
3.4%
^=0.451
10/50 (20%)
28.2%
^=0.080N
11/50(22%)
38.3%
^=0.017
19/50(38%)
53.6%
^=0.276
0/50 (0%)
0.0%
^=0.486N
0/50 (0%)
0.0%
^=0.545N
19/50(38%)
52.1%
^=0.158
62.5
6/50 (12%)
17.7%
^=0.238
5/50 (10%)
14.3%
^=0.021
11/50(22%)
29.6%
^=0.022
5/48 (10%)
19.4%
^=0.011
25/49(51%)
74.2%
^=0.036
12/49 (24%)
42.9%
^=0.006
31/49(63%)
87.9%
;X0.001
1/50 (2%)
4.5%
^=0.739
1/50 (2%)
4.5%
^=0.679
16/48 (33%)
57.0%
^=0.261
200
2/50 (4%)
4.0%
^=0.627
7/50 (14%)
31.9%
;X0.001
9/50(18%)
34.7%
^=0.005
20/48 (42%)
93.3%
;X0.001
21/50 (42%)
100.0%
^=0.061
22/50 (44%)
94.6%
;X0.001
35/50 (70%)
100.0%
;X0.001
7/50 (14%)
51.7%
^=0.012
8/50 (16%)
54.5%
;X0.001
23/48 (48%)
92.2%
^=0.008
625
51/73(70%)
95.4%
;X0.001
4/73 (5%)
10.8%
^=0.043
55/73 (75%)
95.9%
;X0.001
4/73 (5%)
44.6%
;X0.001
3/73 (4%)
59.4%
^=0.492
3/73 (4%)
59.4%
;X0.001
3/73 (4%)
59.4%
;X0.001
2/73 (3%)
40.0%
^=0.446
4/73 (5%)
51.8%
;X0.001
5/72 (7%)
100.0%
^=0.253
-------
Table 6-7. Incidence of primary neoplasms in male B6C3Ft mice exposed to 1,3-butadiene by inhalation for 103 weeks
(continued).
Target organ
Liver (continued)
Harderian gland
Preputial gland
Neoplastic lesion
Hepatocellular carcinoma
Hepatocellular adenoma or
carcinoma
Adenoma
Carcinoma
Adenoma or carcinoma
Carcinoma
Concentration (ppm)
0
4/49 (8%)
10.3%
^=0.178d
21/50 (42%)
47.9%
^=0.067d
6/50 (12%)
14.8%
^<0.001d
0/50 (0%)
0.0%
^=0.720d
6/50 (12%)
14.8%
^<0.001d
0/50 (0%)
0.0%
^<0.001C
6.25
6/49 (12%)
14.5%
^=0.381
23/50 (46%)
53.0%
^=0.375
7/50 (14%)
17.3%
^=0.497
1/50 (2%)
2.6%
^=0.522
7/50 (14%)
17.3%
^=0.497
0/50 (0%)
0.0%
NA
20
8/50 (16%)
25.0%
^=0.141
30/50 (60%)
70.1%
^=0.078
8/50 (16%)
25.8%
^=0.395
1/50 (2%)
4.2%
^=0.425
9/50(18%)
29.5%
^=0.217
0/50 (0%)
0.0%
NA
62.5
9/50(18%)
47.9%
^=0.066
25/48 (52%)
79.2%
^=0.185
19/50(38%)
63.4%
/K0.001
3/50 (6%)
11.7%
^=0.086
20/50 (40%)
64.9%
/K0.001
0/50 (0%)
0.0%
NA
200
8/50 (16%)
82.7%
^=0.006
33/48 (69%)
100.0%
^=0.030
30/50 (60%)
95.4%
/K0.001
2/50 (4%)
6.3%
^=0.352
31/50(62%)
95.5%
/K0.001
5/50 (10%)
45.7%
/K0.001
625
1/80 (1%)
12.5%
^=0.910
5/72 (7%)
100.0%
^=0.450
6/73 (8%)
100.0%
^=0.264
0/73 (0%)
0.0%
NA
6/73 (8%)
100.0%
^=0.002
0/73 (0%)
0.0%
NA
^Overall rate: number of tumor-bearing animals/number of animals examined.
bSurvival-adjusted rate. Kaplan-Meier estimated neoplasm incidence at the end of the study after adjustment for intercurrent mortality.
°Life table test. Beneath the control incidence are the p values associated with the trend test. Beneath the dosed group incidence are the p values corresponding to
pairwise comparison between
the control and dosed groups. The life table analysis regards neoplasms in animals dying prior to terminal kill as being (directly or indirectly) the cause of death.
^Logistic regression test. This test regards the neoplasms as nonfatal.
NA = not applicable; no tumors in these groups.
NOS = not otherwise specified.
N = incidence in dose group is lower than in control group.
Source: NTP, 1993.
-------
Table 6-8. Incidence of primary neoplasms in female B6C3Ft mice exposed to 1,3-butadiene by inhalation for
103 weeks.
Target organ
All organs
Heart
Lungs
Forestomach
Liver
Neoplastic lesion
Malignant lymphoma (lymphocytic,
mixed, NOS, or undifferentiated cell
type)
Histiocytic sarcoma
Malignant lymphoma or histiocytic
sarcoma
Hemangiosarcoma
Alveolar/bronchiolar adenoma
Alveolar/bronchiolar
adenocarcinoma or carcinoma
Alveolar/bronchiolar adenoma,
adenocarcinoma, or carcinoma
Squamous cell papilloma
Squamous cell carcinoma
Squamous cell papilloma or carcinoma
Hepatocellular adenoma
Hepatocellular carcinoma
Concentration (ppm)
0
6/50 (12%)a
14.6%b
^<0.001C
3/50 (6%)
6.9%
^<0.001C
9/50(18%)
20.5%
^<0.001C
0/50 (0%)
0.0%
^<0.001C
4/50 (8%)
10.5%
p=0.002d
0/50 (0%)
0.0%
^<0.001C
4/50 (8%)
10.5%
^<0.001C
0/50 (0%)
0.0%
^<0.001d
0/50 (0%)
0.0%
^<0.001C
0/50 (0%)
0.0%
^<0.001C
11/49(22%)
29.7%
^=0.599N
4/49 (8%)
10.3%
p=0.n&d
6.25
12/50 (24%)
34%
^=0.068
2/50 (4%)
4.5%
^=0.5 18N
14/50 (28%)
37.0%
^=0.136
0/50 (0%)
0.0%
NA
11/50(22%)
30.9%
^=0.039
5/50 (10%)
13.3%
^=0.029
15/50(30%)
39.5%
^=0.004
0/50 (0%)
0.0%
NA
0/50 (0%)
0.0%
NA
0/50 (0%)
0.0%
NA
10/49 (20%)
27.8%
^=0.531N
6/49 (12%)
14.5%
^=0.381
20
11/50(22%)
38.7%
^=0.029
7/50 (14%)
20.0%
^=0.077
18/50(36%)
52.1%
^=0.005
0/50 (0%)
0.0%
NA
12/50 (24%)
40.7%
^=0.013
11/50(22%)
42.9%
/K0.001
19/50(38%)
63.7%
/K0.001
2/50 (4%)
8.3%
^=0.149
1/50 (2%)
4.2%
^=0.414
3/50 (6%)
12.5%
^=0.056
9/50(18%)
30.3%
^=0.519N
8/50 (16%)
25.0%
p=0.\4l
62.5
7/50 (14%)
35.9%
^=0.055
4/50 (8%)
17.7%
^=0.195
11/50(22%)
47.2%
^=0.021
1/49 (2%)
4.8%
^=0.392
17/50 (34%)
64.8%
/K0.001
9/50(18%)
40.8%
;X0.001
24/50 (48%)
78.5%
;X0.001
1/50 (2%)
9.1%
^=0.260
1/50 (2%)
8.3%
^=0.277
2/50 (4%)
16.7%
^=0.044
14/48 (28%)
65.8%
^=0.025
9/50(18%)
39.9%
^=0.066
200
9/50(18%)
39.7%
^<0.001
7/50 (14%)
28.1%
^=0.002
16/50 (32%)
56.7%
;X0.001
21/50 (42%)
100.0%
/K0.001
14/49 (29%)
100.0%
^=0.002
19/49(39%)
100.0%
^<0.001
25/49(51%)
100.0%
;X0.001
3/50 (6%)
100.0%
^=0.078
1/50 (2%)
3.8%
^=0.374
4/50 (8%)
100.0%
^=0.001
15/50(29%)
89.0%
^=0.009
8/50 (16%)
82.7%
^=0.006
625
32/80 (40%)
70.8%
;X0.001
4/80 (5%)
10.3%
^=0.038
36/80 (45%)
73.9%
;X0.001
23/80 (29%)
100.0%
;X0.001
17/78 (22%)
100.0%
^=0.010
8/78 (10%)
100.0%
;X0.001
22/78 (28%)
100.0%
^<0.001
16/80 (20%)
100.0%
^=0.002
6/80 (8%)
70.5%
;X0.001
22/80 (28%)
100.0%
;X0.001
1/80 (1%)
100.0%
^=0.505
1/80 (1%)
12.5%
^=0.910
-------
Table 6-8. Incidence of primary neoplasms in female B6C3Ft mice exposed to 1,3-butadiene by inhalation for
103 weeks (continued).
Target organ
Liver (continued)
Ovary
Mammary gland
Harderian gland
Neoplastic lesion
Hepatocellular adenoma or carcinoma
Benign granulosa cell tumor
Malignant granulosa cell tumor
Benign or malignant granulosa cell
tumor
Adenoacanthoma
Carcinoma
Malignant mixed tumor
Adenoma
Carcinoma
Adenoma or carcinoma
Concentration (ppm)
0
15/49(31%)
39.3%
p=QA91A
1/49 (2%)
2.8%
^=0.030d
0/49 (0%)
0.0%
^=0.068d
1/49 (2%)
2.8%
^=0.006d
0/50 (0%)
0.0%
^=0.025C
0/50 (0%)
0.0%
^<0.001C
0/50 (0%)
0.0%
^<0.001C
8/50 (16%)
20.8%
^=0.046d
0/50 (0%)
0.0%
^=0.873Nd
8/50 (16%)
20.8%
^=0.061d
6.25
14/49 (29%)
34.3%
^=0.504N
0/49 (0%)
0.0%
^=0.517N
0/49 (0%)
0.0%
NA
0/49 (0%)
0.0%
^=0.517N
1/50 (2%)
2.9%
^=0.489
2/50 (4%)
5.8%
^=0.221
0/50 (0%)
0.0%
NA
10/50 (20%)
29.2%
^=0.356
1/50 (2%)
2.7%
^=0.493
10/50 (20%)
29.2%
^=0.356
20
15/50(30%)
45.5%
^=0.441
1/48 (2%)
3.2%
^=0.735
0/48 (0%)
0.0%
NA
1/48 (2%)
3.2%
^=0.735
2/50 (4%)
7.7%
^=0.152
2/50 (4%)
5.7%
^=0.192
0/50 (0%)
0.0%
NA
6/50 (12%)
20.7%
^=0.51 IN
1/50 (2%)
2.3%
^=0.631
7/50 (14%)
22.5%
^=0.575N
62.5
19/50(38%)
74.8%
^=0.027
6/50 (12%)
28.5%
^=0.026
3/50 (6%)
19.3%
^=0.046
9/50(18%)
42.9%
^=0.001
6/50 (12%)
32.5%
/K0.001
6/50 (12%)
16.2%
^=0.008
0/50 (0%)
0.0%
NA
15/50(30%)
61.0%
^=0.016
0/50 (0%)
0.0%
NA
15/50(30%)
61.0%
^=0.016
200
16/50 (32%)
91.7%
^=0.008
6/50 (12%)
100.0%
^=0.020
2/50 (4%)
54.2%
^=0.037
8/50 (16%)
100.0%
^=0.001
4/50 (8%)
13.6%
p=.02l
1 1/50 (22%)
39.1%
^<0.001
0/50 (0%)
0.0%
NA
20/50 (40%)
89.3%
^=0.001
1/50 (2%)
50.0%
^=0.085
20/50 (40%)
89.3%
^=0.001
625
2/80 (3%)
100.0%
^=0.302
6/79 (8%)
27.1%
^=0.303
0/79 (0%)
0.0%
NA
6/79 (8%)
27.1%
^=0.303
0/80 (0%)
0.0%
NA
12/80(15%)
100.0%
;X0.001
4/80 (5%)
29.4%
^=0.003
9/80(11%)
45.2%
^=0.176
0/80 (0%)
0.0%
NA
9/80(11%)
45.2%
^=0.176
"Overall rate; number of tumor-bearing animals/number of animals examined.
bSurvival-adjusted rate. Kaplan-Meier estimated neoplasm incidence at the end of the study after adjustment for intercurrent mortality.
°Life table test. Beneath the control incidence are the p values associated with the trend test. Beneath the dosed group incidence are the p values
pairwise comparison between
the control and dosed groups. The life table analysis regards neoplasm in animals dying prior to terminal kill as being (directly or indirectly) the
^Logistic regression test. This test regards the neoplasms as nonfatal.
NA = not applicable; no tumors in these groups.
NOS = not otherwise specified.
N = incidence in dose group is lower than in control group.
Source: NTP, 1993.
corresponding to
cause of death.
-------
Table 6-9. Incidence of primary neoplasms in male B6C3Ft mice exposed to 1,3-butadiene by inhalation for 9
months and 15 monthsa.
Target organ
All organs
Heart
Lungs
Forestomach
Liver
Harderian gland
Neoplastic lesion
Malignant
lymphoma
(histiocytic,
lymphocytic,
mixed, NOS, or
undifferentiated)
Hemangiosarcoma
Alveolar/
bronchiolar
adenoma,
adenocarcinoma, or
carcinoma
Squamous cell
papilloma or
carcinoma
Hepatocellular
adenoma or
carcinoma
Adenoma or
carcinoma
9 months
1 5 months
9 months
1 5 months
9 months
1 5 months
9 months
15 months
9 months
1 5 months
9 months
1 5 months
Concentration (ppm)
0
1/10
4/10
2/10
6.25
1/1
0/10
1/10
20
1/2
1/10
4/10
2/10
62.5
0/10
2/10
0/10
3/10
4/10
200
1/10
2/10
4/10
1/10
1/10
4/10
3/10
625
1/10
2/7
3/7
3/10
5/7
1/10
3/7
1/10
5/7
3/7
Overall rate: number of tumor-bearing animals/number of animals examined.
NA = not applicable; no tumors in these groups.
NOS = not otherwise specified.
N = incidence in dose group is lower than in control group.
Source: NTP, 1993.
-------
Table 6-10. Incidence of primary neoplasms in female B6C3Ft mice exposed to 1,3-butadiene by inhalation for
9 months and 15 months".
Target organ
All organs
Heart
Lungs
Forestomach
Liver
Ovary
Mammary gland
Harderian gland
Neoplastic lesion
Malignant lymphoma
(lymphocytic, mixed,
NOS, or undifferentiated
cell type)
Hemangiosarcoma
Alveolar/bronchiolar
adenoma,
adenocarcinoma, or
carcinoma
Squamous cell papilloma
or carcinoma
Hepatocellular adenoma
or carcinoma
Benign or malignant
granulosa cell tumor
Adenoacanthoma,
adenocarcinoma,
carcinoma, or malignant
mixed tumor
Adenoma or carcinoma
9 months
1 5 months
9 months
1 5 months
9 months
1 5 months
9 months
1 5 months
9 months
1 5 months
9 months
1 5 months
9 months
1 5 months
9 months
1 5 months
Concentration (ppm)
0
1/10
1/10
2/9
6.25
1/10
1/1
20
0/10
1/1
62.5
3/10
1/10
1/10
1/10
1/10
200
1/10
1/10
2/10
3/10
2/10
3/10
1/10
4/10
2/10
3/10
625
1/8
0/2
2/2
1/8
1/2
1/2
1/2
1/2
1/2
1/5
0/2
oo
Overall rate; number of tumor-bearing animals/number of animals examined.
NA = not applicable; no tumors in these groups.
NOS = not otherwise specified.
N = incidence in dose group is lower than in control group.
Source: NTP, 1993.
-------
As in the previous study (NTP, 1984), exposure of mice to 1,3-butadiene was associated
with the development of malignant lymphocytic lymphomas and to a lesser extent with
histiocytic sarcomas. The incidence of malignant lymphomas, particularly lymphocytic
lymphomas, was significantly increased in males and females exposed to 625 ppm and in
females exposed to 20 and 200 ppm (survival-adjusted) compared with controls. In addition,
there were significant exposure-response trends (/X0.001) in both sexes. The lymphocytic
lymphomas were well differentiated and occurred as early as week 23, peaking before the 15-
month interim evaluation. Many organs, particularly the spleen, lymph nodes, liver, lung, and
kidney, were affected in mice with lymphocytic lymphoma; however, the thymus was involved
in most mice and was the primary organ affected in some. The lymphocytic lymphomas
consisted of uniform populations of small- to medium-sized lymphocytes, whereas the mixed and
undifferentiated lymphomas generally consisted of more heterogeneous populations of
lymphocytes with pleomorphism and atypia. Other histological types of malignant lymphomas
(mixed and undifferentiated), commonly associated with the spontaneous lymphomas in aging
B6C3FJ mice, were seen at low incidence in some groups. The incidences of histiocytic sarcoma
were significantly increased in males and females exposed to 200 and 625 ppm and in males
exposed to 62.5 ppm. The histiocytic sarcomas (previously referred to as reticulum cell
sarcomas or type A sarcomas) were large and monomorphic, with dark basophilic nuclei and
relatively abundant eosinophilic cytoplasm.
Hemangiosarcomas of the heart were observed in male (at >20 ppm) and female (at >62.5
ppm) mice exposed to 1,3-butadiene for 2 years. The incidences of hemangiosarcomas of the
heart were significantly increased in male mice exposed to >62.5 ppm and in female mice
exposed to >200 ppm. There was a significant exposure-response trend in both sexes. The
cardiac hemangiosarcomas were observed in all ventricular locations, but were more frequent in
the left ventricular wall. Typical hemangiosarcomas had solid foci of anaplastic, pleomorphic
spindle cells at the center with a loose arrangement at the periphery. They were occasionally
multifocal and frequently coexisted with foci of endothelial hyperplasia distant and separate from
the main neoplasm. Hemangiosarcomas of the heart are considered uncommon in untreated
B6C3FJ mice (none were observed in 573 male and 558 female historical controls in NTP
inhalation studies). In male mice, the lower incidence of cardiac hemangiosarcoma at 625 ppm
compared with that at 200 ppm was attributed to the early mortality from induction of lethal
lymphocytic lymphoma at 625 ppm. The time-to-tumor detection for all hemangiosarcomas of
the heart ranged from 682 days at 20 ppm to 289 days at 625 ppm for males and from 649 days at
20 ppm to 307 days at 625 ppm for females. When hemangiosarcomas occurred in multiple
organs, the cardiac neoplasms were usually designated as primary, because the incidence of
hemangiosarcomas was highest in the heart and the earliest lesions occurred in the heart.
6-19
-------
However, it could not be determined with certainty if the hemangiosarcomas observed in other
organs were metastases or primary neoplasms. Subcutaneous, splenic, and hepatic
hemangiosarcomas that were found in the absence of cardiac hemangiosarcomas may reflect the
development of spontaneous vascular neoplasms known to occur in B6C3FJ mice.
Exposure of mice to 1,3-butadiene was also associated with an increased incidence of
pulmonary neoplasms in male and female mice. Although the incidence of alveolar/bronchiolar
adenomas was not significantly increased in male mice in the 2-year study, the combined
incidences of alveolar/bronchi olar adenocarcinomas and carcinomas and the combined
incidences of the benign and malignant pulmonary neoplasms were significantly increased at
62.5, 200, and 625 ppm. In female mice, the incidences of the benign and malignant neoplasms
analyzed separately or together were significantly increased in all exposure groups compared
with controls. Thus, even at 6.25 ppm, 1,3-butadiene was carcinogenic to female B6C3FJ mice.
The lower incidence of lung neoplasms at 625 ppm compared with the incidence at 200 ppm was
attributed to the high rate of early deaths due to the competing risks of lymphocytic lymphoma in
female mice exposed to 625 ppm. There was a significant exposure-response trend for combined
adenomas and carcinomas in both sexes. The time to tumor detection for lung tumors combined
ranged from 587 days at 6.25 ppm to 251 days at 625 ppm for males, and from 519 days at 6.25
ppm to 275 days at 625 ppm for females. The spectrum of lung lesions ranged from alveolar
epithelial hyperplasia to adenomas, carcinomas, and adenocarcinomas. Histologically, the
alveolar/bronchiolar adenomas exhibited distortion of the alveolar structure through the
formation of complex, irregular papillary patterns; the alveolar/bronchiolar carcinomas were
similar, but consisted of heterogeneous cell populations with various degrees of cellular
pleomorphism and atypia. The adenocarcinomas were larger, highly anaplastic neoplasms, often
accompanied by hemorrhage or necrosis.
In the forestomach, significant increases in squamous cell papillomas and carcinomas
combined were observed in male mice exposed to >200 ppm and in female mice exposed to
>62.5 ppm compared with controls. There was a significant exposure-response trend for
papillomas and carcinomas combined in both sexes. The combined incidence of squamous cell
papillomas and carcinomas of the forestomach (males, 4/575 [0.7%]; females, 9/561 [1.6%]) for
historical controls suggests that these lesions are relatively uncommon in B6C3FJ mice.
Increased incidences of hepatocellular adenomas and carcinomas were also seen in 1,3-
butadiene-exposed mice (Tables 6-7 and 6-8). The hepatocellular adenomas were discrete,
expansile masses; the carcinomas were larger than the adenomas and consisted of markedly
disorganized hepatocytes. The low incidence of liver neoplasms observed in males and females
at 625 ppm probably reflects increased early deaths from malignant lymphoma. Hepatocellular
adenomas and carcinomas are common neoplasms in B6C3FJ mice, occurring in 196/572 (34%)
6-20
-------
of male and 87/558 (15.6%) of female historical controls in NTP inhalation studies. The data
suggest that 1,3-butadiene has only a weak tumorigenic effect in the livers of male and female
mice. However, a chemical-related effect is supported by the detection of an activated K-ras
oncogene in liver neoplasms obtained from mice exposed to 1,3-butadiene (Goodrow et al.,
1990). According to Reynolds et al. (1987), activated K-ras oncogene had never been detected in
liver neoplasms from untreated B6C3FJ mice.
Although a variety of neoplasms were seen in the ovaries of female mice, only benign and
malignant granulosa cell tumors were definitely attributed to exposure to 1,3-butadiene (Table 6-
5). The ovarian granulosa cell tumors varied from small benign tumors to large cystic tumors
with thick trabeculae and spaces filled with blood or clear fluid. The overall historical control
incidence at NTP for benign and malignant granulosa cell tumors each was 1/548 (0.2%).
Increased incidences of mammary gland neoplasms were seen in female mice exposed to
>62.5 ppm 1,3-butadiene. Mammary tumors included adenoacanthomas, adenocarcinomas, and
malignant mixed tumors, the latter occurring only at 625 ppm. The mammary gland tumors
combined exhibited a significant exposure-response relationship. The adenoacanthomas were
considered variants of adenocarcinomas that have prominent squamous differentiation. The
malignant mixed tumors consisted of epithelial components arranged in glandlike structures and
anaplastic spindle-cell components. Mammary gland adenocarcinomas and adenoacanthomas
were considered uncommon in female B6C3FJ mice; the overall historical incidence at NTP was
21/561 (3.7%) for carcinomas and 1/561 (0.2%) for adenoacanthomas in female control mice.
The Harderian gland was identified as another site of 1,3-butadiene-induced neoplasms in
male and female mice (Tables 6-7 and 6-8), with significant exposure-related increases in
adenomas at 62.5 and 200 ppm and a low incidence of carcinomas in males exposed to 6.25, 20,
62.5, or 200 ppm. The low incidence of Harderian gland tumors at 625 ppm was attributed to
early deaths due to lymphocytic lymphoma, which precluded the development of Harderian
gland tumors. The investigators noted that the occurrence of Harderian gland carcinomas in
mice, particularly males, is unusual. The overall incidence of Harderian gland carcinomas was
2/575 (0.3%) in male and 3/561 (0.5%) in female historical controls at NTP. The 2-year
historical incidence of adenomas and carcinomas (combined) of the Harderian gland for control
groups in NTP inhalation studies was 25/575 (4.3%) for males and 13/561 (2.3%) for females.
Preputial gland carcinomas, also considered to be rare neoplasms in B6C3FJ mice, were
seen in five males (/X0.05) exposed to 200 ppm (none were reported in one survey of NTP
historic control data). These tumors were also thought to be exposure-related lesions. Some
preputial carcinomas were composed of large eosinophilic epithelial cells that were well
differentiated; more frequently, the carcinomas had necrotic cores and a thin layer of very
6-21
-------
anaplastic basophilic epithelial cells that aggressively invaded surrounding tissue and blood
vessels.
Renal tubule adenomas were seen in 2/50 females exposed to 200 ppm 1,3-butadiene and
in 1/50, 3/48, and 1/49 of males exposed to 6.25, 62.5, and 200 ppm, respectively. At the 15-
month evaluation, renal tubular adenoma occurred in 1/7 males exposed to 625 ppm. The
historical incidence of spontaneous renal tubule adenomas in untreated control groups in NTP
inhalation studies was 1/571 (0.2%) for males and 0/559 (0.0%) for females. Histologically, the
renal tubule adenomas contained multiple dilated tubules separated by thin connective tissue
septa. These renal lesions were probably related to exposure to 1,3-butadiene in males and
possibly related to exposure in females.
One neurofibrosarcoma of the subcutaneous tissue was observed in two females exposed
to 625 ppm at the 15-month evaluation. In the 2-year study, the combined incidences of
neurofibrosarcomas and sarcomas of the subcutaneous tissue were significantly increased in
female mice exposed to 62.5 ppm (p=0.017), 200 ppm (p=0.002), and 625 ppm (p=0.013) by the
life table test. Subcutaneous tissue sarcomas (all types) were considered uncommon spontaneous
neoplasms; the historical incidence was 2/561 (0.4%) for female controls at NTP, suggesting that
these subcutaneous tissue neoplasms may have been exposure-related. The historical incidence
for NTP inhalation studies was not reported.
One adenoma and one carcinoma of the Zymbal's gland were seen in females exposed to
625 ppm; one adenoma also occurred in a concurrent control male mouse, but none were
reported in historical controls. The report indicated that these Zymbal's gland neoplasms may be
related to 1,3-butadiene exposure.
Carcinomas of the small intestine, another uncommon tumor in the B6C3FJ mouse, were
seen in two females exposed to 6.25 ppm and in one female exposed to 62.5 ppm. One
carcinoma each was seen in one male each exposed to 6.25, 20, or 62.5 ppm, and in two males
exposed to 200 ppm. The relationship of these neoplasms to exposure to 1,3-butadiene could not
be determined; however, controls did not exhibit proliferative lesions of the intestine.
In supplemental analyses, the authors performed a "Poly-3" quantal response test (Bailer
and Portier, 1988; Portier and Bailer, 1989) as an alternative to the logistic regression analyses,
whose sensitivity was reduced by the decreased survival in the higher exposure groups. For
tumor sites related to butadiene exposure, the "Poly-3" test detected significant responses in
some of the exposure groups that had not been detected by the logistic regression analyses. The
overall results were consistent with those already presented.
The authors also fitted a modified Weibull model (Portier et al., 1986) to the "Poly-3"
survival-adjusted tumor rates to determine the shape parameters for the exposure-response
relationships. About half of the tumor sites associated with butadiene exposure had exposure-
6-22
-------
response relationships consistent with a linear model (i.e., shape parameter of 1). Most of the
other tumor sites exhibited supralinear exposure-response relationships (i.e., steep slope in low-
exposure region; shape parameter significantly <1). These sites were the liver in male mice, the
mammary gland in females, and the Harderian gland and lung in both sexes. Only the malignant
lymphoma in males and heart hemangiosarcoma in females had a shape parameter significantly
greater than 1, suggestive of a sublinear exposure-response relationship.
6.3.2.2. 2-Year Stop-Exposure Study (NTP, 1993)
An additional study with B6C3Fj mice, referred to as "stop-exposure study," was
conducted to assess the relationship between exposure level and duration of exposure to outcome
of 1,3-butadiene carcinogenicity. Groups of 50 male mice were exposed 6 h/day, 5 days/week at
concentrations of (a) 200 ppm for 40 weeks, (b) 625 ppm for 13 weeks, (c) 312 ppm for 52
weeks, or (d) 625 ppm for 26 weeks. After the exposures were stopped, the animals were placed
in control chambers for the remainder of the 103-week studies. The total exposures to 1,3-
butadiene (concentration x duration of exposure) were approximately 8,000 ppm-weeks for
groups exposed to 200 ppm for 40 weeks or 625 ppm for 13 weeks; the total exposures were
approximately 16,000 ppm-week for groups exposed to 312 ppm for 52 weeks or 625 ppm for 26
weeks. No additional controls were included for these studies, because they were run
concurrently with the 2-year studies.
Using the stop-exposure protocol, inhalation exposure to 1,3-butadiene had no effect on
mean body weights. However, exposure to 1,3-butadiene markedly reduced survival in all stop-
exposure groups as a result of the development of neoplasms, particularly malignant lymphomas
and hemangiosarcomas of the heart (Figure 6-3). A comparison of the two groups receiving total
exposures of 8,000 ppm-weeks showed that the survival of mice exposed to 625 ppm (13 weeks)
was similar to that of mice exposed to 200 ppm (40 weeks). By contrast, in the groups exposed
to 16,000 ppm-weeks, survival of mice exposed to 625 ppm (26 weeks) was significantly lower
than that of mice exposed to 312 ppm (52 weeks).
Neoplasms induced in the stop-exposure studies are summarized in Table 6-8. Overall,
the data show that exposure of male mice to 1,3-butadiene using the stop-exposure protocol
induced neoplasms at the same sites as those observed in the 2-year study.
Lymphocyte lymphomas of thymic origin occurred at a markedly increased incidence in
mice exposed to 625 ppm for 13 or 26 weeks. According to the life table test, the incidence of
lymphocytic lymphoma was also significantly increased in mice exposed to 200 ppm for 40
weeks or 312 ppm for 52 weeks. The incidence of histiocytic sarcomas was significantly
increased (life table test) in mice in all stop-exposure groups as well.
6-23
-------
CONTROL
200 PPM 840
625PPMsl3
S25PPMSB26
15
30
45 60 75
WEEKS ON STUDY
go
105
120
Figure 6-3. Kaplan-Meier survival curves for male mice in the stop-exposure inhalation of
1,3-butadiene.
Source: NTP, 1993.
6-24
-------
The lower incidences of lymphocytic lymphomas at 200 ppm (40 weeks) and 312 ppm
(52 weeks) compared to 625 ppm for 13 and 26 weeks, respectively, demonstrate that the
concentration of 1,3-butadiene is a greater contributing factor in the development of this lesion
than the duration of exposure, i.e., a high concentration for a short duration is more effective than
a lower concentration of longer duration. A comparison of the 200 ppm (40 weeks) versus the
625 ppm (13 weeks) and of the 312 ppm (52 weeks) versus the 625 ppm (26 weeks) lymphocytic
lymphoma results using a life table test confirms that the higher concentration/shorter duration
regimen is significantly more effective than the lower concentration/longer duration regimen
within each cumulative exposure grouping (p=0.005 for 8,000 ppm-weeks and/><0.001 for
16,000 ppm-weeks) after survival differences are taken into account.
As observed in the 2-year study, lymphocytic lymphomas occurred very early after
exposure started: as early as 23 weeks in the group exposed to 625 ppm for 26 weeks and as
early as 24 weeks in the group exposed to 625 ppm for 13 weeks. This lesion accounted for 24
and 17, respectively, of the first 25 deaths occurring in these groups by weeks 45 and 79,
respectively. Therefore, early deaths due to lymphocytic lymphoma would have a tremendous
negative effect on the incidence of late-developing lesions.
Hemangiosarcomas of the heart, which also accounted for some of the early deaths, were
significantly increased in most stop-exposure groups compared with the controls. The highest
incidence, which was about twice as high as that of other groups, occurred in the group exposed
to 312 ppm, followed by the groups exposed to 200 ppm and 625 ppm (26 weeks). The lowest
incidence occurred in the group exposed to 625 ppm for 13 weeks. Hemangiosarcomas appeared
at about 9 months in the 200, 312, and 625 ppm (26-week) stop-exposure groups. Comparison
(life table test) of groups having the same total exposures showed that the incidences of
hemangiosarcomas in mice exposed to 625 ppm were significantly lower than that of the
corresponding group exposed to 312 ppm (p=0.032) but not 200 ppm. The incidences of
hemangiosarcomas in both 625-ppm stop-exposure groups were higher than that in the 625-ppm
2-year exposure group, probably because of longer survival of the stop-exposure groups.
The incidences of neoplastic lesions of the lung (alveolar/bronchiolar adenoma,
adenocarcinoma, or carcinoma) were significantly elevated in each exposure group. The highest
incidence occurred in the 200-ppm stop-exposure group, followed by the 312-, 625- (13 weeks),
and 625-ppm (26 weeks) groups. The adenomas developed after week 47 and the
adenocarcinomas and carcinomas developed after week 53; the late appearance of these lesions
relative to lymphocytic lymphomas probably accounted for the lowest incidence of lung
neoplasms occurring in 625 ppm (26 weeks) group. A life table analysis suggested the incidence
of lung lesions in the 625 ppm (26 weeks) group was significantly greater than in the 312 ppm
6-25
-------
(52 weeks) group (p=0.013), but no difference was detected between the 200 ppm (40 weeks)
and 625 ppm (13 weeks) groups.
Mice exposed to 200 ppm 1,3-butadiene for 40 weeks had significantly increased
incidences of hepatocellular adenomas and adenomas/carcinomas combined; the incidences of
hepatocellular carcinomas analyzed alone were not significantly increased. Exposure to 1,3-
butadiene at 312 ppm or 625 ppm (13 or 26 weeks) did not increase the incidence of
hepatocellular neoplasms of any type. The earliest detection of these neoplasms was 67 weeks
for the 625 ppm (13 weeks), 57 weeks for the 200 ppm, 47 weeks for the 312 ppm, and 45 weeks
for the 625 ppm (26 weeks) stop-exposure groups. A logistic regression analysis found no
differences between the 200 ppm and 625 ppm (13 weeks) or the 312 ppm and 625 ppm (26
weeks) groups.
A low incidence of squamous cell papillomas of the forestomach occurred in each of the
groups, and squamous cell carcinomas were seen in mice exposed to 312 ppm or 625 ppm for 13
and 26 weeks. The incidences of squamous cell papillomas were not significantly greater than
those of controls for any group, but the incidences of squamous cell carcinomas were
significantly greater by the life table test, which is considered to be the appropriate test (NTP,
1993) for these fatal neoplasms. A life table analysis also revealed a statistically significant
exposure-rate effect for the squamous cell carcinomas in both of the total exposure groupings
(p=0.019 for 8,000 ppnrweeks and/>=0.015 for 16,000 ppnrweeks), suggesting that the higher
concentration/shorter duration exposures were more potent.
The incidence of adenomas of the Harderian gland was significantly greater in each
exposure group than in the controls by a logistic regression test. A low incidence of Harderian
gland carcinomas occurred in mice exposed to 200 ppm for 40 weeks (not significant), 312 ppm
for 52 weeks (p=0.006), and 625 ppm for 13 weeks (not significant). No Harderian gland
carcinomas were observed in the controls or in mice exposed to 625 ppm for 26 weeks. A
logistic regression analysis did not detect any exposure-rate effects.
Other neoplasms occurred at low incidence in the stop-exposure studies; they were
considered to be related to exposure because of their low spontaneous incidences in NTP
historical control male mice. These neoplasms occurred in the kidney, brain, Zymbal's gland,
and preputial gland. The incidences of these neoplasms are also summarized in Table 6-11.
Renal tubule neoplasms occurred in historical male control mice; the range was 0 to 1%.
The small number of these neoplasms in each of the exposure groups are considered to be related
to administration of 1,3-butadiene because the incidences were greater than the upper range for
historical controls.
6-26
-------
Table 6-11. Incidence of primary neoplasms in male B6C3Ft mice exposed to 1,3-butadiene by inhalation in the
stop-exposure study.
Parameters
Duration of exposures (weeks)
Total exposure (ppnrweeks)
Target organ
Hematopoietic
Heart
Lungs
Neoplastic lesion
Lymphocytic malignant lymphoma
Lymphoma (mixed or NOS)
Histiocytic sarcoma
Malignant lymphoma or histiocytic sarcoma
Malignant lymphoma (lymphocytic, mixed, or NOS)
Hemangiosarcoma
Alveolar/bronchiolar adenoma
Alveolar/bronchiolar
adenocarcinoma or carcinoma
Concentration (duration of exposure)
0 ppm
103
0
2/50a (4%)
4.7%b
2/50 (4%)
5.3%
0/50 (0%)
0.0%
4/50 (8%)
9.8%
4/50 (8%)
9.8%
0/50 (0%)
0.0%
18/50(36%)
46.9%
5/50 (10%)
14.3%
200 ppm
40
8,000
6/50 (12%)
26.7%
^=0.033C
2/50 (4%)
7.8%
^=0.382C
5/50 (10%)
21.3%
p=0.006c
13/50(26%)
46.8%
^<0.001C
8/50 (16%)
32.4%
p=0.023c
15/50(30%)
76.2%
^<0.001C
24/50 (48%)
94.3%
p=0.0\5d
22/50 (44%)
89.5%
^<0.001C
625 ppm
13
8,000
17/50 (34%)
35.8%
/K0.001
5/50 (10%)
34.8%
^=0.010
2/50 (4%)
28.9%
^=0.011
24/50 (48%)
72.1%
/K0.001
22/50 (44%)
58.2%
/K0.001
7/50 (14%)
61.8%
/K0.001
17/50 (34%)
85.3%
^=0.044
18/50(36%)
87.7%
/K0.001
312 ppm
52
16,000
4/50 (8%)
100.0%
^=0.034
4/50 (8%)
58.0%
^=0.005
7/50 (14%)
43.0%
/K0.001
15/50(30%)
100.0%
/K0.001
8/50 (16%)
100.0%
/K0.001
33/50 (66%)
100.0%
/K0.001
26/50 (52%)
100.0%
^=0.001
16/50 (32%)
100.0%
^<0.001
625 ppm
26
16,000
30/50 (60%)
81.5%
^<0.001
3/50 (6%)
43.3%
^=0.002
2/50 (4%)
15.6%
^=0.036
35/50 (70%)
91.2%
;X0.001
33/50 (66%)
89.5%
;X0.001
13/50(26%)
100.0%
;X0.001
12/50 (24%)
100.0%
;X0.001
11/50(22%)
100.0%
^<0.001
to
-------
Table 6-11. Incidence of primary neoplasms in male B6C3Ft mice exposed to 1,3-butadiene by inhalation in the
stop-exposure study (continued).
Parameters
Duration of exposures (weeks)
Total exposure (ppnrweeks)
Target organ
Lungs (continued)
Liver
Forestomach
Harderian gland
Neoplastic lesion
Alveolar/bronchiolar adenoma, adenocarcinoma, or
carcinoma
Hepatocellular adenoma
Hepatocellular carcinomaSum.
Hepatocellular adenoma or carcinoma
Squamous cell papilloma
Squamous cell carcinoma
Squamous cell papilloma or carcinoma
Adenoma
Concentration (duration of exposure)
0 ppm
103
0
21/50 (42%)
54.9%
—Sum.
13/50(26%)
32.1%
11/50(22%)
26.0%
—Sum.
21/50 (42%)
47.9%
1/50 (2%)
2.5%
0/50 (0%)
0.0%
1/50 (2%)
2.5%
6/50 (12%)
14.8%
200 ppm
40
8,000
36/50 (72%)
100.0%
;K0.001cSum
27/49 (55%)
91.1%
;K0.001dSum
14/49 (29%)
50.3%
;?=0.530dSum
33/49 (67%)
93.4%
^=0.004d
3/50 (6%)
21.4%
^=0.195d
0/50 (0%)
0.0%
NA
3/50 (6%)
21.4%
^=0.065C
26/50 (52%)
87.9%
^<0.001d
625 ppm
13
8,000
28/50 (56%)
100.0%
/KO.OOlSum.
19/49(39%)
91.1%
;tf«m.=0.042Sum.
14/49 (29%)
90.9%
^=0.142Sum.
24/49 (49%)
94.4%
^=0.063
4/50 (8%)
28.3%
^=0.260
4/50 (8%)
51.6%
;K0.001d
7/50 (14%)
56.6%
/K0.001
20/50 (40%)
94.3%
^=0.001
312 ppm
52
16,000
32/50 (64%)
100.0%
/KO.OOlSum.
19/50(38%)
100.0%
;tf«m.=0.045Sum.
10/50 (20%)
74.6%
;?=0.453Sum.
24/50 (48%)
100.0%
^=0.169
4/50 (8%)
100.0%
^=0.181
5/50 (10%)
33.1%
/K0.001
9/50(18%)
100.0%
/K0.001
28/50 (56%)
100.0%
/K0.001
625 ppm
26
16,000
17/50 (34%)
100.0%
/K0.001
11/50(22%)
100.0%
;?=0.284Sum.
4/50 (8%)
50.5%
;?=0.393Sum.
13/50(26%)
100.0%
^=0.561
4/50 (8%)
20.1%
^=0.301
6/50 (12%)
40.9%
^<0.001
10/50 (20%)
52.8%
;X0.001
13/50(26%)
100.0%
^=0.046
to
oo
-------
Table 6-11. Incidence of primary neoplasms in male B6C3Ft mice exposed to 1,3-butadiene by inhalation in the
stop-exposure study (continued).
Parameters
Duration of exposures (weeks)
Total exposure (ppm-weeks)
Target organ
Harderian gland
(continued)
Kidney
Brain6
Preputial gland
Zymbal's gland
Neoplastic lesion
Carcinoma
Adenoma or carcinoma
Renal tubule adenoma
Malignant glioma
Malignant neuroblastomas
Carcinoma
Adenoma or carcinoma
Adenoma or carcinoma
Concentration (duration of exposure)
Oppm
103
0
0/50 (0%)
0.0%
6/50 (12%)
14.8%
0/50 (0%)
0.0%
0/50 (0%)
0/50 (0%)
0/50 (0%)
0.0%
0/50 (0%)
0.0%
1/50 (2%)
2.9%
200 ppm
40
8,000
2/50 (4%)
5.6%
p=Q391A
27/50 (54%)
88.3%
^<0.001d
4/48 (8%)
17.4%
^=0.073d
0/50 (0%)
0/50 (0%)
1/50 (2%)
10.0%
^=0.368C
1/50 (2%)
10.0%
^=0.368C
1/50 (2%)
4.8%
^=0.531C
625 ppm
13
8,000
4/50 (8%)
38.8%
^=0.190
23/50 (46%)
100.0%
/K0.001
1/50 (2%)
14.3%
^=0.273
2/50 (4%)
2/50 (4%)
4/50 (8%)
16.9%
^=0.039
5/50 (10%)
22.9%
^=0.013
2/50 (4%)
8.8%
^=0.178
312 ppm
52
16,000
2/50 (4%)
51.5%
^=0.006
30/50 (60%)
100.0%
/K0.001
3/49 (6%)
27.8%
^=0.075
0/50 (0%)
0/50 (0%)
4/50 (8%)
100.0%
/K0.001
4/50 (8%)
100.0%
/K0.001
0/50 (0%)
0.0%
^=0.998N
625 ppm
26
16,000
0/50 (0%)
0.0%
NA
13/50(26%)
100.0%
^=0.046
1/50 (2%)
6.3%
^=0.731
1/50 (2%)
0/50 (0%)
3/50 (6%)
100.0%
^=0.002
3/50 (6%)
100.0%
^=0.002
2/50 (4%)
37.3%
^=0.009
to
VO
"Overall rate, number of tumor-bearing animals/number of animals examined.
""Survival-adjusted rate. Kaplan-Meier estimated neoplasm incidence at the end of the study after adjustment for intercurrent mortality.
'Life table test. The p values for pairwise comparison of exposed groups with controls are beneath the exposed group incidence. N refers to negative association with control group. The life
table analysis regards neoplasms in animals dying prior to terminal kill as being (directly or indirectly) the cause of death.
•"Logistic regression test. This test regards the neoplasms as nonfatal.
eNo statistical analysis.
NA = not applicable.
NOS = not otherwise specified.
N = incidence in dose group is lower than in control group.
Source: NTP, 1993.
-------
Brain neoplasms, including two neuroblastomas and two malignant gliomas observed in
male mice exposed to 625 ppm for 13 weeks and one malignant glioma observed in male mice
exposed to 625 ppm for 26 weeks, may have been related to 1,3-butadiene exposure. Brain
neoplasms are rare in untreated B6C3FJ mice; none have been reported in 574 NTP historical
control male mice. Furthermore, a low incidence of gliomas was also reported in the previous
NTP (1984) study. For these reasons, the brain neoplasms are considered exposure-related
lesions.
A low incidence of preputial gland carcinomas occurred in the exposed groups in the stop-
exposure studies, and none were seen in controls. Compared with the incidence in concurrent
controls, the combined incidences of preputial gland tumors (adenoma and carcinoma) were
significant in male mice exposed to 312 ppm (52 weeks) and to 625 ppm (13 and 26 weeks) by
the life table test. Preputial gland carcinomas were not reported in a survey of NTP historical
control mice, further indicating that these neoplasms are probably related to exposure to 1,3-
butadiene.
One male exposed to 200 ppm for 40 weeks, two males exposed to 625 ppm for 13 weeks,
and two males exposed to 625 ppm for 26 weeks developed Zymbal's gland carcinomas. This
lesion did not occur in male mice exposed to 312 ppm for 52 weeks; one control male, however,
developed an adenoma. The combined incidence of Zymbal's gland adenomas and carcinomas
in animals exposed to 625 ppm for 26 weeks was significantly increased compared with controls
by the life table test. Zymbal's gland neoplasms are rare spontaneous neoplasms that had not
been observed in any NTP historical controls before the only occurrence of this adenoma in the
control male mice for these studies.
To summarize the results of the stop-exposure study pertaining to the relationship between
exposure level and duration of exposure: For lymphocytic lymphomas, there is strong evidence
that higher concentration/shorter duration exposures are more potent than lower
concentration/longer duration exposures for both the 8,000 ppm-weeks and 16,000 ppm-weeks
total exposure groupings. There is also some evidence for a similar exposure-rate effect for
forestomach squamous cell carcinomas in both total exposure groupings. Any exposure-rate
effects at other sites are less clear, especially because it is difficult to distinguish a small apparent
increased potency effect of higher concentration/shorter duration exposures from an effect of
longer potential postexposure follow-up times following the shorter duration exposures.
6.3.2.3. Summary of NTP (1993) Study
The 2-year inhalation study showed that 1,3-butadiene is a potent carcinogen in mice at all
concentrations evaluated. It also demonstrated that exposure to lower concentrations of 1,3-
6-30
-------
butadiene than those used in the previous NTP (1984) study allowed expression of neoplasms at
other sites and provided clearer exposure-response relationships because of increased survival.
Statistically significant increases in the incidences of malignant tumors at one or more sites
occurred in male mice exposed to >20 ppm and in females exposed to >6.25 ppm 1,3-butadiene
(the lowest exposure concentration used) for periods up to 103 weeks. The possibility therefore
exists that lower exposure concentrations would also cause cancer in B6C3FJ mice. The
percentage of animals bearing malignant tumors increased from about 30% in the controls to
nearly 90% in the highest exposure group, 625 ppm. Lymphocytic lymphomas,
hemangiosarcomas of the heart, lung neoplasms, and neoplastic lesions of the forestomach,
mammary gland, ovary, and liver, lesions identified in the NTP (1984) study, were again
increased in this study. In addition, the Harderian gland and preputial gland were identified as
sites of 1,3-butadiene-induced neoplasms. Tumors observed in the kidneys, skin, Zymbal's
gland, and intestine may also have been related to 1,3-butadiene exposure.
The stop-exposure study demonstrated that limited exposure to 1,3-butadiene also induces
neoplasms at multiple organ sites in male B6C3FJ mice. Incidences of lymphocytic lymphomas,
hemangiosarcomas of the heart, alveolar-bronchiolar neoplasms, forestomach squamous cell
neoplasms, Harderian gland neoplasms, and preputial gland neoplasms were increased compared
with controls after exposure to 625 ppm 1,3-butadiene for only 13 weeks. The stop-exposure
study also demonstrated an apparent exposure-rate effect for the induction of lymphocytic
lymphomas by 1,3-butadiene. At equivalent total exposures, the induction of lymphocytic
lymphomas was greater with exposure to a higher concentration of 1,3-butadiene for a shorter
time than for exposure to a lower concentration for a longer duration.
Overall, the NTP (1993) was a very well conducted study with a precise and
comprehensive presentation of the data. Adequate numbers of animals of both sexes were
exposed to multiple concentration levels of 1,3-butadiene for a major portion of their life span.
Comprehensive histopathologic evaluations were performed and mortality and tumor incidences
were analyzed statistically using multiple methods.
6.3.2.4. 1-Year Study (Irons et al, 1989; Irons, 1990)
To elucidate the mechanism of murine leukemogenesis, Irons and co-workers compared
the induction of thymic lymphomas and expression of murine leukemia virus in NIH Swiss male
mice and B6C3Fj male mice by exposing them to 1,250 ppm 1,3-butadiene, 6 h/day, 5
days/week for 52 weeks. Activation of an endogenous esotropic retrovirus has been associated
with spontaneous lymphomas in the B6C3Fj mouse. The NIH mouse strain was used because it
does not express the esotropic murine leukemia viruses expressed in B6C3Fj mice. The
background rate for thymic lymphoma in NIH mice is nearly zero. Although there was a marked
6-31
-------
difference between the incidence of thymic lymphoma/leukemia in B6C3FJ mice (57%) and the
incidence in similarly exposed NIH mice (14%), the study showed that 1,3-butadiene can induce
thymic lymphomas independently of an activated retrovirus. In addition, because these studies
were for only 52 weeks, they did not necessarily allow for a full response for induction of
lymphomas by 1,3-butadiene.
6.4. STUDIES OF 1,3-BUTADIENE METABOLITES
The 1,3-butadiene metabolites l,2-epoxy-3-butene and l,2:3,4-diepoxybutane have been
shown to be carcinogenic in rats when administered by skin application or subcutaneous
injection (Van Duuren et al., 1963, 1966).
6.5. RELATED COMPOUNDS
1,2-Epoxybutane, a related compound that is used as a stabilizer for chlorinated
hydrocarbon solvents, was administered by inhalation 6 h/day, 5 days/week for 24 months at
exposure concentrations of 0, 200, or 400 ppm to F344/N rats and 0, 50, or 200 ppm to B6C3FJ
mice (Dunnick et al., 1988). The treatment and control groups consisted of 50 male and 50
female animals of each species. Exposure-related inflammatory, degenerative, and proliferative
lesions occurred in the nasal cavity of both rats and mice. Neoplastic lesions were restricted to
the respiratory tract in rats. At 400 ppm, nasal papillary adenomas were observed in seven male
rats and two female rats; none were observed in controls. In male rats exposed to 400 ppm, there
was also an increased incidence of alveolar/bronchiolar adenomas or carcinomas (combined)
(5/50) compared with controls (0/50). No exposure-related neoplastic lesions were seen in male
or female mice.
The toxicology and carcinogenicity of 4-vinyl-l-cyclohexene, a dimer of 1,3-butadiene,
were reviewed in the 1985 U.S. EPA 1,3-butadiene report, based on a draft NTP report. The
final NTP report (NTP, 1986) contains the same information; therefore, the data are not
summarized in this update. The basic conclusion was that there was clear evidence of
carcinogenicity of 4-vinyl-l-cyclohexene (by gavage) in female mice on the basis of increased
ovarian neoplasms and equivocal evidence in male mice based on marginal increases of
malignant lymphomas and alveolar/bronchiolar adenomas. In male mice, the sensitivity for
detecting a carcinogenic response was limited by poor survival at the high exposure. In rats,
there was inadequate evidence in males, at least in part because of excessive mortality, and
equivocal evidence in females based on increased neoplasms of the clitoral gland.
Chloroprene (2-chloro-l,3-butadiene) and isoprene (2-methyl-l,3-butadiene) are
structurally related to 1,3-butadiene. In an NTP (1998) inhalation bioassay, chloroprene produced
multiple organ carcinogenic responses in rats and mice, with several sites being the same as those
6-32
-------
targeted by 1,3-butadiene. Groups of 50 male and 50 female F344/N rats and 50 male and 50
female B6C3FJ mice were exposed to 0, 12.8, 32, or 80 ppm chloroprene 6 h/day, 5 days/week,
for 2 years. Chloroprene was carcinogenic to the oral cavity, thyroid gland, and kidney of male
and female rats, as well as the lung of male rats and the mammary gland (fibroadenomas) of
female rats. It also produced tumors of the lung, circulatory system (hemangiomas and
hemangiosarcomas), Harderian gland, and forestomach (papillomas) of male and female mice, as
well as kidney tubular cell adenomas in male mice and tumors of the liver, mammary gland, skin,
mesentery, and Zymbal's gland in female mice. Melnick et al. (1999) calculated EC10s for both
1,3-butadiene and chloroprene for the various treatment-related tumors and concluded that the
overall carcinogenic potency of chloroprene and 1,3-butadiene appeared to be similar in mice. In
particular, female lung tumors, which appear to be the most sensitive tumor response in mice
from both 1,3-butadiene and chloroprene exposures, yielded the same EC10s for both 1,3-
butadiene and chloroprene. With respect to rats, chloroprene has greater carcinogenic potency in
the F344/N rat than 1,3-butadiene has in the Sprague-Dawley rat.
In a different inhalation bioassay (Trochimowicz et al., 1998), male and female Wistar rats
and Syrian golden hamsters were exposed to up to 50 ppm chloroprene for 6 h/day, 5 days/week,
for up to 24 months (rats) or 18 months (hamsters). These investigators reported that "there was
no evidence of carcinogenicity related to (chloroprene) exposure" in either species under these
exposure conditions. However, there was a statistically significant increased incidence of
mammary gland fibroadenomas in the female rat, as was seen in the NTP study.
In a 2-year rat inhalation bioassay of isoprene (NTP, 1997), groups of 50 male and 50
female rats were exposed to 0, 220, 700, or 7,000 ppm isoprene, 6 h/day, 5 days/week, for 104
weeks. In males, there were increased incidences of mammary gland neoplasms (fibroadenomas
plus carcinomas), renal tubule adenomas, and testicular adenomas. In female rats, there were
increased incidences and multiplicities of mammary gland fibroadenomas. A few rare brain
neoplasms in exposed females may also have been treatment-related.
There are no 2-year bioassays of isoprene exposure in mice; however, there are two
inhalation studies that used shorter exposure periods (Placke et al., 1996; Melnick et al., 1996b).
Placke et al. (1996) exposed 12 groups of 50 male B6C3FJ mice to various concentration x time
scenarios, with exposure concentrations up to 2,200 ppm and exposure durations up to 80 weeks.
The mice were then held until week 104, with the exception of 5 exposure groups that had
excessive mortality and were slaughtered after 96 weeks. Groups of 50 female mice were
exposed to 0, 10, and 70 ppm isoprene, 8 h/day for 80 weeks, and similarly held until week 104.
The male mice exhibited most of the same treatment-related tumors as mice exposed to 1,3-
butadiene—tumors of the lung, liver, Harderian gland, and forestomach, as well as heart (and
spleen) hemangiosarcomas and histiocytic sarcomas. As with chloroprene, a major difference
6-33
-------
from the results of 1,3-butadiene exposure was the absence of the early-onset T-cell lymphomas.
The female mice were exposed to lower concentrations and only showed treatment-related
increased incidences of Harderian gland adenomas, pituitary gland adenomas, and possibly
hemangiosarcomas of the spleen. Placke et al. concluded that isoprene appears to be about one
order of magnitude less potent than 1,3-butadiene in mice. Melnick et al. (1996b) exposed male
B6C3FJ mice to isoprene at concentrations up to 7,000 ppm for 26 weeks followed by a 26-week
"recovery" period. Treatment-related tumors were observed in the lung, liver, Harderian gland,
and forestomach, but no hemangiosarcomas or lymphomas were induced under these conditions.
6.6. DISCUSSION AND CONCLUSIONS
The long-term inhalation studies have shown that 1,3-butadiene is carcinogenic in rats and
mice, inducing tumors at multiple organ sites (NTP, 1984, 1993; Owen et al., 1987). Results of
the 1993 NTP mouse study and the Owen et al. rat study are summarized in Table 6-12.
Results of the 1993 NTP mouse bioassay detailed in this report confirmed the
carcinogenicity of 1,3-butadiene in male and female B6C3FJ mice demonstrated in an earlier
1984 NTP study. Of particular interest in the 1993 study were the large number of primary organ
sites of tumor induction by 1,3-butadiene; the early and extensive development of lymphomas;
the induction of uncommon tumors, such as hemangiosarcomas of the heart and squamous cell
neoplasms of the forestomach; and the development of malignant lung tumors at exposure
concentrations as low as 6.25 ppm. Because there were no exposure levels of 1,3-butadiene at
which a carcinogenic response was not induced, it is likely that exposure to concentrations below
6.25 ppm would also cause cancer in mice.
Exposure to 1,3-butadiene at concentrations ranging from 6.25 to 625 ppm for 2 years
caused increased incidences of neoplasms in the hematopoietic system, heart, lung, forestomach,
mammary gland, ovary, and liver, all lesions identified in the NTP (1984) study. The Harderian
gland and preputial glands were identified as additional sites, and tumors in the kidneys, skin,
Zymbal's gland, and intestine were marginally associated with 1,3-butadiene. Because of
increased survival, the study also established clearer concentration-response relationships than
the 1984 study. Competing risks of early-developing lethal lymphocyte lymphomas at high
concentrations preempted the appearance of late-developing neoplasms at some organ sites.
Separate experiments with reduced exposure durations (stop-exposure study) showed that
continued exposure is not necessary for development of neoplasms. The incidences of
lymphocytic lymphomas, hemangiosarcomas of the heart, and tumors of the lung, forestomach,
Harderian gland, and preputial gland were increased in mice exposed for only 13 weeks to 625
ppm 1,3-butadiene, and it is likely that even shorter exposure durations would have produced a
carcinogenic response. The stop-exposure study also suggests that concentration is a greater
6-34
-------
Table 6-12. Significant positive primary tumor incidences in mice and rats exposed
to 1,3-butadiene for 2 years".
Target tissue
Systemic
Heart
Lung
Forestomach
Harderian gland
Liver
Preputial Gland
Ovary
Mammary gland
Brain
Testis
Pancreas
Zymbal gland
Thyroid
Uterus
Neoplasm
Malignant lymphoma
Histiocytic sarcoma
Haemangiosarcoma
Alveolar bronchiolar adenoma or
carcinoma or
adenocarcinoma
Squamous-cell papilloma or carcinoma
Adenoma or carcinoma
Hepatocellular adenoma or carcinoma
Carcinoma
Granulosa cell tumors, benign or
malignant
Adenocanthoma, carcinoma,
malignant mixed tumors
carcinoma, fibroadenoma
Gliomab
Leydig cell tumor
Exocrine adenoma
Carcinoma0
Follicular cell adenoma or carcinoma
Stromal sarcoma0
Exposure concentration (ppm)
6.25
F
20
F
F
62.5
M
M
MF
F
MF
F
F
F
200
F
MF
MF
MF
MF
MF
MF
M
F
F
625
MF
MF
MF
MF
MF
M
F
F
1,000
F
8,000
F
M
M
F
Unshaded column - mouse.
Shaded column - rat.
M - male, F- female.
'Results from the 1993 NTP mouse study and the 1987 Owen et al. rat study. The 1984 NTP mouse study is not included because it was
terminated at 61 weeks because of excessive mortality.
bNo significant increases by pairwise comparison, but increasing trend in male SD rats (p<0.05).
°No significant increases by pairwise comparison, but increasing trend in female SD rats (p<0.05).
6-35
-------
contributing factor in the development of lymphocytic lymphomas than duration of exposure. At
comparable total exposures, the incidence of lymphocytic lymphomas was greater with exposure
to a high concentration of 1,3-butadiene for a short time compared with a lower concentration for
a longer duration.
A morphological continuum of 1,3-butadiene-induced proliferative lesions to neoplasia or
the progression of benign to malignant neoplasms was evident for a number of sites in both the
2-year and the stop-exposure study (NTP, 1993). Increased incidences of proliferative,
nonneoplastic lesions (hyperplasia) of the cardiac endothelium, alveolar epithelium, forestomach
epithelium, germinal epithelium and granulosa cells of the ovaries, mammary gland, and
Harderian gland probably represent treatment-related preneoplastic changes at these target sites.
The distinction between adenoma and carcinoma further reveals the biological progression of the
benign lesions to malignant neoplasia. For example, in the lungs of male mice, progression from
alveolar-bronchiolar adenoma to carcinoma was evident in the 200-ppm exposure group and in
all of the stop-exposure groups.
The mechanism of 1,3-butadiene-induced carcinogenicity is not known; however,
metabolism likely involving two reactive metabolites, l,2-epoxy-3-butene and 1,2:3,4-
diepoxybutane, is thought to be an important factor (Chapters 3 and 4).
The carcinogenic response of 1,3-butadiene in rats and mice is well documented. Also,
studies reviewed have shown that mice are more sensitive to the induction of carcinogenic effects
than are rats (see Table 6-12). The carcinogenic activity in Sprague-Dawley rats exposed to
1,000 or 8,000 ppm 1,3-butadiene was largely limited to endocrine tissues or hormonal
responsive tissues, such as pancreas, Leydig cells of the testis, uterus, Zymbal's gland, mammary
gland, and thyroid (Owen et al., 1987), whereas exposure of B6C3FJ mice to much lower
concentrations of 1,3-butadiene caused significantly increased incidences of mammary gland
neoplasms and granulosa cell neoplasms of the ovary as well as malignant lymphomas,
histiocytic sarcomas, hemangiosarcomas of the heart, alveolar-bronchiolar neoplasms, squamous
cell neoplasms of the forestomach, Harderian gland neoplasms, and hepatocellular neoplasms.
The reason for the species difference is not known, but may in part be due to differences in
toxicokinetics. For example, metabolism studies have shown that blood concentrations of 1,3-
butadiene are higher in mice than in rats, and are lower in monkeys than in either rodent species
(Chapter 3). In vitro studies using liver microsomes have shown that the metabolism of the
reactive intermediate, l,2-epoxy-3-butene, to the non-DNA-reactive l,2-dihydroxybut-3-ene is
the prevalent pathway in human and rat preparations, whereas mouse liver microsomes convert
l,2-epoxy-3-butene to DNA-reactive l,2:3,4-diepoxybutane in addition to the nonreactive 1,2-
dihydroxybut-3-ene (Csanady and Bond, 1991).
6-36
-------
Investigations by Irons and co-workers (Irons et al., 1989; Irons, 1990) to explain the
species differences of 1,3-butadiene-induced carcinogenicity have focused on the possibility that
activation of an endogenous leukemia retrovirus may play a critical role in 1,3-butadiene-induced
lymphoma in B6C3FJ mice. The incidence of thymic lymphomas was greater in B6C3FJ mice
(57%) than in NIH Swiss mice (14%) exposed to 1,250 ppm 1,3-butadiene for 1 year. However,
the NIH Swiss mouse does not express the endogenous leukemia retrovirus and has a very low
background rate for thymic lymphomas. Thus, the finding that exposure to 1,3-butadiene caused
a 14% incidence of thymic lymphomas in NIH Swiss mice suggests that 1,3-butadiene can
induce thymic lymphomas independently of an activated retrovirus.
Identification of activated oncogenes in chemically induced tumors also may provide
information regarding the mechanism of tumor induction by butadiene. For example, because K-
ras is the most commonly detected oncogene in human cancers, tumors from the NTP (1993)
study were evaluated for the presence of K-ras oncogenes (Goodrow et al., 1990). Activated K-
ras oncogenes were detected in 6/9 lung tumors, 3/12 hepatocellular carcinomas, and 2/11
lymphomas obtained from B6C3FJ mice exposed to 1,3-butadiene at concentrations ranging from
62.5 to 625 ppm. A specific codon 13 mutation was found in most of the activated K-ras
oncogenes, suggesting a chemical-specific effect. Activated K-ras genes have not been found in
spontaneously occurring liver tumors or lymphomas (Goodrow et al., 1990) and were observed
in only 1/10 of spontaneous lung tumors in B6C3Fj mice (Goodrow et al., 1990; Reynolds et al.,
1987). Furthermore, it was shown that tumor suppressor genes are inactivated during 1,3-
butadiene carcinogenesis. Soderkvist et al. (1992) identified allelic losses in the p53 tumor
suppressor gene in lung and mammary carcinomas and lymphomas of B6C3Fj mice exposed to
1,3-butadiene, that were analogous to those observed in a variety of human cancers.
Immune-function assays conducted by Thurmond et al. (1986) in which B6C3Fj mice
were exposed by inhalation to 1,250 ppm 1,3-butadiene for 6 or 12 weeks showed that 1,3-
butadiene exerts no significant immunosuppressive effects, suggesting that 1,3-butadiene causes
neoplasia by mechanisms other than compromise of immune function.
In addition to the carcinogenic effects noted in the NTP (1993) study, exposure to 1,3-
butadiene caused hematological changes indicative of a partially regenerative anemia in mice
exposed to >62.5 ppm 1,3-butadiene. Mice exposed to 625 ppm exhibited bone marrow atrophy
and splenic and hepatic extramedullary hematopoiesis. Increases in mean cell volume and mean
cell hemoglobin at 625 ppm 1,3-butadiene suggested that although 1,3-butadiene caused
suppression o-f hematopoiesis in the bone marrow, younger larger cells may have been released
into the blood from extramedullary sites. A macrocytic-megaloblastic anemia was reported in
B6C3Fj mice exposed to 1,250 ppm 1,3-butadiene for 6 weeks (Irons et al., 1986a,b).
6-37
-------
7. EPIDEMIOLOGIC STUDIES OF CARCINOGENICITY
This updated review presents the evaluation of studies published from 1985 through
December 1998. The follow-up proposed by Lemen et al. (1990) of the cohort studied by
Meinhardt et al. (1982) and Downs et al. (1992), an abstract submitted for the International
Symposium, is not reviewed in this evaluation. Lemen et al. (1990) did not present any results,
and no details of study design and analysis were available for Downs et al. (1992). Since 1985,
investigators have conducted studies of workers who produce 1,3-butadiene as a raw material
(monomer production) or who use 1,3-butadiene in styrene-butadiene rubber (SBR) production
(polymer production).
7.1. MONOMER PRODUCTION
7.1.1. Texaco Cohort
7.1.1.1. Downs et al. (1987): Mortality Among Workers at a Butadiene Facility
Investigators examined a cohort of 2,586 permanent male employees who worked a
minimum of 6 months in a Texaco butadiene manufacturing plant (monomer production) that
supplied the raw material to two adjacent SBR plants studied by Meinhardt et al. (1982) and for
which an update has been proposed by Lemen et al. (1990). Data were available for the 37-year
period from January 1, 1943, through December 31, 1979. Vital status of the cohort was
determined through the Social Security Administration (SSA). Individuals whose vital status
was unverifiable through SSA were traced through the Texas Department of Public Safety.
Death certificates were obtained from the health departments of the States where the individual
resided at the time of death. When this effort was unsuccessful, the individual's name was
placed on a list, which was submitted to the health departments of Texas and Louisiana, to obtain
the death certificates. A trained nosologist coded the death certificates using the eighth revision
of the International Classification of Diseases (ICD).
Because quantitative exposure data had not been accumulated for individual workers, the
investigators used department codes to construct a qualitative exposure scale composed of four
groups: Group I, low exposure (included utility, office, and management workers, N = 432);
Group II, routine exposure (included process, laboratory, storage, and transport workers, N =
710); Group III, nonroutine exposure (included skilled maintenance workers, N = 993); and
Group IV, unknown exposures (N = 451). The investigators postulated that Group III workers
may have had exposure to higher concentrations with a lesser frequency than Group II workers.
Of 2,586 employees in the cohort, 175 (6.8%) were black. Scrutiny of death certificates
revealed that 45 blacks (7.5% of total deaths) were improperly coded as whites. At this point,
investigators conducted a preliminary analysis on the total cohort, using both black and white
7-1
-------
national death rates. The standard mortality ratios (SMRs) were higher based on black rates as
compared with white rates for four cause-specific deaths only: all lymphohematopoietic cancers
(SMR = 169 vs. 138), lymphosarcoma (SMR = 336 vs. 220), Hodgkin's disease (SMR = 135 vs.
102), and leukemia (SMR =155 vs. 119). Most of the other SMRs for both cancers and
noncancers were decreased based on black rates. Therefore, using black rates would have
underestimated the risks. Thus, the entire cohort was treated as white, and all further analyses
were conducted using white death rates.
Expected deaths were calculated using two referent populations: U.S. white males
(national comparison) and white males in a seven-county area surrounding the plants (local
comparison). The rates were standardized for age, race, sex, and calendar year. SMRs (labeled
NSMR for national comparisons and LSMR for local comparisons) were calculated in the
customary manner by dividing the observed deaths by the expected deaths and multiplying the
ratio by 100. Under the null hypothesis, the significance of the ratios of observed to expected
deaths was tested assuming that the observed (O) deaths followed a Poisson distribution using a
two-sided test and assuming ap value of <0.05 to be significant. Comparisons between Groups
I, II, and III were done by using the Mantel-Haenzel procedure for computation of relative risks
in follow-up studies with stratified data (Rothman and Boice, 1982), and power calculations were
performed using the normal approximation to the Poisson distribution (Beaumont and Breslow,
1981). The person-years at risk were not accrued until after the sixth month of employment.
A total of 64,800 person-years was accrued for the follow-up period. There were 603
deaths from 1943 through 1979; death certificates were obtained for 579 individuals (96%). The
vital status was unknown for 73 individuals (2.8% of the total cohort).
Results of this investigation indicated lower than expected mortality for these workers
from all causes (NSMR = 80,/X0.05 and LSMR = 96,/?>0.05, O = 603) and from all cancers
(NSMR = 84,/>>0.05 and LSMR = 76,/><0.05, O = 122). However, a site-specific comparison
indicated a statistically significant increase in mortality from lymphosarcoma and
reticulosarcoma (ICD code 200, NSMR = 235, 95% confidence intervals [CI] = 101-463, O = 8)
compared with national rates and a nonsignificant excess (LSMR = 182,/>>0.05) compared with
local rates.
Since the Meinhardt et al. (1982) study had indicated lymphohematopoietic cancers may
be increased among the individuals hired during wartime,1 the Texaco cohort was subdivided
into wartime workers and postwar workers, although no process change took place in monomer
production after the war. A comparison of wartime workers (N = 1,061; 452 deaths) who had
worked for at least 6 months prior to 1945 and postwar workers (N = 1,525; 151 deaths) found an
Meinhardt et al. (1982); the war/postwar comparison was related to process change.
7-2
-------
increase for all lymphohematopoietic cancers among wartime workers (NSMR = 150, 95% CI =
84-247, O = 15) and among postwar workers (NSMR = 134,/>>0.05, O = 6). However,
stratification reduced sample sizes considerably. Of a total of 7 deaths from leukemia, 6 were in
wartime workers while 4.2 were expected.
The analyses by duration of employment on mortality showed an increase among those
who worked <5 years for all lymphohematopoietic cancers (NSMR = 167,/>>0.05, O = 11), with
most of the increase attributed to leukemia (NSMR = 187,/?>0.05, O = 5) and residual
lymphohematopoietic cancers2 (i.e., non-Hodgkin's lymphoma [NHL], multiple myeloma, and
other lymphohematopoietic cancers) (NSMR = 172,/>>0.05, O = 5). Among those who worked
>5 years, a nonsignificant increase was found for all lymphohematopoietic cancers (NSMR =
127, O = 10), mainly due to an increase in residual lymphohematopoietic cancers (NSMR = 200,
0 = 7).
Further analyses were conducted for the four groups identified on the qualitative exposure
scale. For those with routine exposure (Group II), increases were noted for all
lymphohematopoietic cancers, Hodgkin's disease, and residual lymphohematopoietic cancers.
An excess of kidney cancer was also observed in this group based on one case. Similarly, in
those with nonroutine exposure (Group III), excesses were observed for all lymphohematopoietic
cancers, Hodgkin's disease, leukemia (of seven leukemia cases five were in this group), and
residual lymphohematopoietic cancers. For those in the low-exposure group (Group I), excess
mortality was seen for the same cancers (excluding Hodgkin's disease): all
lymphohematopoietic cancers, leukemia, and residual lymphohematopoietic cancers. In general,
use of local southeast Texas coastal rates resulted in lower SMRs for the above three groups
except for Hodgkin's disease in routine and nonroutine exposure groups, which showed slight
increases over national rates. Both of these SMRs were based on one observed case in each
group. None of the excess found in these three groups was statistically significant.
The comparison of Groups II, III, and IV with the low-exposure group (Group I) resulted
in inconsistent findings due to a small number of cause-specific deaths and could not be reliably
interpreted.
Analyses were also done by latency and number of years worked using national rates.
Although the results for number of years worked were inconsistent for total cancers, the SMRs
increased from 80 to 93, with increasing latency for this category. Similarly, excess SMRs for
all lymphohematopoietic deaths were observed in all latency periods (0 to 9, 20 to 29, 30 to 39)
except for 10 to 19 years. The number of years of employment results showed an inverse
relationship for these cause-specific deaths. For cause-specific deaths due to lymphosarcoma and
2Residual lymphohematopoietic cancers include ICD codes 200, 202, 203, 208, and 209.
7-3
-------
reticulosarcoma (ICD code 200), both the latency as well as number of years employed showed
an inverse relationship. The notable finding in this analysis was for workers who had a latency
of 0 to 9 years and had worked for less than 10 years (NSMR = 1,198,/K0.01, O = 4). This
increase was statistically highly significant (tested by the author of this document using the
Poisson distribution).
This is an extensively analyzed cohort mortality study. As correctly acknowledged by the
investigators, there are a few methodological limitations to this study, the major ones being a
lack of industrial hygiene (IH) data and a lack of personal work histories. In addition, half of the
total cohort worked less than 5 years in the plant. Some of the workers from this cohort had also
worked in two neighboring SBR plants. The exposures to other chemicals in the SBR plants and
in prior jobs are the confounders that were not adjusted for in this study. The cohort is relatively
small to start with, but stratification in several subgroups further reduced the power.
The major strength of the study is that it was conducted in a butadiene (monomer)
production facility in a cohort where confounding exposure from styrene was absent. The
excesses observed are in cancers of the lymphohematopoietic system, which are consistent with
cancer findings of the SBR plant workers. Most of the cases of lymphohematopoietic cancers are
concentrated in workers employed for less than 10 years and in wartime workers, which may be
due to the occurrence of higher exposures during wartime years. The exposures during
subsequent periods were lower. Thus, the finding of excess cancer mortality in short-term
employees is not evidence against dose-response relationship.
7.1.1.2. Divine (1990): An Update on Mortality Among Workers at a 1,3-Butadiene
Facility—Preliminary Results
In 1990, Divine reported an updated analysis of the same Texaco plant (monomer
production) cohort. The follow-up on the original cohort was extended through 1985 (adding 6
more years) by updating the information on workers from company data and the SSA. Death
certificates were obtained from the health departments of Texas, Louisiana, Ohio, and
Mississippi and were coded by a trained nosologist according to the eighth revision of the ICD.
The National Death Index (NDI) records were searched for workers for whom the SSA failed to
provide the vital status.
Mortality analyses were performed using Monson's computer program (Monson, 1974).
Again, the white male death rates of the U.S. population were used due to uncertainties about
race information in the company files and because there were few blacks in the cohort. Person-
years were accrued similarly to the Downs et al. (1987) study.
The qualitative exposure categories remained the same. IH sampling data at the time of
this study supported the exposure categories developed earlier. For this study, lymphosarcoma
7-4
-------
(ICD code 200) was reported separately from the cancers of other lymphatic tissues (ICD codes
202, 203, and 208).
A total of 74,219 person-years had accrued through 1985. The number of deaths had
increased to 826, and death certificates were not available for 49 individuals (6%). Of 2,582
employees in the cohort,3 1,708 individuals were still alive and 48 (1.9%) were lost to follow-up.
Overall, the pattern of results was unchanged from the report by Downs et al. (1987) for this
cohort. For the total cohort, the SMRs for all lymphohematopoietic cancers and Hodgkin's
disease were increased but not significantly; however, for lymphosarcoma and reticulosarcoma,
the excess was significantly larger (SMR = 229, 95% CI = 104-435, O = 9) and accounted almost
entirely for the increase in overall lymphohematopoietic cancers. Analyses by various
subcohorts also yielded results similar to those observed in the earlier study (Downs et al., 1987).
The highest increase was observed in lymphosarcoma and reticulosarcoma among workers who
had worked more than 5 years but less than 10 years (SMR = 245, 95% CI = 79-572, O = 5).
Analyses by wartime and postwar subcohorts demonstrated a statistically significant increase
among the wartime subcohort for the same cause-specific deaths (SMR = 269, 95% CI = 108-
555, O = 7), while an excess in the postwar subcohort had two deaths from lymphosarcoma and
reticulosarcoma and was not statistically significant. There was one additional death from
leukemia, bringing the total to eight deaths in the total cohort; this death occurred in the postwar
subcohort.
Among the subcohorts based on exposure levels, the only statistically significant excess
was observed for lymphosarcoma and reticulosarcoma among workers who were ever employed
in the routine exposure category, exposure Group II (SMR = 561, 95% CI = 181-1,310, O = 5).
Among workers who were ever employed in the nonroutine exposure category (Group III), an
excess was observed for all lymphohematopoietic cancers based on 11 deaths, mainly due to an
increase in leukemia (SMR = 185, 95% CI = 68-403, O = 6). Lymphosarcoma in this group was
slightly increased based on two cases.
For the total cohort, no pattern with latency or duration of years worked was observed for
either all deaths or total cancer deaths. For all lymphohematopoietic cancers, excesses were
observed in the latency groups of 30+ years (SMR = 205, O = 8) and 0 to 9 years (SMR = 200,
O = 4). Both of these groups had worked less than 10 years. Deaths from lymphosarcoma were
also increased in the same duration and latency groups. For duration of less than 10 years and
latency of 30+ year and 0 to 9 year groups, the SMRs were 3,333 (O = 2) and 1,333 (O = 4),
respectively. For leukemia the SMRs increased with increasing latency among the workers who
had worked less than 10 years. The SMRs were 125 (O = 1), 182 (O = 2), and 250 (O = 4)
The cohort was reduced to 2,582 from 2,586 because of some double counting and some coding error where a female was coded as male.
7-5
-------
respectively for latency of 10-19 year, 20-29 year, and 30+ year groups, respectively. No
statistical test results were presented for this analysis. Similar analyses by different exposure
groups failed to show any pattern for all lymphohematopoietic deaths and lymphosarcoma deaths
among low-exposure and unknown-exposure groups. Among routinely exposed groups, the
excesses were observed for the same two latency and duration groups as for the total cohort,
whereas for nonroutine exposure the excesses were observed only for 20 to 29 and 30+ years'
latency groups who had worked for less than 10 years. All of these excesses were based on <3
deaths in each group, making interpretation of these findings by exposure levels very difficult.
The investigator stipulated that the increased lymphosarcoma and reticulosarcoma among
wartime workers may be due to higher exposures during wartime, or that this group was
somehow different from those hired at other times; perhaps they were not eligible for the draft
because of age or health reasons. It should be noted that although the excesses in leukemia
deaths were not statistically significant, they occurred in workers who had worked less than 10
years and increased with increasing latency.
This also is a well-conducted study; unfortunately, the same methodological limitations
that were present in the Downs et al. (1987) study are applicable to this study. However, the
findings of this study are consistent with the earlier study as well as with other SBR plant
studies.
7.1.1.3. Divine et al. (1993): Cancer Mortality Among Workers at a Butadiene Production
Facility
This update added another 5 years of follow-up to the earlier cohort of monomer workers
(Divine, 1990). Cohort inclusion criteria remained the same but were extended from December
31, 1979, to December 31, 1990. This yielded additional workers, resulting in a total cohort of
2,749 individuals. The four exposure groups were similar to those used in earlier studies, with
slight changes as follows. The background exposure group (included office utility, warehouse,
and transportation workers, N = 347). This group was called the low-exposure group in the
previous two studies (Downs et al., 1987; Divine, 1990). The low-exposure group (included
workers from operating units, planners and engineers, welders, carpenters, and workers from
brick masons, N = 958). This group was a combination of some of the low-exposure and all of
the unknown-exposure group from the previous two studies. The nonroutine exposure group
(included skilled maintenance workers such as pipefitters, tinsmiths, instrument and electrical
workers, and insulators, N = 865). The routine exposure group (included process, lab, storage,
and transport workers, N = 1,056). The last two categories appeared to be the same as in the
earlier two studies; the change in the number of individuals in these categories was due to the use
of complete work histories rather than the last assigned department. For this study, the
7-6
-------
investigators reviewed the results of the IH data and information obtained from plant personnel
and found that the main difference between the routine and nonroutine exposure groups was in
the frequency and not the intensity of exposure.
Monson's computer program (Monson, 1974) was used for analysis of this study also.
All the analytical methods included use of white male death rates of the U.S. population (because
there were very few blacks in the study, they were assumed to be white for the analysis) and
calculation of person-years. The follow-up procedures and acquisition of death certificates were
the same as in an earlier study by Divine (1990).
A total of 83,591 person-years was accrued. At the end of the follow-up period, 1,660
individuals were still alive, 38 were lost to follow-up, and 1,051 were deceased (death certificates
were obtained for 1,036 individuals).
The overall results observed in this study were similar to the earlier two studies. The
statistically significant elevated SMRs were observed for lymphatic and hematopoietic cancers
(SMR =171, 95% CI = 104-264, O = 20) and for lymphosarcoma and reticulosarcoma for
workers employed for less than 5 years (SMR = 286, 95% CI = 104-622, O = 6). There were 3
additional leukemia deaths in total cohort, bringing the total to 11 deaths. For workers who were
employed for less than 5 years, the SMR for leukemia was 170 (95% CI = 73-335, O = 8).
Analysis by wartime and postwar subcohorts found an increase in lymphosarcoma and
reticulosarcoma in the wartime employees (SMR = 254, 95% CI = 102-523, O = 7). There were
seven leukemia deaths in this group versus four in the postwar subcohort. The analysis by
exposure group showed an increase for lymphosarcoma and reticulosarcoma in the routine
exposure group (SMR = 452, 95% CI = 165-984, O = 6). There were six leukemia deaths in the
nonroutine exposure group. Analysis by latency and duration of employment yielded the largest
increase in 0 to 9 years' latency for individuals employed for less than 5 years (possibly wartime
employees) for lymphosarcoma and reticulosarcoma. The SMR was 3,333 based on 2 observed
cases. No statistical test results were presented for this analysis.
Again, the investigators stipulated that the observed excesses in all lymphatic and
hematopoietic cancers, Hodgkin's disease, and leukemia were among those employees who were
hired during the war and among the nonroutinely exposed group who had worked for less than 5
years. They observed that this group was employed in both the period and the exposure category
in which exposures were highest. They also acknowledged that although the complete work
histories were available, the IH data were available only for the last 10 years of the study.
7-7
-------
7.1.1.4. Divine and Hartman (1996): Mortality Update of Butadiene Production Workers
This recent follow-up of the same cohort added 46 more individuals to the cohort (a total
of 2,795) by extending the inclusion criteria and the follow-up period through December 31,
1994. The person-years accrued increased to 85,581. Of the 2,795 individuals, 999 were still
alive, 574 were lost to follow-up (all but 28 were known to be alive), and 1,222 were deceased
(death certificates were obtained for 1,202 individuals). The follow-up procedures and analytical
techniques (for SMR analysis) were the same as for earlier studies. The exposure categories
were changed to background-exposure group (same as 1993 follow-up), low-exposure group
(same as 1993 follow-up), and varied-exposure group (comprised nonroutine and routine
exposure groups of 1993).
Based on IH data available since 1980, each employee's potential exposure to butadiene
was estimated by separating the employee's work history by job categories into 1-year segments.
Two variables were used to calculate the estimated exposure (job categories and calendar time
periods). There were six exposure classes based on job categories: 0, 1, 2, 3, 4, and 5 with 0,
0.1, 0.2, 0.3, 0.4, and 0.5 weights (wt), respectively, and five calendar time periods: <1946 (wt =
10), 1946-59 (wt = 8), 1960-76 (wt = 4), 1977-85 (wt = 2), and 1986-94 (wt = 1). The
cumulative exposure was obtained for each individual by summing the scores for all the years of
employment. These exposure estimates were used to conduct survival analyses for (1) total
lymphohematopoietic cancer, (2) lymphosarcoma, (3) NHL, (4) multiple myeloma, and (5)
leukemias.
Three different models were used for the survival analysis: a Cox proportional hazard
model with a time-dependent estimate of cumulative exposure, a person-time logistic regression
model with a time-dependent estimate of cumulative exposure, and a nested case-control model
using conditional logistic regression. Each case had 10 matched controls (by date of birth
[+2 years]). The selection of controls without replacement was from noncases at the time of the
occurrence of each case.
The results of the SMR analyses were very similar to the two earlier follow-up studies of
this cohort (Divine, 1990; Divine et al., 1993). The SMRs for the all lymphatic and
hematopoietic cancers were increased in the total cohort (SMR = 147, 95% CI = 106-198,
O = 42) and in the employees who worked for less than 5 years (SMR = 162, 95% CI = 101-245,
O = 22). Although the SMR for lymphosarcoma and reticulosarcoma for this group was still
elevated (SMR = 261, 95% CI = 95-568, O = 6), it was no longer statistically significant.
Similarly, in wartime workers the elevated SMR for lymphosarcoma and reticulosarcoma barely
missed statistical significance (SMR = 241, 95% CI = 97-497, O = 7). It is not explained why
the number of wartime employees changed from 1,066 in 1990 (Divine) to 1,050 in 1993 (Divine
et al.) and to 1,048 in 1996 (Divine and Hartman). The analysis by exposure groups showed
7-8
-------
excess SMRs for all lymphatic and hematopoietic cancers (SMR = 172, 95% CI = 117-244) and
for lymphosarcoma and reticulosarcoma (SMR = 249, 95% CI = 100-513, O = 7), both in varied
exposure group. There were 2 additional deaths from leukemia, bringing the total number of
deaths to 13, of which 11 were in the varied-exposure group (SMR = 154, 95% CI = 77-257).
All of the leukemia decedents were first employed in the 1940s, and eight had worked for less
than 5 years.
Except for the risk factor "Hire-age" in the Cox model, the survival analyses failed to
show any significant increase in the risk ratios, in any cause-specific cancer, by any of the three
methods. The Cox model, which used time-dependent estimates of exposures, showed
significant excesses in relative risks (RR) for Hire-age and occurrence of all lymphatic and
hematopoietic cancers (RR = 1.08, 95%CI= 1.05-1.12), leukemia (RR= 1.08, 95% CI = 1.02-
1.15), lymphosarcoma (RR-1.07, 95% CI = 1.00-1.14), and NHL (RR = 1.09, 95% CI = 1.04-
1.14).
Moreover, the investigators point out that the observed statistically significant excess
SMR in the total cohort for all lymphatic and hematopoietic cancers is due to increased SMRs
for lymphosarcoma and cancer of other lymphatic tissue. They state that there were no
additional deaths from lymphosarcoma since 1981. Of nine lymphosarcoma deaths eight
belonged in persons first employed before 1950, six had worked for less than 5 years, seven were
hired during WW II, and seven belonged in the varied-exposure group.
In contrast, the investigators found an additional five deaths of cancers of other lymphatic
tissue since their last follow-up in 1993. Four of these deaths were due to lymphomas, bringing
the total to 15 deaths from this cause. Fourteen of the decedents were employed prior to 1950.
The investigators further state that the observed excess SMR for this category in the varied-
exposure group increased with increasing length of employment.
The investigators go on to expostulate that per new diagnostic criteria lymphosarcoma is
now classified as lymphoma, and thus is included under the category of NHL.4
Under the previous classification (8h ICD, Adapted), lymphohematopoietic cancers comprised the following subcategories: lymphosarcoma and
reticular sarcoma, Hodgkin's disease, leukemia, and other lymphatic tissue cancers. In 1994, the International Lymphoma Group's Revised
European-American Lymphoma (REAL) classification was proposed for the lymphohematopoietic cancers, and is being adopted into the ICD-O
(Berard and Hutchison, 1997). This classification is based on new ideas evolving in the fields of molecular biology, genetics, and immunology,
which have rendered the old classification for lymphohematopoietic cancers obsolete. The REAL classification comprises the following
subcategories: B-cell neoplasms, T-cell and putative natural killer (NK)-cell neoplasms, Hodgkin's disease, and unclassified lymphomas. It should
be noted that both leukemias and lymphomas that are produced by B-cells are included under B-cell neoplasms, and leukemias and lymphomas
produced by T-cells and NK-cells are included under T-cell and NK-cell neoplasms. It should also be noted that any lymphoma (such as B-cell, T-
cell, and NK-cell) that is not classified as Hodgkin's disease is also included under non-Hodgkin's lymphoma.
Furthermore, the Leukemia Society of America (1999) defines lymphohematopoietic cancers as follows: "Leukemia, Lymphoma, Hodgkin's
disease, and Myeloma are cancers of the body's blood forming and immune systems: the bone marrow and lymph nodes. They are considered to
be related cancers because they involve the uncontrolled growth of cells with similar functions."
7-9
-------
Thus they computed the SMR of 176 (95% CI = 103-282) for the observed 17 deaths from NHL.
This was based on the assumption that 50% of the deaths from cancers of other lymphatic tissue
were from lymphomas. Out of 17 decedents, 16 were hired before 1950, 6 had worked for more
than 5 years, and 12 belonged in the varied-exposure group.
Just as in the earlier two follow-ups (Divine, 1990; Divine et al., 1993), the investigators
stipulate that the above findings were consistent with the hypothesis that high exposures to 1,3-
butadiene during wartime may be associated with NHL. Since the investigators did not find any
association between cumulative exposure and occurrence of non-Hodgkin lymphoma in their
survival analysis, they suggest that peak exposures rather than cumulative exposures may be
associated with the observed increase. Furthermore, they acknowledge that there are no data or
means available to estimate the peak exposures. In addition, they note that extremely high peak
exposures did exist during the 1950s and 1960s. Also, there is a lack of any information on other
confounding factors.
Although the investigators have done a good job of estimating the exposure and have
conducted various analyses, by their own admission they may have used an incorrect category of
cumulative exposures. The survival analysis using the Cox model did show statistically
significant risk ratios for the risk factor "Hire-age" in four out of five cancer types evaluated.
Time-dependent exposure estimates were used for the Cox model. With every update, the
cohort inclusion date has been extended, adding more employees. Addition of recent employees,
younger in age and with shorter follow-up, may have resulted in dilution of observed risk.
There are a few limitations to this study, some of which are correctly pointed out by the
investigators. IH data are available only from 1980. Fifty percent of the cohort worked for less
than 5 years. That means they left because they were old (during wartime, mainly the older
population was available to work), or sick, or found other jobs and were exposed to other
chemicals. The cohort is small, and the cause-specific mortality of NHL is relatively rare. The
investigators assert the occurrence of extremely high peak exposures in the 1950s and 1960s has
been mentioned, but it is not known how these peak exposures compared with those in wartime.
Moreover, since the 1950s, exposure to 1,3-butadiene in the monomer facility has declined.
Therefore, considering the very low overall exposure to 1,3-butadiene and the modest size of this
cohort, there may not be enough power in the study to detect a leukemia excess.
However, this is still the largest cohort exposed to monomer. Qualitative measures of
exposure and IH data since 1980 are available. Upon completion of this study, this cohort will
have 52 years of follow-up.
7-10
-------
7.1.2. Shell Oil Refinery Cohort
7.1.2.1. Cowlesetal. (1994): Mortality, Morbidity, and Hematological Results From a
Cohort of Long-Term Workers Involved in 1,3-Butadiene Monomer Production
Shell Oil's Deer Park Refinery produced a butadiene monomer from 1941 to 1948 and
1970 to the present. The cohort consisted of male workers who had a minimum of 5 years of
employment in jobs with potential exposure to butadiene or at least 50% of their total duration of
employment (minimum of 3 months) in these jobs. This facility also had several other refinery
operations and chemical production units. Three different analyses were performed on this
cohort: (1) mortality, (2) morbidity, and (3) hematological.
1. Mortality Analysis:
A total of 614 employees composed the cohort. The follow-up period was from 1948 to
December 31, 1989. Vital status was assessed from company records, SSA, master beneficiary
files, and the NDI. Death certificates were obtained for all the deceased workers and coded by a
trained nosologist according to the revision of the ICD in effect at the time of death. Mortality
rates of Harris County, TX, were used to compute the age-, race-, and calendar year-adjusted
SMRs, using the Occupational Cohort Mortality Analysis Program (OCMAP) from the
University of Pittsburgh.
A total of 7,232 person-years were accrued. Of 614 employees, 589 were still alive, 1
was lost to follow-up, and 24 were dead. No excess mortality, either for total deaths or total
cancers (including cause-specific cancers), was observed.
2. Morbidity Analysis:
Original cohort members who were active at some time between January 1, 1982, and
December 31, 1989, qualified for the morbidity study. Morbidity data were obtained from the
Shell Health Surveillance System. The follow-up period was from 1982 to 1991. Causes of
morbidity were coded according to the 9th revision clinical modification of the ICD. Morbidity
ratios (SMbRs) were calculated by using the internal comparison group of employees who were
active during the same time period and had no exposure to butadiene.
A total of 438 employees were included in this analysis. No excess morbidity by any
cause was observed.
3. Hematological Data Analysis:
Of 438 individuals included in the morbidity study, periodic hematological data were
available for 429 individuals. These hematological data reveal that seven hematological
outcomes were measured (between 1985 and 1991). The most recent laboratory test results were
7-11
-------
used for the analysis. Comparisons were done with similar results from 2,600 nonexposed
employees. No differences were observed between butadiene-exposed vs. nonexposed groups.
This study has quite a few methodological limitations. The cohort is small, and deaths
are few. No exposure information was available. The number of employees selected for this
study from the time period 1941-1948, when exposure was probably higher, is unclear. Over
50% of the cohort was hired in 1970 or later, with an average follow-up of 12 years. This means
that the cohort was still young, showing "healthy worker" effect, and enough latent period had
not elapsed to show increases in cancers, which usually have a long latent period. Moreover, the
power calculations computed by the investigators showed that the study size and duration were
sufficient to detect a fivefold excess in lymphohematopoietic cancer. Thus, the absence of
positive results in this study does not provide any negative evidence toward the causal
association between butadiene and occurrence of cancer.
7.1.3. Union Carbide Cohort
7.1.3.1. Wardetal. (1995): Mortality Study of Workers in 1,3-Butadiene Production Units
Identified From a Chemical Workers Cohort
Ward et al. (1996c): Mortality Study of Workers Employed in 1,3-Butadiene
Production Units Identified From a Large Chemical Workers Cohort
The study cohort was selected from 29,139 workers at three Union Carbide Corporation
facilities in the Kanawha Valley, West Virginia. A total of 527 male workers who had worked
between 1940 and 1979 were identified from work history records as having ever worked in the
departments where there was a potential for butadiene exposure. Only the individuals who
worked in these departments during the butadiene production period were selected for the study
(i.e., 364 individuals). The vital status was determined through December 31, 1990, using the
NDI. Death certificates were obtained for decedents and coded according to the revision of the
ICD codes in effect at the time of death. Both U.S. and Kanawha County mortality rates were
used for comparison. A modified life table analysis developed by the National Institute for
Occupational Safety and Health (NIOSH) was used to compute the SMRs.
Of 364 workers, 176 were alive, 3 were lost to follow-up, and 185 were dead at the end of
1990. The SMR for all causes was 91, while for all cancers it was 105. Neither of them were
statistically significant. The only statistically significant increase was observed for
lymphosarcoma and reticulosarcoma, based on four cases (SMR = 577, 95% CI = 157-148). A
county-based comparison yielded a similar result. By duration of employment and latency, a
statistically significant excess of the SMR was observed among workers who were employed for
more than 2 years and with more than 30 years of latency (SMR = 1980, 95% CI = 408-5,780, O
7-12
-------
= 3). Of these four cases, three had started their employment between 1942 and 1946 (i.e.,
wartime), while one had started in 1952.
The investigators stated that except for butadiene exposure, there were no common
exposures to other chemicals in the four individuals who had died of lymphosarcoma and
reticulosarcoma, although two of them had been assigned to an acetaldehyde unit for some time.
Investigators considered the possibility that the risk of lymphosarcoma and reticulosarcoma may
be attributable to exposure to acetaldehyde. Hence they identified a subcohort of 233 individuals
who were ever assigned to an acetaldehyde unit. Of 48 deaths identified from this subcohort,
except for the two deaths of lymphosarcoma and reticulosarcoma previously identified among
the workers included in the butadiene production cohort, there were no additional deaths found in
this subcohort from lymphosarcoma and reticulosarcoma.
This study has a few methodological limitations. The cohort is very small, no
adjustments for confounding exposures to other chemicals were done, and no exposure
information is available. The qualitative exposure is assumed based on the department coded for
potential exposure to butadiene. It is still interesting to note that the exposure in these plants was
to butadiene monomer alone, either in the production process or the recovery from the olefin
cracking process, and not to styrene-butadiene polymer. The only other cohort exposed to
butadiene monomer (Downs et al., 1987; Divine, 1990; Divine et al., 1993; Divine and Hartman,
1996) also found excess in lymphosarcoma and reticulosarcoma in the wartime subcohort.
Although by itself this study would not have furnished compelling evidence of positive
association between the exposure to 1,3-butadiene monomer and occurrence of lymphosarcoma,
it supports the Texaco cohort findings.
Studies in monomer production workers are summarized in Table 7-1.
7.2. POLYMER PRODUCTION
7.2.1. Cohort Identified by Johns Hopkins University (JHU) Investigators
7.2.1.1. Matanoski and Schwartz (1987): Mortality of Workers in Styrene-Butadiene
Polymer Production
This cohort mortality study of SBR polymer production workers from eight plants (seven
U.S. and one Canadian) was reviewed in a 1985 document (U.S. EPA, 1985). At that time, this
study was submitted to the U.S. Environmental Protection Agency but was not published.
Because the findings of the published study are essentially the same, it will not be reviewed
again.
7-13
-------
Table 7-1. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—monomer production.
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and
limitations
Downs etal. (1987)
2,586 permanent male employee
cohort mortality
Worked for at least a minimum
of 6 months from January 1,
1943-December31, 1979
Follow-up from 1943 through
1979 (37 years)
Comparison group U.S.
population (national) and 7
counties surrounding the plants
(local)
Four exposure groups based
on qualitative exposure scale:
Group I, low (N = 432)
Group II, routine (N = 710)
Group III, nonroutine
(N = 993)
Group IV, unknown
(N = 451)
SS NSMR = 235 and SNS
LSMR = 182 for
lymphosarcoma and
reticulosarcoma for total
cohort
SS NSMR R= 1,198 for
lymphosarcoma and
reticulosarcoma for latency
of 0-9 years and <10 years of
employment
Cohort of monomer production
workers, a major strength
Lack of IH data
!/2 the cohort worked less than 5
years in the plant
Relatively small cohort; therefore
hard to interpret results after
further stratification
Lack of adjustment for
confounding for people who
worked in SBR plant too
Divine (1990)
Update of the cohort from
Downs et al. (1987)
Cohort reduced to 2,582
Follow-up extended through
1985
Comparison group U.S.
population
Same exposure groups as the
earlier study
For lymphosarcoma and
reticulosarcoma:
SSSMR = 229 for total
cohort
SS SMR = 269 for prewar
subcohort
SS SMR = 561 for routinely
exposed for less than 10
years
No pattern with latency or
duration of employment
Same methodologic limitations as
the earlier study
-------
Table 7-1. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—monomer production (continued)
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and
limitations
Divine etal. (1993)
Update of the cohort from
Divine (1990)
Cohort increased to 2,749 as the
inclusion period extended
through December 31, 1990
Follow-up extended through
1990
Comparison group U.S.
population
Similar exposure groups as
earlier with some
redistribution of workers
Group I, background
(N = 347)
Group II, low (N = 958)
Group III, nonroutine
(N = 865)
Group IV, routine
(N= 1,056)
For lymphosarcoma and
reticulosarcoma
SS SMR = 254 for prewar
subcohort
SS SMR = 286 for workers
employed less than 5 years
Same strengths and limitations as
earlier study
Divine and Hartman (1996)
Update of the cohort from
Divine etal. (1993)
Cohort increased to 2,795 as the
inclusion period extended
through December 31, 1994
Follow-up extended through
1994
Comparison group U.S.
population
Internal comparison
Based on IH data and work
histories using
6 exposure classes
5 calendar periods
Individual exposures were
estimated for each worker
Three different models used
for the survival analysis
Results of SMR analysis
were similar to earlier
studies
Analysis by NHL category
resulted in SS SMR=176
Except for "Hire-age" in Cox
analysis which showed SS
RRs for four out of five
cancers, survival analysis
failed to show any SS excess
in any cause-specific cancer
by other analytical methods
52 years' follow-up
Exposure estimation useful
Major limitation is no peak
exposure estimation available in
war subcohort, which has the SS
lymphosarcoma excess
-------
Table 7-1. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—monomer production (continued).
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and
limitations
Cowlesetal. (1994)
Cohort of monomer production
workers from 1941-1948 and
from 1970-1994
5 years or 50% of total duration
worked in jobs with potential to
1,3-butadiene exposure
Mortality follow-up from 1948-
1989 (614 employees)
Morbidity follow-up from 1982-
1989 (438 employees)
Hematologic data analyses
1985-1991 (429 employees)
None
No excess observed in either
mortality or morbidity study
No hematologic differences
found between the exposed
and nonexposed employees
Very small cohort
Exposure is not certain
Deaths are very few
50% of cohort hired after 1970
when exposures were low
Not enough latent period has
elapsed
Ward et al. (1995, 1996c)
Cohort of 364 male employees
who had worked between 1940
and 1979
Employees who had worked in
monomer production during
World War II
Follow-up through December
31, 1990
U.S. population
Kanawha County population
Departments where
production of 1,3-butadiene
occurred
For lymphosarcoma and
reticulosarcoma
SS SMR = 577
SS SMR = 1980 for more
than 2 years of employment
and 30 years of latency
Small cohort
Adjustment for work in
acetaldehyde unit showed no
association with observed
lymphosarcoma deaths
No adjustments for confounding
exposures to other chemicals
No exposure information
available
SS = Statistically significant.
SNS = Statistically nonsignificant.
NSMR = National standard mortality ratios.
LSMR = Local standard mortality ratios.
IH = Industrial hygiene.
-------
7.2.1.2. Matanoski et al (1989): Epidemiologic Data Related to Health Effects of
1,3-Butadiene
Matanoski et al. (1990): Mortality of a Cohort of Workers in the Styrene-Butadiene
Polymer Manufacturing Industry (1943-1982)
These two publications essentially reported the same updated reanalysis of the earlier
cohort. In addition, Matanoski et al. (1989) presented the results of the nested case-control study
in this population. Three methodological differences in the original analysis (Matanoski et al.,
1987) and the reanalysis presented in these two publications should be noted: extension of
follow-up through 1982, fewer workers whose vital status was unknown (3.4% vs.
6.6% in the earlier report), and deletion of workers from the Canadian plant who had relatively
short-term exposure (i.e., workers who had worked for less than 10 years or who had not reached
the age of 45 during employment). Analytical methods were essentially unchanged from the
earlier analysis.
In addition to information received from the SSA and the Motor Vehicle Administration,
follow-up through local plant beneficiary records and the NDI was done to assess the vital status
of the study cohort. Follow-up procedures for Canadian workers were similar to the earlier
study. Death certificates were obtained from local health departments. The total cohort was
reduced from 13,920 to 13,422 in this study. Of 12,113 workers for whom vital status was
successfully traced, 23% (2,784) were still working in the plants, 53.4% (6,472) were alive but
not working in the plants, 20.2% (2,441) had died, and vital status was unknown for 3.4% (416).
The racial distribution was 75% whites, 10% blacks, 15% unknown (presumed to be white for
the analysis), and less than 1% other.5 Death certificates were obtained for 97.2% of the
deceased individuals and were coded by a trained senior nosologist, using the eighth revision of
the ICD.
Data analyses were done using age, race, calendar time, and cause-specific U.S.
population rates. A modified life-table program by Monson (1974) was used. The person-years
were calculated through December 31, 1982. First-year work experience was omitted from
person-years because one of the inclusion criteria was that an individual had to have worked for
at least 1 year. A total of 251,431 person-years were accrued, of which 226,475 were contributed
by whites.
Statistically significant lower SMRs for all causes of death (81) and for all cancers (85)
were virtually the same as in earlier studies. The SMRs for all causes of death by 5-year calendar
period demonstrated increasing SMRs with increasing time period, indicating a "healthy worker"
effect in earlier calendar years. Blacks showed higher SMRs than whites in later years. A
statistically significant excess for all causes of deaths was observed for blacks in the last 3 years
The percentages, which are quoted from the paper, add up to 101. This is due to rounding of the numbers by the authors of the paper.
7-17
-------
of follow-up (SMR = 134, 95% CI = 101-175, O = 54). Most of the cause-specific cancer SMRs
showed deficits in both races. A few cancer sites demonstrated excess mortality in both races.
Among whites, excesses were observed for esophageal cancer, kidney cancer, Hodgkin's disease,
and other lymphohematopoietic cancers. Among blacks, excesses were observed for stomach,
liver, and prostate cancer; all lymphohematopoietic cancers; lymphosarcoma; leukemia; and
other lymphohematopoietic cancers. None of the excesses were statistically significant.
Because the risks for kidney, digestive, and lymphohematopoietic system cancers
approached those of the reference population, which was unusual for an occupational cohort with
low overall risks, investigators further analyzed the data by work areas. For production workers,
deaths from lymphohematopoietic cancers, Hodgkin's disease, and leukemia were
nonsignificantly increased for the total cohort and among whites (except for leukemia). The only
significant excess observed for the total cohort was for other lymphohematopoietic cancers,
which included NHL and multiple myeloma (SMR = 260, 95% CI = 119-494, O = 9). Among
blacks, however, statistically significant excesses were observed for all lymphohematopoietic
cancers (SMR = 507, 95% CI = 187-1,107, O = 6) and leukemia (SMR = 655, 95% CI = 135-
1,906, O = 3). The other two excesses observed among blacks for lymphosarcoma and other
lymphohematopoietic cancers (including NHL and multiple myeloma) were based on one and
two cases, respectively, none being statistically significant (p>0.05).
Among white maintenance workers, no excesses of lymphohematopoietic cancers were
found, with the exception of Hodgkin's disease (SMR = 170, 95% CI = 35-495), based on only
three deaths. However, rates were nonsignificantly increased for digestive tract malignancies
(i.e., esophagus, stomach, and large intestine). Among black maintenance workers,
nonsignificant excesses were observed for cancer of the rectum and stomach. For utility
workers, the numbers were reported to be too small to reach firm conclusions about risks. For
the "other" category of workers (including laboratory workers, management, and administrative
workers), excesses were observed for Hodgkin's (SMR = 130, 95% CI = 16-472, O = 2) and
leukemia (SMR =116, 95% CI = 43-252, O = 6) among whites and for leukemia (SMR = 246,
95% CI not given, O = 1) among blacks. Nonsignificant increased SMRs for the digestive
system among blacks were also observed for the stomach, liver, and pancreas, all of which were
based on fewer than five cases.
Analysis by duration of work or latency for the total cohort did not show an increase in
hematopoietic cancers.
This is still the largest cohort of SBR workers. The increased follow-up, better tracing,
and exclusion of short-term workers from the Canadian plant have resulted in demonstrating the
excess mortality from malignancies of the lymphohematopoietic system, digestive system, and
kidney. However, the limitations of the earlier study of this cohort (i.e., lack of exposure data
and inclusion of less than 50% of the population in the follow-up cohort) still exist. The
7-18
-------
magnitude of the bias introduced by exclusion of workers (2,391) due to missing information on
total work history or crucial information such as date of birth could be substantial. Although an
attempt was made to correct for race, the race was unknown for 15% of the eligible cohort, and
this segment was assumed to be white for the analysis. This would result in an overestimation of
rates in blacks and an underestimation of rates in whites. No explanation was given as to how
the total eligible population of 13,422 was reduced to 12,113. No data were presented by
individual plants, but as indicated in the earlier study, only four plants had follow-up starting
from 1943, whereas in the other four plants the starting dates of the follow-up ranged from 1957
to 1970; thus, these latter four plants may not have had long enough follow-up for malignancies
to develop.
7.2.1.3. Matanoski et al (1989): Epidemiologic Data Related to Health Effects of
1,3-Butadiene
Santos-Burgoa et al. (1992): Lymphohematopoietic Cancer in Styrene-Butadiene
Polymerization Workers
To elucidate the separate contributions of 1,3-butadiene and styrene to the cancers
identified in the updated cohort study, a nested case-control study of this cohort of SBR workers
was conducted using estimates of exposure to 1,3-butadiene and to styrene for each job.
Fifty-nine cases and 193 controls (matched for duration of work) were included in the analysis.
Among the case group were 26 cases of leukemia; 18 of other lymphatic cancers, which included
10 multiple myelomas and 7 NHLs; 8 Hodgkin's lymphomas; and 6 lymphosarcomas.
Cases (workers who had lymphohematopoietic cancer as either the underlying or
contributory cause of death on death certificates) arose from the original eight plants with the
same selection criteria for the eligibility of that cohort (13,422), with the exception of the
Canadian plant. For the Canadian plant, the restriction of either 10 years of work or those who
had reached age 45 during employment was dropped from the selection of cases, which added
two more cases to lymphohematopoietic cancers. Another four cases were added in which
individuals had died of another cause of death but had a lymphohematopoietic cancer at the time
of their death. Two cases were deleted from the final analysis, one lymphosarcoma due to lack
of any controls and one NHL due to lack of job records from which exposure could be identified.
Controls included workers from the same cohort who were alive or had died of any cause
other than malignant neoplasms. Controls were individually matched to cases by plant; age; hire
year; employment as long as or longer than the case; and, if the control was dead, then survival to
the death of the case. Based on these criteria, an average of 3.3 controls per case was selected
instead of 4 controls per case as intended by the investigators. This average of 3.3 controls per
one case had more than a 90% chance of detecting the twofold risk from exposure to 1,3-
7-19
-------
butadiene. Both cases and controls had about 15 years of employment and were hired at 36 to 37
years of age, somewhat older than usually seen in occupational populations.
Exposures to 1,3-butadiene and styrene were calculated from the job records of each
subject, the number of months each job was held, and an estimate of the 1,3-butadiene and
styrene exposure levels associated with that job. Both the job identification and exposure
estimation were done independently and without knowledge of case or control status of the
subjects. To estimate 1,3-butadiene and styrene exposures, all jobs within the rubber industry
were ranked from 0 to 10 by a group of senior engineers with many years of experience in the
industry. One-third of the jobs were determined to have no routine exposure, but almost all jobs
were thought to have intermittent exposure. Cumulative dose for both styrene and 1,3-butadiene
was calculated using the score and duration for each job in the participants' work history.
Because the distribution of exposure scores was skewed to the right, a log transformation of the
scores was used in the analyses. As the logarithmic transformation approached normal
distribution, only the transformed exposure variables were used for the analyses.
Analyses were done by using "ever/never exposed" categories to both butadiene and
styrene and using high-exposure vs. low-exposure groups (based on mean log exposure
cumulative rank for each substance determined by combining cases and controls). Both
conditional (matched) and unconditional (unmatched) logistic regression analyses were
performed. Odds ratios (OR) for matched sets were then calculated based on maximum
likelihood estimates of the OR, and test-based confidence limits around the OR were calculated.
Unadjusted for the presence of the other chemicals and unmatched, analyses by
"ever/never exposed" to butadiene and styrene found significantly increased relative odds for
leukemia for both high and low exposures. Relative odds for butadiene were 6.82 (95% CI =
1.10-42.23) and for styrene were 4.26 (95% CI = 1.02-17.78).
Nonsignificant excesses were also observed for all lymphohematopoietic and other
lymphohematopoietic cancers for exposures to both butadiene and styrene. Other excesses were
for Hodgkin's disease among workers exposed to butadiene and lymphosarcoma among workers
exposed to styrene.
Matched analyses demonstrated that risk for all lymphohematopoietic neoplasms was
significantly increased among workers exposed to butadiene (OR = 2.30, 95% CI = 1.13-4.71).
Separate evaluation of these neoplasms revealed that most of the association could be explained
by a significant excess risk for leukemia (OR = 9.36, 95% CI = 2.05-22.94), but other cancers in
this group were not significantly elevated. Leukemia also showed a threefold increase associated
with styrene exposure (OR = 3.13, 95% CI = 84-112).
Conditional logistic regression was used to separate the risks associated with each of
these substances. Again, there was a significant excess of leukemia associated with butadiene
(OR = 7.61, 95% CI = 1.62-35.64) and a nonsignificant excess of leukemia associated with
7-20
-------
styrene exposures (OR = 2.92, 95% CI = 0.83-10.27). When exposures to both chemicals were
evaluated in the model as dichotomous variables, only butadiene was found to be associated with
leukemia (OR = 7.39, 95% CI = 1.32-41.33).
To determine if specific jobs within the SBR industry might explain some of the risk of
leukemia, the investigators categorized each worker according to the longest job held. A
mixed-job category that combined utilities, operation services, and laboratory jobs was
associated with relative odds of 3.78 (95% CI = 1.2-11.9). When butadiene was added to the
model, the OR increased to 6.08 for the mixed-job category (95% CI = 1.56-23.72). The relative
odds were 13.3 (95% CI = 2.24-78.55) for association between butadiene exposure and risk of
leukemia adjusted for mixed jobs in this model. Thus, both the mixed-job category and exposure
to butadiene seem to contribute to the risk of leukemia. The trend test for increasing risk of
leukemia with increasing exposure levels of butadiene (0 through 8) was statistically significant
(trend = 3.76, />=0.05). A similar trend was not found for styrene. The higher risk of leukemia
seen in the original cohort for black workers could not be evaluated adequately because race was
partially controlled in this nested case-control study.
Unlike the mortality study of this cohort, the case-control study did not show other
lymphomas to be associated with production jobs, but the number of cases was small.
Interestingly, when each chemical was analyzed by stratification, there was an excess risk for
butadiene exposure when exposure to styrene was low (OR = 6.67, 95% CI = 1.06-42.7). A
similar nonsignificant increase also was observed for styrene when butadiene exposure was low.
This might have resulted from small numbers of NHLs or multiple myelomas included together
with potentially different etiologies or correlated exposure data. Thus, investigators suggest that
further evaluation of each cancer in this other-lymphoma category should be performed
separately.
Investigators also caution that estimated exposures in this study were crude and not
substantiated by monitoring data. As they correctly pointed out, the original ordinal rank does
not create a perfect exposure scheme. The distribution of ranks was skewed to the right and had
to be log-transformed to differentiate between no exposure and low exposure. Matching on
duration of work may have overmatched the dose and resulted in underestimation of the risk.
Validation of diagnosis of lymphohematopoietic malignancies was not done in this study. The
panel ranked 71% of the jobs in ranks of two or less; thus misclassification of exposure based on
the estimated exposure by job as judged by the panel members is quite possible. Because the
panel members were blind concerning the status of the individual being the case or control, the
distribution of misclassification should be the same in cases and controls.
7-21
-------
7.2.1.4. Matanoski et al. (1993): Cancer Epidemiology Among Styrene-Butadiene Rubber
Workers
This was an effort to verify the findings of the investigators' earlier nested case-control
study among styrene-butadiene production workers (Santos-Burgoa et al., 1992). This study had
shown statistically significant elevated relative odds for leukemias. The results from the analysis
conducted with a new set of three controls per case were similar to the results from the earlier
study. The new controls were matched to all the variables except duration of work with the case.
Comparability between the previous and new controls was checked by reviewing the information
on cases and controls from the earlier study. To verify that the cause of death was correctly
coded on the death certificates, hospital records for cases were obtained. Of their initial 59 cases,
diagnosis was verified for 55 cases by hospital records review. Records were available for 54
cases and the diagnosis was correct in all cases. Two cases were omitted for incorrect diagnosis
and one was added that was identified by hospital record review. These omissions and additions
were in the categories of NHL (ICD 202) and lymphosarcoma (ICD 200). Records were
available for 25 out of 26 leukemia cases and were found to be correctly coded on the death
certificates.
Exposure estimation was done based on measurements provided by seven rubber plants,
the International Institute of Synthetic Rubber Producers, and NIOSH. Although there was
variability among plants, a significant correlation was observed between the log transformed data
provided by the company and the ranks of 464 job- and area-specific titles. Of the seven plants
that provided exposure measurements for butadiene, three had geometric means. Thus, using the
geometric means, the cohort data were reanalyzed for these three plants. The workers who were
hired before 1960 and had 10 or more years of service showed excesses for all
lymphohematopoietic cancers (SMR = 163, 95% CI = 113-227, O = 34) and leukemia and
aleukemia (SMR= 181, 95% CI = 101-299, O = 15).
This reanalysis of earlier data with new information on exposure estimation validates the
earlier results of these investigators.
7.2.1.5. Matanoski et al. (1997): Lymphohematopoietic Cancers and Butadiene and Styrene
Exposure in Synthetic Rubber Manufacture
This is a further analysis of exposures (in synthetic rubber production plants) to the case-
control population that was studied earlier by Matanoski et al. (1993). Selection of 1,242
controls was based on the frequency distribution of population sizes across the eight plants.
Potential for exposure for controls was slightly longer since they had to live at least as long as or
longer than the cases. Thus the variable of "duration worked" was included in the model.
Of eight plants, exposure measurements were obtained from seven plants. Several
exposure measurements for 1,3-butadiene (3,952) and styrene (3,649) were obtained, most being
7-22
-------
personal monitoring data. Based on these, job exposures were calculated, assuming that relative
exposures of jobs were similar across the plants. Using z-score transformation, exposure
estimation was carried out for all jobs with missing measurements, for each plant. Thus, using
these job- and plant-specific exposure levels, as well as job histories of cases and controls,
cumulative exposures were calculated in ppm multiplied by number of months exposed (ppm-
months). Time-weighted average (TWA) exposures in ppm were also calculated based on total
cumulative exposures in ppm-months divided by the total time employed for both styrene and
1,3-butadiene for each individual.
Using SPSS unconditional logistic regression, analysis was done. Beta values were
converted into corresponding odds ratios for TWA exposure of log normal (In) ppm for each
chemical. The variables used in the multivariate models included birth year, year of hire (prior to
or after 1950), age at hire, race, and duration worked. Step-down analysis was done.
TWA exposure of In (ppm + 1) of 1,3-butadiene was associated with leukemia (OR = 1.5,
95% CI= 1.07-2.1) and Hodgkin's lymphoma (OR = 1.73, 95% CI = 0.99-3.02). On the other
hand, average time-weighted exposure of In (ppm + 1) of styrene was significantly associated
with all lymphohematopoietic cancers (OR = 2.2, 95% CI = 1.46-3.33): NHLs, which included
ICD codes 200 and 202 (OR = 2.67, 95% CI = 1.22-5.84); lymphosarcoma (OR = 3.88, 95% CI
= 1.57-9.59); and myeloma (OR = 3.04, 95% CI = 1.33-6.96).
The category of all leukemias was significantly associated with 1,3-butadiene TWA
exposure of In (ppm + 1) as well as longest job in "labor, or laboratory," with ORs of 1.49 (95%
CI = 1.05-2.11) and 2.64 (95% CI = 1.21-5.30), respectively. The subtype lymphoid leukemia
was associated with 1,3-butadiene TWA exposure of In (ppm + 1), while myeloid leukemia was
associated with the category of longest job in "services, labor, or laboratory." Both were
statistically significant. The analysis by cumulative exposure found higher risks for workers who
had their cumulative dose in a shorter period (i.e., higher intensity) than for workers who had the
same cumulative dose in a longer period of time.
This analysis confirms the association between TWA exposure and cumulative exposure
to 1,3-butadiene and occurrence of leukemia. In this study, an association between styrene and
NHL is found for the first time. Investigators are unable to explain this association successfully,
but they qualify it by stating that probably both styrene and 1,3-butadiene are associated with
lymphohematopoietic cancers. Since exposures to these two chemicals are closely related in the
SBR industry, it is difficult to tease out the differences between these two; thus styrene may be
masking the effect of 1,3-butadiene and its association with NHL. The authors correctly point
out that exposure measures represent the more recent exposures in the industry, the IH data being
available only for the past 15 to 20 years. There may have been misclassification due to
individual recollection in assigning the job exposures. Furthermore, the authors assert that
7-23
-------
relative exposure differences for specific jobs have changed greatly over time and that further
work is required to correct these differences.
7.2.2. Cohort Identified by University of Alabama (UAB) Investigators
7.2.2.1. Delzelletal (1996): A Follow-Up Study of Synthetic Rubber Workers
A retrospective cohort mortality study was conducted by Delzell et al. (1996) of synthetic
rubber workers employed in seven U.S. and one Canadian plant. Of the eight plants, seven
plants (including the Canadian plant) were studied by JHU (Matanoski and Schwartz, 1987;
Matanoski et al., 1989, 1990, 1993; Santos-Burgoa et al., 1992) and one (two initial plants
combined into one) by Meinhardt et al. (1982). Of the seven plants studied by JHU, one located
in Texas that opened in 1970 was not included in UAB study. The cohort comprised all the male
workers who had worked for at least 1 year between January 1, 1943, and January 1, 1992 (49
years), which was the end of the follow-up period. The follow-up period was shorter for plants
1, 2, and 6 because the complete records of employees from these plants were available much
later than 1943. The Canadian plant (plant 8) also had a shorter follow-up period because
follow-up of men who had left employment before 1950 was not feasible.
Since the inclusion criteria for this study were different, there were some additions and
deletions to the earlier study cohort. The vital status was assessed by using plant records; the
SSA's death master file; the NDI; DMV records of Texas, Louisiana, and Kentucky for the U.S.
plants; and plant records and record linkage with the Canadian Mortality Data Base for the
Canadian plant.
Death certificates were acquired from plant and corporate offices and from state vital
records. The underlying cause of death was coded by a trained nosologist using the ninth
revision of the ICD. Any cancer was coded as a contributory cause of death. For the Canadian
decedents, the underlying cause of death was used from Canadian death registration and coded
according to the ICD revision in effect at the time of death. All ICD codes were converted to
eighth revision codes for analysis. The Ontario Cancer Registry provided the information on
incident cancer cases (including the date of diagnosis, primary site, ninth revision ICD code, and
histologic classification) for the study period.
Mortality analysis included computation of SMRs using the U.S. male general and state
population rates and Ontario male rates; SMRs by quantitative exposure (cumulative ppm-years
and peak ppm-years) to 1,3-butadiene, styrene, and benzene; and stratified internal comparisons.
Various within-cohort analyses were conducted using Poisson regression models.
This study included exposure estimation for each individual. A detailed description of
this estimation appears in Section 7.2.2.2, Macaluso et al., 1996. Complete work histories were
available for 97% of the cohort. Analysis for process group was conducted on the workers from
7-24
-------
all the plants. Subgroup analyses were restricted to 6 plants (1,354 workers from 2 plants were
excluded from the analyses due to the lack of information on specific work areas).
Of 15,649 males who had worked in SBR and related processes, 13,586 were white and
2,063 were black. Vital status assessment indicated that 10,939 (70%) workers were alive, 3,976
(25%) were dead, and 734 (5%) were lost to follow-up. Death certificates were acquired for
3,853 (97%) individuals. A total of 386,172 person-years (336,532 for whites and 49,640 for
blacks) was accrued.
Total cohort analysis found SMRs of 87 and 93 for all causes and all cancers,
respectively. The SMR for leukemia was 131 based on 48 observed deaths (95% CI = 97-174).
The SMRs for lymphosarcoma and other lymphopoietic cancers were close to null. Subcohorts
of whites, blacks, ever-hourly, and never-hourly showed a similar pattern of below-null results
for both all causes and all cancer deaths.
Ever-hourly was the only subcohort in which statistically significant excesses were found
for leukemia. The SMR was 143 (95% CI = 104-191, O = 36) for this subcohort. For white
ever-hourly workers, the SMR was 130 (95% CI = 91-181, O = 36), while for blacks the SMR
was 227 (95% CI = 104-431, O = 9). Further analyses of this ever-hourly subcohort by year of
death (<1975, 1975-84, 1985+), year of hire (<1950, 1950-59, 1960+), and age at death (<55
years, 55-64 years, 65+ years) showed statistically significant leukemia excess SMRs for 1985+
year of death (SMR = 187, 95% CI = 111-296, O = 18), 1950-59 year of hire (SMR = 200, 95%
CI = 122-310, O = 20), and <55 years at death (SMR = 179, 95% CI = 104-287, O = 17). The
lymphosarcoma SMR for this subcohort was 102 based on 4 cases, while the SMR for other
lymphopoietic cancers was 106 based on 17 cases. Neither of these excesses was statistically
significant.
When this subcohort of ever-hourly workers was further restricted to >10 years of
employment and >20 years since hire, leukemia SMRs of 224 (95% CI = 149-323, O = 28) for
all workers, 192 (95% CI = 119-294, O = 21) for whites, and 436 (95% CI = 176-901, O = 7) for
blacks were observed. Analysis by presence of solution polymerization versus absence of such
process in this restricted cohort resulted in leukemia SMRs of 209 (95% CI = 100-385) and 228
(95% CI = 135-160), respectively.
When analyses were done by various process groups, more than twofold increases were
observed for leukemia in polymerization process, SMR = 251 (95% CI = 140-414, O = 15);
coagulation process, SMR = 248 (95% CI = 100-511, O = 7); maintenance labor, SMR = 265
(95% CI = 141-453, O = 13); and laboratory workers, SMR = 431 (95% CI = 207-793, O = 10).
Analysis by further restricting the process groups by 5+ years of employment and 20+ years
since hire in each group showed excesses in leukemia SMRs in the same processes as above.
Analyses by mutually exclusive process groups showed excesses for ever in
polymerization and never in maintenance labor or laboratories (O/E = 8/4.7), ever in maintenance
7-25
-------
labor and never in polymerization or laboratories (O/E = 6/3.7), and ever in laboratories and
never in polymerization or maintenance labor (O/E = 8/1.6). Within the labor group, leukemia
increase was observed for workers ever in maintenance labor and never in production labor (O/E
= 11/3.8). On the other hand, for workers in production labor and never in maintenance labor,
the leukemia excess was negligible (O/E = 2/1.4). No excess mortality from leukemia was
observed among ever in finishing and never in polymerization process workers (O/E = 4/4.5).
In six out of eight plants, increased SMRs for leukemia were observed in ever-hourly
workers, exhibiting an internal consistency among plants. None of the excess was statistically
significant.
A statistically nonsignificant deficit was observed for lymphosarcoma in total cohort
based on 11 cases (SMR=80). Although some subcohort groups showed SMR excesses for this
specific cause, i.e., in ever-hourly white men during 1985-1991 (SMR=284, O=4), in field
maintenance workers (SMR=219, O=7), in production laborers (SMR=263, O=2), and in
maintenance laborers (SMR=188, O=3), none was statistically significant and most were based
on small numbers. Among the 42 deaths observed from other lymphohematopoietic cancers, 17
were NHL. As per International Classification of Diseases for Oncology (ICD-O) guidelines
(see footnote 4), the diagnosis of lymphosarcoma is no longer in use; instead they are classified
as lymphomas (Berard and Hutchinson, 1997). Thus, addition of 17 cases of NHL observed in
this study with 11 cases of lymphosarcoma results in 28 cases of lymphomas. Furthermore, it
should be noted that survival rates for NHL are high. A mean 5-year survival rate for males
diagnosed in 1974-1989 was reported to be 48% by Miller et al. (1993).
An unpublished report by the same authors (Delzell et al., 1996) submitted to the
International Institute of Synthetic Rubber Producers (IISRP) in October 1995 (Delzell et al.,
1995) included many more results of analyses of this cohort that are relevant to this assessment.
A review of the unpublished results is presented in the following paragraphs.
Various analyses by estimated 1,3-butadiene and styrene exposures were conducted. The
RRs calculated by Poisson regression for 1,3-butadiene ppm-years adjusted for styrene ppm-
years, age, years since hire, calendar period, and race for 0, >0-19, 20-99, 100-199, and 200+
ppm-years were 1, 1.1, 1.8, 2.1, and 3.6, respectively. When analysis was restricted to leukemia
as the underlying cause of death and person-years 20+ years since hire, the results were similar.
Analysis restricted to ever-hourly also showed positive results for butadiene. Various analyses
were conducted by using alternate ppm-years categories of exposure. All the analyses
consistently showed similar results, strengthening the association between 1,3-butadiene and
occurrence of leukemias. It is interesting to note that all the leukemia subjects who were
exposed to 1,3-butadiene were also exposed to styrene. Only two leukemia cases had exposure
to styrene but not to 1,3-butadiene.
7-26
-------
Analysis by 1,3-butadiene peak-years and styrene peak-years demonstrated an association
with 1,3-butadiene peak-years and occurrence of leukemia when adjusted for styrene peak-years,
1,3-butadiene and styrene ppm-years, and other covariates. The association, however, was
irregular. A similar analysis for styrene peak-years was weak and imprecise.
The investigators also conducted a cancer incidence study in the Canadian plant.
Information was obtained from the Ontario Cancer Registry from 1965 to 1992. Standard
incidence ratios (SIRs) were calculated by using the male general population of Ontario. No
increased incidence was found for any cancer in this study.
This is a well-designed, -conducted, and -analyzed study. The main strengths of the
study are large cohort size; long follow-up period (49 years); availability of exposure estimations
on each individual, processes, and tasks; and in-depth analyses using both the general population
and internal comparison groups.
There are a few limitations, as correctly pointed out by the investigators. The cause of
death on death certificates was not confirmed by medical records. Histologic typing was not
available for leukemias. These limitations may have led to misclassification. Furthermore, as
pointed out in the Macaluso et al. (1996) study, there may have been misclassification of
exposure, but this was thought to be nondifferential by investigators. Two plants were
eliminated from the final analysis due to the lack of detailed work histories. Although this may
have resulted in fewer uncertainties, valuable data may have been lost. Nevertheless, the
association between exposure to butadiene and occurrence of leukemia was present among both
white and black workers and was fairly consistent across plants.
7.2.2.2. Macaluso et al. (1996): Leukemia and Cumulative Exposure to Butadiene, Styrene,
and Benzene Among Workers in the Synthetic Rubber Industry
A cohort mortality study conducted in synthetic rubber workers by Delzell et al.
(1996) (Section 7.2.2.1) had a component of exposure estimation. The exposures to 1,3-
butadiene, styrene, and benzene were estimated by Macaluso et al. (1996).
An exposure estimation was conducted on each and every worker based on detailed work
histories, work area/job specification, IH monitoring survey records, IH recommendations,
various records from the plants, historical aerial pictures, use of protective and safely equipment,
walk-through surveys, and interviews with plant management as well as long-term employees in
specific areas/jobs. The quantitative exposure estimation was based on process analysis, job
analysis, and exposure estimation for specific tasks in different time periods. The job-exposure
matrices (JEMs) were computed for 1,3-butadiene, styrene, and benzene, which were linked to
work histories of each employee to compute the cumulative exposure indices.
Quantitative estimates of exposure to 1,3-butadiene and styrene were based on
background exposure plus task-specific exposure, using multiple exposure and point source
7-27
-------
models, respectively. Input variables for these models were derived from several information
sources described earlier. Limited validation of exposure estimates was attempted by comparing
the available IH data from the 1970s and 1980s as well as actually measuring the air
concentrations of 1,3-butadiene and styrene under controlled conditions. The latter method
showed good agreement among the methods of sampling, whereas the comparison of IH data
indicated overestimations of 1,3-butadiene exposure.
For each job, 8-h TWA intensities and the number of peak exposures (15-min exposures
over 100 ppm) were calculated. Based on job exposures, a JEM database was developed that
was linked with individual work histories to develop individual quantitative work exposure
estimates. For each individual, the exposure indices were multiplied by the length of
employment in that particular process or job and were added up for the total employment period
in various jobs to estimate the cumulative exposure.
Mortality analysis was done by calculating the SMRs and RRs using estimated
quantitative exposures to 1,3-butadiene, styrene, and benzene. Both cumulative ppm-years and
peak-years were calculated for each individual in the study. Person-year data were grouped by
1,3-butadiene, styrene, and benzene ppm-years for both SMR analyses as well as RR analyses.
Comparability between cohort mortality rates and general population reference mortality rates
was assured by limiting the SMR analysis to the individuals whose underlying cause of death
was listed as leukemia (51 people). Risk ratios were computed by the Mantel-Haenszel method
and the 95% CI were computed by the Breslow method. Poisson regression models were used
for adjustment of multiple confounders and to compute within-cohort mortality rates, and the X2
test for linear trend was used to examine the dose response.
Work histories were available for 97% of the population. Fifty-two in-depth interviews
with plant management and long-term employees identified 446 specific tasks/work areas with
potential for 1,3-butadiene, styrene, and benzene (3 plants only) exposure. Eight-hour TWAs for
1,3-butadiene, styrene, and benzene were 0 to 64 ppm, 0 to 7.7 ppm, and <1 ppm, respectively,
the median exposures being <2 ppm for 1,3-butadiene and 0.5 to 1.1 ppm for styrene.
Exposure analysis found that 75% of the cohort was exposed to 1,3-butadiene, 83% was
exposed to styrene, while only 25% was exposed to benzene. The median cumulative exposure
to 1,3-butadiene, styrene, and benzene was 11.2, 7.4, and 2.9 ppm-years, respectively. The
exposure prevalence as well as median cumulative exposure were higher in individuals who had
died of leukemia. Among the leukemia decedents, 85% had exposure to 1,3-butadiene, with their
median cumulative exposure being 36.4 ppm-years. This exposure was two times higher
compared with all decedents and three times higher compared with all other employees. The
exposure to styrene was present in 90% of leukemia decedents, with median cumulative exposure
being 22.4 ppm-years, two times and three times higher compared with all the decedents and all
other employees, respectively. Benzene exposure was found to be less frequent among leukemia
7-28
-------
decedents compared with all other employees. Analysis by benzene exposure showed no
association with the occurrence of leukemia after adjustment for 1,3-butadiene and styrene.
Leukemia SMRs increased with increasing cumulative exposure to 1,3-butadiene as well
as styrene. Mortality RRs computed for cumulative 1,3-butadiene exposure adjusted for race,
age, and cumulative styrene exposure also showed increasing RRs for increasing cumulative
exposure to 1,3-butadiene. The adjusted RRs for cumulative exposures of butadiene of 0, <1,
1 to 19, 20 to 79, and 80+ ppm-years were 1, 2.0, 2.1, 2.4, and 4.5, respectively. The linear X2
test for trend was statistically significant (p=0.01). When similar RRs were computed for styrene
exposure, neither showed a consistent pattern nor a trend of increasing risk with increasing
exposure. A similar trend test was statistically not significant.
Analysis by exclusion of the nonexposed population resulted in RRs of 1, 1.5, and 1.7 for
0.1 to 19, 20 to 79, and 80+ ppm-years of the cumulative exposures of 1,3-butadiene. The linear
trend test was statistically significant (p=0.03), substantiating the earlier finding of increasing
risk of leukemia with increasing cumulative exposure to 1,3-butadiene. Although the same
analysis suggested increasing risk of leukemia with increasing cumulative exposure to styrene
after adjustment for 1,3-butadiene and other covariates, the results were imprecise and
statistically nonsignificant.
Neither a positive nor negative interaction was found between the cumulative exposures
to 1,3-butadiene and styrene.
For the last decade or so, epidemiologists have been including exposure estimation in
their studies. The methods used and efforts made to do exposure estimations are improving but
variable. This study is one of the best efforts of exposure estimations to date. The investigators
have used many available methods to come up with best estimates of exposures of 1,3-butadiene,
styrene, and benzene. They also have validated these estimates on a smaller scale. Although this
is considered the best effort, it should be noted that these are estimates and not actual
measurements. Two plants were eliminated from the analysis because detailed work histories
were lacking. Thus it is possible that individuals may have been misclassified with respect to
process or job, resulting in either over- or underestimations of exposure. However, there is no
reason to believe that the misclassification of exposure occurred only in individuals who had died
of leukemia.
7.2.2.3. Sathiakumar et al. (1998): Mortality from Cancer and Other Causes of Death Among
Synthetic Rubber Workers
This paper presented further analysis of the cohort studied by Delzell et al. (1996) and
Macaluso et al. (1996). Thus, the follow-up period, methods of data collection for exposure, and
health endpoints are the same as Delzell et al. (1996). Statistically significant deficits were
observed for all causes of deaths, all cancers, and quite a few other cause-specific cancer and
7-29
-------
noncancer deaths in the total cohort. Except for statistically significant elevated SMR for
leukemia in the subcohort of ever-hourly workers, no other excess was observed for any other
cancer or noncancer endpoint. When this ever-hourly subcohort was further restricted to worked
for > 10 years and > 20 years since hire by race, in addition to excess leukemia in both whites
and blacks, statistically significant increased SMR was observed for cancer in the large intestine
among blacks (SMR = 231, 95% CI = 119-402, O = 12). Investigators examined the cause-
specific deaths for cancers of the lung and large intestine in the ever-hourly subcohort by
different characteristics such as year of death, years since hire and years worked, and year of hire.
The only statistically significant excess was found for lung cancer among whites who had died
between 1975 and 1984. Among blacks, excesses for lung cancer were found in workers who
had died after 1985 and workers who had more than 20 years since hire and had worked for less
than 10 years. Excess for cancer of the large intestine was found in black workers who had more
than 20 years since hire and had worked for more than 10 years. When analysis was done by
process group, excess lung cancer was observed for the maintenance group among both whites
and blacks while most of the other groups had deficit for this cancer. Laryngeal cancer was
increased only in whites among production, maintenance, and other operation workers.
The limitations listed in Delzell et al. (1996) and Macaluso et al. (1996) about
misclassification apply to this study too. Adjustment for smoking was not performed; thus, it is
difficult to know whether the observed excesses in some subgroups for lung and laryngeal cancer
had any causal association with 1,3-butadiene.
Please note that the results for lymphohematopoietic cancer reported in this paper are not
included in this writeup to avoid repetition.
Studies in polymer production workers are summarized in Table 7-2.
7.3. POPULATION STUDY
7.3.1. Loughlin et al. (1999): Lymphatic and Haematopoietic Cancer Mortality in a
Population Attending School Adjacent to Styrene-Butadiene Facilities, 1963-1993
This study was conducted in the Port Neches-Groves High School, southeastern Texas, to
address the communities' concern for the students' health. The Texas Department of Health had
conducted a retrospective cohort mortality study of female graduates of the same high school in
1980 (unpublished report authored by R. K. Donelson) in which no excess of
lymphhematopoietic cancer was observed. The current investigation by Loughlin et al. (1999) is
a follow-up of the earlier Health Department study. The investigators included students who
graduated between 1963 and 1993, including males, and followed them through 1995.
The cohort was assembled by including all the seniors pictured or listed in the high
school yearbooks for the school years 1962-63 to 1992-93. Any student who attended the school
7-30
-------
Table 7-2. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—polymer production.
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and limitations
Matanoski et al. (1989 and
1990)
Update of the cohort from
Matanoski and Schwartz
(1987)
Cohort mortality of 8 SBR
polymer production plant
workers
Reduced cohort of 13,422
followed through 1982
Worked for at least 1 year
Comparison group U.S.
population
Divided in four major
areas based on the
longest job held:
Production workers
Utility workers
• Maintenance
workers
• All other work sites
Among production workers:
• SS SMR = 260 for other
lymphohematopoietic cancers
in whites
• SS SMR =507 for all
lymphohematopoietic cancers
• SS SMR = 655 for leukemia
in blacks
No relation observed with
latency or duration of
employment
Largest cohort mortality study of SBR
workers
Lack of exposure data
Exclusion of 50% of the population in
the follow-up
Four plants had follow-up ranging from
12 years to 25 years; may not be enough
time for malignancies to develop
Matanoski etal. (1989)
Santos-Burgoa et al. (1992)
Nested case-control study
Cases:
• Of leukemia 26
• Of other lymphatic
cancers 18
Controls matched on:
• Plant, age, hire year,
employment duration,
survival to the death of
the case
• An average 3.3 controls
(instead of intended 4)
were selected
Exposure to 1,3-
butadiene and styrene
was done by job
identification and levels
associated with that job
Estimations of job and
exposure levels were
done independently of
the status of the case or
control
The jobs were ranked
from 0 to 10
Cumulative dose was
calculated using the
score and duration for
eachjob
For 1,3-butadiene:
• Ever/never exposure
- SS OR = 6.82 (high) and
4.26 (low) were found for
leukemia
• Matched analyses
-SS OR = 2.3 for all
lymphohematopoietic cancer
- SS OR = 9.36 for leukemia
• Conditional analyses
• SS OR = 7.61 for leukemia
One of the strengths is that an attempt
was made to estimate actual exposure
Matching may have overmatched the
dose
Lack of validation of diagnosis of
hematopoietic malignancies may have
resulted in misclassification
Misclassification of exposure based on
job categories
-------
Table 7-2. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—polymer production (continued)
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and limitations
Matanoskietal. (1989),
Santos-Burgoa et al. (1992)
(continued)
Log transformation of
scores used in analyses
Analyses were done by:
• Ever/never exposed
• High vs. low
exposure
• Both matched
(conditional) and
unmatched
(unconditional)
SS OR = 6.67 for other
lymphoma when styrene
exposure was low
• SS trend of 3.76 was found
for increased risk of
leukemia with increasing
exposure levels of butadiene
Matanoskietal. (1993)
to
Same as nested case-control
study
A new set of 3 controls per
case
Cause of death verified by
hospital records
Cohort data reanalysis
Exposure estimation
done based on
measurements provided
by seven plants, IISRP,
andNIOSH
• Similar results with new
controls
• Reanalysis of cohort data for
three plants
-SS SMR = 163 for all LHC
-SSSMR= 181 for
leukemia and aleukemia
Verification of cause of death
New set of controls validates earlier
results
-------
Table 7-2. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—polymer production (continued).
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and limitations
Matanoski et al. (1997)
Case-control population of
Matanoski et al. (1993)
Based on personal
monitoring exposure
measurements:
1,3-butadiene - 3,952
styrene - 3,649
Time-weighted average
(TWA) exposures in log
normal (In) ppm
calculated for each
chemical
TWA of In (ppm+1) was
associated with:
• SSOR= 1.5 for leukemia
• SS OR = 2.64 leukemia
among labor or laboratory
SS associations were found
with styrene TWA of In
(ppm+1) for:
• OR = 2.2 lymphohemato-
poietic cancer
• OR = 2.67 non-Hodgkin's
lymphoma
OR = 3.88 lymphosarcoma
OR = 3.04 myeloma
- Confirms the association between 1,3-
butadiene and leukemia excess
- association between styrene and non-
Hodgkin's lymphoma found for the first
time is not explained fully
-------
Table 7-2. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—polymer production (continued)
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and limitations
Macalusoetal. (1996)
Cohort of seven U.S. and
one Canadian SBR workers
mortality study
Worked for at least 1 year
between January 1, 1943,
and January 1, 1992
Follow-up period through
January 1, 1992
15,649 male workers
U.S. population Respective
State populations where the
plants were located
Ontario male rates for
Canadian plant
Internal comparison using
Poisson regression
Exposure estimation
based on several
information sources
including IH
Quantitative exposure
estimates on
background, task-
specific, multiple
exposure, and point
sources models for 1,3-
butadiene, styrene, and
benzene
Peak exposures
8-h time-weighted
intensities
Cumulative exposures
Exposures estimated for
each individual
Adjusted RRs for cumulative
exposure to 1,3-butadiene of
0, <1, 1-19, 20-79, and 80+
ppm-years were 1, 2.0, 2.1,
2.4, and 4.5.
Trend test was SS
Exclusion of nonexposed
population also had similar
results with SS trend test
Methods used and efforts made for
exposure estimation are best efforts to
date
Misclassification with respect to job may
be possible but unlikely to be only in
leukemia deaths
-------
Table 7-2. Epidemiologic studies of the health effects of exposure to 1,3-butadiene—polymer production (continued).
Authors
Population studied
1,3-Butadiene exposure
assessment
Results
Strengths and limitations
Delzelletal. (1996)
Same as Macaluso et al.
(1996)
Same as Macaluso et al.
(1996)
Analysis by ever-hourly
and never-hourly
Analysis by process
groups
Ever-hourly workers
showed for leukemia
-SSSMR=143forall
ever-hourly workers
-SSSMR = 227 for blacks
-SSSMR= 187 for 1985+
year of death
-SS SMR = 200 for 1950-
59, year of hire
-SSSMR= 179for<55
years age at death
-SSSMR = 224for>10
years of employment and
>20 years since hire
(SMR = 192 for whites and
SMR = 436 for blacks,
both SS)
Various process groups
showed for leukemia
-SS SMR = 251 for
polymerization process
-SS SMR = 265 for
maintenance labor
-SSSSMR = 431for
laboratory worker
Cancer incidence study in
Canadian plant did not
show any increased
incidence for any cancer
Same as Macaluso et al. (1996)
Cause of death not verified
Histologic typing of leukemia not
available, thus leading to
misclassification
SS= Statistically significant, SMR= Standard mortality ratio.
IH=Industrial hygiene, RR=Risk ratios.
OR=Odds ratio, LHC=Lymphohematopoietic cancers.
-------
for 3 months was included and was considered to have attended the school for that academic
year. The school records were abstracted for the students' names, dates of birth, and years they
attended the school. The name changes were noted if there were any due to adoption or
marriage. A total of 15,553 students was identified. Sex and name change were verified by
accessing several ancillary data sources. Vital status was ascertained by using the NDI, SSA
death master files, and records of the Texas Department of Health. The name changes were
checked by using the marriage databases from 1966 to 1994 and creating an algorithm to match
with one or more marriages resulting in name change. A total of 15,722 possible last names was
identified for the 7,591 women who had qualified for the cohort. The final cohort consisted of
15,403 students after the elimination of 145 students with unknown date of birth, 2 with
incongruous date of birth, and 3 with unknown sex.
A total of 310,254 person years was accrued by 7,882 males and 7,521 females. Deficits
in SMRs were observed for all causes in both males and females. For all cancers, females had a
deficit, while excess was observed in males (SMR = 1.22). Neither the deficits nor the excesses
were statistically significant A subcohort of males was examined by years of attendance at Port
Neches High School. Males who had attended the high school for < 2 years showed increases for
all causes (SMR= 1.42, 95% CI= 1.11-1.80, O = 71), for all cancers (SMR = 2.11, 95% CI =
0.96-4.00, O = 9), for leukemia/aleukemia (SMR = 5.39, 95% CI = 1.09-15.64, O = 3), and for
all external causes of death (SMR = 1.58, 95% CI = 1.17-2.10, O = 48). For males who had
attended the high school for >3 years the only statistically significant excess was observed for
benign neoplasms (SMR = 6.05, 95% CI = 1.65-15.48, O = 4). Similar analysis in females
showed excess in all causes of deaths among females who attended the school for < 2 years
(SMR =1.64, 95%CI= 1.14-2.29).6
Analysis by year of graduation, graduation either before 1980 or after 1980, did not show
any clear pattern. Use of the Texas population for comparison ensued in similar results.
This study has quite a few methodologic limitations. First, no environmental
measurements of either styrene or 1,3-butadiene for the school are available. Second, the cohort
is young; a person who entered the high school at the age of 14 in 1963 would be age 46 in 1995.
Since students up to 1992-93 are included, some will be barely 14 to 18 years old. Still, it is
interesting to note that males who had attended the high school for < 2 years had excesses for all
causes, all cancers, and leukemias. Third, as correctly pointed out by the authors, an incidence
study would have been more appropriate because the survival for leukemias is high; thus a
mortality study may result in underestimating the problem. Fourth, no other information such as
occupation, lifestyle, etc., was available. Therefore, it is difficult to establish a causal link
The paper does not provide the number of observed deaths on which this SMR was based.
7-36
-------
between the environmental exposure to 1,3-butadiene and observed excess in leukemias in men
who had attended the high school for < 2 years.
A comment on this paper submitted to the editors by Hanks and Townsley (2000), the
lawyers representing 14 cases, points out that only four of their clients were dead by the cutoff
date and were counted in the study. Since the end of the follow-up date of this study, there have
been two deaths from leukemia and two deaths from Hodgkin's disease in males (one died from
small-cell carcinoma of the lung presumed to be secondary to extensive chemotherapy of
Hodgkin's disease). Among the six who are still alive, two males have Hodgkin's disease and
one has NHL, while two females have leukemias and one has Hodgkin's disease. Furthermore,
they mention that they are aware of several other students who died from lymphohematopoietic
cancers who were not included in the study cohort because they attended the high school prior to
1963.
7.4. SUMMARY AND DISCUSSION
1,3-Butadiene is a colorless, odorless gas and due to its extremely volatile nature
exposure to this gas is entirely via inhalation. 1,3-Butadiene becomes diluted in ambient air and
is eliminated by photooxidation. Occupational populations are exposed to 1,3-butadiene in the
production/recovery of 1,3-butadiene monomer and the production of synthetic rubber (polymer),
resins, and plastics. The general population is exposed to butadiene in ambient air, the major
sources of its release being automotive exhaust and cigarette smoke. 1,3-Butadiene has been
shown to be mutagenic as well as carcinogenic in animals and humans. Data in animals,
particularly in mice, show that 1,3-butadiene is a multisite carcinogen even at the lowest dose of
6.25 ppm (NTP, 1993). Therefore, its potential to cause cancer in humans has become an
important public health issue.
It is difficult to study the health effects of exposure to 1,3-butadiene in the general
population because exposure to 1,3-butadiene is ubiquitous, "unexposed" reference populations
used in occupational cohort studies are likely to contain a substantial number of individuals who
are exposed to 1,3-butadiene nonoccupationally. Furthermore, the issue of health measurement
is complicated by the fact that occupational cohorts tend to be healthier than the overall general
population and have below-average mortality, which is referred to as the "healthy worker effect."
Thus, the SMRs observed in occupational cohorts, computed using the general population as the
reference group, are underestimations of real risk.
7.4.1. Monomer Production
To evaluate the carcinogenicity of 1,3-butadiene, cohorts from monomer and polymer
production plants were studied by several investigators. The largest cohort of monomer
7-37
-------
production workers was initially studied by Downs et al. (1987) with three follow-ups by Divine
(1990), Divine et al. (1993), and Divine and Hartman (1996). The cohort initially included
2,586 workers, which increased to 2,795 individuals in the last follow-up due to an extended
time period for the inclusion criteria. The four exposure groups were identified by Downs et al.
(1987) based on a qualitative exposure scale. The groups remained the same in Divine's (1990)
follow-up and were similar but slightly changed in Divine et al. (1993). In their last follow-up,
based on IH data, Divine and Hartman (1996) estimated the potential exposure to 1,3-butadiene
for each employee by their work histories (in 1-year segments), using job categories and calendar
time periods. Cumulative exposures were obtained by summing the scores for all the years of
employment.
The findings of all four investigations were essentially the same even after 52 years of
follow-up. Deficits were observed for mortality from all causes and all cancers. The only
statistically significant excess observed was for lymphosarcoma (ICD code 200). Downs et al.
(1987) observed this excess for the total cohort and for the subcohort of workers who had worked
for less than 10 years and latency of 0 to 9 years. This excess was seen in the wartime subcohort
in all three follow-up studies (SMR = 269 [Divine, 1990], and SMR = 254 in two latter follow-
ups: Divine et al. [1993]; Divine and Hartman [1996]). Furthermore, in the last follow-up, the
SMR of 176 (95% CI=103-282) was computed for NHL, using new diagnostic criteria that
classified lymphosarcoma as lymphoma and thus included it in NHL. Of 17 cases on which this
excess was based, 16 were hired prior to 1950. Survival analysis using estimated cumulative
exposures failed to show any association between the exposure and occurrence of NHL.
Hence, the investigators thought that peak exposures rather than cumulative exposures
may be associated with the observed increase in NHL. No information on peak exposure levels
was available for this period, but it was believed that the peak exposures were high during the
war period. When analyses were done by years of employment and latency excess for
lymphosarcoma, mortality was always found to be in individuals employed for less than 10 years
with a latency of 0 to 9 years. It should be noted that after 52 years of follow-up, except in the
survival analysis using the Cox model (which showed statistically significant excess in leukemia
for the risk factor "Hire-age"), no statistically significant elevated mortality was observed for
leukemia in this cohort (because the study population may not have enough power to detect a
leukemia excess), which was the main finding in the SBR workers.
A small cohort of 364 individuals was identified from 29,139 workers at three Union
Carbide Corporation plants who had potential exposure to 1,3-butadiene (Ward et al., 1995,
1996c). The exposure to 1,3-butadiene was assumed based on department codes, and no
adjustments for confounding by other chemicals were done. Analysis of the subcohort of
workers who had also worked in the acetaldehyde unit did not show any excess in deaths from
7-38
-------
lymphosarcoma independent of exposure to 1,3-butadiene. As observed in the Divine studies
(Divine, 1990; Divine et al., 1993; Divine and Hartman, 1996), a statistically significant excess
for lymphosarcoma (SMR = 577) was observed in this cohort based on four cases. Of these four
cases, three had started their employment between 1942 and 1946, that is, during wartime.
A third cohort of 614 workers exposed to monomer was studied by Cowles et al. (1994),
but the study failed to show any excess mortality or morbidity. Due to several methodologic
limitations, this study failed to provide any negative evidence to support the causal association
between exposure to 1,3-butadiene and the occurrence of lymphosarcoma that was observed in
the other two cohorts.
7.4.2. Polymer Production
A further follow-up and reanalysis of a large SBR polymer production workers' cohort
study (Matanoski and Schwartz, 1987) was conducted by Matanoski et al. (1989, 1990). This
follow-up added 3 years to the earlier study, but the findings of this follow-up were essentially
the same as those of the earlier study. The only statistically significant excesses were among
production workers. Among whites, the excess was for other lymphohematopoietic cancers
(SMR = 260), and among blacks the excesses were for all lymphohematopoietic cancers (SMR =
507) and leukemia (SMR = 655). Analyses by duration of work and latency did not show any
increases in hematopoietic cancers. There were no exposure measurements or estimations done
in this study.
A nested case-control study from this cohort (Matanoski et al., 1989, 1990) was
conducted by the same investigators and reported by Matanoski et al. (1989) and Santos-Burgoa
et al. (1992). Fifty-nine cases of lymphohematopoietic cancers and 193 matched controls were
identified. Exposures to 1,3-butadiene and styrene were estimated for these individuals using
job records and levels of exposure to 1,3-butadiene and styrene associated with those jobs
independent of the case or control status. The jobs were ranked and cumulative dose was
calculated for each case and control. Analyses were conducted using log transformed scores.
The relative odds were increased for high (OR = 6.82) and low (OR = 4.26) exposures in the
ever/never exposed analysis, matched analysis (OR = 9.36), and conditional analysis (OR = 7.61)
for leukemia. All the increases were statistically significant. A statistically significant trend also
was observed for increasing risk of leukemia with increasing levels of exposure to 1,3-butadiene.
In 1993, Matanoski et al. (1993) reevaluated the analysis of the nested case-control study
because their findings had been questioned by Acquavella (1989) and Cole et al. (1993),
particularly with respect to the base cohort study. Even though the investigators chose a new set
of three controls per case, and verified the cause of death by obtaining hospital records, the
findings of the new analysis were similar to those of the earlier analysis. Furthermore, the
7-39
-------
investigators estimated the exposures to the cohort based on measurements provided by seven
rubber plants, IISRP, and NIOSH. In an analysis of the subcohort from the three plants, for
which they had geometric means of exposure, statistically significant excesses were observed for
all lymphohematopoietic cancers (SMR = 163) as well as for leukemia and aleukemia (SMR =
181).
In Matanoski et al. (1997), the exposure refinement was done by the authors using several
exposure measurements for 1,3-butadiene and styrene. These included personal monitoring, job-
and plant-specific exposure levels, job histories, and time spent in each job to calculate the
cumulative exposure in ppm-months. Statistically significant excesses were observed for a
TWA of 1 ppm of 1,3-butadiene and leukemia (OR =1.5) and Hodgkin's lymphoma (OR = 1.73).
Statistically significant associations also were observed for all leukemias as well as lymphoid
leukemia for a TWA of 1 ppm and for all leukemias and myeloid leukemia for the longest job in
"service, labor, or laboratory." The TWA of 1 ppm of styrene was significantly associated with
all lymphohematopoietic cancers (OR=2.2), NHL (OR=2.67), lymphosarcoma (OR=3.88), and
myeloma (OR=3.04). The investigators were unable to explain the observed association between
NHL and styrene.
Delzell et al. (1996) and Macaluso et al. (1996) reported separately the two components
of the follow-up study of synthetic rubber workers. These investigators studied the seven plants
studied by Matanoski and Schwartz (1987), Matanoski et al. (1989, 1990, 1993), and Santos-
Burgoa et al. (1992) and one plant (two initial plants combined into one) by Meinhardt et al.
(1982). The follow-up period was 49 years. Investigators estimated the exposures to 1,3-
butadiene, styrene, and benzene for each worker. This was done by using various means such as
job histories, work areas, IH data, historical plant data, aerial pictures, interviews with long-term
employees and managers, walk-through surveys, etc. Quantitative exposures were calculated,
and limited validation of exposure estimates were attempted using available 1970s and 1980s IH
data. Cumulative and peak exposures were calculated for each worker. Comparison with the
U.S. population resulted in statistically significant excesses for leukemia in ever-hourly workers
(SMR = 143) and its subcohort of black workers (SMR = 227). The excesses also were found in
the ever-hourly cohort for year of death (SMR = 187 for 1985+), year of hire (SMR = 200 for
1950-59), age at death (SMR = 179 for <55 years), and for more than 10 years employment and
more than 20 years since hire (SMR =192 for whites and SMR = 436 for blacks). Laboratory
workers, maintenance workers, and polymerization workers also showed increased SMRs of 431,
265, and 251, respectively. All these analyses were done adjusting for styrene and benzene.
When internal comparison was done using the estimated ppm-years exposure data, relative ratios
increased with increasing exposures. The trend test was statistically significant.
7-40
-------
The incidence study conducted in the Canadian plant employees did not show any
increases in any cause-specific cancers.
These investigators found a nonsignificant deficit in lymphosarcoma in the total cohort
for the specific site observed in monomer workers. However, they did find nonsignificant
excesses in ever-hourly workers during the time period 1985-1991 (field maintenance workers, in
production laborers, and maintenance laborers). In Sathiakumar et al. (1998), these investigators
reported the results for NHL, which includes lymphosarcoma per ICD-O (Berard and Huchinson,
1997). Of 28 NHLs observed in the cohort, 26 occurred in ever-hourly workers (SMR = 99).
Although the SMR was just below 100 and statistically nonsignificant, the authors noted that due
to high survival rates for NHL, the association between the employment in SBR industry and
NHL may be obscured.
7.4.3. Relevant Methodologic Issues and Discussion
Throughout this chapter, various methodologic issues, including strengths and limitations,
are discussed. Except for a morbidity analysis in one monomer study (Cowles et al., 1994) and
an incidence analysis in the Canadian plant of the cohort identified by UAB, all of the studies
reviewed here are mortality studies. The 5-year relative survival rates for leukemia by year of
diagnosis between 1989 and 1994, among whites were 44% and among blacks were 31%. For
NHL the survival rates were 52% for whites and 41% for blacks for the same calendar period
(Landis et al., 1999). Thus, these mortality studies may not be the appropriate measure to
evaluate the carcinogenicity of 1,3-butadiene. The major concerns of the mortality studies
reviewed here are small cohort sizes in monomer studies, lack of data on confounding variables,
and latency analysis in one study. Furthermore, death certificates were used by all the
investigators, which could lead to misclassification bias. Validation of diagnosis of
lymphohematopoietic cancer was not done in any of the studies except Matanoski et al. (1993).
Although it is a minor limitation, it is a methodologic concern given the fact that recording of
lymphohematopoietic cancer, particularly leukemia, on death certificates is unreliable (Percy et
al., 1981).
Lack of exposure information is another major limitation of Cowles et al. (1994) and
Ward et al. (1995, 1996c). Cowles et al. (1994) did not attempt any exposure estimation. The
cohort was very small, there were very few deaths, and more than 50% of the cohort had an
average follow-up of 12 years.
Ward et al. (1995, 1996c) did not attempt any exposure estimation either. This cohort
also was very small and was restricted to workers who had worked in 1,3-butadiene production.
The high SMR for lymphosarcoma and reticulosarcoma observed in this study was based on only
four cases. They used employment of 2+ years as surrogate for exposure and stated that there
7-41
-------
were no other common exposures to other chemicals. Two cases also had exposure to
acetaldehyde, which has never shown any lymphohematopoietic carcinogenesis before.
Furthermore, it should be noted that only the morbidity study involving acetaldehyde exposure,
by Bittersohl (1974), reported nine cancer cases: five squamous cell carcinomas of the bronchi,
two squamous cell carcinomas of the mouth cavity, and one adenocarcinoma of the stomach and
cecum each. No lymphosarcoma or reticulosarcoma was observed in this study. Thus, it is
unlikely that lymphosarcoma observed in the Union Carbide cohort is due to the confounding
effect of the acetaldehyde exposure. In animal studies the only statistically significant excesses
observed were for nasal tumors in male and female rats and laryngeal tumors in male and female
hamsters after inhalation exposure to acetaldehyde (U.S. EPA, 1987).
Considering that the cohort was small and only four deaths occurred from
lymphosarcoma and reticulosarcoma, it should be noted that this finding is consistent with the
finding of the other monomer facility studied by Divine (1990), Divine et al. (1993), and Divine
and Hartman( 1996).
A monomer cohort study conducted by Downs et al. (1987) and followed by Divine
(1990) and Divine et al. (1993) also lacked exposure information, although the surrogate
exposure grouping was done by qualitative exposure information based on job descriptions/work
areas. The investigators estimated the exposures in their last follow-up (Divine and Hartman,
1996) and found that except for an excess observed for lymphosarcoma and reticulosarcoma in
the wartime subcohort, there were no excesses in any cause-specific cancer mortality. They go
on to discuss that, per new diagnostic criteria, lymphosarcoma is now classified as lymphoma
and thus is included under the category of NHL with other lymphomas. Thus they computed the
SMR of 176 (95% CI = 103-282) for the observed 17 deaths from NHL. This was based on the
assumption that 50% deaths in cancers of other lymphatic tissue were from lymphomas. Out of
17 decedents, 16 were hired before 1950, 6 had worked for more than 5 years, and 12 belonged
in the varied-exposure group.
Just as in the earlier two follow-ups (Divine, 1990; Divine et al., 1993) the investigators
stipulate that the above findings were consistent with the hypothesis that high exposures to 1,3-
butadiene in wartime may be associated with NHL. Since they did not find any association
between cumulative exposure and occurrence of non-Hodgkin lymphoma in their survival
analysis, they suggest that peak exposures rather than cumulative exposures may be associated
with the observed increase. Furthermore, they acknowledge that there are no data or means
available to estimate the peak exposures. In addition, they note that extremely high peak
exposures did exist during the 1950s and 1960s. They also assert that the lack of association
may be due to a high survival rate for NHL. Also, there is a lack of any information on other
confounding factors. After 52 years of follow-up and extensive analyses, this cohort has not
7-42
-------
observed any statistically significant excess in mortality from leukemia except in survival
analysis by Cox model, for the risk factor "Hire-age," that was observed in SBR workers.
Nonetheless, the finding of excess mortality from lymphosarcoma and reticulosarcoma is
consistent with the findings of Ward et al. (1995, 1996c), and, although not statistically
significant (SMR = 224), of Meinhardt et al. (1982).
Matanoski and Schwartz (1987) and Matanoski et al. (1989, 1990) did not have any
exposure information available. The cohort was distributed in four major areas based on longest
jobs held and the qualitative exposure information used as surrogate. When the nested case-
control study was undertaken by these investigators (Matanoski et al., 1989; Santos-Burgoa et
al., 1992), exposure estimation was done by using various sources only for selected cases and
controls. They observed a statistically significant high excess from leukemia mortality, which
the authors concluded as being causally associated with exposure to 1,3-butadiene.
Matanoski et al. (1993) validated their earlier results of the nested case-control study by
using a new set of three controls per case. They also verified the cause of death noted on the
death certificates and the diagnosis noted on the hospital charts. They found that the diagnosis
noted on 25 out of 26 charts agreed with the cause of death noted on the death certificates. The
results of this study were similar to the earlier nested case-control study.
This finding of a high excess of leukemia mortality in the case-control study was
questioned by Acquavella (1989) and Cole et al. (1993) because no excess leukemia mortality
was found in the base cohort study from which the cases and controls were selected. Their
argument that the results of the case-control study were statistically incompatible with the results
of the cohort study was based on the calculations of number of leukemias that should have been
seen in the cohort study, based on the relative odds observed in the case-control study. When
Cole et al. (1993) applied the relative odds of 7.6 observed by Santos-Burgoa et al. (1992) to the
60% of the cohort exposed to 1,3-butadiene, their calculation resulted in approximately 104
leukemia cases. When the leukemia rates of the general population were applied to the
remaining 40% of the cohort that was not exposed to 1,3-butadiene, an additional 9.2 leukemia
cases were expected. Thus, a total of 113 leukemia cases should have been observed for this
cohort instead of the 22 leukemia cases that actually were found by the investigators. Variability
in both the prevalence of exposure and the relative odds was looked at by these authors (Cole et
al., 1993), and they concluded that there was no reasonable combination that resolved the
incompatibility between the findings of the cohort and case-control studies.
Matanoski and Santos-Burgoa (1994) disagreed with this criticism. They asserted that the
60% exposure observed among the controls in the case-control study overestimated the
prevalence of exposure for the cohort population and that the matching criteria may have skewed
the control selection and produced controls who were not representative of the base cohort.
7-43
-------
The disagreement about the cohort and case-control study is still not resolved.
Nonetheless, it should be noted that the nested case-control study was the first to demonstrate a
strong association between exposure to 1,3-butadiene and occurrence of leukemias.
The main limitations of the cohort study were that more than 50% of the population was
excluded due to lack of work histories or start date and lack of exposure data. The follow-up for
four plants where the starting date was 1957 to 1970 may not have been long enough for
malignancies to develop. As far as the nested case-control study is concerned, as pointed out by
the authors, the estimated exposures were crude and not substantiated by IH data. The exposure
misclassification may have occurred based on the estimated exposure by job if the jobs were
incorrectly identified for higher or lower exposure. However, the panel members were blind
toward the status of cases and controls; thus the distribution of misclassification should be the
same in cases and controls.
In their latest study, Matanoski et al. (1997) estimated the exposures for each case and the
control. Leukemia was found to be associated at a TWA of In (ppm + 1) of butadiene. The other
associations observed in this study were Hodgkin's lymphoma with exposure to 1,3-butadiene
and NHL with styrene.
The Delzell et al. (1996) and Macaluso et al. (1996) cohort study is one of the best efforts
of exposure estimation to date. Some misclassification of exposure may have occurred with
respect to certain jobs, but it is unlikely to have occurred only in leukemia cases. The
investigators also did some validation of exposure estimates based on IH data. They pointed out
correctly that the excess mortality observed for leukemia was based on death certificates and was
not verified by medical records. Histologic typing of leukemia was also not available. This may
have resulted in misclassification. Two plants were eliminated from the final analysis due to the
lack of work histories, which may have resulted in the loss of valuable data.
In a commentary paper, Irons and Pyatt (1998) raised an issue of potential confounding
exposure to dithiocarbamates (DTC), which include compounds such as dimethyldithiocarbamate
(DMDTC) and diethyldithiocarbamate (DEDTC) and their respective oxidized dithiuram
counterparts such as thiram and disulfiram, in SBR production workers. When SBR production
was changed from hot process to cold process, DMDTC was used as a stopping agent in cold
process for SBR production from 1950 to 1965. No exposure data for DMDTC have been
presented in the Irons and Pyatt (1998) paper. Based on circumstantial evidence Irons and Pyatt
claim that since 87% of the polymerization, coagulation, and laboratory workers who developed
leukemia were employed during the 1950 to 1965 time period (Delzell et al., 1996), there was
more likelihood that exposure to DMDTC could be associated with the occurrence of leukemia in
this cohort.
7-44
-------
It is arguable whether DTCs are really confounding the results of Delzell et al. (1996). It
should be noted that to be a confounding factor the agent should be associated with exposure as
well as outcome. Dithiocarbamates have been in use since the 1940s as fungicides and to treat
parasitical skin diseases. Disulfiram has been in use for treatment of alcoholism. So far there is
not even a case report of leukemia in the literature in reference to any of the DTCs. Hence, at
this time it is conjecture on the authors' part that DTCs are causally associated with leukemia.
The main exposure route for DMDTC in SBR production facilities was dermal. Therefore, it is
inconceivable that workers could have had constant exposure to DMDTC 8 hours a day, 5 days a
week. Furthermore, since it was not known whether the workers who developed the leukemia
were actually exposed to DMDTC, the circumstantial evidence presented by the authors may be
an ecological fallacy. No data are available for DMDTC in animals. Pyatt et al. (1998) found
DMDTC to be a potent immunosuppressive compound in vitro. Relevance of this finding in
vivo is not known at this time.
When data on other related compounds, i.e., the oxidized dithiuram counterparts, were
reviewed, it was found that rats developed tumors when disulfiram was administered
simultaneously with nitrite via ingestion, but when it was administered alone, they did not
develop any tumors (PDR, 1998). An inhibitory effect of disulfiram on carcinogenesis has been
demonstrated in rats and mice for tumors induced by other chemicals at different sites (IARC,
1976). Moreover, both DEDTC and disulfiram, when administered to animals, inhibit the
cytochrome P-450 containing monooxygeneses and decrease the level of cytochrome P-450
detectable as its carbon monoxide complex. These two compounds also inhibit the cytochrome
P-450 when incubated with hepatic microsomes in vitro in the presence (but not absence) of
NADPH. This suggests that the inhibition reflects a primary effect of metabolism of the
compound, rather than secondary effect of liver damage (Neal and Halpert, 1982).
In their review of thiram, IARC (1991) reported on the only human data available to them
from the USSR. In a group of 223 workers who were employed in the manufacture of thiram for
more than 3 years, 1 case of thyroid cancer was reported among 105 workers examined. It
should be noted that this study was published in 1971. There are no other human data available
on this compound since then.
Thiram was tested via diet for 104 weeks (all the animals were killed at 112 weeks) in
SPF Fischer rats. The incidence of leukemia was significantly decreased in both male (10/50
control, 4/49 low dose -/X0.05, and 2/50 high dose -/X0.05) and female (14/49 control, 6/50
low dose -/><0.05, and 2/50 high dose -/><0.01) rats. In another diet study for 104 weeks
(experiment terminated at 130 weeks) in Fischer 344 rats, there was a significant decrease in
monocytic leukemia in both male and female treated rats (IARC, 1991).
7-45
-------
Villani et al. (1999), in their experiment with oral administration of thiram to B6C3FJ
male mice, found no significant increase of micronucleated splenocytes. A borderline increase of
DNA damage was found by Comet assay on circulating lymphocytes, but a negative result was
observed for splenocytes. On the other hand, in vitro, positive results with both genetic
endpoints were reported in assays with human lymphocytes in the dose ranges of 0.5 to 24 |J,/mL
and 0.1 to 8 |J,/mL for micronucleus and Come assays, respectively. Based on these results, the
authors concluded that despite its established genotoxicity in vitro, thiram is devoid of
appreciable clastogenic and/or aneugenic activity in vivo after oral administration to mice at the
maximum tolerated dose.
Therefore, the current data exhibit that DTCs are probably protective against causation of
cancer rather than the other way around.
Based on these monomer and polymer production worker cohorts, it is observed that an
increased number of lymphohematopoietic cancers occur in these populations. However, a clear
difference is becoming apparent. Increased NHLs develop in workers exposed to the monomer
(Downs et al., 1987; Divine, 1990; Divine et al., 1993; Divine and Hartman, 1996; Ward et al.,
1995, 1996c), while excess leukemias occur in workers exposed to the polymer (Matanoski et al.,
1990, 1993; Santos-Burgoa et al., 1992; Delzell et al., 1996; Macaluso et al., 1996, Sathiakumar
et al., 1998). Furthermore, in the monomer workers, probably the high exposures over a shorter
period of time (wartime) rather than low exposures over a longer period of time were associated
with observed excess of NHLs. As stipulated in their last follow-up (Divine and Hartman, 1996),
since most of the workers who developed NHLs were hired prior to the 1950s, when very high
peak exposures occurred, the peak exposures and not the cumulative exposures may be
associated with the observed excess of NHLs. The investigators also asserted that very high peak
exposures existed in the 1950s and 1960s, but neither any information about the frequency of
these peak exposures (compared to prior to the 1950s) nor any information about the variations
in intensity of peak exposures for these different time periods is available.
In the case of 1,3-butadiene, the scientific community is divided in two schools of
thought. One school believes that both NHLs and leukemias should be considered as tumors of
the same organ system since they both are mesodermic in origin, i.e., in the
lymphohematopoietic system, whereas the other school believes that they should be considered
as two separate site-specific tumors. This second group proposes that the excess observed for
leukemia in polymer production workers is due to the SBR process, and therefore the SBR
process and not 1,3-butadiene should be considered carcinogenic to humans. The Leukemia
Society of America defines "Leukemia, Lymphoma, Hodgkin's disease, and Myeloma are
cancers of the body's blood-forming and immune systems: the bone marrow and lymph nodes.
They are considered to be related cancers because they involve the uncontrolled growth of cells
7-46
-------
with similar function." Recently the International Lymphoma Group revised the classification of
the lymphohematopoietic system. The Revised European American Lymphoma (REAL) was
proposed in 1994 and is being adapted into the International Classification of Disease of
Oncology (ICD-O) (Berard and Hutchison, 1997). This classification is based on new ideas
evolving in the fields of molecular biology, genetics, and immunology that have rendered the old
classifications for lymphohematopoietic neoplasms obsolete (see footnote 4).
These monomer and polymer production worker cohorts demonstrate an excess number
of lymphohematopoietic cancers in the occupationally exposed workers. Increased NHLs are
reported for monomer production workers, whereas excess leukemias occur predominantly in
polymer production workers. There are several possible explanations for this apparent
discrepancy between the monomer and polymer workers. It has been hypothesized that the
observed excess of NHLs in the monomer production workers may be related to exposure
intensity, i.e., the excess risk may result from the high (peak) exposures during wartime rather
than the much lower exposures currently encountered by monomer production workers or the
likewise comparatively lower exposures encountered by polymer production workers. The
absence of a significant leukemia excess in these same monomer workers may be attributable to
low statistical power in the monomer studies. It should be noted that there is some suggestion of
excess leukemias in the monomer production workers, although these were not statistically
significant. The Union Carbide cohort had a leukemia excess of 23% based on 2 cases, and the
Texaco cohort has had several follow-ups since it was first studied by Downs et al. (1987).
Downs et al. studied the monomer workers cohort from 1943 through 1979 and reported a total
of seven leukemia cases. In a subsequent follow-up through 1986, eight deaths from leukemia
were reported by Divine (1990). In a further follow-up of this cohort through 1990, Divine et al.
(1993) observed 11 leukemia deaths, while Divine and Hartman (1996) found an elevated risk of
leukemias of 13% based on 13 cases through 1994. Though the Texaco cohort is a relatively
large monomer production cohort, it has low power to detect a statistically significant excess for
leukemias. Furthermore, with every follow-up, the investigators of the Texaco cohort increased
the calendar period for the worker inclusion criteria, hence adding many younger workers with
little cumulative exposure, shorter follow-up periods, and inadequate latency periods, thereby
diluting the risk. In addition, 1,3-butadiene is produced at the end of the monomer production
process, and current 1,3-butadiene exposures are very low in these workers. Finally, the workers
may lack exposure to a necessary co- or modifying factor that may be present in the polymer
production, resulting in the development of leukemias, although the findings of Delzell et al.
(1996) and Macaluso et al. (1996) show no evidence of confounding by exposure to other
chemicals.
7-47
-------
In fact, the apparent discrepancy may be largely an artifact. Under the latest classification
system for lymphohematopoietic cancers, all lymphomas not classified as Hodgkin's disease are
now included under NHL (see footnote # 4). Using this classification, an excess of NHL of 37%
(based on 15 cases; not statistically significant) was reported for workers who had worked > 10
years and with >20 years since hire in the UAB (polymer) cohort (Sathiakumar et al., 1998;
previously lymphosarcomas and NHL were reported separately for this cohort). Furthermore, as
these investigators report, their evaluation of NHL relations was limited by their reliance on
death certificates. NHL has high survival rates and may, in later clinical stages, transform into
leukemia; therefore leukemia may be reported on the death certificates. In addition, as discussed
above, nonsignificant excesses of leukemia were observed in two monomer studies. Thus,
excesses of both leukemia and NHL have been observed for both monomer and polymer
workers, and it may be that the increased risk of NHL is primarily observed among workers
exposed to high concentrations of 1,3-butadiene (mostly wartime monomer workers), whereas
the polymer production studies have greater power to detect a significant leukemia excess among
SBR workers who have modest to low exposures. In any event, leukemias and lymphosarcomas
are related tumor types and can both be classified as lymphohematopoietic cancers (see footnote
#4).
Finally, an alternate explanation is that the monomer workers may lack exposure to a
necessary co- or modifying factor that may be present in polymer production, resulting in the
development of leukemias, although the findings of Delzell et al. (1996) and Macaluso et al.
(1996) show no evidence of confounding by exposure to other chemicals.
7.4.4. Criteria of Causal Inference
In most situations, epidemiologic data are used to delineate the causality of certain health
effects. Several cancers have been causally associated with exposure to agents for which there is
no direct biological evidence. Insufficient knowledge about the biological bases for diseases in
humans makes it difficult to identify exposure to an agent as causal, particularly for malignant
diseases when the exposure was in the distant past. Consequently, epidemiologists and
biologists have provided a set of criteria that define a causal relationship between exposure and
health outcome. A causal interpretation is enhanced for studies that meet these criteria. None of
these criteria actually prove causality; actual proof is rarely attainable when dealing with
environmental carcinogens. None of these criteria should be considered either necessary (except
temporality of exposure) or sufficient in itself. The absence of any one or even several of these
criteria does not prevent a causal interpretation. However, if more criteria apply, they provide
credible evidence for causality.
7-48
-------
Thus, applying the criteria of causal inference to the monomer and polymer cohort
mortality studies and one nested case-control study, in which an increased risk of
lymphohematopoietic cancers was demonstrated, resulted in the following:
• Temporality. There is temporality of exposure to 1,3-butadiene prior to the occurrence
of NHLs in the monomer workers and leukemias in the SBR workers.
Strength of association. Strength of association between exposure and the occurrence
of NHLs in the wartime workers ranged from 154% to 477% higher risk among
workers exposed to the monomer as compared with the nonexposed general population
(Divine, 1990; Divine et al., 1993; Divine and Hartman, 1996; Ward et al., 1995,
1996c). The excess risk of leukemia ranged from 43% to 127% higher among workers
exposed to the SBR in ever-hourly workers compared with the general population
(Delzell et al., 1996). Internal comparison of the SBR worker population resulted in a
4.5-fold increased leukemia risk among the highest exposure group in the same cohort
(Macaluso et al., 1996). The nested case-control study from the SBR cohort showed a
7.6-fold increase in the risk of leukemia (Matanoski et al., 1989, 1993; Santos-Burgoa
etal., 1992).
• Consistency. Two cohort studies in monomer workers showed an increased risk of
NHLs (Divine, 1990; Divine et al., 1993; Divine and Hartman, 1996; Ward et al., 1995,
1996c), while one cohort study (Delzell et al., 1996; Macaluso et al., 1996) (with a
cohort derived from seven U.S. plants and one Canadian plant) and one nested case-
control study (Matanoski et al., 1989, 1993; Santos-Burgoa et al., 1995) showed an
excess risk of leukemia in SBR workers. The SBR workers cohort defined by Delzell
et al. (1996) showed a fairly consistent association between exposure to butadiene and
occurrence of leukemia across the plants included in the study. Furthermore, there is a
major population overlap between the SBR cohorts identified by JHU and by UAB.
These two groups of investigators have independently found excess leukemias in the
SBR workers.
• Specificity. Increased lymphohematopoietic cancers are observed in both the monomer
(NHLs) and polymer (leukemias) production workers. It should be noted that exposure
to a particular chemical (or drug or radiation) may cause more than one type of
leukemia or another type of hematopoietic cancer (Linet, 1985). Hays et al. (1997), in
7-49
-------
their Chinese cohort exposed to benzene, found that development of leukemia and NHL
was dose related.
• Biological gradient. The biological gradient, which refers to the dose-response
relationship, was observed only in SBR workers. Both the nested case-control study
and the cohort study showed increasing risk of leukemia with increasing exposures.
Such a relationship was not observed in monomer workers. The reason may be that a
very small number of people were exposed to high peak levels of 1,3-butadiene for a
shorter period of time who showed the occurrence of NHL. Due to a lack of
information on intensity and frequency of exposure levels they could not be further
stratified to evaluate the dose response. As noted earlier, the monomer studies did not
have enough power to detect a statistically significant excess of leukemia.
• Biological plausibility. As described in Chapter 4, hemoglobin adducts have been
detected in humans exposed to 1,3-butadiene (Osterman-Golkar et al., 1993; Sorsa et
al., 1996). Significantly increased frequencies ofhprt mutant lymphocytes were
observed in high-exposure groups by Legator et al. (1993) and Ward et al. (1994).
Mutations, chromosomal aberrations, and cell transformations, all well-established
steps in the process of carcinogenesis, were observed in human and animal studies.
This makes a convincing argument for the biological plausibility of occurrence of
leukemia in SBR workers and lymphosarcoma in monomer workers.
In summary, the findings of excess lymphohematopoietic cancers in polymer and
monomer production workers are consistent with a causal association with exposure to 1,3-
butadiene. As demonstrated above, the causality criteria of temporality, strength of association,
consistency, specificity, and biological gradient are satisfied. In addition, as discussed in the
earlier sections, 1,3-butadiene is metabolized by humans and other species to genotoxic
metabolites and is carcinogenic in mice and rats, thus fulfilling the criterion of biological
plausibility as well. Thus, the human evidence is considered sufficient.
7-50
-------
8. HAZARD CHARACTERIZATION
8.1. INTRODUCTION
The U.S. Environmental Protection Agency (EPA) first published a health assessment of
1,3-butadiene in 1985 (U.S. EPA, 1985). The 1985 assessment concluded that 1,3-butadiene was
a probable human carcinogen and calculated an upper bound cancer potency estimate of
0.25/ppm based on mouse data. Since then, a number of new studies on 1,3-butadiene have been
completed in various disciplines such as epidemiology, toxicology, and pharmacokinetics. The
purpose of this document is to review the new information and determine if any changes are
needed to the earlier conclusions.
The current reassessment focuses primarily on the major health effects, as requested by
EPA's Office of Transportation and Air Quality (formerly the Office of Mobile Sources):
carcinogenicity, mutagenicity, and reproductive/developmental toxicity. A detailed exposure
assessment was not requested nor conducted.
The major findings of this report are based on extensive carcinogenicity, toxicology,
mutagenicity, and pharmacokinetic data. Epidemiologic data consist of studies in monomer and
polymer production workers and include estimated exposures to 1,3-butadiene, styrene, and
benzene in some studies. Animal studies reviewed for the report consist of acute, subchronic,
and chronic assays, primarily in rodents.
This chapter briefly summarizes and integrates the critical data on which these hazard
findings are based and discusses the strengths and weaknesses of those data and the resulting
confidence in the findings. In addition, this chapter discusses the limited data on potentially
sensitive or more highly exposed subpopulations and suggests future research needs. With the
exception of the section on special subpopulations, all of the sections in this chapter discuss
material presented in the earlier chapters of this assessment.
1,3-Butadiene (CH2=CH-CH=CH2, CAS No. 106-99-0) is a colorless, odorless,
noncorrosive gas with a boiling point of -4.4°C and a vapor pressure of 1,900 mm/Hg at 20°C
(Kirshenbaum, 1978).
Approximately 12 billion pounds of 1,3-butadiene are produced annually worldwide, with
3 billion pounds produced in the United States (Morrow, 1990; USITC, 1990). It is used as an
intermediate in the production of polymers, elastomers, and other chemicals. The major uses of
1,3-butadiene are in the manufacture of styrene-butadiene rubber (SBR)—that is, synthetic
rubber—and of thermoplastic resins. 1,3-Butadiene is not a component of gasoline or diesel fuel,
it is formed as a by-product of incomplete combustion. The primary way the 1,3-butadiene is
released in the environment is via emissions from gasoline- and diesel-powered vehicles and
equipment. 1,3-Butadiene emissions from vehicles are reduced by catalytic converters; total
8-1
-------
emissions may decline as older cars without converters are removed from service. Other releases
occur from the combustion of other fossil fuels and biomass. Minor releases occur in production
processes, tobacco smoke, and vapors from the burning of plastics as well as rubber (Miller,
1978).
1,3-Butadiene is highly volatile and has a low solubility in water; thus environmental
release results in atmospheric contamination. Atmospheric destruction of 1,3-butadiene occurs
primarily by photoinitiated reactions. A significant amount of destruction also occurs by the gas
phase reaction with ozone and reaction with nitrate radicals at nighttime, particularly in urban
areas (U.S. DHHS, 1992). The major photooxidation products of 1,3-butadiene are acrolein and
formaldehyde (Maldotti et al., 1980). Although it is degraded rapidly in the atmosphere, 1,3-
butadiene is almost always present at low concentrations in urban and suburban areas. 1,3-
Butadiene is not found in significant amounts in food, soil, water, plants, fish, or sediment.
Therefore, the predominant pathway of exposure is via inhalation.
Monitoring done from 1987 to 1994 by Aerometric Information Retrieval System at more
than 20 different urban and suburban locations detected ambient air levels of 1,3-butadiene
ranging from 0.22 to 1.02 |-lg/m3 (0.10 to 0.46 ppb). Indoor air levels are likely to be higher than
ambient levels when smoking occurs. 1,3-Butadiene emissions from cigarettes have been
measured to be 200 to 400 |J,g/cigarette, and levels in smoke-filled bars have been found to range
from 2.7 to 19 |J,g/m3(1.2 to 8.4 ppb) (Lofroth et al., 1989; Brunnemann et al., 1990).
8.2. HAZARD ASSESSMENT
8.2.1. Metabolism and Pharmacokinetics
1,3-Butadiene is metabolized into several genotoxic metabolites by experimental
animals and humans. Metabolic activation is required for 1,3-butadiene carcinogenicity, and
there is evidence that 1,3-butadiene is metabolized to at least three genotoxic metabolites: a
monoepoxide (l,2-epoxy-3-butene, EB), a diepoxide (l,2:3,4-diepoxybutane, DEB), and an
epoxydiol (3,4-epoxy-l,2-butanediol, EBD) (see Figure 3-1 for a schematic of the 1,3-butadiene
metabolic pathways). Qualitatively, the metabolism of 1,3-butadiene is similar among species.
The enzymes responsible for the metabolic activation of 1,3-butadiene to these epoxide
metabolites, as well as the enzymes responsible for the detoxification of these reactive
metabolites, exist in humans as well as in mice and rats. EB and DEB have been measured in the
blood of rats, mice, and monkeys after 1,3-butadiene exposure, and their production by human
tissues has been observed in vitro. Formation of EBD has been observed in vitro using tissues
from mice, rats, and humans. Recently, EBD has also been detected indirectly in vivo in rats,
mice, and humans by the analysis of DNA and hemoglobin adducts.
8-2
-------
Although the metabolic pathways of 1,3-butadiene biotransformation are similar across
species, there are quantitative differences in the metabolic rates for various pathways between
different species. These rate differences result in quantitative differences in the blood and tissue
concentrations of the various metabolites across species that, in turn, likely account for some of
the differences in carcinogenicity and other toxicity observed between the mouse and rat.
Activation rates for 1,3-butadiene to EB and EB to DEB are higher in the mouse than in the rat.
In the limited number of human tissue samples that have been examined, activation rates in
humans exhibit a high degree of variability and appear to nearly span the range between mice and
rats, although most of the rates are well below those in mice. Detoxification pathways also differ
quantitatively, with glutathione conjugation being the predominant pathway for the removal of
EB and DEB in the mouse and hydrolysis being the major detoxification pathway in humans. In
the rat, EB and DEB in the liver are removed at a slightly greater rate by glutathione transferase
than by epoxide hydrolase. Overall, the greater rates of activation compared with detoxication
for biotransformation of 1,3-butadiene in mice versus rats is reflected by higher blood and tissue
concentrations of EB and DEB in vivo.
Hemoglobin adducts, which can serve as a measure of tissue dose, have been detected in
exposed workers (Osterman-Golkar et al., 1993; Sorsa et al., 1996) as well as in the blood of
exposed mice and rats.
8.2.2. Mutagenicity
1,3-Butadiene, through its metabolism, has exhibited mutagenic and genotoxic activity
in a wide variety of in vitro and in vivo assay systems. The genetic toxicology literature on 1,3-
butadiene, EB, and DEB consists of more than 450 publications with positive genotoxic findings
in viruses, bacteria, plants, and animals. Consistent with (more accurately, as a consequence of)
the differences in metabolism described in Chapter 3, mutagenic responses in mice are
consistently greater than the responses in rats when animals were exposed to 1,3-butadiene. The
observation that the species difference in hprt response was greater after 4 weeks exposure than
after 2 weeks (Meng et al., 1999a) suggests that the differences would continue to diverge with
the length of exposure. A range of studies, both in vitro and in vivo, on two major metabolites
indicate that EB primarily induces point mutations and small deletions, while DEB, as a
bifunctional alkylating agent, also results in larger deletions. Additionally, DEB is mutagenically
active in in vitro systems at concentrations that are similar to those found in the blood of mice
exposed to 1,3-butadiene by inhalation. EB requires concentrations 10-fold to 100-fold greater in
vitro than those found in the blood of 1,3-butadiene-exposed mice. However, a recent study with
EB and DEB measuring hprt mutations in female mice and rats revealed that DEB was about
threefold more mutagenic in the Fischer 344 rats than in C3B6F1 mice (Meng et al., 1999b).
-------
EBD has been less extensively studied, but recent evidence suggests that most of the
trihydroxybutyl-guanine adducts in mice and rats exposed to 1,3-butadiene are derived from
EBD, and that this is also a major metabolite in vivo. Positive dominant lethal and heritable
translocation studies in mice indicate that exposure to 1,3-butadiene can result in germ cell
mutation and heritable risk.
Significantly increased frequencies ofhprt mutant lymphocytes were observed in high-
exposure workers in a butadiene production plant by Ward et al. (1994, 1996b), but not in rubber
plant workers in China (Hayes et al., 1996) or in a Czech 1,3-butadiene plant (Tates et al., 1996).
Both positive (Sram et al., 1998) and negative (Sorsa et al., 1994) findings for chromosome
damage have also been reported in exposed workers. Although the reasons for the differences in
these findings are not clear, several in vitro studies indicate that genetic polymorphisms for
enzymes involved in the metabolism of 1,3-butadiene affect the induction of genetic damage
(Uuskula et al., 1995; Norppa et al., 1995; Bernardini et al., 1998).
These mutagenicity and molecular data suggest that 1,3-butadiene is both a somatic and
germ cell mutagen in mammals, possibly including humans.
8.2.3. Characterization of Reproductive and Developmental Effects
A variety of reproductive and developmental effects have been observed in mice
exposed to 1,3-butadiene by inhalation. There are no human data on reproductive or
developmental effects. The most sensitive short-term developmental endpoint was decreased
fetal weight in the mouse. Decreases were observed at the lowest exposure concentration (40
ppm, 6 h/day, gestation days 6-15); thus there was not a no-observed-adverse-effect level
(NOAEL) for this effect. No developmental toxicity was observed in rats.
The most sensitive reproductive endpoint observed in subchronic exposure studies was
fetal deaths in dominant lethal studies of mice (i.e., male mice were exposed to 1,3-butadiene and
effects on litters were measured after mating to unexposed females). The dominant lethal
responses are believed to represent a genotoxic effect. Dominant lethal effects in humans would
likely be manifested as infertility (due to reduced fertility or very early deaths) or spontaneous
abortions.
From chronic exposure studies (2-year bioassays), the most sensitive reproductive effects
were ovarian atrophy in female mice and testicular atrophy in male mice. Testicular atrophy was
primarily a high-exposure effect. Ovarian atrophy, on the other hand, was observed at the lowest
exposure level (6.25 ppm, 6 h/day, 5 days/week for 2 years). Uterine atrophy was also observed
in the highest exposure groups; however, this is likely to be a secondary effect of the ovarian
atrophy. The mechanisms of ovarian atrophy are unknown, although there is strong evidence that
the effect is mediated by the diepoxide metabolite (see Chapter 5). It is further expected, based
8-4
-------
on metabolic data, that humans would produce lower concentrations of this metabolite than do
mice (see Chapter 3). Thus, it is likely that humans are less sensitive to 1,3-butadiene-induced
ovarian atrophy than are mice.
8.2.4. Cancer Characterization
Cancer characterization is a synthesis of all pertinent information addressing the question
of how likely an agent is to be a human carcinogen. EPA's 1986 Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 1986) provided a classification system for the characterization of the
overall weight of evidence for potential human carcinogen!city based on human evidence, animal
evidence, and other supportive data. This system included Group A: Human Carcinogen, Group
B: Probable Human Carcinogen, Group C: Possible Human Carcinogen; Group D: Not
Classifiable as to Human Carcinogenicity; and Group E: Evidence for Noncarcinogenicity to
Humans.
As part of the guidelines development and updating process, the Agency has developed
revisions to the 1986 guidelines to take into account knowledge gained in recent years about the
carcinogenic processes. With regard to the weight-of-evidence evaluation for potential human
carcinogenicity, EPA's 1996 Proposed Guidelines for Carcinogen Risk Assessment (U.S. EPA,
1996) and the subsequent revised external review draft Guidelines for Carcinogen Risk
Assessment (U.S. EPA, 1999b) emphasize the need for characterizing cancer hazard, in addition
to hazard identification. Accordingly, the question to be addressed in hazard characterization is
expanded to how likely an agent is to be a human carcinogen, and under what exposure
conditions a cancer hazard may be expressed. The revised guidelines also stress the importance
of considering the mode(s) of action information for making an inference about potential cancer
hazard beyond the range of observation, typically encountered at levels of exposure in the general
environment. "Mode of action" refers to a series of key biological events and processes that are
critical to the development of cancer. This is contrasted with "mechanisms of action," which is
defined as a more detailed description of the complete sequence of biological events at the
molecular level that must occur to produce a carcinogenic response. Furthermore, the 1999
guidelines (U.S. EPA, 1999b) have extensive discussion on children's health and sensitive
populations and will serve as EPA's interim guidance to risk assessors (Federal Register, 2001).
To express the weight of evidence for potential human carcinogenicity, EPA's proposed
guidelines utilize a hazard narrative in place of the classification system. However, in order to
provide some measure of consistency, standard hazard descriptors are used as part of the hazard
narrative to express the conclusion regarding the weight of evidence for potential human
carcinogenicity.
3-5
-------
The sections to follow evaluate and weigh the individual lines of evidence and combine
all evidence to make an informed judgment about the potential human carcinogenicity of 1,3-
butadiene. A conclusion regarding carcinogenicity in accordance with EPA's 1999 Guidelines
for Carcinogen Risk Assessment (U.S. EPA, 1999b) is provided, which also includes a hazard
narrative along with an appropriate hazard descriptor.
8.2.4.1. Cancer Evidence From Experimental Animals
1,3-Butadiene is an animal carcinogen. Several chronic inhalation bioassay studies
have been conducted with 1,3-butadiene: a 2-year rat study (HLE, 1981b); two lifetime mouse
studies (NTP 1984, 1993)—the first terminated early because of excessive mortality and the
second one using lower exposure concentrations; a 2-year stop-exposure study with male mice
(NTP, 1993); and a 1-year study comparing the induction of thymic lymphomas in two different
strains of male mice (Irons et al., 1989). These studies provide unequivocal evidence that 1,3-
butadiene is a multisite carcinogen in both rats and mice (see Chapter 6). These studies also
demonstrate that the mouse is more sensitive than the rat to 1,3-butadiene-induced
carcinogenicity and develops tumors at different sites, although the reasons for these interspecies
differences are not fully understood at this time. The most sensitive site was the female mouse
lung, which exhibited significantly increased tumor incidence at the lowest exposure
concentration tested (6.25 ppm). Other tumors that exhibited increased incidences in the mouse
include lymphomas, heart hemangiosarcomas, histiocytic sarcomas, and tumors of the lung,
forestomach, Harderian gland, liver, preputial gland, ovary, and mammary gland. In the rat,
increases were observed in tumors of the brain, testis, pancreas, Zymbal gland, thyroid gland,
uterus, and mammary gland. The mammary gland (female) is the only site with significant tumor
increases common to both mice and rats.
8.2.4.2. Cancer Evidence in Humans
Sufficient evidence exists of increased lymphohematopoietic cancer in workers exposed
to 1,3-butadiene to consider it carcinogenic to occupationally exposed populations. The
conclusion of "sufficient evidence" to consider 1,3-butadiene carcinogenic to occupationally
exposed populations is based mainly on epidemiologic studies examining workers exposed to
1,3-butadiene. An excess of lymphohematopoietic cancers was observed in 1,3-butadiene
polymer production workers (studied by two different investigators) and monomer production
workers in North America. An excess of leukemias was observed in polymer production workers
while an excess of non-Hodgkin's lymphomas (NHL) (previously diagnosed as lymphosarcoma
and reticular sarcoma, but now included in NHL per the new classification in the International
Classification of Diseases of Oncology [ICD-O]} was observed in monomer workers. In light of
8-6
-------
the revised classification adopted by the International Lymphoma Group's Revised European-
American Lymphoma (REAL) classification and the Leukemia Society of America, these cancers
are included in the category of lymphohematopoietic cancers, and thus will be so used by the
authors of this assessment document.1
The epidemiologic studies in polymer and monomer production workers reviewed in
Chapter 7 are summarized in Table 8-1.
The strongest evidence comes from a retrospective cohort study of 15,000 synthetic
rubber workers (UAB cohort) with 49 years of follow-up Delzell et al. (1996). Quantitative
exposures (cumulative and peak) to 1,3-butadiene, styrene, and benzene were estimated for each
worker (Macaluso et al., 1996). Limited validations of exposure estimates were attempted by
various means. Significant excesses ranging from 43% to 336% were found for leukemia in
ever-hourly workers as compared with the general population, after adjusting for styrene and
benzene. An internal comparison, using estimated ppm-years of 1,3-butadiene exposure resulted
in increasing risk ratios for leukemia with increasing exposures. This trend was statistically
significant. A fairly consistent association between exposure to 1,3-butadiene and occurrence of
leukemia across the six plants was also found.
The major strengths of this study are as follows. First, the study had detailed and
comprehensive quantitative exposure estimations for 1,3-butadiene, styrene, and benzene for
each individual. Second, the cohort was large, with a long follow-up period of 49 years. Third,
both external and internal comparison showed similar results. Fourth, adjustments for potential
confounding factors were carried out. Fifth, analyses by duration of employment and for latency
were conducted.
The study had some limitations. First, some misclassifications of exposure may have
occurred with respect to certain jobs, but it is unlikely to have occurred only in leukemia cases
because the exposures were calculated a priori. Second, the excess mortality observed for
leukemia was based on death certificates and was not verified by medical records. This may
have resulted in misclassification. Third, histologic typing of leukemia was also not available, so
currently it is not known whether a single cell type or more than one cell type is associated with
Under the previous classification (8th ICD, Adapted), lymphohematopoietic cancers comprised the following subcategories: lymphosarcoma and
reticular sarcoma, Hodgkin's disease, leukemia, and other lymphatic tissue cancers. In 1994, the International Lymphoma Group's REAL
classification was proposed for the lymphohematopoietic cancers and is being adopted into the ICD-O (Berard and Hutchison, 1997). This
classification is based on new ideas evolving in the fields of molecular biology, genetics, and immunology, which have rendered the old
classification for lymphohematopoietic cancers obsolete. The REAL classification comprises the following subcategories: B-cell neoplasms, T-cell
and putative natural killer (NK)-cell neoplasms, Hodgkin's disease, and unclassified lymphomas. It should be noted: Both leukemias and
lymphomas that are produced by B-cells are included under B-cell neoplasms, and leukemias and lymphomas produced by T-cells and NK-cells are
included under T-cell and NK-cell neoplasms. Please note that any lymphoma (such as B-cell, T-cell, and NK-cell) that is not classified as
Hodgkin's disease is also included under NHL.
Furthermore, the Leukemia Society of America defines lymphohematopoietic cancers as follows: "Leukemia, Lymphoma, Hodgkin's disease,
and Myeloma are cancers of the body's blood forming and immune systems: the bone marrow and lymph nodes. They are considered to be related
cancers because they involve the uncontrolled growth of cells with similar functions."
8-7
-------
Table 8-1. Summary of key epidemiologic studies.
Plants
7 U.S. and 1
Canadian polymer
production plants
(UAB cohort)3
7 U.S. and 1
Canadian polymer
production plants
(JHU cohort)3
1 U.S. monomer
production plant
(Texaco cohort)
3 U.S. monomer
production plants
(Union Carbide
cohort)
1 U.S. monomer
production plant
(Shell Oil Deer
Park cohort)
Number of
workers,
dates studied
15,000,
1943-1994
13,500,
1943-1985
2,800,
1943-1994
364,
1940-1990
614,
1948-1989
Authors
Delzelletal., 1996
Macaluso et al.,
1996
Matanoski and
Schwartz, 1987
Matanoski et al.,
1989, 1990, 1993,
1997
Santos-Burgoa et
al., 1992
Downs etal., 1987
Divine, 1990
Divine etal., 1993
Divine and
Hartman, 1996
Ward etal., 1995,
1996a
Cowlesetal., 1994
Approach
Cohort study using
quantitative exposure
estimates for 1,3-
butadiene, styrene,
and benzene for each
worker
Cohort study using
qualitative
exposures;
case-control study
using estimated
quantitative
exposures for each
case and control
Cohort study using
qualitative
exposures, last
follow-up includes
quantitative exposure
estimates
Cohort study using
qualitative exposures
Cohort study using
qualitative exposures
Significant findings
— Excess mortality for leukemia
observed in ever-hourly
workers, SMR = 143 to 436
— 4.5-fold increased leukemia
risk among the highest exposure
group with internal comparison
— Excess leukemia observed
consistently across the plants
— Leukemia risk increased with
increasing exposure level
— Excess mortality due to
lymphohematopoietic cancers
observed in cohort study
— 7-fold increase for leukemia
observed in case-control study
— Leukemia risk increased with
increasing exposure level in
case-control study
— Excess mortality due to
NHL, SMR = 176 in wartime
workers
— Earlier studies report excess
mortality due to
lympho sarcoma, SMR = 254 to
269 in wartime workers
— Excess mortality due to
lympho sarcoma13
— SMR = 577 in workers hired
between 1942 and 1952
— No increase in cancer
mortality or cancer morbidity
observed
a Six U.S. plants and one Canadian plant were common in Johns Hopkins University (JHU) and University of Alabama,
Birmingham (UAB), studies.
bNow included under NHL.
SMR: standard mortality ratio.
-------
the exposure to 1,3-butadiene. Fourth, two plants were eliminated from the final analysis due to
the lack of work histories, which may have resulted in the loss of valuable data. Finally, Irons
and Pyatt (1998) have recently raised an issue of potential confounding exposure to
dithiocarbamates (DTCs). The DTCs comprise thiono-sulfur compounds. The representative
compounds include dimethyldithiocarbamate (DMDTC) and diethyldithiocarbamate (DEDTC)
and their oxidized counterparts thiram and disulfiram, respectively. DTCs have been in use since
the early 1940s as fungicides and treatments for parasitic skin diseases. The only human study in
thiram manufacturing workers (one of the DTC) reported a single case of thyroid cancer (IARC,
1991). Available animal studies have not provided any evidence that DTC causes
carcinogenesis. As a matter of fact, the studies in animals observed decreased incidence of
leukemia that was dose related (higher the dose, lower the incidence) as compared with control
animals. Two different diet studies where thiram was administered for 104 weeks (SPF Fischer
rats and Fischer 344 rats) observed significantly decreased incidence of leukemia in both male
and female treated rats (IARC, 1991). Disulfiram has been in use for the treatment of
alcoholism. So far there is not even a case report of leukemia in the literature in reference to any
of the DTCs. Therefore, at this time it is conjecture that DTCs are causally associated with
leukemia, hence confounding the results of Delzell et al. (1996). A detailed discussion on DTCs
can be found in Section 7.4.3.
Additional evidence is provided by the studies conducted earlier by Matanoski and
Schwartz (1987), Matanoski et al. (1989, 1990), Santos-Burgoa et al. (1992), and Matanoski et
al. (1993) in many of these same polymer workers (13,500 individuals; JHU cohort).2 A
significant excess of lymphohematopoietic cancer but not leukemia was observed in the cohort
studies, while a significantly increased odds ratio of 7 for leukemia was observed in a nested
case-control study of this cohort (Santos-Burgoa et al., 1992; Matanoski et al., 1993). A
significant trend of increasing risk of leukemia with increasing exposure level of 1,3-butadiene
was also observed. The finding of a high excess of leukemia mortality in the case-control study
was questioned by Acquavella (1989) and Cole et al. (1993) because no excess leukemia
mortality was found in the base cohort study from which the cases and controls were selected.
These authors asserted that the results of the case-control study were statistically incompatible
with the results of the cohort study. Matanoski and Santos-Burgoa (1994) disagreed with this
assertion. The debate about the findings of the cohort study and the case-control study is still
unresolved. Nonetheless, it should be noted that this nested case-control study was the first to
demonstrate a strong association between exposure to 1,3-butadiene and occurrence of
One Canadian plant and six U.S. plants were common in the JHU and the UAB cohorts.
8-9
-------
leukemias. A detailed discussion of the differing views about the findings of the cohort study
and the case-control study can be found in Section 7.4.3.
Three different cohorts of monomer production workers have been studied. The strongest
evidence of human carcinogenicity is provided by the largest cohort of approximately 2,800
workers in a Texaco plant studied by several investigators (Downs et al., 1987; Divine, 1990;
Divine et al., 1993; Divine and Hartman, 1996). The only significant excess mortality observed
was for lymphosarcoma (now included in NHL) in the wartime subcohort of workers (154% to
169% higher than the general population). The investigators estimated exposures for each
individual in their last follow-up (Divine and Hartman, 1996) and found that except for an excess
observed for NHL (76% higher than the general population) in the wartime subcohort, there were
no excesses in any cause-specific cancer mortality. Since they did not find any association
between cumulative exposure and occurrence of NHL in their survival analysis, they indicated
that it may be peak exposures rather than cumulative exposures that are associated with the
observed increase. No information about peak exposure levels was available for the wartime
period that was found to be associated with the occurrence of NHL; however, it is believed that
exposures were high.
The major strengths of this study are, first, it is a relatively large cohort of monomer
workers. Second, it had a long follow-up period of 52 years. Third, analyses by duration of
employment and for latency, as well as adjustment for potential confounding factors, were
conducted. Fourth, the exposures in each individual were estimated in the last follow-up.
Except for "Hire-age" in survival analysis using Cox model, after 52 years of follow-up this
study did not find any statistically significant excess in leukemia (observed in SBR workers);
however, this study may not have enough power to detect the leukemia increase.
Some of the limitations of the study are as follows. First, although the exposures were
estimated in their last follow-up, there were no data or means available to the investigators to
estimate the peak exposures that were hypothesized to be associated with the observed increase
in NHL in wartime workers. Second, although the authors state that extremely high peak
exposures did exist during the 1950s and 1960s, neither any information about the frequency of
these peak exposures (as compared with prior to the 1950s) nor any information about the
variations in intensity of peak exposures for these different time periods was available. Third,
there is a lack of information on other confounding factors. Fourth, although the cohort is
relatively large, it had low power to detect excess leukemias (power calculation resulted in 25%
power to detect the observed 13% increase). Nonetheless, the finding of excess mortality from
lymphosarcoma (now included in NHL) is consistent with findings of Ward et al. (1995, 1996c)
and, although not statistically significant, of Meinhardt et al. (1982).
8-10
-------
A small cohort of 364 individuals who had potential exposure to 1,3-butadiene at three
Union Carbide plants was studied by Ward et al. (1995, 1996a). A statistically significant excess
for lymphosarcoma (477% higher than the general population) was found based on four cases.
The main limitations of this study are that the cohort was small and that exposures were assumed
based on department codes. In addition, there was no analysis for latency or adjustments for
potential confounding by exposure to other chemicals.
Cowles et al. (1994) studied the third cohort of 614 workers. No excess cancer was
observed either in mortality or morbidity studies. Due to several methodologic limitations (such
as small cohort size, short follow-up, lack of exposure information), the absence of positive
results in this study does not provide any negative evidence towards the causal association
between exposure to 1,3-butadiene and occurrence of cancer.
All the epidemiologic studies, cohort and nested case-control, evaluated for this
assessment are observational studies in occupationally exposed populations. As such, they have
various methodologic strengths and limitations as discussed above. A common limitation to all
the studies is the use of death certificates, which could lead to misclassification bias. Validation
of diagnosis of lymphohematopoietic cancer was not done in any of the studies except in
Matanoski et al. (1993).
The scientific community is divided into two schools of thought about the evaluation of
carcinogenicity of 1,3-butadiene. One school believes that both NHL and leukemias should be
considered as tumors of the same organ system since they are both mesodermic in origin, that is,
lymphohematopoietic system (see footnote 1 on p. 8-6), while the other school believes that they
should be considered as two separate site-specific tumors. This second group proposes that the
excess observed for leukemia in polymer production workers is due to the SBR process, and
therefore the SBR process and not 1,3-butadiene should be considered carcinogenic to humans.
These monomer and polymer production worker cohorts demonstrate an excess number
of lymphohematopoietic cancers in these occupationally exposed workers. Increased
lymphosarcomas are reported for monomer production workers, whereas excess leukemias occur
predominantly in polymer production workers. There are several possible explanations for this
apparent difference between the monomer and polymer workers. It has been hypothesized that
the observed excess of lymphosarcomas in the monomer production workers may be related to
exposure intensity, i.e., the excess risk may result from the high (peak) exposures during wartime
rather than the much lower exposures currently encountered by monomer production workers or
the likewise comparatively lower exposures encountered by the polymer production workers.
The absence of a significant leukemia excess in these same monomer workers may be
attributable to low statistical power in the monomer studies. There is some suggestion of excess
leukemias in the monomer production workers, although these were not statistically significant.
8-11
-------
The Union Carbide cohort had a leukemia excess of 23% based on 2 cases, and the Texaco
cohort had an elevated risk of leukemias of 13% based on 13 cases. Even the Texaco cohort,
which is a relatively large monomer production cohort, has low power to detect a statistically
significant excess for leukemias, and with every follow-up, the investigators of the Texaco cohort
increased the calendar period for the worker inclusion criteria, hence adding many younger
workers with little cumulative exposure, shorter follow-up periods, and inadequate latency
periods, thereby diluting the risk. Furthermore, it should be noted that in every follow-up since
the initial study by Downs et al. (1987) additional cases of leukemia (7 to 13) have been reported
by investigators. In addition, 1,3-butadiene is produced at the end of the monomer production
process and current 1,3-butadiene exposures are very low in these workers. Finally, the workers
may lack exposure to a necessary co- or modifying factor that may be present in the polymer
production, resulting in the development of leukemias, although the findings of Delzell et al.
(1996) and Macaluso et al. (1996) show no evidence of confounding by exposure to other
chemicals.
In fact, the apparent difference may be largely an artifact. Under the latest classification
system for lymphohematopoietic cancers, all lymphomas not classified as Hodgkin's disease are
now included under NHL (see footnote #1). Using this classification, an excess of NHL of 37%
(based on 15 cases; not statistically significant) was reported for workers who had worked > 10
years and with > 20 years since hire in the UAB (polymer) cohort (Sathiakumar et al., 1998;
previously lymphosarcomas and NHL were reported separately for this cohort). Furthermore, as
these investigators report, their evaluation of NHL relations was limited by their reliance on
death certificates. NHL has high survival rates and may, in later clinical stages, transform into
leukemia, therefore leukemia may be reported on the death certificates. In addition, as discussed
above, nonsignificant excesses of leukemia were observed in two monomer studies. Thus,
excesses of both leukemia and NHL have been observed for both monomer and polymer
workers, and it may be that the increased risk of NHL is primarily observed among workers
exposed to high concentrations of 1,3-butadiene (mostly wartime monomer workers), whereas
the polymer production studies have greater power to detect a significant leukemia excess among
SBR workers who have modest to low exposures. In any event, leukemias and lymphosarcomas
are related tumor types and can both be classified as lymphohematopoietic cancers (see footnote
#1).
Thus, applying the modified "Hill criteria" of causal inference to the monomer and
polymer cohort mortality studies and nested case-control study, in which an increased risk of
lymphohematopoietic cancers was demonstrated, results in the following:
8-12
-------
Temporality. There is temporality of exposure to 1,3-butadiene prior to the
occurrence of NHL in the monomer workers and leukemias in the SBR workers.
Strength of association. Strength of association between exposure and the
occurrence of NHL in the wartime workers ranged from 154% to 477% higher risk
among workers exposed during monomer production as compared with the
nonexposed general population (Divine, 1990; Divine et al., 1993; Divine and
Hartman, 1996; Ward et al., 1995, 1996c). The excess risk of leukemia ranged from
43% to 336% higher among ever-hourly workers exposed during SBR production as
compared with the general population (Delzell et al., 1996). Internal comparison of
the SBR worker population resulted in a 4.5-fold increased leukemia risk among the
highest exposure group in the same cohort (Macaluso et al., 1996). The nested case-
control study from the SBR cohort showed a sevenfold increase in the risk of
leukemia (Matanoski etal., 1989, 1993; Santos-Burgoa et al., 1992).
Consistency. Two cohort studies in monomer workers showed an increased risk of
NHL (Divine, 1990; Divine et al., 1993; Divine and Hartman, 1996; Ward et al.,
1995, 1996c), while one cohort study (Delzell et al., 1996; Macaluso et al., 1996; with
a cohort derived from seven U.S. plants and one Canadian plant) and one nested case-
control study (Matanoski et al., 1989, 1993; Santos-Burgoa et al., 1992) showed an
excess risk of leukemia in SBR workers. The SBR workers cohort defined by Delzell
et al. (1996) showed a fairly consistent association between exposure to 1,3-butadiene
and occurrence of leukemia across the plants included in the study. Furthermore,
there is a major population overlap between the SBR cohorts identified by JHU and
by UAB. These two groups of investigators have independently found excess
leukemias in the SBR workers.
Specificity. Increased lymphohematopoietic cancers are observed in both the
monomer (NHL) and polymer (leukemias) production workers. It should be noted
that exposure to a particular chemical (or drug or radiation) may cause more than one
type of leukemia or another type of lymphohematopoietic cancer (Linet, 1985).
Hayes et al. (1997) in their Chinese cohort exposed to benzene found that
development of leukemia and NHL were dose related.
8-13
-------
• Biological gradient. The biological gradient, which refers to the dose-response
relationship, was observed only in SBR workers. Both the nested case-control study
and the cohort study showed increasing risk of leukemia with increasing exposures.
Such a relationship was not observed in monomer workers. The reason may be
because only a small number of people who were exposed to high peak levels of 1,3-
butadiene for a short period of time were probably at increased risk of NHL. Due to a
lack of information of intensity and frequency of exposure levels they could not be
further stratified to evaluate the dose response. As noted earlier, the monomer studies
did not have enough power to detect statistically significant excess of leukemia.
• Biological plausibility. Extensive data suggest that the metabolism of 1,3-butadiene
is qualitatively similar across species investigated, although there are differences in
the amounts of the active metabolites (detailed in Chapter 3). Hemoglobin adducts
have been detected in exposed workers. In addition, some studies have reported
significantly increased frequencies ofhprt mutant lymphocytes and chromosome
damage in exposed workers.
In summary, the findings of excess lymphohematopoietic cancers in polymer and
monomer production workers are consistent with a causal association with exposure to 1,3-
butadiene. As demonstrated above, the causality criteria of temporality, strength of association,
specificity, biological gradient (in SBR workers only), and consistency are satisfied. In addition,
as discussed in the earlier sections, 1,3-butadiene is metabolized by humans and other species to
genotoxic metabolites and is carcinogenic in mice and rats, thus fulfilling the criterion of
biological plausibility as well. Thus, the human evidence is considered sufficient.
8.2.4.3. Cancer Characterization Conclusion
1,3-Butadiene is carcinogenic to humans by inhalation. This characterization is
supported by the total weight of evidence provided by the findings discussed above: (1)
epidemiologic studies showing increased lymphohematopoietic cancers in workers
occupationally exposed to 1,3-butadiene (by inhalation), (2) laboratory studies showing that 1,3-
butadiene causes a variety of tumors in mice and rats by inhalation, and (3) studies demonstrating
that 1,3-butadiene is metabolized into genotoxic metabolites by experimental animals and
humans. The specific mechanisms of 1,3-butadiene-induced carcinogenesis are unknown;
however, it is virtually certain that the carcinogenic effects are mediated by genotoxic
8-14
-------
metabolites of 1,3-butadiene. Under EPA's 1999 Guidelines for Carcinogen Risk Assessment
(U.S. EPA, 1999b), 1,3-butadiene is characterized as "carcinogenic to humans" by inhalation.
8.2.4.4. Recent Evaluations of 1,3-Butadiene by Other Organizations
In recent years, various organizations have arrived at different classifications of
carcinogenicity or hazard characterizations for 1,3-butadiene. The conclusions of the evaluations
conducted over the past 5 years are as follows:
1. 9th Report on Carcinogens, 2000, by the National Toxicology Program (NTP): In their
most recent report (NTP, 2000), mandated by the U.S. Congress, 1,3-butadiene is listed as
known to be a human carcinogen. This listing was based on sufficient evidence of
carcinogenicity from studies in humans, including epidemiologic and mechanistic
information, to indicate a causal relationship between occupational exposure to 1,3-
butadiene and excess mortality from lymphatic and hematopoietic cancers.
2. International Agency for Research on Cancer (IARC): In their re-evaluation of some
organic chemicals (IARC, 1999), 1,3-butadiene was classified asprobably carcinogenic
to humans - Group2A. This classification was based on limited evidence in humans and
sufficient evidence in animals for the carcinogenicity of 1,3-butadiene, and sufficient
evidence in experimental animals for the carcinogenicity of l,2:3,4-diepoxybutane.
3. List of MAK and BAT Values of Germany: In their maximum concentrations and
biological tolerance values at the workplace, Germany's Commission for the
Investigation of Health Hazards of Chemical Compounds in the Work Area (Deutsche
Forschungsgemeinschaft, 1999) has classified 1,3-butadiene as Category 1, a substance
that causes cancer in humans and that can be assumed to make a significant contribution
to cancer risk. The basis for this classification was that epidemiologic studies provide
adequate evidence of an association between the exposure of persons and the
development of cancer. Alternatively, inadequate epidemiologic data can be substituted
with information about the mechanism of action in humans.
4. Occupational Safety and Health Administration (OSHA), 1996: OSHA, part of the U.S.
Department of Labor, in its final 1,3-butadiene rule (OSHA, 1996), listed 1,3-butadiene
as a potential human carcinogen. OSHA concluded that on the basis of its analysis of
animal studies and epidemiologic studies, there is strong evidence that workplace
8-15
-------
exposure to 1,3-butadiene poses increased risk of death from cancers of the
lymphohematopoietic system. The epidemiologic findings supplemented the findings
from the animal studies, which demonstrated a dose response for multiple tumors,
particularly for lymphomas in mice exposed to 1,3-butadiene.
5. Canadian Environmental Protection Act, 1998: Health Canada (1998) stated that 1,3-
butadiene is considered highly likely to be carcinogenic in humans. This
characterization was based on the evidence of an association between exposure in the
occupational environment and leukemia that fulfills several of the traditional criteria for
causality of association observed in epidemiologic studies, supporting limited data on
genotoxicity in human populations and the overwhelming weight of evidence of
carcinogenicity and genotoxicity at relatively low concentrations in some species of
experimental animals.
All of the evaluations described here were based on the same database. Some of the
interpretation differences were because of various uncertainties and data gaps in the current 1,3-
butadiene literature. These evaluations are summarized in Table 8-2.
8.3. SPECIAL SUBPOPULATIONS
8.3.1. Sensitive Subpopulations
It is uncertain whether children or other subpopulations have greater susceptibility to
exposure to 1,3-butadiene than does the general population. There is no information available
in humans on the potential carcinogenicity in children from exposure to 1,3-butadiene at this
time. Occurrence of leukemia is causally associated with exposure to 1,3-butadiene in adults,
and leukemia is one of the most common cancers in children. Furthermore, leukemia risk in
children has been shown to increase with simultaneous exposure to multiple risk factors (Gibson
et al., 1968). Thus, exposure to 1,3-butadiene may be an additional risk factor increasing the
leukemia risk further in children.
Both the rat and mouse studies suggested that females are more sensitive to
1,3-butadiene-induced carcinogenicity than males, and the female mammary gland was the only
1,3-butadiene-related tumor site common to both species. The underlying mechanism for the
observed gender differences in rodents is unknown. Breast cancer rates are increasing in women
of all ages. The known risk factors explain only a small proportion of the occurrence of breast
8-16
-------
Table 8-2. Carcinogenicity assessments of 1,3-butadiene.
Agency (year)
Cancer classification
Remarks
NTP (2000)
"Known to be a human carcinogen" —based on
sufficient evidence of carcinogenicity in humans,
including epidemiologic and mechanistic
information to indicate causal relationship
Using NTP classification system
IARC (1999)
"2A-probably carcinogenic to humans"—based on
limited evidence in humans, sufficient evidence in
animals, and sufficient evidence in experimental
animals for the carcinogenicity of 1,2:3,4-
diepoxybutane
Cancer classification using IARC
system.
Deutsche
Fo rs chungsgemeins chaft
(1999)
"Category 1" —based on epidemiologic studies
provide adequate evidence between exposure and
the development of cancer
MAK and BAT guidance for
classification
OSHA (1996)
"Potential occupational carcinogen" —based on
strong evidence in workers exposed at workplace
poses increased risk of death from cancer and dose-
response for multiple tumors in animals
Using OSHA standards of carcinogen
characterization
Health Canada (1998)
Proposed "highly likely to be carcinogenic to
humans" —based on fulfillment of several
causality criteria in humans, limited data on
genotoxicity in humans, and overwhelming
evidence of carcinogenicity and genotoxicity in
animals at low doses
Cancer characterization system of
Canadian Environmental Protection
Act
cancer. If exposure to 1,3-butadiene is causally associated with breast cancer in women, then it
is an additional breast cancer risk factor, the risk from which could be reduced by reducing
exposure to 1,3-butadiene.
Tobacco smoke contains 1,3-butadiene as well as other carcinogens, and a few studies
suggest that parental smoking increases the risk of leukemia or lymphoma in children (John et
al., 1991; Stjernfeldt et al., 1986). The overall evidence, however, is inconclusive because other
studies observed no increased risk. Furthermore, if there is an effect in children from parental
smoking, it is unclear whether it is attributable to preconception effects on fathers' sperm, in
utero exposure of the fetus, and/or postnatal exposure to environmental tobacco smoke.
Because metabolic activation of 1,3-butadiene to epoxide metabolites is believed to be
necessary for carcinogenicity, it is possible that genetic differences in activation or detoxification
8-17
-------
enzymes could result in different risks to different human subpopulations. For example,
investigators have observed that polymorphism in glutathione-S-transferase genes confers
differential susceptibility to the induction of sister chromatid exchanges by 1,3-butadiene
metabolites in cultured human lymphocytes. However, the critical/rate-limiting mechanistic
steps are unknown at present; thus, it is unknown whether actual human subpopulations may
have notably different susceptibility to 1,3-butadiene.
There is no information available in humans on the developmental effects of 1,3-
butadiene exposure at this time. The animal data, particularly in the mouse, indicate an effect
from prenatal exposure on growth and an effect from paternal exposure on prenatal death. Only
the Anderson et al. (1993, 1995, 1996) study followed animals postnatally after paternal
exposure, and no significant effects were detected. The postnatal effects of prenatal exposures
have not been evaluated, but the fact that fetal weight is the critical effect for this type of
exposure suggests the possibility of greater sensitivity of the developing animal than the adult,
and the long-term consequences of 1,3-butadiene exposure may indicate that they are at even
greater risk. Thus, exposure to 1,3-butadiene may be an additional risk factor for developmental
disorders in children.
The effects of chronic exposure on ovarian, uterine, and testicular atrophy in mice suggest
that females may be at greater risk than males because of the greater sensitivity of the ovary to
the effects of 1,3-butadiene exposure. The ovarian atrophy effects suggest that exposure to 1,3-
butadiene may increase the risk for women of early menopause. Early menopause has been
shown to result from cigarette smoking, of which 1,3-butadiene is a component. Further
elucidation of species differences in the levels of the diepoxide metabolite, which has been
shown to be a major factor in ovarian atrophy, is important in determining the risks of early
menopause in women.
8.3.2. Highly Exposed Subpopulations
Some subpopulations may be at greater risk than the general population as a result of
higher exposure to 1,3-butadiene. Heavy smokers may be highly exposed to 1,3-butadiene due
to its formation in tobacco smoke. Cigarette smoke has been shown to be a risk factor for
various types of leukemias. It should be noted, however, that known and suspected
leukemogenic constituents of tobacco smoke include benzene, polonium-210, nitrosamines, and
hydrocarbons in addition to 1,3-butadiene (Schottenfeld and Fraumeni, 1996).
8-18
-------
8.4. FUTURE RESEARCH NEEDS
Although there is sufficient evidence to characterize 1,3-butadiene as carcinogenic to
humans in this assessment, there are some data gaps in various areas that, if filled, would refine
the assessment. The specific research needs are as follows:
8.4.1. Epidemiology
• Exposures in the monomer production workers should be estimated using the same
methodology that was used for the SBR production workers by the University of
Alabama investigators. If it turns out that the exposures to 1,3-butadiene in the
monomer and SBR production are fairly comparable, then that would suggest that co-
exposures to other factor(s) in the SBR production (not 1,3-butadiene alone) are
associated with the occurrence of leukemia. If, on the other hand, the exposures to
1,3-butadiene in the monomer production workers are considerably lower than in the
SBR production workers, then the lack of a statistically significant excess leukemia in
the monomer cohort (13) would not nullify the interpretation of the causal association
between exposure to 1,3-butadiene and excess leukemia in the SBR production
workers.
• Dithiocarbamate exposure should be estimated for the UAB cohort and its
relationship to leukemias analyzed.
• The medical records for the leukemia cases in the studies by Delzell et al. (1995,
1996) and Macaluso et al. (1996) should be reviewed to verify the cell types of
leukemias.
• Further followup of the SBR cohort will serve two goals: (1) to determine whether
any excesses in cardiovascular disease or other site-specific malignancies with a
longer latency period are associated with exposure to 1,3-butadiene, and (2) to
examine the subcohort hired after 1965 when the DMDTC exposures were drastically
reduced and use of protective gear became mandatory. The investigators also will
have to consider whether the exposures to 1,3-butadiene were also reduced at that
time. Only then can the association between DMDTC and occurrence of leukemia
deaths be interpreted.
8-19
-------
• Mortality studies should be conducted in other monomer and SBR production worker
cohorts to examine whether the results of the North American studies can be
duplicated.
• All epidemiologic studies to date have examined male cohorts. Some 1,3-butadiene
production facilities around the world (e.g., China) employ women in their
laboratories. If the number of women in these facilities is large enough, a cancer
study might help determine whether women have an increased risk of getting breast or
ovarian cancers, as observed in rodents. Similarly, a reproductive/developmental
study might help determine whether there is an increased risk of any other
reproductive effects or of developmental effects in exposed fetuses.
• A reproductive study of exposed males is also needed to examine potential dominant
lethal effects in humans.
• Biomarkers of exposure have been recently identified in a transitional study. A health
outcome study in these workers would be valuable.
8.4.2. Toxicology
• A bioassay should be performed to observe the effects of combined exposures to 1,3-
butadiene and DMDTC.
• Bioassays should be performed to evaluate if there are any neurological effects,
especially developmental neurotoxicity effects, associated with exposure to 1,3-
butadiene.
• Further research is needed on the three active metabolites of 1,3-butadiene to
elucidate the mechanism of action for testicular atrophy, abnormal sperm heads,
ovarian atrophy to ovarian neoplasms, fetal weight loss, and dominant effects.
• Elucidation of the mechanisms responsible for the interspecies differences in
sensitivity to 1,3-butadiene could assist in resolving questions about the human risk
for reproductive effects and for cancer at sites for which the Delzell et al. (1995,
1996) study may have had insufficient power to detect an effect.
8-20
-------
• Postnatal evaluation of pre- and perinatal exposure to 1,3-butadiene and
multigeneration studies might elucidate the longer term consequences of
developmental exposures, as well as the reproductive toxicity in both males and
females, including early ovarian atrophy in females.
8.4.3. Molecular Biology
• Once the mechanisms of 1,3-butadiene-induced health effects are better understood,
information on polymorphisms in human metabolic enzymes (or DNA repair
enzymes, etc.) could help define sensitive subpopulations.
8.5. SUMMARY AND CONCLUSIONS
The purpose of this assessment is to review the new information that has become
available since EPA's 1985 health assessment of 1,3-butadiene (U.S. EPA, 1985) and to
determine if any changes were needed to the earlier conclusions.
1,3-Butadiene is a gas used commercially in the production of styrene-butadiene rubber,
plastics, and thermoplastic resins. The major environmental source of 1,3-butadiene is the
incomplete combustion of fuels from mobile sources (e.g., automobile exhaust). Tobacco smoke
can be a significant source of 1,3-butadiene in indoor air.
This assessment concludes that 1,3-butadiene is carcinogenic to humans by inhalation,
based on the total weight of evidence: (a) epidemiologic studies showing increased
lymphohematopoietic cancers (leukemias in polymer and NHL in monomer workers) in workers
occupationally exposed to 1,3-butadiene (by inhalation), (b) studies showing that 1,3-butadiene
causes a variety of tumors in mice and rats by inhalation, and (c) studies demonstrating that 1,3-
butadiene is metabolized into genotoxic metabolites by experimental animals and humans. The
specific mechanisms of 1,3-butadiene-induced carcinogenesis are unknown; however, it is
virtually certain that the carcinogenic effects are mediated by genotoxic metabolites of
1,3-butadiene.
Animal data suggest that females may be more sensitive than males for cancer effects;
nevertheless, there are insufficient data from which to draw any conclusions on potentially
sensitive subpopulations.
In summary, the primary change in EPA's conclusions about the health effects of 1,3-
butadiene from the 1985 document is that the cancer characterization has been changed from
probable human carcinogen to carcinogenic to humans by inhalation.
8-21
-------
9. PHARMACOKINETIC MODELING
9.1. INTRODUCTION
Several physiologically based pharmacokinetic (PBPK) models of 1,3-butadiene
metabolism and disposition have been developed to attempt to explain the interspecies
differences in the potency and site specificity of the carcinogenic response between mice and rats
and to provide a corresponding dosimetric basis for quantitatively extrapolating carcinogenic
potency from rodents to humans. PBPK models use species-specific physiological parameters
such as alveolar ventilation rates and blood flow rates, chemical-specific distribution parameters
such as blood:air and tissue:blood partition coefficients, and species- and chemical-specific
metabolic rates to elucidate the pharmacokinetics (i.e., the uptake, distribution, metabolism, and
excretion) of a chemical.
Ideally, such models provide species-specific target tissue doses of the lexicologically
active form(s) of the chemical. Carcinogenic risks from bioassay data can then be extrapolated
to humans on the basis of equivalent effective doses, reducing some of the uncertainties that
occur when interspecies extrapolation is based simply on exposure to the parent compound,
especially when nonlinear physiological processes are involved. Assumptions must still be made
to the effect that the mechanisms of action of the active form(s) of the compound at the target
tissue(s) are the same across species and that the tissues of different species are equally sensitive.
If these assumptions are not valid, pharmacodynamic data and modeling would be required for
more precise risk assessment.
PBPK models that fall short of describing target tissue doses of the active form(s) of a
chemical may still be useful for improving the dosimetric basis of interspecies extrapolation for
quantitative risk assessment. For example, it is well established that metabolic activation of
1,3-butadiene is necessary for its carcinogenic action (Chapter 4). Therefore, a PBPK model
describing the production and disposition of l,2-epoxy-3-butene (EB), the first product of
metabolic activation of 1,3-butadiene, may be able to provide a better dose metric than the
default methodology of using exposure to 1,3-butadiene itself.
This chapter reviews and analyzes the PBPK models for 1,3-butadiene that are currently
available and assesses their usefulness for quantitative risk assessment of 1,3-butadiene based on
interspecies extrapolation. These models generally assume, for simplicity, that the transfer of
1,3-butadiene to tissues is blood flow-limited and that each tissue compartment is "well mixed."
9-1
-------
9.2. PBPK MODELS FOR 1,3-BUTADIENE
9.2.1. Hattis and Wasson (1987)
The first PBPK model for 1,3-butadiene was the unpublished model of Hattis and
Wasson (1987). They defined the effective dose of 1,3-butadiene as the amount that is
metabolically converted to EB and used this dose as a basis for a risk assessment of occupational
1,3-butadiene exposure. Their model consists of three compartments: a fat compartment; a
muscle compartment; and a liver and vessel-rich compartment, which includes the brain, heart,
kidneys, and other small visceral organs. The transfer of 1,3-butadiene between blood and
tissues is assumed to be blood flow-limited. Metabolism to the monoepoxide is ascribed to the
entire liver and vessel-rich compartment and is assumed to follow simple Michaelis-Menten
kinetics. No further metabolism of EB is considered.
The only chemical-specific parameter values then available were whole-body maximal
metabolic rates for mice and rats inferred from the chamber study data of Kreiling et al. (1986b).
These data provided the Vmax and preliminary KM estimates for the liver and vessel-rich
compartment. Tissue:blood and blood:air partition coefficients were estimated from chemical
structure and solubility data using empirical relationships (e.g., Fiserova-Bergerova and Diaz,
1986). Model simulations were then run, adjusting KM and the blood:air partition coefficient to
fit the blood 1,3-butadiene concentration data of Bond et al. (1986), to derive "best estimates" for
these parameters. Human metabolic rates were estimated by allometric scaling of the mouse and
rat rates because no PBPK data were available for human metabolism of 1,3-butadiene. The
parameter values used by Hattis and Wasson (1987) are summarized in Table 9-1.
No additional data were available at that time for an independent validation of this model.
A minimal sensitivity analysis was conducted by varying KM and the blood:air partition
coefficient among a few values and observing the effect on the ultimate risk estimates. Hattis
and Wasson (1987) claimed that their model is not very sensitive to reasonable differences in
partition coefficients. Similarly, the model is insensitive to the precise value of the metabolic
parameters because, given the blood:air partition coefficient values that were used, metabolic
conversion in their model is limited by blood flow to the liver and vessel-rich compartment.
The Hattis and Wasson (1987) model is not discussed further here because it has been
superseded by new data and other modeling efforts.
9.2.2. Hallenbeck (1992)
Hallenbeck (1992) reported having done a PBPK-based cancer risk assessment for
1,3-butadiene; however, he provided no details of the PBPK model that he used. Furthermore,
he used the area under the 1,3-butadiene concentration-versus-time curve for the lung as his
9-2
-------
Table 9-1. Parameter values used in the Hattis and Wasson (1987) PBPK model.
Parameter
Physiological parameters
Alveolar ventilation
(L/min)
Weight (kg)
Qf (L/min)
Qm (L/min)
Qlvr (L/min)
Vf(L)
Vm(L)
Vlvr (L)
Partition coefficients0
Blood:air
P™
Metabolic parameters
Vmax (mol/min)
KM (mol/L)f
Rat
0.15
0.40
0.0136
0.0226
0.1042
0.028
0.300
0.036
0.35
118.2
5.26
5.4
1.47E-6d
5E-6f
Mouse
0.0233
0.028
0.00192
0.00319
0.01617
0.0028
0.0196
0.00308
0.35
118.2
5.26
5.4
1.87E-7d
5E-6f
Human
11.38a
4.8b
70
0.69a
0.35b
2.61a
5.09a
4.35b
14.024
34.756
8.513
0.35
118.2
5.26
5.4
8.0E-56
5E-6f
aAwake.
bAsleep.
°The blood:air partition coefficient of 0.35 is the "best estimate" value from "fitting" the model. The
tissue:blood partition coefficients (P) are from functions of the blood:air partition coefficient for which the
"best estimate" value of 0.35 was used. Partition coefficients are assumed to be the same across species.
dFrom Kreilmg et al. (1986b).
Trom allometric scaling of the rodent values.
f"Best estimate" from "fitting" the model.
Subscripts f, m, and Ivr designate the fat, muscle, and liver and vessel-rich compartments (tissues), respectively.
Q: tissue blood flow rate.
V: tissue volume.
P: tissue:blood partition coefficient.
tissue-dose surrogate, taking no account of metabolic activation. As presented, this model
contributes nothing to the current state of knowledge regarding the pharmacokinetic modeling of
1,3-butadiene.
9-3
-------
9.2.3. Kohn and Melnick (1993)
The PBPK model of Kohn and Melnick (1993) focuses on the disposition of EB in the
mouse, rat, and human. This model incorporates additional tissues (compartments) and
metabolic reactions based on experimental data that were not available at the time of the Hattis
and Wasson (1987) model; however, it also relies on theoretically derived partition coefficients.
The Kohn and Melnick model is blood flow-limited and consists of six compartments: lung,
blood, fat, liver, other rapidly perfused tissues (viscera), and slowly perfused tissues (muscle).
Metabolism occurs in the liver, lung, and viscera compartments. The metabolic reactions include
conversion of 1,3-butadiene to EB, the conversion of EB to l,2:3,4-diepoxybutane (DEB), the
enzymatic hydrolysis of EB, and the enzymatic conjugation of EB with glutathione.
With the exception of the partition coefficients, which were derived theoretically from
published methodologies, all of the mouse, rat, and human parameter estimates were from
experimental data in the literature. None of the parameter values were obtained by optimization.
The parameter values used by Kohn and Melnick (1993) are summarized in Table 9-2.
Blood:tissue partition coefficients for 1,3-butadiene were from Hattis and Wasson (1987). The
blood:air partition coefficients reported by Csanady et al. (1992) for 1,3-butadiene and EB were
used as lung:air partition coefficients. The fatblood partition coefficient for EB was calculated
using an empirical relationship from Lyman et al. (1990), whereas the tissue:blood partition
coefficients of EB for the other tissues were derived using the method of Fiserova-Bergerova and
Diaz (1986). These are essentially the same procedures used by Hattis and Wasson (1987).
Michaelis-Menten kinetics were used to describe the oxidation of 1,3-butadiene and EB
by the cytochrome P-450 isozyme CYP2E1, the hydrolysis of EB by epoxide hydrolase, and the
glutathione S-transferase-catalyzed conjugation of EB with glutathione. KM and Vmax values for
each of these reactions in the liver and lung of the mouse, rat, and human were taken from the in
vitro data of Csanady et al. (1992). The lung values were also assumed to apply to the viscera
compartment. Csanady et al. reported kinetic constants for the oxidation of EB to DEB only for
mouse liver preparations. Therefore, Kohn and Melnick (1993) included DEB production only in
the mouse liver compartment and only as a disappearance route for EB; the distribution of DEB
was not further modeled. 1,3-Butadiene and EB were treated as competitive inhibitors of each
other in the rate equations for mouse liver CYP2E1. Finally, although glutathione was treated as
saturating for glutathione S-transferase in the mouse, rat, and human liver, glutathione
conjugation with EB in human lung and viscera was assumed to be first order.
To validate their model, Kohn and Melnick (1993) compared predicted 1,3-butadiene
absorption and blood concentrations for mice and rats with the measurements of Bond et al.
(1986). They also modified the model to include a chamber compartment and compared
predicted EB concentrations in the chamber and maximum metabolic elimination rates with the
9-4
-------
Table 9-2. Parameter values used in the Kohn and Melnick (1993)
PBPK model.
Parameter
Physiological parameters"
Body weight (kg)
Cardiac output (L/h)
Ventilation rate (L/h)
Fraction blood
Fraction fat
Fraction liver
Fraction viscera
Fraction muscle
Fat flow fraction
Liver flow fraction
Viscera flow fraction
Muscle flow fraction
Partition coefficients0
Air partition BD
Fat partition BD
Liver partition BD
Viscera partition BD
Muscle partition BD
Air partition EB
Fat partition EB
Liver partition EB
Viscera partition EB
Muscle partition EB
Biochemical parameters for
1,3-butadiene oxidation*
Liver V cytl (nmol/h/mg)
Liver Km cytl (mM)
Lung V cytl (nmol/h/mg)
Lung Km cytl (mM)
Mouse
0.028
1.044
2.64
0.05
0.04
0.062
0.05
0.78
0.05
0.16
0.52
0.19
Rat
0.4
7.32
15.6
0.054
0.08
0.05
0.083
0.59
0.07
0.16
0.40
0.36
Human
70
660b
l,200b
0.077
0.144
0.025
0.037
0.547
0.036
0.16
0.446
0.361
1.5
118.2
5.49
5.34
5.26
60
1.8083
0.6545
0.6348
0.6533
155.4
0.002
138.6
0.00501
35.4
0.00375
9.6
0.00775
70.8
0.00514
9
0.002
9-5
-------
Table 9-2. Parameter values used in the Kohn and Melnick (1993)
PBPK model (continued).
Parameter
Biochemical parameters'1
Liver V cyt2 (nmol/h/mg)
Liver Km cyt2 (mM)
Liver V EH (nmol/h/mg)
Liver Km EH (mM)
Liver V GST (nmol/h/mg)
Liver Km GST (mM)
Lung k hydr (h'Vmg)
Lung V GST (nmol/h/mg)
Lung Km GST (mM)
Lung k GST (liVmg)
Protein concentrations
Liver micro prot (mg/L)d
Liver cyto prot (mg/L)d
Lung micro prot (mg/L)e
Lung cyto prot (mg/L)f
Mouse
12
0.0156
347.4
1.59
30,000
35.3
0.1116
6,380
36.5
11,600
82,800
3,000
82,800
Rat
148.8
0.26
14,460
13.8
0.0792
2,652
17.4
16,800
108,000
3,000
108,000
Human
1,110
0.58
2,706
10.4
0.1914
0.1536
14,500
58,000
3,000
58,000
^Compartment volumes are given as fractions of body weight; compartment blood flow rates are given as
fractions of cardiac output.
bHuman cardiac output at rest: 336 L/h; human ventilation rate at rest: 240 L/h.
°Lung:air and tissue:blood; assumed same for all species.
dData from Csanady et al. (1992).
ePersonal communication from R. Philpot, NIEHS.
fAssumed same as for liver.
BD: 1,3-butadiene; EB: l,2-epoxy-3-butene.
V: Vmax;Km: KM.
cytl denotes oxidative metabolism of butadiene to EB; cyt2 denotes oxidative metabolism of EB.
EH: epoxide hydrolase.
GST: glutathione S-transferase.
micro prot: microsomal protein; cyto prot: cytoplasmic protein.
k hydr: apparent first-order rate constant for EB hydrolysis; k gst: apparent first-order rate constant for glutathione
conjugation.
9-6
-------
Laib et al. (1990) results for mice and rats. Kohn and Melnick claimed that their model
predictions are comparable to the experimental results except for overestimates in the blood 1,3-
butadiene concentrations, which they ascribed to inadequacies in the model or experimental
sources of error in the blood concentration measurements.
To assess the sensitivity of the model to the values of various parameters, relative
sensitivity coefficients for different model variables were estimated by finite differences, as given
by Frank (1978). The physiological parameters to which the model was the most sensitive were
the lung:air partition coefficient and the cardiac output. Because the ventilation rate is greater
than the rate of 1,3-butadiene absorption, the lung:air partition coefficient and the cardiac output
are the major parameters governing 1,3-butadiene uptake. Predicted 1,3-butadiene concentrations
were not very sensitive to variations in the biochemical parameters; however, EB levels were
somewhat more sensitive to the parameters describing hepatic glutathione S-transferase and
epoxide hydrolase kinetics.
Based on their model simulations, Kohn and Melnick (1993) reported that 1,3-butadiene
uptake and the disposition of EB are controlled to a greater extent by physiological parameters
than by biochemical parameters. The model further suggests that storage in fat is a significant
fraction of retained 1,3-butadiene, especially in rats and humans. Kohn and Melnick also found
that predicted EB tissue concentrations do not correlate with tumor incidences in mice and rats,
and they concluded that other factors are crucial in 1,3-butadiene-induced carcinogenesis. These
other factors may include pharmacokinetic variables that were not part of the model, such as
accumulation of the diepoxide or formation of other metabolites or mechanistic
(pharmacodynamic) phenomena, such as formation of DNA adducts or efficiency of DNA repair.
The Kohn and Melnick (1993) model appears to have a reasonable basic structure, in
terms of the compartments and metabolic reactions included, given the biochemical parameters
that were available at the time. A major strength of their model is that none of the parameter
estimates is adjusted to fit experimental data. Two important drawbacks of the model are the use
of empirically derived partition coefficients and the lumping of various tissues with different
metabolic capabilities (Chapter 3) into a viscera compartment, which is assumed to have the same
metabolic activity as the lung. Partition coefficients for 1,3-butadiene and EB have since been
measured by Johanson and Filser (1993) and Medinsky et al. (1994), and experimental values for
the 1,3-butadiene partition coefficients are substantially less than the empirically derived
estimates, which suggests that the specific results reported by Kohn and Melnick may not be
relevant. For example, the role of physiological parameters in controlling 1,3-butadiene uptake
and the amount of 1,3-butadiene storage in fat may not, in fact, be as great as the Kohn and
Melnick model predicts (Medinsky et al., 1994). More recent revisions to this model are
discussed in Sections 9.2.7.1 and 9.2.10.
9-7
-------
9.2.4. Johanson and Filser (1993)
Johanson and Filser (1993) developed a PBPK model for 1,3-butadiene and EB
disposition in rats and mice. Their model is blood flow-limited and consists of four main
physiological compartments—lungs and arterial blood, muscle and vessel-rich tissues, fat, and
liver—as well as a chamber compartment and an intrahepatic subcompartment. Metabolism is
assumed to take place exclusively in the liver. The metabolic reactions include oxidation of 1,3-
butadiene to EB; hydrolysis of EB; intrahepatic first-pass hydrolysis of EB; conjugation of EB
with glutathione, which is described by a "ping-pong" mechanism; and the turnover and depletion
of hepatic glutathione.
In contrast with the previous PBPK modeling efforts for 1,3-butadiene, Johanson and
Filser (1993) conducted in vitro studies of rat homogenates to obtain empirical values for the
tissue:air partition coefficients for 1,3-butadiene and EB. All physiological parameters were taken
from Arms and Travis (1988), except the alveolar ventilation rates, which were reduced to 60% of
those suggested by Arms and Travis on the basis of generalized observations of uptake rates of
various gases in closed-chamber experiments (Johanson and Filser, 1992).
For the oxidative metabolism of 1,3-butadiene, the model uses the Vmax values from the in
vitro studies of Filser et al. (1992). A KM value was derived by fitting the model to the in vivo
data of Lieser (1983) for the rat and Kreiling (1986b) for the mouse because the model could not
reproduce the results observed in these closed-chamber studies using the KM values of either
Filser et al. (1992) or Csanady et al. (1992). Values for the metabolic parameters pertaining to the
conjugation of EB with glutathione and to the hydrolysis of EB were taken from the in vitro data
of Kreuzer et al. (1991). The value of the "intrinsic KM" for the intrahepatic hydrolysis of EB (see
below) was set to 20% of the "apparent KM" value of Kreuzer et al. because the model then fit
various in vivo data. The flow rate between the hepatic and intrahepatic compartments was
estimated from the kinetic parameters. The physiological and biochemical parameter values used
by Johanson and Filser (1993) are summarized in Table 9-3.
In terms of the metabolic reactions involved, the Johanson and Filser (1993) model differs
from the Kohn and Melnick (1993) model in that further oxidation of EB to DEB is not included,
conjugation of EB with glutathione is described by the two-substrate ordered sequential ping-
pong mechanism (reviewed by Mannervik, 1985) rather than by Michaelis-Menten kinetics, and
glutathione turnover and the intrahepatic first-pass hydrolysis of EB are incorporated. Given
the KM values for glutathione conjugation used in the model, the conjugation of EB becomes rate-
limited by glutathione only when glutathione is almost completely depleted. Cytosolic
glutathione turnover is depicted by zero-order production and first-order elimination. Intrahepatic
first-pass hydrolysis of EB is hypothesized to occur, based on the observations of
9-8
-------
Table 9-3. Parameter values used in the Johanson and Filser (1993) PBPK model.
Parameter
Physiological data
Body weight (g)
Alveolar ventilation (mL/min)
Cardiac output (mL/min)
Blood flows
(% of cardiac output)
Compartment volumes
(% of body weight)
Partition coefficients'3
1,3 -Butadiene
1 ,2-Epoxy-3 -butene
Standard animal
Simulations
Standard animal
Simulations
Standard animal
Simulations
Muscle and VRG
Fat
Liver
Lung and arterial
Muscle and VRG
Fat
Liver
Lung and arterial, muscle and VRG, liver
Fat
Blood
Lung and arterial, muscle and VRG, liver
Fat
Blood
Mouse
25
27.5
15
proportional to bw2'3
17
proportional to bw2'3
66
9
25
1
75
10
5.5
Rat
250
157.5-217.5a
70.2
83
66
9
25
1
80
7
4
0.25
7.23
3.03
0.706
1.89
83.4
VO
-------
Table 9-3. Parameter values used in the Johanson and Filser (1993) PBPK model (continued).
Parameter
Metabolic constants
1,3 -Butadiene oxidation
EB hydrolysis
EB conjugation
Glutathione kinetics
Microsomal protein (mg/g liver)
Vmax (nmol-mm'-mg"1)0
KM (|imol/L air)d
Microsomal protein (mg/g liver)
Vmax (nmol-mm'-mg"1)6
Apparent KM (mmol/L)e
Intrinsic KM (% of apparent KM)d
Cytosolic protein (mg/g liver)
Vmaj/KMofEB(nL-min-1-mg-1)e
KM toward EB (mmol/L)e
KM toward glutathione (mmol/L)f
Initial steady-state concentration (mmol/L)
Elimination rate constant (h"')f
Mouse
30
3.22
5
30
19
1.5
20%
95
15
100
0.1
8.31g
5.5h
0.15
Rat
30
2.17
5
30
17
0.7
20%
95
11
100
0.1
5.56g
4.2h
0.15
VO
o
^Depending on experiment simulated.
bTissue:blood and blood:air. measured for rats; assumed same for mice.
Trom Filser (1992).
dObtained by best fit.
eKreuzer et al. (1991).
fAverage of literature data.
8Deutschmann and Laib (1989).
hKreilmg et al. (1988).
VRG: vessel-rich tissue group.
EB: l,2-epoxy-3-butene.
bw% = (body weight)%.
-------
Filser and Bolt (1984), because of proximity of the monooxygenase to the epoxide hydrolase in
the endoplasmic reticulum. Newly formed EB within this intrahepatic compartment will be more
readily hydrolyzed than EB that must diffuse in from outside the compartment, as reflected by a
lower KM in the intrahepatic compartment.
To attempt to validate the model, Johanson and Filser (1993) compared simulated results
with the data from various in vivo experiments. In addition to the 1,3-butadiene kinetics data
used to fit the KM for 1,3-butadiene oxidation and the EB kinetics data of Filser and Bolt (1984)
for the rat and Kreiling (1987) for the mouse that were used to fit the intrinsic KM for intrahepatic
first-pass hydrolysis, the model apparently reproduces the EB concentrations appearing in
chamber air as a result of 1,3-butadiene exposure in the experiments of Rolzhauser (1985) for the
rat and Kreiling (1987) for the mouse. However, it is not clear from the text whether these
experimental data were also used to fit the intrinsic KM. The model also reproduces the
glutathione concentrations observed by Deutschmann (1988) in rat and mouse liver after 1,3-
butadiene exposure, and Johanson and Filser claimed that no model parameters were fitted to
these data. Finally, simulated blood concentrations of EB approximate those observed by Bond
et al. (1986) in the mouse but are slightly higher than those observed in the rat.
No sensitivity analysis for the model parameters was reported.
The results of Johanson and Filser's (1993) model simulations suggest that the internal
dose of EB, expressed as the concentration of EB or the area under the concentration-time curve
in the venous blood, the other compartments, or the whole body, is at most about three times
greater in the mouse than in the rat for a given exposure concentration. The greatest differences
in internal dose of EB between the two species result from 1,3-butadiene exposure concentrations
of above 1000 ppm, when glutathione depletion occurs in the mouse but not in the rat after 6 to 9
h of exposure. Once again, the relatively small interspecies differences in body burden of EB
indicated by PBPK modeling cannot explain the striking differences in cancer response between
mice and rats exposed to 1,3-butadiene. Johanson and Filser suggested that differences in the
kinetics of DEB or nonmetabolic factors, such as differences in immune response or in the
expression of oncogenes, may be responsible for the interspecies differences in cancer response.
A major advancement found in the PBPK model of Johanson and Filser (1993) is the use
of experimentally derived partition coefficients, especially because these values differ
substantially from the theoretically estimated values. A further strength of their analysis is that
they compared the simulation results with data from several different experiments. The Johanson
and Filser model also incorporates hepatic glutathione turnover and depletion as well as
intrahepatic first-pass hydrolysis of EB, although the significance of these refinements is
unknown. Some of the limitations of the model include the exclusion of extrahepatic metabolism
and of further metabolism of EB to DEB. In addition, the values of the KM for 1,3-butadiene
9-11
-------
oxidation and of the intrinsic KM for intrahepatic first-pass hydrolysis of EB were obtained by
fitting in vivo data. Finally, no sensitivity analysis was reported, although, for example, it was
acknowledged that wide ranges of glutathione concentrations and turnover rates have been
observed. Therefore, it is unknown how sensitive the model is to changes in these and other
parameters. More recent modifications to this model are discussed in Sections 9.2.7.2 and
9.2.7.3.
9.2.5. Evelo et al. (1993)
Evelo et al. (1993) present a PBPK model for the uptake, distribution, and metabolic
clearance of 1,3-butadiene in mice and rats. Their stated objective was to investigate the relative
importance of liver and lung metabolism at different 1,3-butadiene exposure concentrations. The
Evelo et al. model has six physiological compartments: liver, fat, muscle, a vessel-rich group,
the bronchial area of the lung, and the alveolar area of the lung. A chamber compartment is also
included for validation against the data from closed-chamber experiments. 1,3-Butadiene
metabolism is assigned to both the alveolar and bronchial areas of the lung and to the liver. Gas
exchange occurs in the alveolar area of the lung.
Values for the standard physiological parameters were allometrically scaled from the data
of Travis (1988). Volumes and blood flows for the two separate lung compartments were taken
from Greep and Weis (1977). Tissue:blood and blood:air partition coefficients were theoretically
estimated using the regression analysis method of Fiserova-Bergerova and Diaz (1986), as was
done previously by Hattis and Wasson (1987).
To describe the oxidation of 1,3-butadiene to EB, Evelo et al. (1993) calculated the ratios
of the maximum metabolic activity between the liver and the lung from the in vitro data of
Schmidt and Loeser (1985) for the mouse and the rat. Then, the total (whole-body) maximum
metabolic activities, the KMs, and the "most probable distribution" of metabolic activity between
the alveolar and bronchial areas of the lung were derived by optimizing the model against the
closed-chamber data of Kreiling et al. (1986b) for the mouse and Bolt et al. (1984) for the rat.
The only options considered for the distribution of the metabolic activity of the lung were that all
the metabolism took place in either one of the two areas, that it was equal in each area, or that it
was distributed relative to the volumes of each area; the best fit was found using the latter
distribution. The values of the physiological and metabolic parameters used in the Evelo et al.
model are summarized in Table 9-4.
The only independent validation of the model was against the whole-body extraction
ratios reported by Dahl et al. (1990). Evelo et al. (1993) calculated extraction ratios of 8.4% for
the mouse and 5.2% for the rat, whereas Dahl et al. found ratios of 12.8% for the mouse and
4.3% for the rat. Evelo et al. also noted that the whole-body Vmax value obtained for the rat by
9-12
-------
Table 9-4. Parameter values used in the Evelo et al. (1993) PBPK model.
Parameter
Physiological parameters
Body mass (kg)
Cardiac output (mL/min)
Alveolar ventilation (mL/min)
Blood flows (mL/min):
Liver
Fat
Muscle
Vessel-rich tissue
Bronchial lung area
Alveolar lung area
Volumes (mL):
Liver
Fat
Muscle
Vessel-rich tissue
Bronchial lung area
Alveolar lung area
Partition coefficients'1
Blood:air
Fat:blood
Liverblood
Muscle :blood
Kidney :blood
Lung:blood
Brain:blood
Vessel rich:bloodb
Metabolic parameters
Vmax,total (jimol-h-'-kg-1)
VmaxJlver (nmol-h-'-kg-1)
Vma3,bronclM (Hmol-h- -kg )
Vmax,alveolai (nmol-h- -kg )
KM (HM)
Mice
0.0275
24.83
24.5
6.14
2.34
3.81
10.75
1.79
23.04
1.65
2.94
19.09
1.17
0.2
0.18
Rats
0.215
75.93
118.7
19.17
6.52
11.13
33.60
5.514
70.42
8.63
14.0
162.7
9.49
1.29
1.63
0.894
32.362
2.675
1.871
1.690
1.272
2.355
2.02
465
318
77
70
8
200
171
13
16
5
aSame for all species.
bMean value of kidney :blood and brain:blood.
9-13
-------
fitting the model to the data of Bolt et al. (1984) does not fall within the range of values allowed
by experimental error based on the gas-uptake studies of Laib et al. (1992).
Evelo et al. (1993) stated that sensitivity analyses found the model optimization to be
relatively insensitive to variability in the value of KM. No other sensitivity analysis results are
reported.
The model simulations of Evelo et al. (1993) suggest that the relative importance of 1,3-
butadiene metabolism in the mouse lung is greater than the distribution of metabolic activity
would imply, especially at exposure concentrations of less than 200 ppm and for KM values of
less than the "best fit" value. Evelo et al. concluded that there is a strong first-pass effect in the
mouse lung. At higher concentrations, alveolar metabolism is saturated, and liver metabolism
becomes relatively more important. The relative importance of lung metabolism also increases
with decreasing exposure concentration for the rat and human, especially with lower values of
KM; however, unlike for the mouse, the lung metabolism never exceeds the liver metabolism.
Evelo et al. suggested that the higher rate of metabolic activation in the mouse lung could be
responsible for the mouse's greater sensitivity to developing lung carcinomas and heart
hemangiosarcomas from exposure to 1,3-butadiene.
The Evelo et al. (1993) model suffers from a number of serious weaknesses. Several
important parameters are not empirically derived. The partition coefficients are estimated
theoretically, and the whole-body Vmax and KM are optimized. For the rat, this exercise generated
a Vmax value that was inconsistent with other in vivo data. Furthermore, sensitivity analyses
revealed that the optimization was insensitive to variability in the value of KM, so there is
considerable uncertainty in the actual value of this parameter. The results pertaining to the
relative importance of lung metabolism, however, are highly sensitive to the value of KM. The
separation of the lung into alveolar and bronchial areas and the "optimized" distribution of lung
metabolism between the two areas also appear tenuous. Other limitations of the model are that
metabolism is limited to the lung and liver and that further metabolism of EB is not incorporated.
In addition, the model was not adequately validated, and only limited sensitivity analyses are
described. Finally, results for humans are discussed; however, the parameters used for the
human model are not fully reported.
9.2.6. Medinsky et al. (1994)
The model of Medinsky et al. (1994) describes 1,3-butadiene and EB uptake and
metabolism in mice and rats. The Medinsky et al. model is a venous equilibration, flow-limited
model with six physiological compartments—liver, lung, fat, slowly perfused tissue group,
rapidly perfused tissue group, and blood—and a compartment representing the air in closed-
chamber experiments. The model describes the oxidative metabolism of 1,3-butadiene in the
9-14
-------
liver and lung, as well as hydrolysis and glutathione conjugation of EB in the liver. In the
mouse, hepatic oxidation of EB is also included. In addition to measuring actual partition
coefficients, Medinsky et al. conducted closed-chamber experiments of 1,3-butadiene uptake
with both mice and rats to test the predictions of their model.
Medinsky et al. (1994) measured partition coefficients for 1,3-butadiene and EB
experimentally in vitro for both mouse and rat tissues. They found no significant differences
between the two species, except for the muscle:air partition coefficient for 1,3-butadiene and the
fat:air coefficient for EB (although the ultimate fatblood coefficient was not significantly
different). Organ and body weights were taken from specific experiments on 1,3-butadiene. The
remaining physiological parameters were based on average literature values, with the exception
of alveolar ventilation rate. Alveolar ventilation rates, conventionally defined as 70% of
measured total ventilation rates, yielded overestimates of 1,3-butadiene uptake at low
concentrations, consistent with observations by Johanson and Filser (1992) for other volatile
organic chemicals. Therefore, "apparent" alveolar ventilation rates were obtained by
optimization to provide rates that yielded the best fit of the model to the EB uptake data. The
optimized rates represented 63% of alveolar ventilation for both rats and mice.
Oxidation of 1,3-butadiene and EB (the latter in mouse liver only) and hydrolysis of EB
were described using Michaelis-Menten kinetics. Glutathione conjugation of EB was assumed to
be first order, based on the large KM value reported by Csanady et al. (1992).
Rate constants for the metabolism of 1,3-butadiene and EB were taken from the in vitro
data of Csanady et al. (1992). Apparent enzyme affinities (KM) measured in vitro were used
directly, whereas maximum metabolic rates (Vmax) were scaled to the whole organs. However,
when the organ microsomal concentrations reported by Csanady et al. are used to scale the
metabolic rates similarly reported by Csanady et al., "[1,3-butadiene] uptake from the closed
chamber is underestimated." Therefore, Medinsky et al. (1994) used literature values that were
two to six times greater for microsomal concentrations in the liver and lung in order to
successfully simulate the chamber study results. The parameter values used in the Medinsky et
al. model are summarized in Table 9-5.
For validation of the model components pertaining to EB uptake and metabolism, model
predictions were compared with the EB uptake data from the closed-chamber experiments of
Filser and Bolt (1984) for rats and Kreiling et al. (1987) for mice, although these were the same
data used to optimize the alveolar ventilation rates. The model predictions were deemed
"adequate," although EB uptake was overestimated at the highest exposure concentration,
especially for the rats (3,000 ppm). Medinsky et al. (1994) then compared model simulations of
1,3-butadiene uptake to their own closed-chamber data for mice and rats exposed to 1,3-
butadiene and to data from the closed-chamber experiments of Bolt et al. (1984) for rats and
9-15
-------
Table 9-5. Parameter values used in the Medinsky et al. (1994) PBPK model.
Parameter
Physiological parameters
Alveolar ventilation (L/hr/kg)a
Cardiac output (L/hr/kg)b
Body weight (kg)0
Blood flows (fraction of cardiac output)
Liver
Fat
Lung
Slowly perfused tissues
Rapidly perfused tissues
Organ volumes (fraction of body weight)
Liver
Fat
Lung
Slowly perfused tissues
Rapidly perfused tissues
Partition coefficients for 1,3-butadiene
Blood:air
Liverblood
Lung(& rapid) :blood
Muscle(& slow):blood
Fat:blood
Partition coefficients for EB
Blood:air
Liverblood
Lung(& rapid) :blood
Muscle(& slow):blood
Fat:blood
Tissue concentrations
Liver microsomal concentration (mg/g liver)
Lung microsomal concentration (mg/g lung)
Liver cytosolic concentration (mg/g liver)d
Rat
17
17
0.215-0.475
0.25
0.09
1.0
0.15
0.51
0.05
0.09
0.0053
0.71
0.0347
1.49
0.799
0.617
0.987
14.9
50.4
1.43
1.09
0.393
2.74
35
20
108
Mouse
41
41
0.028-0.035
0.25
0.09
1.0
0.15
0.51
0.0624
0.10
0.005
0.70
0.0226
1.34
1.01
1.10
2.99
14.3
36.6
1.15
1.54
0.645
2.49
35
20
82.8
9-16
-------
Table 9-5. Parameter values used in the Medinsky et al. (1994) PBPK model
(continued).
Parameter
Rate constants for oxidative metabolism of
l,3-butadiened
Liver Vmax (jamol/kg/h)
KM (nmol/L)
Lung Vmax (nmol/kg/h)
KM (nmol/L)
Rate constants for EB metabolism in the liver*1
Oxidation Vmax (^mol/kg/h)
KM (nmol/L)
Hydrolysis Vmax (^mol/kg/h)
KM (nmol/L)
Glutathione conjugation K (L/kg/h)
Rat
62
3.75
1.01
7.75
260
260
5.66
Mouse
338
2.00
21.6
5.01
26
15.6
754
1590
4.36
^Obtained by optimization.
bVentilation/perfusion = 1.
TDepending on experiment simulated.
dFrom Csanady et al. (1992), with Vmax values scaled to whole organ using above microsomal concentrations.
EB: l,2-epoxy-3-butene.
Kreiling et al. (1986b) for mice, and concluded that the model adequately predicted the in vivo
uptake results. Medinsky et al. also compared model predictions with the 1,3-butadiene retention
data of Bond et al. (1986) and found the results similar for exposure concentrations up to about
100 ppm. At higher concentrations, the model overestimated 1,3-butadiene retention observed in
mice. Furthermore, the blood concentrations of EB following 1,3-butadiene exposure, as
reported by Bond et al., were overestimated by the model for both mice (except at the lowest
exposure) and rats by about two- to fourfold, although Medinsky et al. suggested that the
discrepancy might be attributable to EB loss from the blood during sampling.
No comprehensive sensitivity analysis for the model parameters was reported. Medinsky
et al. (1994) did note that use of the microsomal concentrations reported by Csanady et al. (1992)
resulted in underestimation of the 1,3-butadiene uptake from chamber studies. In addition, they
investigated whether the model was sensitive to the different values obtained for the muscle:air
partition coefficients for the mouse and rat and determined that the species-specific coefficients
provided the best fits to their 1,3-butadiene uptake results for the two species. Medinsky et al.
also determined that the inclusion of lung metabolism improves the model fit for the mouse,
especially at lower exposure concentrations, but has little effect for the rat.
9-17
-------
Based on their model simulations, Medinsky et al. (1994) suggested that lung metabolism
may play an important role in 1,3-butadiene uptake and carcinogenesis. Their model predicts
locally generated concentrations of EB that are 15 times greater in the mouse lung than in the rat
lung, for a 6-h exposure to 10 ppm. Medinsky et al. recommended that more research be done to
characterize 1,3-butadiene metabolism and target cells in the mouse lung and to understand the
pharmacokinetics of DEB in different species. They further claimed that "quantitation of the
concentrations of [1,3-butadiene], [EB], and [DEB] in target and non-target tissues of rats and
mice after exposure to [1,3-butadiene] is essential for validation of existing models before these
models can be applied to predict behavior in humans."
One of the major strengths of the Medinsky et al. (1994) model is that these investigators
experimentally measured partition coefficients and confirmed the results of Johanson and Filser
(1993), suggesting that the empirical values for the partition coefficients for 1,3-butadiene differ
significantly from the theoretical values used in previous models. Medinsky et al. also
conducted closed-chamber experiments to obtain validation data for their model and investigated
the role of lung metabolism in 1,3-butadiene uptake. Some limitations of the model include the
fact that metabolism was restricted to the liver and lung, although other tissues are known to
metabolize 1,3-butadiene as well (Chapter 3). In addition, the alveolar ventilation rates were
determined by fitting experimental closed-chamber data, and there are uncertainties about the
actual values for organ microsomal contents. Finally, only 1,3-butadiene oxidation was
described in the lung, although rate constants for further metabolism of EB are also available
from Csanady et al. (1992). Further developments of this model are discussed in Sections 9.2.7.4
and 9.2.9.
9.2.7. Blaine Presentations
The results of four PBPK modeling exercises for 1,3-butadiene were presented at the
International Symposium on the Evaluation of Butadiene and Isoprene Health Risks held in
Blaine, Washington, in June 1995. The proceedings of the symposium were published in a
special issue of the scientific journal Toxicology (vol. 113) in 1996. None of the models was
described in detail. The published results of these modeling efforts are summarized briefly
below.
9.2.7.1. Kohn andMelnick (1996)
Kohn and Melnick modified their 1993 model for mice and rats by (a) distributing blood
among arterial and venous spaces and tissue capillary beds, (b) separating the GI tract from the
viscera compartment in order to more realistically model liver perfusion, (c) adding a
compartment for the alveolar space in the lung, and (d) including GSH depletion and resynthesis.
9-18
-------
Parameter values that were not available in the literature were obtained by optimization. These
investigators found that the inclusion of the alveolar space eliminated the need to decrease
alveolar ventilation rates by 40% as is done in some of the other models. They further noted that
subcompartmentalization of the blood necessitated considerable extrahepatic metabolism in order
to reproduce observed uptakes of 1,3-butadiene and EB. While generally reproducing uptake
rates, the model overestimated blood EB concentrations. In addition, these investigators
observed that the model was sensitive to the range of values that have been reported for some of
the parameters, in particular the blood:air partition coefficient and certain enzymatic activities.
They conclude that "[a] valid and useful toxicokinetic model must have reliable physiological
and enzymological data for [1,3-butadiene] biotransformation before it can be credibly used for
human risk assessment." Kohn's further refinements of this model are discussed in Section
9.2.10.
9.2.7.2. Johamon and Filser (1996)
This presentation basically summarized Johanson and Filser's 1993 model, which was
discussed above. Some results of extending the model to humans were also presented, although
the parameter values used for humans were not reported and it is not clear how some of them
were obtained. The extended model predicts relative internal doses of EB (based on steady-state
concentrations or AUCs for mixed venous blood) for mice, rats, and humans in the ratio
1.6:1.0:0.3.
9.2.7.3. Csanddy et al (1996)
Csanady and Filser and colleagues presented a modified version of Johanson and Filser's
model. The modifications are (a) Vmax for 1,3-butadiene was based on in vivo rather than in vitro
data, (b) disposition of DEB for the mouse only was included using the Csanady et al. (1992) in
vitro data for EB oxidation and DEB hydrolysis and GSH conjugation, and (c) production of
hemoglobin-EB adducts was incorporated. The modified model predicts relative internal doses
of EB (AUC for blood) for mice, rats, and humans in the ratio 1.3:1.0:0.3. Predicted Ftb-EB
adduct levels for mice and rats were in good agreement with experimental measurements. Filser
and colleagues maintain that inclusion of an intrahepatic first-pass effect is necessary to resolve
the problem occurring in the Medinsky et al. and Kohn and Melnick models whereby the models
can simulate 1,3-butadiene uptake data, but then blood levels of EB are overpredicted. The
investigators are currently refining this model further.
9-19
-------
9.2.7.4. Sweeney et al (1996)
Sweeney et al. presented a modified version of the 1994 Medinsky et al. model. The
reported modifications were (1) addition of a GSH submodel that included diurnal variations, (2)
addition of a DEB submodel, and (3) incorporation of nonenzymatic elimination of EB.
However, it was not clear how parameter estimates for modifications 2 and 3 were obtained and
whether or not the DEB submodel included further metabolism of DEB or merely its production
and distribution. Sweeney et al. also investigated the possibility that EB transfer from blood to
liver enzymes is diffusion-limited rather than flow-limited, and examined the effects of using
optimized parameters for the phase I oxidation of 1,3-butadiene to EB and for phase II reactions
involving EB. (The phase I parameters were rejected because they yielded results inconsistent
with observations.) These investigators found that while their base modified model
overestimates blood EB concentrations by about an order of magnitude, the model variations
with diffusion limitation and optimized parameters both adequately simulate observed
concentrations. In addition, the base and diffusion-limited models were able to reasonably
simulate measured liver EB concentrations taking into account the 3-5 minute lag time between
cessation of exposure and tissue removal, while the model variation with optimized parameters
apparently overestimated phase II metabolism. Thus, the authors concluded that the parameter
values derived from in vitro rates were more appropriate than the optimized values. They further
concluded that, overall, the diffusion-limited model yielded results most consistent with
experimental data. A more recent refinement of this model was published in 1997 and is
discussed in Section 9.2.9.
9.2.8. Reitz et al. (1996)
In 1996 Reitz et al. prepared an unpublished PBPK model for the Chemical
Manufacturers Association, which presented the model to EPA for consideration. The Reitz et
al. model was based on the Medinsky et al. (1994) model, with modifications as well as inclusion
of DEB elimination and extension of the model to humans. A major difference between the two
models is that because the Medinsky et al. model overestimates EB blood levels compared to
measurements by Himmelstein et al. (1994), and overestimates EB exhalation compared to data
obtained by Medinsky and colleagues from their chamber experiments, Reitz et al. chose not to
use the in vitro constants for EB metabolism measured by Csanady et al. (1992) but rather to
estimate the EB metabolic parameters from in vivo data. Reitz et al. state that they have "found
it necessary to 'calibrate' in vitro measurements with selected in vivo studies in order to obtain
accurate simulations of in vivo data" for PBPK models of other chemicals as well. A further
difference between the two models is that the Reitz et al. model does not include lung
metabolism but rather assumes that all metabolism occurs only in the liver. These investigators
9-20
-------
provide a number of unconvincing reasons for removing lung metabolism from the model. More
to the point, the alternative approach they have taken to estimating the EB metabolism
parameters using in vivo data really provides estimates only of whole-body metabolism, and it
would be difficult to credibly partition the whole-body metabolism to specific tissue
compartments.
Reitz et al. derived new parameter estimates for EB metabolism in the mouse as follows.
First, they assumed that oxidation of EB to DEB is insignificant compared to EB hydrolysis or
conjugation with GSH, so the oxidation pathway was ignored during the derivation of rate
constants for the other pathways. Second, they modeled GSH conjugation using Michaelis-
Menten kinetics. Vmax was set at 350 |J,mol/h/kg based on the in vivo clearance results of Kreiling
et al. (1987). A Km value of 4.2 |J,mol/L was derived by optimization, fitting the data of
Himmelstein et al. (1994). Note that these values differ substantially from those used by
Medinsky et al. (1994) for the liver (Vmax=154,000 |J,mol/h/kg and Km=35,300 |J,mol/L, yielding
a first-order clearance rate [i.e., Vmax/Km] of 4.36). Reitz et al. claim that the first-order rate
constant used by Medinsky et al. was left unchanged during the optimization for Km and that
once Vmax and Km were obtained, the first-order rate constant was adjusted. This is unclear,
however, because the first-order clearance rate used by Medinsky et al. was for GSH
conjugation, while the first-order clearance rate used in the Reitz et al. model is for hydrolysis.
Reitz et al. assume first-order hydrolysis for all species based on the results of Kreiling et al. for
rats, a species in which hydrolysis plays a key role in EB elimination, suggesting that EB
metabolism was not saturated at exposure levels up to 5000 ppm. Reitz et al. then adjusted the
"first-order rate constant" so that 20% of EB metabolism from a 6-h exposure to 8000 ppm
occurs via the hydrolysis pathway to fit the urinary metabolite data of Sabourin et al. (1992).
The first-order clearance rate was eventually set to 6.3 L/kg/h, again significantly different from
the value of 0.47 derived from the Vmax and Km values used by Medinsky et al.
For EB oxidation, Reitz et al. rely on Vmax and Km values based on in vitro data provided
by the Chemical Industry Institute of Toxicology (CUT; i.e., Medinsky and colleagues) for rats,
mice, and humans—presumably from liver preparations, although this is not stated. It is unclear
why the in vitro values presented for the mouse differ from those reported by Csanady et al.
(1992) (Vmax of 1.3 nmol/mg protein/min vs. 0.20 and Km of 140 |J,mol/L vs.15.6, respectively).
The in vitro Vmax values were then normalized for protein content, apparently using the same
protein concentration values used for the Medinsky et al. model. Because of the differences in
the in vitro measurements, the parameter values for EB oxidation used in the Reitz et al. mouse
model differ from those used by Medinsky et al. (Vmax of 170 |J,mol/h/kg vs. 26 and Km of 140
|j,mol/L vs. 15.6, respectively). These CUT data for EB oxidation appear to be the same data
published by Seaton et al. in 1995 and used in the Sweeney et al. (1997) model; although
9-21
-------
Sweeney et al. corrected the Vmaxs to account for hydrolyzed DEB and obtained values of 176.6
and 57.1 |J,mol/h/kg for the mouse and rat, respectively, vs. 170 and 42 used by Reitz et al.
Then, Reitz et al. used a parallelogram approach for interspecies extrapolation to derive
new parameter values for EB hydrolysis and conjugation with GSH in the rat. Under this
approach, in vivo metabolic activities for the rat are estimated from in vitro activities using the
same relationship between in vitro and in vivo activities observed for the mouse. (Recall that
Reitz et al. assumed that all metabolism occurring in vivo is from liver activity.) Reitz et al.
assumed that the apparent first-order rate constants (i.e., Vmax/Km) derived from the Csanady et
al. (1992) in vitro data reflected enzyme content in each species. They then estimated in vivo
rate constants for the rat by multiplying the in vivo rate constants for the mouse by the ratio of
the in vitro first-order rate constants for the liver and by the ratio of liver weights (as a fraction of
body weight). Thus, an in vivo first-order clearance rate for EB hydrolysis of 13.2 L/kg/h was
derived for the rat by multiplying 6.3 by ([2.48/0.26]/[5.79/1.59]) (from Csanady et al., 1992,
Table II) and by (0.05/0.0624). Similarly, an in vivo Vmax of 345 |J,mol/h/kg was estimated for
GSH conjugation in the rat by multiplying 350 by ([241/13.8]/[500/35.3]) (from Csanady et al.,
1992, Table III) and by (0.05/0.0624). The Km value for GSH conjugation was assumed to be the
same across species, even though the Km values used by Reitz et al. for EB oxidation range from
140 in mice to 500 in humans. Parameter estimates for EB oxidation (Vmax and Km) for the rat
were derived from in vitro rat data provided by CUT, as discussed above for the mouse.
The basis for the human 1,3-butadiene oxidation parameter values is not revealed in Reitz
et al.'s report; however, the values were apparently obtained in the same manner as those used by
Medinsky et al. for the rat and mouse. For the rat and mouse, Medinsky et al. used the in vitro
data of Csanady et al. (1992) but used different microsomal protein concentration values. For
humans, Reitz et al. apparently also used the in vitro data of Csanady et al. as well as the liver
protein concentrations used by Medinsky et al. for the rat and mouse.
Human metabolic parameter values for EB conjugation with GSH and hydrolysis were
derived according to the parallelogram approach described above. For GSH conjugation, an in
vivo Vmax of 44.6 |J,mol/h/kg was estimated for humans by multiplying the mouse value of 350
by the ratio of the in vitro first-order rate constants for the liver from Csanady et al. (1992)
([45. l/10.4]/[500/35.3]) and by the ratio of liver weights (0.026/0.0624). For hydrolysis, Reitz et
al. used the midpoint of the range of normalized first-order rate constants for the 12 human liver
samples reported in Csanady et al. (1992) and compared that to a corresponding rate constant for
mice, yielding a ratio of 12.1:1. Had they used the midpoint of the actual first-order rate
constants reported for three human livers, which included those with the highest and lowest
enzyme activities, the ratio would have been about 20% lower, 9.6:1. Reitz et al. then
mistakenly multiplied the ratio of 12.1 for enzyme activity by the in vivo first-order rate constant
9-22
-------
for hydrolysis of 13.2 L/kg/h derived for the rat rather than the value of 6.3 estimated for the
mouse and by the ratio of liver weights. Had they used the correct in vivo estimate, they would
have obtained a parameter value of 31.8 L/kg/h for the first-order rate constant for hydrolysis in
humans rather than the value of 66.6 that they use in their model. Parameter estimates for EB
oxidation (Vmax and Km) for humans were derived from in vitro human data provided by CUT, as
discussed above for the mouse.
To extend the model to include DEB elimination, Reitz et al. estimated DEB partition
coefficients and metabolic parameters as follows. Partition coefficients were estimated from data
provided to Reitz et al. by Thornton-Manning of the Inhalation Toxicology Research Institute
(ITRI) on blood and tissue levels of DEB in female rats and mice exposed to 62.5 ppm 1,3-
butadiene for 4-6 h. (These data were apparently later published by Thornton-Manning et al.,
[1997].) Reitz et al. report postexposure blood levels of 13.6 pmol DEB/g in rats and 345 pmol/g
in mice. They also claim that within a given species, similar levels of DEB were detected in
bone marrow, mammary tissue, ovary tissue, and fat (6.3-10.5 pmol/g in rats and 169-264 pmol/g
in mice). On the basis of these data, Reitz et al. calculated an average tissue:blood concentration
ratio of 0.6, which they used as the DEB tissue:blood partition coefficient for all tissues of mice,
rats, and humans. However, Sweeney et al. (1997) later published experimentally derived
tissue:blood partition coefficients for DEB in the mouse that ranged from 1.41 for the kidney
(i.e., richly perfused tissue) to 2.20 for fat. Based on reports by Bond of CUT that the vapor
pressure of DEB was too low for calculation of the blood:air partition coefficient using vial
equilibrium techniques, Reitz et al. arbitrarily set the blood:air partition coefficient to 1,000, a
high value, for all species in their model.
Reitz et al. estimated the kinetic constants for hydrolysis and GSH conjugation of DEB
using in vitro Vmax and Km values provided by CUT for rats, mice, and humans. On the basis of
the relatively high Km values, they tentatively assume that both DEB hydrolysis and conjugation
with GSH are first-order processes, "pending development of better in vivo data," noting that in
vitro Kms obtained for EB did not accurately predict in vivo behavior. Thus, DEB elimination
was described using a single first-order clearance term reflecting the sum of the first-order rate
constants for hydrolysis and conjugation, which were purportedly derived from the in vitro data.
However, the first-order terms that Reitz et al. report could not be replicated. The CUT data for
EB hydrolysis and conjugation with GSH appear to be the same data published in 1996 by
Boogaard and Bond and by Boogaard et al., respectively, except that Reitz et al. apparently
misreported the rat Km for GSH conjugation as 240 |J,M rather than 24,000 |J,M. Using the
corrected Km value, it appears that Reitz et al. mistakenly used the microsomal protein
concentrations for normalizing the GSH conjugation results. Using the cytosolic protein
concentrations for GSH conjugation yields corrected first-order clearance terms (summing
9-23
-------
hydrolysis and conjugation) of 4.5 and 8.4 L/kg/h for the rat and mouse, respectively (vs. the
2.80 and 3.84 for rat and mouse, respectively, reported by Reitz et al.). It is unknown whether or
not the value presented by Reitz et al. for humans is also in error.
Parameter values used in the Reitz et al. model are summarized in Table 9-6. Except for
the EB and DEB metabolic parameters and DEB partition coefficients, which were estimated as
described above, mouse and rat parameter values were identical to those used by Medinsky et al.,
with the exception of mouse lung and rapidly perfused tissues organ volumes, which differed
slightly. For humans, blood flows and partition coefficients were assumed to be the same as for
the mouse, with the exception of the muscle:blood partition coefficient for 1,3-butadiene, where
the rat value was used. No reference was cited for human values for the physiological
parameters and organ volumes.
For validation of the new metabolic constants for EB hydrolysis and conjugation with
GSH (with EB oxidation set to 0) for the mouse, the modified model was used to predict the
blood EB concentrations measured by Himmelstein et al. (1994), although these data had been
used for parameter optimization. Predicted levels ranged from 20% too high to 80% too low.
These predictions are better than those from the original Medinsky et al. model, which had
overestimated blood EB concentrations by more than fourfold. Reitz et al. also evaluated the
ability of the modified model to predict data from CUT on EB exhalation during 1,3-butadiene
exposure. The modified model performed better than the original Medinsky et al. model, which
had overpredicted EB exhalation at all exposure levels; however, the modified model
underestimated exhalation at higher exposures and overestimated exhalation at lower exposures.
The modified model was also used to simulate the EB gas uptake data of Kreiling et al. (1987); it
gave a reasonable fit to these data, as had the original Medinsky et al. model.
Similarly, the new metabolic constants for EB hydrolysis and conjugation with GSH
(with EB oxidation set to 0) for the rat were used to predict the blood EB concentrations
measured by Himmelstein et al. (1994), and the modified model gave results within a range of
0.5-1.4 times the observed levels, which were an improvement over the overpredictions of the
Medinsky et al. model. The modified rat model also provided a reasonable fit to the EB gas
uptake data of Filser and Bolt (1984), as had the original Medinsky et al. model.
For validation of the extended model including DEB metabolism, Reitz et al. simulated
the blood concentration data of Himmelstein et al. (1994) for mice exposed to 71, 602, or 1282
ppm 1,3-butadiene for 6 h and blood concentration data from ITRI for mice exposed to 62.5 ppm
1,3-butadiene for 4 h. Predicted values of blood DEB ranged from 6.5 times lower at 71 ppm to
1.1 times higher at 1282 ppm compared to the Himmelstein et al. observations; the predicted
value was 2.2 times lower than the ITRI value at 62.5 ppm. Together, these comparisons suggest
that the model may be underestimating blood DEB for mice at lower exposure levels.
9-24
-------
Table 9-6. Parameter values used in the Reitz et al. (1996) PBPK model.
Parameter
Physiological parameters*
Alveolar ventilation (L/h/kg)
Cardiac output (L/h/kg)
Body weight (kg)
Blood flows (fraction of cardiac output)8
Liver
Fat
Lung
Slowly perfused tissues
Rapidly perfused tissues
Organ volumes (fraction of body weight)3
Liver
Fat
Lung
Slowly perfused tissues
Rapidly perfused tissues
Partition coefficients for l,3-butadieneb
Blood:air
Liverblood
Lung(& rapid) :blood
Muscle(slow) :blood
Fat:blood
Partition coefficients for EBb
Blood:air
Liverblood
Lung(& rapid) :blood
Muscle(slow) :blood
Fat:blood
Partition coefficients for DEB°
Blood:air
Liverblood
Lung(& rapid) :blood
Muscle(slow) :blood
Fat:blood
Mouse
41
41
0.028
0.25
0.09
1.0
0.15
0.51
0.0624
0.1
0.0078
0.71
0.0198
1.34
1.01
1.10
2.99
14.3
36.6
1.15
1.54
0.645
2.49
1000
0.6
0.6
0.6
0.6
Rat
17
17
0.325
0.25
0.09
1.0
0.15
0.51
0.050
0.09
0.0053
0.72
0.0347
1.49
0.80
0.62
0.99
14.9
50.4
1.43
1.09
0.393
2.74
1000
0.6
0.6
0.6
0.6
Human
4.96
4.96
70.0
0.25
0.09
1.0
0.15
0.51
0.026
0.21
0.0077
0.60
0.0563
1.34
1.01
1.10
0.99
14.3
36.6
1.15
1.54
0.645
2.49
1000
0.6
0.6
0.6
0.6
9-25
-------
Table 9-6. Parameter values used in the Reitz et al. (1996) PBPK model (continued).
Parameter
Rate constants for oxidative metabolism of
l,3-butadiened
Vmax (nmol/kg/h)
KM (nmol/L)
Rate constants for EB metabolism0
Oxidation Vmax (^mol/kg/h)
KM (nmol/L)
GSH Conjugation Vmax (nmol/kg/h)
KM (nmol/L)
First order Hydrolysis K (L/kg/h)
Rate constants for DEB metabolism0 f
First order elimination K (L/kg/h)
Mouse
338
2.0
170.0
140
350.0
4.2
6.3
3.84
Rat
62.0
3.75
42.0
150
345
4.2
13.2
2.80
Human
64.4
5.14
51.0
500
44.6
4.2
66.6e
2.17
"Mouse and rat values same as Medinsky et al. (1994) (see Table 9-5), except for mouse lung and rapidly perfused tissue organ volumes which
differ slightly. Source of human values not reported.
bMouse and rat values from Medinsky et al. (1994). Human values set equal to mouse values, except for muscle:blood partition coefficient for 1,3-
butadiene where rat value was used.
"Derived as described in text.
dMouse and rat values same as Medinsky et al. (1994) liver values. Human values derived as described in text.
"This value appears to be in error (see text). The corrected value is thought to be 31.8.
fThe mouse and rat values appear to be in error (see text). The corrected values are thought to be 8.4 and 4.5, respectively. It is unclear whether or
not the human value is also in error.
And those simulations were made using an apparently incorrect value for the first-order clearance
term for DEB (see above). The corrected clearance term is over twice as large, suggesting that
model predictions using the corrected term may underestimate DEB blood levels to an even
greater extent. For validating the extended model in rats, the only data used were ITRI blood
concentrations from a 4-h exposure to 1,3-butadiene, sampled at 2, 4, and 4.5 h. Predicted values
of blood DEB ranged from 1.9 times higher at 2 h to 6.2 times higher at 4.5 h compared to the
observed values.
There are no human data appropriate for validating the human model.
No comprehensive sensitivity analysis for the model parameters was reported. The only
sensitivity analysis mentioned was for the blood:air partition coefficient for DEB, which was set
to an arbitrary high value of 1000. Reitz et al. report that simulations with values ranging from
100 to 10,000 gave similar results.
A strength of the Reitz et al. model is that it was able to simulate observed levels of blood
EB and DEB following 1,3-butadiene exposure fairly well, although the mouse blood EB results
were used for parameter optimization, so a good fit to those data is to be expected, and the DEB
9-26
-------
validation would need to be repeated using corrected clearance terms. A concern, however, is an
apparent trend across the mouse blood DEB predictions, suggesting that the model results may
deviate (underestimate actual levels) more at lower exposures, which are of greater
environmental relevance. A trend also occurs with the rat blood EB predictions, although in this
case the trend suggests the model overestimates blood EB levels at the lower exposures. The rat
blood DEB data used for model validation were for one exposure level only; thus nothing about
trends can be inferred.
Some limitations of the Reitz et al. model include the assumption that all metabolism
occurs only in the liver, and the uncertainties in the metabolic constants derived from the
unvalidated partitioning of EB clearance amongst the various metabolic pathways. In addition,
values for some of the model parameters were obtained by optimization, and it is thus unclear
what these parameters actually represent. Furthermore, the model raises serious issues about the
use of in vitro versus in vivo data. In some cases the model uses metabolic constants derived
from in vitro data, while in other cases, the in vitro data are deemed inadequate because the
resulting model cannot replicate other experimental results. In the latter cases, metabolic
constants were derived by fitting in vivo data, yielding parameter estimates that differ drastically
from those based on the in vitro data and that suggest different structural forms for some
metabolic pathways (e.g., first-order vs. saturable). For example, the Km values for GSH
conjugation of EB differed by almost four orders of magnitude. In sum, the Reitz et al. model is
a dubious patchwork of metabolic rate constants, some fitted and some independently measured,
some derived from in vitro data and some from in vivo data, some based on whole body
clearance data and some based only on liver data.
An additional major limitation is the absence of validation for the human model. The
parallelogram approach used for extrapolation to the rat can conceptually be used for
extrapolation to humans, but given the uncertainties in the model, the unresolved discrepancies
between the Reitz et al. model and other PBPK models for 1,3-butadiene, and the limited
validation of both the mouse and rat versions, the human model has little credibility.
Furthermore, while a model capable of predicting blood levels of EB and DEB may be a
dosimetric improvement over default dosimetry assumptions, these blood levels alone do not
explain species differences in target organ carcinogenic responses. Given the large number of
questionable assumptions made by Reitz et al. in ascribing all metabolism to the liver and in
partitioning whole-body EB clearance among various metabolic pathways, other tissue
compartment metabolite concentrations obtained from their model would be highly unreliable.
9-27
-------
9.2.9. Sweeney et al. (1997)
In 1997, Sweeney et al. (CUT) published their own further revisions of the 1994
Medinsky et al. model for mice and rats. The revised model includes the disposition of DEB and
some other metabolic reactions not depicted in the original model, such as nonenzymatic
reactions of EB and DEB as well as GSH depletion and synthesis. The final model also includes
an alternative pathway for 1,3-butadiene oxidation. The model contains the original six
physiological compartments: liver, lung, fat, slowly perfused tissue, rapidly perfused tissue, and
blood; however, the blood compartment was subdivided into arterial and venous compartments
for modeling intravenous (iv) administration of EB and DEB.
The revised model uses the same physiological parameters, including blood flows and
organ volumes, and the same partition coefficients for 1,3-butadiene and for EB as the original
Medinsky et al. model (see Table 9-5). Arterial and venous blood subcompartments were
assigned values of 2% and 4% of body weight, respectively. The same microsomal and cytosolic
protein concentrations were also reportedly used; however, there appears to be a discrepancy
with the mouse lung DEB hydrolysis Vmax (see below). The source and value of lung cytosolic
protein concentrations were not reported. A value of 61 mg protein/g lung, consistent with that
cited in Boogaard et al. (1996), was apparently used for the rat; however, a value of about 50 mg
protein/g lung seems to have been used for the mouse, rather than the 32 mg/g cited in Boogaard
et al. Tissue:blood partition coefficients for DEB were experimentally determined using hexane
as the partitioning medium because of the low vapor pressure of DEB. These coefficients and
the metabolic constants used in the Sweeney et al. (1997) model are reported in Table 9-7.
The Sweeney et al. (1997) model describes the reactions depicted in Figure 9-1.
Oxidative metabolism of 1,3-butadiene to EB, GSH conjugation of EB and DEB, hydrolysis of
DEB, and depletion and synthesis of GSH are described in both the liver and the lung. Oxidation
of EB to DEB and EB hydrolysis are described only in the liver, although Sweeney et al.'s
Figure 2 depicts EB oxidation in the lung as well. Nonenzymatic first-order reactions of EB and
DEB are modeled in all tissues except the slowly perfused compartment. In the final model, an
alternative pathway for 1,3-butadiene oxidation in the liver and lung and a first-order elimination
pathway in the slowly perfused compartment were also incorporated. All of these reactions are
included in both the mouse and rat models.
The revised model uses the same rate constants for oxidative metabolism of 1,3-butadiene
(to all products), based on the in vitro data of Csanady et al. (1992), as the original model. The
Csanady et al. Vmax and Km values for EB hydrolysis and conjugation with GSH were also used;
however, instead of using the latter values to derive a first-order rate constant as in the original
model, GSH conjugation was described assuming a ping-pong mechanism and a Km of 100
9-28
-------
Table 9-7. Parameter values used in Sweeney et al. (1997) PBPK modeP.
Parameter
Partition coefficients for DEBb
Liver(& lung):blood
Kidney (& rapid) :blood
Muscle(& slow):blood
Fat:blood
Rate constants for oxidative metabolism of
1,3-butadiene
Liver (total) Vmax (nmol/kg/h)c
^ (nmol/L)c
Liver (to EB) Vmax (nmol/kg/h)d
^ (nmol/L)d
Liver (to other volatiles) Vmax (|amol/kg/h)e
Kn (nmol/L)e
Lung (total) Vmax (nmol/kg/h)c
K,,, (nmol/L)c
Lung (to EB) Vmax ([imol/kg/hf
Kn (nmol/L)d
Lung (to other volatiles) Vmax (nmol/kg/h)e
K,,, (nmol/L)e
Fraction to EB (2 -product model/
Rate constants for metabolism of EB
Liver oxidation
One enzyme Vmax (|amol/kg/hr)g
K,,, (nmol/L)B
Two enzyme Vmaxl (|amol/kg/hr)h
K,,,, (nmol/L)
Vmax2 (nmol/kg/hr)1
K^ (nmol/L)g
Liver hydrolysis Vmax (nmol/kg/hr)c
K,,, (nmol/L)c
Liver GSH conjugation Vj^ (nmol/kg/hry
K,,, (umol/Ly
K,,,0*1 (nmol/L)k
Lung GSH conjugation Vmax (nmol/kg/hr)1
K,,, (umol/Ly
K,,,0*1 (nmol/L)k
Mouse
1.41
1.53
1.82
2.20
338
2.0
97
0.88
243
2.72
21.6
5.01
6.4
1.6
16.1
9.5
0.192
176.6
145
32.5
15.6C
144.1
145
754
1590
154,000
35,300
100
4088
36,500
100
Rat
1.41
1.53
1.82
2.20
62
3.75
8.2
1.54
54
4.36
1.01
7.75
0.13
3.18
0.88
9.14
0.241
57.1
141
10
141g
47.1
141
260
260
78,100
13,800
100
819
17,400
100
9-29
-------
Table 9-7. Parameter values used in Sweeney et al. (1997) PBPK model3
(continued).
Parameter
Rate constants for metabolism of DEB
Liver hydrolysis Vmax (nmol/kg/h)m
^ (nmol/L)m
Liver GSH conjugation Vmax (nmol/kg/h)n
^ (nmol/L)n
Lung hydrolysis Vmax (nmol/kg/h)m
^ (nmol/L)m
Lung GSH conjugation Vmax (nmol/kg/h)n
1^ (nmol/L)n
Nonenzymatic reaction rate constants'1
for EB (L/h)
Blood
Liver
Lung
Fat
Muscle(& slow)
for DEB (L/h)b
Blood
Liver
Kidney(& rapid)
Fat
Muscle(& slow)
First-order clearance rate for slowly perfused
compartment*1
For EB (L/kg/h)
For DEB (L/kg/h)
Mouse
4193
8100
50,342
6400
466.1
7500
577
1700
0.558
4.14
2.70
1.56
0
0.189
3.15
4.1
2.8
0
0.697
0.762
Rat
5555
2700
60,264
24,000
122.7
7100
332
4170
0.582
4.94
6.07
1.72
0
0.189
3.15
4.1
2.8
0
1.47
3.07
a For physiological parameters and 1,3-tmtadiene and EB partition coefficients see Table 9-5. For parameters used in GSH submodel see Sweeney
etal. (1997).
b Measured for mouse; assumed equivalent for rat.
0 Same as Medinsky et al. (1994), based on Csanady et al. (1992) in vitro data.
d Obtained by optimization, based on mass balance data of Bond et al. (1986) (see Sweeney et al. (1997) for details).
" Obtained by optimization, using data reported by Himmelstein et al. (1997).
'Obtained by optimization, using Himmelstein et al. (1994) blood EB data.
8 Based on Seaton et al. (1995) microsome data, with Vmax corrected to account for hydrolyzed DEB. Rat and mouse Kms are transposed compared
to those reported by Seaton et al.; however, the values are similar so the effects are likely to be inconsequential.
h Based on Seaton et al. (1995) cDNA-expressed CYP 2E1 data.
1 Difference between one-enzyme Vmax and Vmaxl.
' From Csanady et al. (1992), although it's unclear where the lung cytosolic protein concentrations used for conversion are from.
k From Johanson and Filser (1993).
111 Based on Boogaard and Bond (1996) microsome data, using Medinsky et al. (1994) microsomal protein concentration values, except for the
mouse lung where a value of about 31 was apparently used rather than 20.
° Based on Boogaard et al. (1996) cytosol data; although a value of 50 was apparently used for the mouse lung cytosolic protein concentration
rather than the value of 32 reported in Boogaard et al.
p Determined experimentally by Sweeney et al. (1997).
q Obtained by optimization for the rat using DEB blood clearance data of Valentine et al. (1997). Mouse values were estimated using the clearance
rate ratios for the rat.
9-30
-------
1,3-Butadiene
CYP ^-- i Detoxification products
Aldehyde products (?)
EH
CYP
1,2-Epoxy-3-butene -
CYP
Precursors
Degradation products
Detoxification products
,x GSufathione
GSutalhione Precursors
1,2:3,4-DIepoxybutane—^^
EH/ NE/ GST
Degradation products
Detoxification products Detoxification products
Figure 9-1. Reactions modeled in the PBPK model of Sweeney et al. (1997)a.
aReaction pathways are indicated by solid lines; inhibitory feedback is denoted by dashed lines. CYP = cytochrome P-
450. EH = epoxide hydrolase. GST = glutathione S-transferase. NE = nonenzymatic reactions.
Source: Sweeney etal., 1997.
|j,mol/L for GSH as proposed in the Johanson and Filser (1993) model. Sweeney et al. explain
that while a simpler description might be adequate in scenarios dealing with 1,3-butadiene
exposure, they wanted a more general description that would also apply to EB exposures
resulting in low levels of GSH.
To derive parameters for EB oxidation in the liver, Sweeney et al. used the kinetic data of
Seaton et al. (1995) from mouse, rat, and human liver microsomes instead of the Csanady et al.
data, for reasons that are not stated. First, using the kinetic data for DEB hydrolysis of Boogaard
and Bond (1996), the EB oxidation rates were corrected to take into account any DEB that may
have undergone hydrolysis before it could be measured as oxidation product. Then, to derive
parameters for a "two-enzyme" EB oxidation scenario, the Michaelis-Menten rate equation was
split into two terms to portray EB oxidation by both a high-affinity, low-capacity isozyme and a
low-affinity, high-capacity isozyme. CYP 2E1 was considered to be the high-affinity enzyme
(subscript "1" in Table 9-7) and was assigned a Vmax based on the EB oxidation rate measured by
Seaton et al. for cDNA-expressed CYP 2E1 and, for the mouse, the smaller Km for liver
microsomes observed by Csanady et al. The low-affinity enzyme (subscript "2") was assigned
the larger Km observed by Seaton et al. for liver microsomes and its Vmax was estimated by
subtracting Vmax for the high-affinity enzyme from the corrected Vmax based on the Seaton et al.
9-31
-------
microsome data. For the rat, the Seaton et al. Km was used for both enzymes. The oxidation
reactions for 1,3-butadiene and EB were modeled as noncompetitive.
Parameter estimates for DEB hydrolysis were derived from the in vitro work of Boogaard and
Bond (1996) on human, rat, and mouse liver and lung microsomes. However, the Vmax values
were converted using the protein concentration values used by Medinsky et al. (1994)
rather than the values reported by Boogaard and Bond, which were identical to those used by
Csanady et al. (1992). Furthermore, the mouse lung hydrolysis Vmax value reported by Sweeney
et al. appears to be either incorrect or the result of using a different value for lung microsomal
protein concentrations than that used in the original model. The reported Vmax is 50% higher than
that which one would obtain using the original protein concentrations. Parameter estimates for
conjugation of DEB with GSH were taken from the in vitro work of Boogaard et al. (1996) on rat
and mouse liver and lung cytosolic fractions. The cytosolic protein concentrations used are
consistent with those cited by Csanady et al. and Boogaard et al. except for the mouse lung (see
above). DEB hydrolysis was modeled using simple Michaelis-Menten kinetics, while DEB
conjugation with GSH was described assuming a ping-pong mechanism, as was done for EB
conjugation.
Nonenzymatic reactions of EB and DEB were suggested by the results of the vial equilibration
techniques used to determine the partition coefficients. EB and DEB losses were observed in all
tissues except muscle even after GSH depletion, epoxide hydrolase inhibition, and heat
inactivation of other enzymes. First-order rate constants for these nonenzymatic processes were
estimated from the time course data for EB and DEB disappearance.
The submodel for GSH synthesis and degradation in the liver and lung is apparently the same
submodel discussed in Sweeney et al. (1996), which was based on the model of D'Souza et al.
(1988) with some modifications to account for diurnal variation. Some of the parameter
estimates come from D'Souza et al. and others are estimated by optimization. It is not clear
which datasets were used for optimization. According to Sweeney et al. (1996), declines in GSH
observed by Himmelstein et al. (1995) in control and exposed animals were simulated well.
Sweeney et al. (1997) attempted to validate the DEB disposition part of their model against
DEB blood clearance data from rats after intravenous administration of DEB (Valentine et al.,
1997); however, they found that the DEB reactions already described in the model were
insufficient to account for the rate of disappearance of DEB from the blood. To remedy this
situation, Sweeney et al. determined that adding a first-order clearance term for DEB to the
slowly perfused compartment, but not to any other compartments, substantially improved the fit
of the model to the DEB clearance data. The value of the first-order clearance term was obtained
by optimization. The biological relevance of this additional term is unclear, although Sweeney et
al. note that epoxide hydrolases have been found in many tissues, including the skin, which is a
9-32
-------
component of the slowly perfused compartment, although a minor component. These
investigators further found that the same first-order clearance term for the slowly perfused
compartment was needed to fit the rat blood clearance data of Valentine et al. following iv
administration of EB, and an estimate of this clearance rate was also obtained by optimization.
Sweeney et al. observed that the value of the first-order clearance term for rats was 1.47 times the
ratio of Vmax to Km for liver hydrolysis for both EB and DEB, and thus used this formulation to
estimate clearance rates for the mouse.
Model predictions of blood DEB peak concentrations and AUCs were then computed for
comparison with unpublished data of Himmelstein obtained after tail vein administration of EB
in mice. Simulations were conducted using both the one-enzyme and two-enzyme models for
EB oxidation (see above). Predictions from the two-enzyme model were closer to experimental
observations.
This revised model still substantially overpredicted blood EB levels for both rat and mouse
following exposure to 1,3-butadiene. The incorporation of diffusional limitation in the liver that
was investigated in the Sweeney et al. 1996 version of the model was rejected because it resulted
in underpredictions of the mouse DEB blood levels observed after iv administration of EB and
underpredictions of liver GSH levels. Thus, Sweeney et al. explored the possibility that there is
an alternative oxidative pathway for 1,3-butadiene that does not yield EB. These investigators
suggest that this possibility is consistent with data of Bond et al. (1986) showing that following
rodent exposure to [14C] 1,3-butadiene, a large proportion of radiolabeled 14C is exhaled as CO2,
by a biotransformation mechanism that has not yet been explained. In addition, Sweeney et al.
cite the urinary metabolite results of Nauhaus et al. (1996) as indicating that there could be a
substantial flux of 1,3-butadiene through a non-EB-producing pathway that Nauhaus et al.
propose involves metabolism to acrolein and acrylic acid. (See Figure 3-1 for some possible
pathways.)
Sweeney et al. experimented with two different methods for determining the proportion of 1,3-
butadiene metabolism going to each pathway. For both approaches, total metabolism of 1,3-
butadiene was assumed to be described by the rates used in the original Medinsky et al. (1994)
model, as those rates were able to simulate the 1,3-butadiene uptake experiments. The first
method used the single Michaelis-Menten rate equation for total metabolism and assumed that
two products were produced. This two-product model assumed that the fraction metabolized to
EB was independent of 1,3-butadiene concentration and the same for both the liver and lung.
Parameter estimates for the value of the fraction were obtained by optimization using the blood
EB concentration data of Himmelstein et al. (1994) for mice and rats exposed to 1,3-butadiene.
The optimal value for the fraction of 1,3-butadiene metabolized to EB was 0.241 for the rat and
0.192 for the mouse, suggesting that the majority of 1,3-butadiene metabolism might be through
9-33
-------
the hypothesized alternate pathway. Using this optimal fraction for the mouse, the model also
provided a good fit to blood DEB and liver GSH data.
The second method assumed two separate pathways for 1,3-butadiene oxidation and involved
partitioning total metabolism into separate rate equations for each pathway using blood 1,3-
butadiene data reported by Himmelstein et al. (1997) and the mass balance data of Bond et al.
(1986) to fit the parameters, although the Bond et al. data do not include any metabolites exhaled
during the 6-h exposure period. Simulations using this approach suggest that the fraction of 1,3-
butadiene metabolized to EB increases with decreasing 1,3-butadiene exposure.
The models using both approaches were also validated against additional data on 1,3-
butadiene, EB, and DEB blood concentrations from Bechtold et al. (1995) and Thornton-
Manning et al. (1995a). The degree of comparability of predictions with observations varies
across species, experiment, metabolite, time, and exposure level; however, in general, it appears
that at the lowest 1,3-butadiene exposure level evaluated (62.5 ppm), the two-product model
provided better estimates for mouse blood levels than the two-pathway model, which tended to
overestimate levels, especially for EB. Both methods overestimated rat blood levels of EB and
DEB from exposure to 62.5 ppm 1,3-butadiene by about 2-6 times. Differences between the two
approaches are likely to be more dramatic at lower exposure levels, as the fraction of 1,3-
butadiene metabolized to EB increases in the two-pathway model; however, there are no blood
data for validation at lower exposures.
While different approaches to modeling some of the mechanistic pathways were explored, no
comprehensive sensitivity analysis for any of the model parameters was reported.
On the basis of the model simulations of blood levels following iv administration of EB and
DEB, Sweeney et al. conclude that extrahepatic and extrapulmonary metabolism is required to
describe the clearance of these compounds. They suggest that this conclusion is also consistent
with the discrepancy between tissue:blood partition coefficients and the ratio of tissue and blood
concentrations measured by Thornton-Manning et al. following in vivo exposure to 1,3-
butadiene. The lower concentration ratios for tissues such as heart and spleen, which are not
considered major organs for xenobiotic metabolism, suggest that EB and DEB are being further
metabolized throughout the body. Sweeney et al. note that "[uncertainty as to the nature of
these reactions makes extrapolation to other species tenuous."
Another major conclusion of Sweeney et al. is that observations of blood EB and DEB
concentrations in mice and rats are best explained by the existence of an alternative oxidative
pathway for 1,3-butadiene metabolism that does not generate EB. These investigators suggest
that such a pathway might also explain the urinary metabolite data of Nauhaus et al. (1996) and
the exhaled 14CO2 data of Bond et al. (1986). Sweeney et al. further suggest that some support
for the two-pathway model for 1,3-butadiene oxidation can be found in experiments in which
9-34
-------
mice were exposed to both 1,3-butadiene and styrene (Leavens et al., 1996). Styrene coexposure
was found to result in a decreased rate of 1,3-butadiene metabolism, while EB blood
concentrations were increased. Sweeney et al. propose that styrene might be competing with 1,3-
butadiene for a non-EB-producing pathway, forcing more 1,3-butadiene metabolism to EB, even
while the total rate of 1,3-butadiene metabolism decreases. They suggest that if this is the case,
an individual's specific complement of P-450 isozymes may be an important determinant of that
individual's risk from 1,3-butadiene exposure.
One of the strengths of the work of Sweeney et al. is that these investigators conducted actual
experiments to obtain measurements for some of the parameters (e.g., partition coefficients for
DEB) as well as to generate additional data for model validation. They also evaluated different
hypotheses for resolving blood EB overpredictions in the original model. Their final model,
using either approach for describing the proposed alternative pathway for 1,3-butadiene
oxidation, generally yields fairly good predictions of blood EB and DEB concentrations in mice
and rats exposed to 1,3-butadiene. Model validation was conducted using multiple independent
data sets.
The main limitations of the Sweeney et al. model are that several of the reactions described are
hypothetical and a substantial number of the parameters have been estimated by optimization.
The work of these investigators provides a good example of how PBPK modeling is a valuable
tool for generating hypotheses that can be tested, and that result in a better understanding of
underlying mechanisms that can in turn lead to the development of more definitive models. At
this point, however, the model is inadequate for use in risk assessment. While the model
generally predicts blood EB and DEB concentrations in mice and rats exposed to 1,3-butadiene
fairly well, there are serious uncertainties regarding some of the processes being modeled and
some of the parameter values used that make extrapolation of the model to humans and to lower
exposure levels of 1,3-butadiene highly uncertain, especially given the absence of human data
needed to validate any such extrapolations.
9.2.10. Kohn(1997)
Kohn's 1997 paper does not present a comprehensive new 1,3-butadiene PBPK model, but
rather summarizes his comparisons of different assumptions for modeling 1,3-butadiene uptake
data. Kohn observes that the consequences of the mathematical simplifications used to model
complex biological systems are rarely explored, and in this paper he investigates the assumption
of flow limitation commonly used in gas inhalation PBPK models.
The assumption of flow limitation presupposes that the permeability of a tissue to a chemical is
sufficiently high that the levels of the chemical in the tissue itself and in the blood in the tissue
capillary space are nearly at equilibrium and the tissue and its capillary space can be treated as
9-35
-------
one compartment. As long as the permeability is high with respect to the blood flow rate to the
tissue, the uptake of the chemical from the blood to the tissue is limited by the blood flow rate.
A single-pass extraction ratio, e, is defined as the rate of uptake of a chemical by the tissue
divided by the sum of the rates of transport from capillary blood to tissue and venous spaces and
from the tissue into capillary blood.
Kohn compares three models of 1,3-butadiene uptake and disposition by the mouse in closed
inhalation chambers. The first model ("consolidated blood" model) is a classical PBPK model in
which blood is consolidated into explicit arterial and venous blood compartments. The model
also includes compartments for lung, liver, kidney, fat, rapidly perfused tissues ("viscera"), and
slowly perfused tissues ("muscle"). 1,3-Butadiene in the chamber air was treated as partitioning
into the arterial blood, and the distribution of 1,3-butadiene between compartments was assumed
to be flow-limited.
The second model is based on the work of Anderson et al. (1987). It infers a gas exchange
compartment, representing the lung capillary bed, which equilibrates with air inhaled from the
chamber. This implicit compartment, along with implicit venous and arterial blood
compartments, is treated as having no volume and is not represented in the mass balance
equations. No storage of 1,3-butadiene occurs in these blood compartments, and blood
concentrations of 1,3-butadiene are assumed to be in equilibrium with the tissues. With the
exception of the blood compartments, the tissue compartments are the same as in the first model.
1,3-Butadiene distribution is again assumed to be flow-limited.
The third model ("distributed blood" model) distributes the blood among explicit
compartments for arterial and venous blood and for the capillary beds for each individual tissue
represented in the model (lung, liver, kidney, fat, viscera, muscle, and gastrointestinal tract). The
GI tract was separated from the viscera compartment to better depict hepatic perfusion. The
model also includes an explicit alveolar space.
The values of the physiological parameters, partition coefficients, and metabolic constants are
identical for all three models, as applicable, and are summarized in Table 9-8. Tissue volumes,
blood flow rates, alveolar ventilation, and cardiac output are reportedly consensus values taken
from the literature. Capillary bed volumes were taken from Altman and Dittmer (1971). Mouse
tissue:blood partition coefficients were taken from Medinsky et al. (1994) with the exception of
the muscle value, which was deemed unreliable because of concerns that the mouse muscle tissue
may have been contaminated by some attached fat; instead Medinsky et al.'s rat muscle:air value
was divided by their mouse blood:air value to get a muscle:blood coefficient. Kohn reports that
the blood:air partition coefficient of Medinsky et al. could not reproduce the equilibrium
chamber concentration data of Kreiling et al. (1986b) for mice treated with dithiocarb to inhibit
P450 activity; thus an average of the Medinsky et al. (mouse and rat) and the Johanson and Filser
9-36
-------
(1993) (rat) values was used. The kidney partition coefficient was apparently derived by
dividing the rat kidney:air partition coefficient of Johanson and Filser by the mouse blood:air
coefficient of Medinsky et al., and values for the viscera and GI tract were similarly derived from
Johanson and Filser's rat spleen:air coefficient. The extraction ratio, e, occurring in the
distributed blood model, was allowed to vary but was assumed to have the same value for all
tissues.
Metabolic constants for oxidation of 1,3-butadiene in the liver and lung were derived from the
in vitro data of Csanady et al. (1992). A Vmax value for oxidation in the kidney was derived from
data of Sharer et al. (1992). No Km value was available for the kidney, so the lung value was
used. A first-order rate constant for absorption onto the chamber walls and animal fur was taken
from Medinsky et al. (1994).
The models were first used to simulate the 1,3-butadiene uptake data from the chamber
experiments of Kreiling et al. (1986b). All three models fit the data for dithiocarb-treated mice,
with e values between 0.5 and 0.99 providing adequate fit for the distributed blood model. Then,
Kohn investigated the effect of varying e on the fit of the distributed blood model to the Kreiling
et al. data for uninhibited mice exposed to 100 ppm 1,3-butadiene. An extraction ratio of 0.5
gave the best fit; however, even with a value of 0.99, the distributed blood model did not
overpredict uptake as much as did the flow-limited models. The flow-limited models
overpredicted uptake for each 1,3-butadiene exposure level, whereas the distributed blood model
with e=0.5 fit all the data fairly well, although it did start to overpredict uptake for the 5000 ppm
exposure group after 6 h of exposure. Kohn also explored and rejected the hypothesis that
overpredictions of 1,3-butadiene uptake by the flow-limited models might be remedied if 1,3-
butadiene were sequestered in red blood cells.
The models were also used to simulate the mouse blood 1,3-butadiene levels following 6 h of
exposure observed by Himmelstein et al. (1994). Each model reproduced the data within 20%
error, although the distributed blood model provided a better fit at the highest exposure. Kohn
notes that the consistency of results across the three models for the blood data is not unexpected
because steady-state levels are largely determined by the partition coefficients.
A sensitivity analysis was conducted to ascertain which parameters most influence the
predicted chamber concentrations of 1,3-butadiene following 1 h of exposure. Cardiac output
and ventilation rate were among the most influential parameters. In the distributed blood model
the extraction ratio was also an important parameter. Of the partition coefficients, the blood:air
coefficient was determined to have the most influence. Sensitivities to Vmax for the metabolizing
enzymes increased with increasing exposure level and at the lowest exposure of 100 ppm were
much less influential than the extraction ratio.
9-37
-------
Table 9-8. Parameter values used in Kohn (1997) PBPK models.
Parameter
Alveolar ventilation (L/h/kg° 7)a'b
Cardiac output (L/h/kg° 7)a
Body weight (kg)0
Tissue compartment volumes
(% body weight)3
Arterial bloodd
Venous bloodd
Lung
Alveoli
GI tract
Liver
Kidney
Fat
Viscera
Muscle
Capillary bed volumes
(% tissue volume)6
Lung blood
GI tract blood
Liver blood
Kidney blood
Fat blood
Viscera blood
Muscle blood
Blood flow rates
(% cardiac output)3
GI tract
Liver
Kidney
Fat
Viscera
Muscle
Model
Consolidated blood
20.2
15.3
0.028
1.25
3.75
0.6
5.5
1.67
6
11.43
64.5
22.5
16.3
5
22.4
33.8
Anderson et al. (1987)
20.2
15.3
0.028
0.6
5.5
1.67
6
11.43
64.5
22.5
16.3
5
22.4
33.8
Distributed blood
20.2
15.3
0.028
0.661
1.983
0.6
5
7.5
5.5
1.67
6
3.93
64.5
11
2.9
11
10.2
3
7.1
1.3
18.1
4.4f
16.3
5
22.4
33.8
9-38
-------
Table 9-8. Parameter values used in Kohn (1997) PBPK models (continued).
Parameter
Partition coefficients for
1,3-butadiene
Bloodiair8
Lung:bloodh
Liver:bloodh
Kidney :blood'
Fat:bloodh
Visceraiblood1
Muscle :blood'
Rate constants for oxidative
metabolism of 1,3-butadiene
Liver Vmax (mmol/L/hy
KmO-iMy
Lung Vmax (mmol/L/hy
KmCHMy
Kidney Vmax (mmol/L/h)k
K,,, (|aM)m
Microsomal protein
concentrations
(mg protein/g tissue)
Liver11
Lungp
Kidneyq
Model
Consolidated blood
1.95
1.10
1.01
0.687
14.3
0.649
1.10
4.66
2
1.33
5.01
12.9
5.01
30
9
9
Anderson et al. (1987)
1.95
1.10
1.01
0.687
14.3
0.649
1.10
4.66
2
1.33
5.01
12.9
5.01
30
9
9
Distributed blood
1.95
1.10
1.01
0.687
14.3
0.649
1.10
4.66
2
1.33
5.01
12.9
5.01
30
9
9
" Consensus values from literature sources.
b 70% of minute volume per Arms and Travis (1988).
0 Average from experiments of Kreiling et al. (1986b).
d Blood outside of capillary spaces was distributed between arteries and veins in a 1:3 ratio (Menzel et al., 1987).
' From Altman and Dittmer (1971).
f Hepatic artery only.
8 Average of Medinsky et al. (1994) rat and mouse values and Johanson and Filser (1993) rat value.
h From Medinsky et al. (1994).
1 See text.
j From Csanady et al. (1992).
k From Sharer et al. (1992).
m Set equal to lung value.
11 From Alberts et al. (1983).
p From Smith and Bend (1980).
i From Coughtrie et al. (1987).
9-39
-------
Kohn concludes that improper assumptions and optimized parameters can hinder identification
of the correct model. He recommends verification of the accuracy of the mathematical
representations of inhalation, distribution, and metabolism to reduce uncertainties about the
PBPK models.
Although limited to 1,3-butadiene disposition, Kohn's 1997 modeling experiments provide
some interesting observations pertaining to the 1,3-butadiene PBPK model assumptions and
parameter estimates. A strength of this work is that Kohn attempted to estimate all parameter
values from the literature rather than by optimization; although specific values were not always
available and sometimes values for parameters assumed to be similar (e.g., from other tissues or
species) had to be substituted. Then he explored the implications of the basic assumption of flow
limitation using various model representations.
9.3. SUMMARY
1,3-Butadiene is metabolized to three epoxide intermediates and potentially to other reactive
metabolites as well. Comparisons of tissue levels of EB and DEB between mice and rats suggest
that DEB may be an important determinant of the differences in cancer potency between mice
and rats (Thornton-Manning et al., 1995a). However, EB is also a reactive metabolite, and it has
been suggested that EB may be important, for example, in the rat mammary gland tumorigenicity
of 1,3-butadiene (Thornton-Manning et al., 1998). In addition, Perez et al. (1997) have
suggested that the epoxydiol metabolite, 3,4-epoxy-l,2-butanediol, forms more hemoglobin
adducts than EB in both rats and humans exposed to 1,3-butadiene, yet this important metabolite
has been neglected in the current PBPK models. While much progress has been made in recent
years towards elucidating the differences in cancer potency between mice and rats, a complete
picture of the relative potencies of the various reactive metabolites at different tissue sites is still
far from clear.
Furthermore, uncertainties in the existing PBPK models and data make them unreliable for use
in human risk assessment. Serious uncertainties exist pertaining to the model structures and
assumptions, parameter values, and validation. There is a clear progression in the models over
time, as more experimental data have become available for both parameter estimation and model
validation and as different hypotheses have been tested. However, even among recent models,
there are important discrepancies that need to be resolved. In fact, as the models have gotten
more complex, models from different teams of investigators have become more divergent in
certain respects.
The earliest models were limited by the available experimental data, and some parameters were
estimated theoretically. Later models were able to incorporate more experimentally derived
parameters; however, several nonetheless rely on optimization to estimate critical parameters.
9-40
-------
Furthermore, as blood EB data became available for validation, it became apparent that models
that adequately simulated 1,3-butadiene and EB uptake experiments nevertheless overestimated
blood EB levels. The next generation of models proposed a variety of different hypotheses for
dealing with this problem, and none of these divergent hypotheses has been adequately tested.
For example, Johanson and Filser suggest that intrahepatic metabolism resolves the issue, while
Reitz et al. claim the problem lies in extrapolation from in vitro measures of metabolic constants
and Sweeney et al. have proposed that there is an alternative pathway responsible for a
substantial proportion of the oxidative metabolism of 1,3-butadiene.
Other discrepancies among the models arise in the treatment of extrahepatic and
extrapulmonary metabolism and the structural forms of the various phase I and phase II
reactions. The recent work of Kohn (1997) also raises questions about the basic assumption of
flow limitation. In addition, only the most recent models include disposition of the diepoxide
metabolite DEB, and none of the existing models include the disposition of the epoxydiol or
potentially reactive nonepoxide metabolites (see below).
With respect to the parameter values, there are disagreements about the ventilation rate, which
is a key parameter for determining 1,3-butadiene delivery, and about metabolic parameters. For
example, the Vmax and Km values used by Sweeney et al. for conjugation of EB with GSH in the
liver differ by orders of magnitude from those used by Reitz et al. for both the mouse and the rat.
Even the Vmax/Km rates differ by more than 14-fold. Similarly, the Vmax/Km rates for EB
hydrolysis differ by more than 13-fold. Recall that the Sweeney et al. rates are based on in vitro
data, while Reitz et al. rejected the in vitro data and estimated their parameter values from in
vivo data. The validity of the in vitro metabolic data is a critical issue for any PBPK-based
extrapolation of carcinogenic risk from rodents to humans because there are no appropriate
human in vivo PBPK data for 1,3-butadiene and thus interspecies extrapolation must rely on in
vitro data or allometric scaling.
Furthermore, the human in vitro data available for extension of the rodent PBPK models to
humans are limited. The measurements that have been made suggest there is a high degree of
inter-individual variability in metabolic enzyme activities. For example, Seaton et al. (1995)
reported a 60-fold variation in the initial rate of EB oxidation to DEB among 10 human liver
samples. (See also Table 3-4.) If the human in vitro data used as a basis for interspecies
extrapolation in PBPK modeling do not represent the variability in the human population, the
final results could be seriously biased. In addition to standard interindividual variability in the
activities of phase I and phase II enzymes, humans are known to have important polymorphisms
in the types of isoforms of glutathione-S-transferase and epoxide hydrolase that they express, and
these polymorphisms may have substantial effects on 1,3-butadiene metabolism. For example,
9-41
-------
sensitivity to DEB-induced genotoxicity in human lymphocyte cultures has been shown to be
highly associated with lack of the GSTT1 gene (e.g., Norppa et al., 1995; see also Chapter 4).
Another area of uncertainty is model validation. The existing models have been subjected to
limited validation. Virtually all of the model reports claim that the existing models adequately fit
the validation data, despite important differences among the models. In some cases, this is not
surprising because some of the model parameters have been determined by optimization against
data similar to those being used for validation. In other cases, it appears that gas uptake data are
relatively insensitive to various features of the models and might be of limited use in and of
themselves for model validation. Ideally, for the PBPK models to be more reliable, they should
be validated against tissue concentration data for various metabolites in various tissues over time
and for a variety of exposure concentrations, especially at the low- exposure end. More recently,
some of these data have become available (Chapter 3), although they must be interpreted with
caution because it appears that metabolites in some of the tissues are subject to further
metabolism during the lag time between the termination of exposure and the measurement of
tissue concentrations. Blood metabolite measurements have thus far provided the most useful
data for model validation. Because there are no human in vivo PBPK data for validating any 1,3-
butadiene PBPK model extended to humans, it is especially critical that the rodent models be
correct in their descriptions of the underlying biology and be well validated.
Recent studies suggest some additional considerations that might affect the accuracy and utility
of 1,3-butadiene PBPK models. For example, the work of Perez et al. (1997) suggests that it
might be important to include the epoxydiol metabolite, 3,4-epoxy-l,2-butanediol, in future
models. These investigators determined that epoxydiol-hemoglobin adduct levels were
substantially higher than EB-hemoglobin adduct levels in butadiene-exposed rats and humans.
Furthermore, Kemper et al. (1998) suggest that the 3-butene-l,2-diol metabolite, which is the
hydrolysis product of EB and an important metabolite in humans, can be metabolized to other
reactive intermediates in addition to the epoxydiol. Moreover, Richardson et al. (1998) conclude
from their studies of urinary metabolites following EB administration that the biotransformation
of 1,3-butadiene to DEB or the epoxydiol metabolite should be considered an integrated step.
Richardson et al. suggest that this might explain the inability of the PBPK models, which
generally model the reactions separately, to predict blood EB and DEB levels.
Recent studies also demonstrate that there is stereoselectivity in the metabolism of 1,3-
butadiene and its metabolites, and there may be differences in the mutagenic and carcinogenic
potency of the metabolite stereoisomers as well (e.g., Krause and Elfarra, 1997; Richardson et
al., 1998; Oe et al., 1999). If stereoselectivity in 1,3-butadiene metabolism is important, then
parameter estimates derived from in vitro and in vivo studies of commercial 1,3-butadiene
metabolites might not reflect what happens in vivo following 1,3-butadiene exposure.
9-42
-------
Another issue that has not yet been addressed in any of the 1,3-butadiene PBPK models is
metabolism in the bone marrow, which is a target organ in both mice and humans. Maniglier-
Poulet et al. (1995) have measured 1,3-butadiene metabolism to EB in vitro by mouse and human
bone marrow cells. EB formation was stimulated by hydrogen peroxide but not by NADPH,
suggesting peroxidase-mediated metabolism. These investigators suggest that "[although the
extent of conversion is markedly lower than in either rat or mouse hepatic microsomes, in-situ
conversion of [1,3-butadiene] to [EB] in bone marrow should be considered in future models of
[1,3-butadiene] metabolism and pharmacokinetics."
9.4. CONCLUSIONS
As discussed above, the existing PBPK models and data cannot yet adequately explain the
interspecies differences in 1,3-butadiene carcinogenicity. Uncertainties in the model structures
and parameter values also prohibit their use in refining risk assessment dosimetry at this time.
Some areas in which more research is needed include (a) investigation of the alternative
oxidation pathway for 1,3-butadiene hypothesized by Sweeney et al. (1997), (b) resolution of the
issue of the validity of the in vitro metabolic data for extrapolating to in vivo exposure, (c)
evaluation of the necessity for including intrahepatic first-pass hydrolysis of EB and possibly
other integrated reactions, (d) assessment of the impacts of stereoselective metabolism on the
PBPK models, (e) investigation of the validity of the flow limitation assumption, (f) evaluation
of the kinetics of the epoxydiol metabolite and possibly other metabolites in rodents and humans,
(g) clarification of the values of various physiological parameters, in particular the ventilation
rate, (h) more measurements of extrahepatic and extrapulmonary metabolism, (i) better
characterization of the distribution of values for the human metabolic rates, and (j) more time-
course measurements of blood and tissue concentrations of metabolites for model validation.
There is strong evidence that ovarian atrophy in the mouse is attributable specifically to DEB,
the diepoxide metabolite of 1,3-butadiene (Chapter 5). Therefore, a PBPK model describing
DEB disposition following 1,3-butadiene exposure could conceivably be used to extrapolate the
risk of ovarian atrophy from mice to humans; however, the existing PBPK models are inadequate
for this purpose given the substantial uncertainties and unresolved discrepancies described above.
With respect to carcinogenicity, the state of the science is much less certain. More information
on the specific roles of the various metabolites in the various target tissues would be necessary
before any dosimetric predictions from PBPK modeling of 1,3-butadiene could be translated into
human cancer risk estimates.
In conclusion, the existing PBPK models and data are inadequate for developing a reliable
alternative to the default methodology of using exposure to the parent compound as a dose
surrogate for extrapolation of the carcinogenic or reproductive risks from animals to humans.
9-43
-------
And, given the dramatic interspecies differences in cancer response between the mouse and rat,
any attempt to extrapolate the cancer risk in rodents to humans using the default methodology
would involve far greater uncertainties than basing a cancer risk assessment on the occupational
data of Delzell et al. (Chapter 7). Ideally, a reliable, well-validated PBPK model with parameter
values for humans could also be applied to analyzing different human exposure scenarios (e.g.,
extrapolating from occupational to environmental exposures). However, there are too many
uncertainties in the PBPK modeling for that to be practicable at this time.
9-44
-------
10. QUANTITATIVE RISK ASSESSMENT FOR 1,3-BUTADIENE
This chapter presents the quantitative risk analyses performed for the critical health
effects reviewed in this document. Section 10.1 provides the quantitative cancer risk assessment
based on the leukemia results from the Delzell et al. occupational epidemiology study reviewed
in Chapter 7. Section 10.2 presents quantitative analyses based on the cancer results from the
1993 NTP mouse bioassay described in Chapter 6. Section 10.3 presents the results of
benchmark dose modeling of several of the reproductive and developmental endpoints observed
in experimental animals, reviewed in Chapters 5 and 4 (dominant lethal effects). Section 10.4
develops Reference Concentrations (RfCs) for certain reproductive and developmental effects,
which are the most sensitive noncancer endpoints that have been observed in available toxicity
studies. Finally, Section 10.5 summarizes the key quantitative risk estimates.
10.1. EPIDEMIOLOGICALLY BASED CANCER RISK ASSESSMENT
10.1.1. Exposure-Response Modeling
For quantitative risk assessment, it is generally preferable to use high-quality
epidemiologic data, when available, over toxicologic data. In the past, available epidemiologic
data on 1,3-butadiene have been inadequate for quantitative cancer risk assessment because of
the lack of historical exposure information, and all but the most recent previous risk assessments
relied primarily on models based on the NTP mouse bioassay studies.
The retrospective mortality study by Delzell et al. (1995) of a cohort of synthetic rubber
production workers exposed to 1,3-butadiene (reviewed in Chapter 7) presents an opportunity to
perform a quantitative cancer risk assessment based on human data. The investigators developed
a job exposure matrix (JEM) for 1,3-butadiene, styrene, and benzene based on industrial hygiene
data, which contained estimates of the average daily exposure (in ppm based on the 8-h TWA)
and the number of annual peaks (defined as > 100 ppm for 1,3-butadiene and 50 ppm for styrene)
for each area and job code for each study year. The investigators were then able to estimate
cumulative exposures (ppm*years) and number of peak exposures (peak years) for each
individual worker by linking the JEM with the study subjects' work histories.
Delzell et al. (1995) investigated the relationship between cumulative exposure to 1,3-
butadiene and leukemia mortality using Poisson regression analysis (Frome and Checkoway,
1985). The models controlled for the potentially confounding effects of age (40-49, 50-59, 60-
69, 70-79, 80+), years since hire (10-19, 20-29, 30+), calendar period (1950-59, 1960-69, 1970-
79, 1980-89, 1990-91), and race (black, white, other). Plant was considered as a possible
confounder but was dropped from the final models because it did not affect the estimated
parameters for 1,3-butadiene or styrene. Few subjects were exposed to benzene, and benzene did
10-1
-------
not appear to confound the relationship between 1,3-butadiene or styrene exposure and leukemia
mortality. Hence, the model results presented in the report did not control for benzene exposure.
Different functional forms of the relationship between the relative rate (RR) and measures
of exposure were evaluated by Delzell et al. (1995), including the following:
(1) Log-linear: RR = epx
(2) Power: RR = ep[ln(1+X)]
(3) Linear excess: RR = 1 + PX
(4) Polynomial excess: RR = 1 + pjX" + p2Xq+....
where X represents the 1,3-butadiene or styrene exposure categories using the midpoints of the
intervals, P represents the estimated model parameters, and the powers "p" and "q" are fixed real
numbers. Although many polynomial functions (model 4) were considered, only the results from
a square root model were presented because this was considered to provide the best fit. This
model may be represented as:
(5) Square root: RR = 1 + pjX'72.
A positive exposure-response relationship between cumulative exposure to 1,3-butadiene
or styrene and leukemia mortality was observed from the Poisson regression analyses. This
relationship was evident both in models that represented these exposures as categorical variables
(see Table 59 in Delzell et al., 1995) and in models where exposure was represented using
continuous variables as described above. The regression parameter for 1,3-butadiene cumulative
exposure (adjusted for styrene) was found to be statistically significantly greater than 0 (p<0.05)
in all of the models evaluated, whereas nonsignificant relationships were observed for styrene
(adjusted for 1,3-butadiene). The 1,3-butadiene and styrene exposures among exposed study
subjects were found to be moderately correlated (Spearman's rank correlation, r=0.53). The
relationship between 1,3-butadiene cumulative exposure and leukemia mortality appeared to be
independent of the styrene exposure and was not appreciably altered by inclusion of styrene
cumulative exposure in the model. On the other hand, the relationship between styrene
cumulative exposure and leukemia mortality was weakened and irregular when 1,3-butadiene
cumulative exposure was controlled for. These findings suggest that 1,3-butadiene cumulative
exposure is a more likely explanation for the leukemia excess observed in this cohort than
styrene cumulative exposure.
10-2
-------
Analyses of peak years of 1,3-butadiene exposure indicated an association between this
variable and leukemia mortality even after controlling for cumulative exposure, but this
relationship was irregular in the categorical regression analyses. Excluding exposures that
occurred within 5 or 10 years of death (i.e., lagging exposures) only slightly increased the
exposure-response relationship for 1,3-butadiene cumulative exposure, whereas excluding
exposures within 20 years of death weakened and almost eliminated the relationship (see Table
63inDelzelletal., 1995).
Subsequently, Health Canada obtained the individual data on this cohort from Delzell et
al. and performed additional analyses (Health Canada, 1998). Their approach was similar to that
of Delzell et al. (1995), but some of the refinements made by Health Canada are noteworthy.
These refinements include finer stratification of age, calendar year, and years since hire, as well
as use of the actual mean cumulative exposure of 362.7 ppm-years for the highest exposure group
of "200+ ppm-years" rather than the apparently arbitrary value of 250 ppm-years used by Delzell
et al. The Health Canada report fit the same models as Delzell et al. with the exception that the
model with the square root of cumulative exposure (model 5 above) was replaced by one that
included cumulative exposure raised to the power a, which was estimated. This is a more
general model, which contains the "square root" model of Delzell et al. (1995) as a special case
[CC=0.5], with CC describing the shape of the relationship. This model is subsequently referred to
as the "shape" model. These refinements provide a more attractive basis for quantitative risk
assessment, and thus the models fitted by Health Canada were adopted for this analysis.
The results reported by Health Canada (1998) on the effect of cumulative 1,3-butadiene
exposure on the leukemia mortality rate under several models of the relative rate are summarized
together with results from Delzell et al. (1995) in Table 10-1. The smaller deviance values
reported by Health Canada reflect the fact that they used more levels in their stratification. The
differences in the regression coefficients from the two analyses were modest, and the Health
Canada report indicates that these differences were not found to be statistically significant. The
power and the shape models appear to provide the best fit (i.e., lowest deviance). However, the
difference in goodness of fit between these models is slight and thus it is difficult to select one
model as the "best" model on this basis. The Health Canada models are presented graphically in
Figure 10-1.
10.1.2. Prediction of Lifetime Excess Risk of Leukemia Mortality
The relative rate models presented in the report by Health Canada (1998), summarized in
Table 10-1, were used as a basis for predicting the excess risk of leukemia mortality for varying
levels of continuous environmental exposures to 1,3-butadiene. These risk estimates were made
using the relative rate estimates and an actuarial program that accounts for the effects of
10-3
-------
Table 10-1. Results from exposure-response models of cumulative exposure to 1,3-
butadiene using alternative structural forms reported by Health Canada (1998) and
Delzell et al. (1995)a.
Structural
model formb
Log-linear:
RR = epx
Linear:
RR = 1 + PX
Power:
RR = eP[ln(l + X)]
Shape:
RR = 1 + PXK
Health Canada (1998)
Model
deviance
176.721
174.713
171.547
171.994
P estimate
(S.E.)C
0.0029d
(0.0014)
0.00996
(0.0065)
0.28506
(0.0976)
0.4548f (P)
(0.8222)
0.3999f
(a)
(0.2733)
Delzell et al. (1995)
Model
deviance
486.0
486.0
485.6
485.6
P estimate
(S.E.)C
0.00416
(0.0019)
0.00686
(0.0050)
0.20286
(0.0972)
0.1293e(P)
(0.1024)
0.5 (a)
(O)8
"Adapted from Table 11-3 in Health Canada (1998) and Table 67 in Delzell et al. (1995). Additional information on the LRTp-values for the
Health Canada parameter estimates was provided by Health Canada (personal communication from Michael Walker to Leslie Stayner, 16 June
1999). Results presented are adjusted for age, calendar year, years since hire, race, and cumulative styrene exposure.
bX denotes cumulative 1,3-butadiene exposure (ppm-yrs).
0 S.E. is the standard error for the exposure parameter estimates.
d likelihood ratio test (LRT) p-value=0.0578.
eLRTp-value<0.05.
fLRTp-value<0.05 for model with both K and p versus null model.
8 a = '/2 was fixed by Delzell et al. (1995).
competing causes of death.1 U.S. age-specific mortality rates for all race and gender groups
combined (NCHS, 1993) were used to specify the leukemia and all-cause background rates in the
actuarial program. The risks were computed up to age 85 for continuous exposures to 1,3-
butadiene. The occupational 1,3-butadiene exposures in the epidemiologic study were converted
to continuous environmental exposures by multiplying the occupational exposure estimates by a
factor to account for differences in the number of days exposed per year (365/240 days) and
This program is an adaptation of the approach previously used in BEIR IV. Health Risks of Radon and Other Internally Deposited Alpha Emitters.
(1988) Washington, DC: National Academy Press; pp. 131-134.
10-4
-------
CM i
o _
03 -
CD -
co
or
CD
-l-»
CD
or
O -
RR = (1+x)A0.29
RR = 1 + 0.46*xA0.39
RR = exp(0.0029*x)
RR = 1+0.0099*x
\
0
100 200 300
Mean cumulative butadiene exposure (ppm-years)
Figure 10-1. Predicted rate ratios from Health Canada (1998) Poisson regression models".
a Open circles with error bars represent rate ratios and 95% confidence intervals from regression analyses with 1,3-butadiene exposure expressed as a
categorical variable. The four lines represent different exposure-response models, as defined in the legend, with 1,3-butadiene exposure as a continuous
variable. Additional information was provided by Health Canada for the calculation of the confidence intervals (personal communication from Michael
Walker to Leslie Stayner, 18 June 1999).
-------
another factor to account for differences in the amount of air inhaled per day (20/10 m3). The
reported standard errors for the 1,3-butadiene regression coefficients were used to compute the
upper 95% confidence limits for the relative rates based on a normal approximation.
Point estimates and one-sided upper 95% confidence limits for the excess risk of
leukemia mortality associated with varying levels of environmental exposure to 1,3-butadiene
based on the alternative model forms are illustrated in Figures 10-2 to 10-5. Estimates of
leukemia mortality risks and exposure levels corresponding to levels of risk of potential
regulatory interest are presented in Tables 10-2 and 10-3. These estimates vary by several orders
of magnitude depending on the model used, particularly at low levels of risk. For example, at the
1-in-a-million risk level, the maximum likelihood estimates for continuous lifetime 1,3-butadiene
exposure range from 0.26 parts per billion (ppb) (based on the loglinear model) to 1.1 x 10"8 ppb
(based on the shape model).
Consistent with the proposed EPA cancer risk assessment guidelines (U.S. EPA, 1996,
1999), these results were also used to estimate the exposure levels (ECp, "effective
concentration") and their associated 95% lower confidence limits (LECp) corresponding to
varying levels of excess risk (p) ranging from 0.1% to 10%. These estimates are summarized in
Table 10-4. Although the proposed EPA guidelines emphasize the derivation of exposure levels
associated with a 10% excess risk level, this does not seem reasonable in this instance. The 10%
level of risk is associated with exposure levels that are higher than most of the exposures
experienced by the workers in this epidemiologic study. Furthermore, based on the actuarial
program described above, a relative rate of 19 would be required for adults over the age of 20 to
achieve a lifetime excess risk of leukemia death of 10%, but the leukemia standardized mortality
ratios (SMRs) reported by Delzell et al. (1995) were considerably lower.2 Hence, these
considerations suggest that using a 10% risk level would be an upward extrapolation in this case.
A 1% or even lower (e.g., 0.1%) excess risk level would seem to be a more reasonable choice in
this circumstance. Based on the actuarial program, the relative rates for excess risks of 1% or
0.1% are 2.7 and 1.2, respectively. These rates better correspond with the set of SMRs reported
by Delzell et al. (1995). The exposure levels corresponding to a 1% excess risk level are
illustrated for the four relative rate models in Figures 10-2 to 10-5. When a 1% excess risk level
is used, the LEC01 from these analyses ranges from 0.037 to 0.87 ppm, depending on the model.
The maximum reported SMR was 13.33. This SMR was based on two leukemia deaths among black men from plant #2 with at least 10 years of
work (not all of which was salaried) and at least 20 years of elapsed time since hired. (See Table 29 of Delzell et al., 1995.)
10-6
-------
0.09
0.08 -
0.07 -
0.06 -
0.05 -
0.04 -
0.03 -
0.02 -
0.01
LEC01=0.87
One-sided 95% UCL .-
Maximum Likelihood Estimate
1 1.5
Exposure Concentration (ppm)
Figure 10-2. Excess risk and 95% upper confidence limit estimates for leukemia mortality
based on the log-linear model reported by Health Canada (1998)a.
aLog-linear model: RR=epx
10-7
-------
0.1
0.09 -
0.08 -
0.07 -
0.06 -
0.05 -
0.04 -
0.03 -
0.02 -
0.01 -
One-sided 95% UCL
Maximum Likelihood Estimate
0.5
1 1.5
Exposure Concentration (ppm)
Figure 10-3. Excess risk and 95% upper confidence limit estimates for leukemia mortality
based on the power model reported by Health Canada (1998)a.
"Power model: RR=epllr
10-8
-------
0.07
0.06-
0.05-
0.04-
0.03-
0.02-
0.01
LEC01=0.375
One-sided 95% UCL
Maximum Likelihood Estimate
1 1.5
Exposure Concentration (ppm)
Figure 10-4. Excess risk and 95% upper confidence limit estimates for leukemia mortality
based on the linear model reported by Health Canada (1998)a.
"Linear model: RR = 1 +
10-9
-------
1 1.5
Exposure Concentration (ppm)
Figure 10-5. Excess risk estimates for leukemia mortality based on shape model reported
by Health Canada (1998)a.
aShape model: RR=l+pxa; upper confidence limits could not be calculated.
10-10
-------
Table 10-2. Maximum likelihood estimates (MLEs) of excess leukemia mortality
risk with 95% upper confidence limits (95% UCL) for continuous lifetime exposures
to varying concentrations of 1,3-butadiene based on several models reported by
Health Canada (1998) and U.S. population rates.
Model
Log-linear:
RR=ePx
Power:
RR=eP[ln(l+x)]
Linear:
RR = 1 + pX
Shape:
RR = 1 + PXK
Concentration
(ppm)
l.Ox 1C'4
l.Ox ID'3
l.Ox 1C'2
l.Ox 1C'4
l.Ox 1Q-3
l.Ox 1C'2
l.Ox 1C'4
l.Ox 1Q-3
l.Ox 1C'2
1.0 x 1C'4
l.Ox 1Q-3
l.Ox 1C'2
MLE
excess risk
3.8 x 1Q-7
3.8 x 1Q-6
3.8 x 1Q-5
3.7 x 1Q-5
3.5 x 1Q-4
2.4 x 1Q-3
1.3 x lO'6
1.3 x ID'5
1.3 x ID'4
6.1 x ID'4
1.5 x ID'3
3.8 x 1Q-3
95% UCL
excess risk
6.8 x 1Q-7
6.8 x 1Q-6
6.8 x 1Q-5
5.8 x 1Q-5
5.5 x 1Q-4
4.0 x 1Q-3
2.7 x 1Q-6
2.7 x 1Q-5
2.7 x 1C'4
NAa
NA
NA
a Confidence intervals were not calculated for the "Shape" model because the Health Canada report did not include the
variances and covariances of its two exposure parameter estimates.
10-11
-------
Table 10-3. Maximum likelihood estimates (MLEs) and 95% lower confidence
limits (95% LCL) of continuous exposure concentrations associated with varying
excess risk levels for leukemia mortality based on relative rate results of several
models reported by Health Canada (1998) and U.S. population rates.
Model
Log-linear:
RR=ePx
Power:
RR=eP[ln(l+X)]
Linear:
RR = 1 + pX
Shape:
RR = 1 + PXK
Excess
risk
1 x 1Q-6
1 x 1Q-5
1 x 1C'4
1 x 1Q-6
1 x 1Q-5
1 x 1C'4
1 x 1Q-6
1 x 1Q-5
1 x 1C'4
1 x 1Q-6
1 x 1Q-5
1 x 1C'4
MLE
(ppm)
2.6 x 1C'4
2.6 x 1Q-3
2.6 x 1C'2
2.7 x 1Q-6
2.7 x 1Q-5
2.7 x 1C'4
7.7 x 1Q-5
7.7 x 1C'4
7.7 x 1C'3
1.1 x 1Q-11
3.5 x 1C'9
1.1 x 1Q-6
95% LCL
(ppm)
1.5 x 1C'4
1.5 x 1Q-3
1.5 x 1C'2
1.7 x 1Q-6
1.7 x 1Q-5
1.7 x 1C'4
3.7 x 1Q-5
3.7 x 1C'4
3.7 x 1Q-3
NAa
NA
NA
a Confidence intervals were not calculated for the "Shape" model because the Health Canada report did not include the
variances and covariances of its two exposure parameter estimates.
10-12
-------
Table 10-4. Maximum likelihood (ECp) and 95% lower-bound (LECp)
estimates of the continuous exposure concentrations associated with varying
levels of excess risk (p) of leukemia mortality.
Structural
model form
Log-linear model:
RR = epx
Power:
££ = eP[ln(l+X)]
Linear model:
RR = 1 + PX
Shape:
RR = 1 + PXK
Percentage
excess risk
(P)
10
i
0.1
10
1
0.1
10
1
0.1
10
1
0.1
1,3-Butadiene exposure
levels (ppm)
Maximum
likelihood
(ECp)
4.6
1.6
0.24
1.2 x 102
0.13
3.3x 10'3
8.34
0.780
0.0775
42
0.11
3.5 x 10'4
Lower 95%
bound
(LECp)
2.6
0.87
0.14
3.2
3.7 x 10'2
1.9 x 10'3
4.01
0.375
0.0373
NA
NA
NA
Ratio3
3.8 x 10'2
1.1 x 10'2
7.1 x 10'3
3.1 x 10'2
2.7 x 10'1
5.3 x 10'1
2.5 x 10'2
2.7 x 10'2
2.7 x 10'2
NAb
NA
NA
The ratio is the excess risk (p/100%) divided by the one-sided lower 95% confidence limit on the exposure estimate
(LEG,).
bConfidence intervals were not calculated for the "Shape" model because the Health Canada report did not include the
variances and covariances of its two exposure parameter estimates.
10-13
-------
10.1.3. Sources of Uncertainty
It is apparent from the results presented in Tables 10-2 through 10-4 that one major
source of uncertainty is the choice of the model for the prediction of risk. The range of values of
the LEG at the 10% excess risk level are within 2-fold, while the range at the 1% excess risk
level spanned just over one order of magnitude, and the range for the 0.1% level spanned nearly
two orders. In this instance, it seems most reasonable to utilize the results at the 1% risk level
because this corresponds to exposures within the range of this epidemiologic study. However, it
is not possible to clearly choose one of the relative rate models as the best for risk assessment
purposes because none of the models has a biologic basis. Furthermore, all the models
summarized in Table 10-1 fit the observed data nearly as well. Moreover, for a given linear
extrapolation, the ratios in Table 10-4 show that the sensitivity of the result to the choice of
excess risk level varies considerably for these models, with the linear model, as expected, being
the least sensitive.
Another major source of uncertainty in this analysis is the potential misclassification of
exposures in the study by Delzell et al. (1995). This is a frequent limitation of nearly all
epidemiologic studies of this type for quantitative risk assessment purposes. The exposures of
this study were based on modeling a relatively extensive set of data. However, questions have
been raised concerning the accuracy of exposure estimates, particularly for some ill-defined tasks
(letter from Elizabeth Moran, CMA, to Aparna Koppikar, March 25, 1996). For example, the
work histories of maintenance laborers do not indicate whether they were vessel cleaners (a high-
exposure category) or building cleaners (a low-exposure category). The full impact of this
potential exposure misclassification is unknown, but preliminary re-analyses by Delzell et al.
suggest that "the impact would have been to dampen and, possibly, to distort dose-response
relationships" (letter from Delzell and Macaluso, University of Alabama, to Aparna Koppikar,
April 2, 1996).
Another concern about the study has been expressed regarding the assignment of peak
exposures in the analysis, which were defined as average exposures equal to or greater than 100
ppm over 15 minutes. It has been suggested that there were tasks with extremely high peak
exposures (thousands of ppm) over very short time periods (seconds to a few minutes) (letter
from Delzell and Macaluso, University of Alabama, to Aparna Koppikar, April 2, 1996). The
models used in this risk assessment were based on cumulative exposure and did not consider the
potential for the effects of peak exposures. The potential impact of work area assignments and
butadiene peaks on leukemia mortality in this study population is an active area of research
among the investigators at the University of Alabama at Birmingham (UAB), who conducted the
study by Delzell et al. (1995).
10-14
-------
In addition, there is the possibility that the excess of leukemia observed in the UAB study
is related to confounding by another chemical also used at the plants. Initially, the analyses
conducted by the UAB considered potential confounding by benzene and styrene. Benzene was
only weakly associated with leukemia in this analysis, and its effect was eliminated when the
analysis was controlled for 1,3-butadiene and styrene exposures (Macaluso et al., 1996).
Leukemia mortality did appear to increase with styrene exposure, although not as strongly and as
consistently as with exposure to 1,3-butadiene. Studies of workers exposed to styrene alone have
not demonstrated an increased risk of leukemia (Bond et al., 1992; Okun et al., 1985; Wong,
1990; Frentzel-Beyme et al., 1978; Kogevinas et al., 1993). Thus, it appears highly unlikely that
either benzene or styrene exposure was a significant confounder in this study. It was recently
suggested that exposure to dithiocarbamates may potentially confound the observed association
between 1,3-butadiene exposure and leukemia in the study by Delzell et al. (Irons and Pyatt,
1998). Dithiocarbamates, however, have not been associated with leukemia risk, and at this time
the suggestion is little more than conjecture (see discussion in Section 7.3.3).
Other uncertainties are not so much inherent in the dose-response modeling or in the
epidemiologic data themselves, but rather occur in the process of obtaining more general Agency
risk estimates from these specific results. EPA cancer risk estimates are typically derived to
represent an upper bound on increased risk of cancer incidence for all sites affected by an agent
for the general population. For experimental animal studies, this is accomplished by using tumor
incidence data and summing across all the tumor sites demonstrating significantly increased
incidences. Furthermore, the most sensitive sex and species is customarily used in order to be
protective of the general human population. However, in estimating comparable risks from the
Delzell et al. epidemiologic data, certain limitations are encountered. First, the Delzell et al.
study is a retrospective mortality study, and cancer incidence data are not available for the entire
cohort. Second, the epidemiologic data are for a healthy male worker cohort and may not be
representative of the general population. Finally, 1,3-butadiene is a multi-site carcinogen in
rodents (see Chapter 6), and, while increased risk of leukemia was observed in the epidemiologic
study, the study may not have had sufficient statistical power to detect tumor increases at all
sites. The issue of study power is discussed further below, and how these issues were dealt with
quantitatively in deriving the final cancer risk estimates is addressed in the next section.
As discussed above, the rodent bioassay data clearly show that 1,3-butadiene is a
multisite carcinogen. This raises a concern that there may not have been enough statistical power
in the epidemiology study to detect increases for all the possible tumor sites. On the other hand,
this was a large cohort study (over 15,000 subjects), with a long follow-up time (up to 49 years;
25% of the subjects had died by the end of the follow-up), so for most tumor sites, this should
not be a problem. The main tumor site that might be at issue is the lung, which was the most
10-15
-------
sensitive site in both male and female mice. Lung cancer is fairly common in humans; this
reduces the power of an epidemiology study to detect an increase in lung cancer. A crude
"power" calculation is presented below to estimate the lung cancer results one might expect to
have observed in the epidemiology study if humans are as sensitive as female mice to the lung
cancer effects of 1,3-butadiene. The calculation is based on the average employment and
exposure characteristics of the Delzell et al. cohort, exposure estimates and number of subjects
available for 6 of the 8 plants, and the MLE of lung cancer risk for the female mouse (i.e.,
0.1/ppm). It is also assumed that there is no confounding by smoking and that equivalent excess
risks occur from equivalent cumulative exposures, with no latency. The epidemiology data are
for lung cancer mortality, whereas the mouse unit risk estimate is for excess lung cancer
incidence. However, lung cancer has a low survival rate, so this discrepancy should not
substantially impact the calculation.
10.1.3.1 Power Calculation for Lung Cancer for the Delzell et al. Cohort
SBR + related activities: 15649 subjects from 8 plants
median years worked = 7.8
average person-years/subject = 24.7
59% started working before age 30
1. Calculate average cumulative exposure (ppm*years) for the 14,295 subjects in the six plants
with exposure estimates (Delzell et al., 1995, Table 58).
Median cumulative exposure for 12,412 subjects with nonzero exposure = 11.3 ppm*years.
12412 . . ,
x(ll.3) = (0.87) X 11.3 = 9%ppm* years
14295
2. Calculate continuous partial lifetime exposure for period of follow-up.
9.8»»?w * years 1 $>years 240days 10m3 I day
—— x x — x : — = 0.130 BOOT
7$>years 24.7'years 365days 20m I day
3. Calculate MLE of risk for partial human lifetime based on female mouse lung cancer data.
10-16
-------
from TOXRISK: EC10 = 7.1918 ppm
for human lifespan = 70 years
exposure starting at age 30 years
exposure ending at age 55 years (i.e., 30 + 24.7)
risk at age 55
0.1 ,
= Q.Q \4ppm~1
1.191&ppm
4. Calculate partial lifetime individual excess risk per person.
Q.llQppm xO.O\4ppm~l = 0.00181
5. Calculate expected number of excess lung cancer cases.
14295x0.00181 = 26
Background expected lung cancer cases from the six plants = 312 (3 14 observed)
6. The statistical power to detect a predicted RR of 338/3 12 = 1 .083 is given by Beaumont and
Breslow(1981)as:
= Za -
Z^ = 1.645 -2(Vl.083 - \}J312 = 0.208 .
From cumulative normal frequency distribution tables, Z = 0.208 corresponds with A = 0.08.
From Snedecor and Cochrane (1989), probability of a value > Z (Z negative) = 0.5 + A
(Z positive) = 0.5 - A
A = 0.08 and Z positive. Therefore, power = 0.5 - 0.08 = 0.42 = 42%
i.e., the power to detect a statistically significant increase in lung cancer if true SMR = 108 is
42%.
10-17
-------
Thus, according to the above calculation, one might have expected to see 26 excess lung
cancer cases in the epidemiology study. In fact, only 2 excess lung cancer cases were observed in
the workers from the 6 plants over a background of 312 expected cases. However, as calculated
above, the study has low statistical power to detect such a small proportional excess (power to
detect a statistically significant increase in risk if the true SMR = 338/312 = 108 is estimated to
be 42%).
10.1.4. Summary and Conclusions
Risk estimates for environmental exposures to 1,3-butadiene are derived from an analysis
by Health Canada (1998) of the Delzell et al. (1995) occupational retrospective cohort mortality
study of approximately 16,000 workers in 6 North American styrene-butadiene rubber
manufacturing plants. The study results reflect follow-up from 1943 through 1991, with an
average follow-up of 25 years, and with about 25% of the cohort deceased. Although overall
mortality and all cancer mortality were below expected values based on general population
regional rates, the increase in leukemias was statistically significant (SMR = 1.43, 95% CI =
1.04-1.91) for all ever-hourly men (Delzell et al., 1996). The consistency of this leukemia result
across plants and with findings from previous epidemiology studies of 1,3-butadiene plus other
data led to the conclusion that this observed increase was caused by 1,3-butadiene. The
availability of high-quality epidemiologic data based on individual worker exposure estimates
and the significant exposure-response results for 1,3-butadiene observed by Delzell et al.
motivated the decision to base a quantitative cancer risk assessment on this database.
Although this cohort had been previously studied (Matanoski et al., 1989, 1990;
Matanoski and Schwartz, 1987; Matanoski and Santos-Burgoa, 1994), the Delzell et al. update
and analyses are noteworthy for their extensive work on exposure estimation based on detailed
reviews of individual job histories and a job exposure matrix (Delzell et al., 1995; Macaluso et
al., 1996). The careful work on exposure allowed better estimates of risk and exposure response.
Exposure metrics included cumulative ppm-years and years multiplied by annual number of peak
exposures of at least 100 ppm for at least 15 minutes. Additional individual worker exposure
information on both styrene and benzene allowed analyses to be adjusted for these potential
confounding exposures. The Delzell et al. (1995) report includes these analyses.
The Delzell et al. study used Poisson regression analysis with nine categories of
cumulative exposure for 1,3-butadiene and nine categories of exposure for styrene. The analysis
also included, as covariates, adjustments for age, race, calendar year, and years since hire.
Benzene was ruled out as a potential confounder and was not included in the relative rate models.
Relative rate models within the Poisson regression analyses included the (a) log-linear, (b)
power, (c) linear, and (d) square root models. The parameter representing cumulative 1,3-
10-18
-------
butadiene exposure was found to be statistically significant in all the models evaluated, and all
models fit the data adequately in the observable range. The cumulative styrene exposure
parameter was positive for all the models, but not statistically significant. Results with the peak
exposures metric were irregular and are not given further consideration here.
Subsequently, Health Canada obtained the data on this cohort from Delzell and performed
additional analyses of cumulative 1,3-butadiene exposure (Health Canada, 1998). Their
approach was similar to that of Delzell et al. (1995), but several refinements made by Health
Canada are noteworthy. These include finer stratification of age, calendar year, and years since
hire, as well as use of the actual mean cumulative exposure for the highest exposure group. The
Health Canada report fit the same models as Delzell et al., with the exception that the model with
the square root of cumulative exposure was replaced with one that included cumulative exposure
raised to the power CC, which was estimated.
The quantitative risk analysis presented here uses the results of the Health Canada
analyses to extrapolate risk from occupational work-time exposure to environmental continuous
exposure. This was simply done by calculating the cumulative exposure to 1,3-butadiene for
continuous instead of intermittent occupational conditions to obtain the relative rate of leukemia
mortality predicted for each year up to age 85. These predictions, together with U.S. population
rates for leukemia and all-causes mortality, were used to estimate excess risk up to age 85 using
an actuarial method that adjusted for competing risks. This does not mean that an average
lifetime of 85 years was assumed, rather the methodology uses actual age-specific survival rate
data, up to age 85. These techniques have been used before by EPA as well as other government
agencies.
After calculation of the exposure-response relationship, the low-exposure extrapolation is
done in two ways reflecting the different approaches used in EPA's 1986 Guidelines for
Carcinogen Risk Assessment (U.S. EPA, 1986) and in the proposed guidelines (U.S. EPA, 1996,
1999), which, as of this writing, are undergoing revision based on public comments and
comments from review by an EPA Science Advisory Board panel. Although neither version of
the guidelines provides specific guidance for modeling human data, the guidelines differ in their
general approach to low-dose extrapolation. EPA has generally used a linear model to fit human
data, and, under the 1986 Guidelines, the risk estimates are calculated as a potency or slope
factor derived from applying the same linear model to lower exposure concentrations. Under the
proposed guidelines, the risk estimates are obtained by first calculating a "point of departure" at
the low end of the range of observation, using any of the appropriate models, and then
extrapolating to 0 by means of a straight line (linear extrapolation default). The LED10 (i.e., 95%
lower confidence limit on a dose associated with 10% extra risk) is proposed as the standard
10-19
-------
point of departure. However, the LEC01 is used here because 1% is within the observable range
of increased leukemia deaths for the different 1,3-butadiene exposure groups in the Delzell et al.
study, and, in fact, a 10% extra risk level would require upward extrapolation. Furthermore,
exposure levels are expressed as exposure concentrations rather than doses; thus an LEG is used
rather than an LED.
The results of the extrapolations using the four relative rate models were listed in Tables
10-2 through 10-4 and presented in Figures 10-2 to 10-5. The results show that in the observable
excess risk range of 1%, the MLEs of continuous exposure (EC01) vary about 15-fold from 0.11
to 1.6 ppm, and the LEC01 estimates range from 0.037 to 0.87, or about 24-fold. As the
extrapolation extends 10-fold to a 0.001 excess risk level, the MLEs of exposure from the
various models diverge much more, almost a 700-fold range from 0.00035 ppm to 0.24 ppm. At
the 10"6 risk level, the MLEs of exposure diverge by more than seven orders of magnitude.
Clearly, the final risk estimates based on the 1986 guidelines' extrapolation procedures are highly
dependent on the choice of model.
As mentioned above, under the 1986 guidelines approach, the default model of choice
for human data has been the linear relative rate model. This choice is based more on historical
precedence and biological plausibility arguments than on statistical fit or conservatism. In fact,
for a 10"6 risk level, using the MLE from the linear model for 1,3-butadiene would be much less
protective of public health than would be the MLE from either the shape or power models. In
addition, EPA has historically used MLEs for cancer risk estimates from human data rather than
upper bounds as used with animal data. Using this approach with the Delzell et al. leukemia
mortality data, the linear model yields a maximum likelihood potency (slope) estimate of 1.3 x
lO'Vppm.
Using the linear extrapolation default approach under the 1996 and 1999 proposed
guidelines, the 1% excess risk level is chosen for determining the point of departure because it is
within the observable response range of leukemia deaths, and 10% is actually above the
observable range. At the 1% level, the different models provide MLE exposure estimates (i.e.,
EC01) ranging from 0.11 to 1.6 ppm, and the 95% LCLs (i.e., LEC01) on exposure range from
0.037 to 0.87 ppm. Without specific guidance for choice of model from the proposed guidelines,
potency estimates based on linear extrapolation from points of departure from each of the models
examined by Health Canada (1998) are presented in Table 10-5. The linear extrapolation default
is used based on the known genotoxicity of 1,3-butadiene, via its metabolites.
The cancer potency estimates, using EC01s as the point of departure, range from 6.3 x
10~3/ppm (log-linear model) to 9.1 x 10"2/ppm (shape model). The cancer potency estimates,
based on the available LEC01s from three models, range from 0.011/ppm (log-linear) to 0.27/ppm
(power), with the linear model yielding 0.027/ppm.
10-20
-------
Of the various estimates presented in Table 10-5, the potency estimate of 0.027/ppm was
selected to represent the excess leukemia mortality risk from the occupational data. Use of the
95% lower bound reflects some of the uncertainties inherent in the epidemiologic data, as well as
the policy position expressed in the 1996 and 1999 proposed guidelines. In addition, although
various models adequately fit the observed data, there is no compelling reason to deviate from
the linear model historically used for human data.
However, as discussed in Section 10.1.3 above, we actually wish to estimate excess risk
of cancer incidence, not mortality. Therefore, another calculation was done using the same linear
relative rate model and age-specific leukemia incidence rates for 1994-1998 (NCI, 2001; Table
Xffi-2: all leukemias, all races, males & females) from SEER (Surveillance, Epidemiology and
End Results program of the National Cancer Institute) in place of the leukemia mortality rates in
the actuarial program. This calculation assumes that leukemia incidence and mortality have the
same exposure-response relationship for 1,3-butadiene exposure and that the incidence data are
for first occurences of leukemia or that relapses provide a negligible contribution. The
calculation also relies upon the fact that the leukemia incidence rates are small compared to the
all-cause mortality rates. The resulting 1,3-butadiene exposure concentration estimates for an
excess risk of 1% are an EC01 of 0.528 ppm and an LEC01 of 0.254 ppm. Linear extrapolation
from the LEC01 (i.e., 0.01/LEC01) yields a unit risk estimate of 0.04/ppm for leukemia incidence.
An adjustment factor of 2 was then applied to this unit risk estimate to reflect evidence
from rodent bioassays suggesting that extrapolating the excess risk of leukemia in a male-only
occupational cohort may underestimate the total cancer risk from 1,3-butadiene exposure in the
general population. First, studies in both rats and mice indicate that 1,3-butadiene is a multisite
carcinogen. It is possible that humans exposed to 1,3-butadiene may also be at risk of cancers
other than leukemia and that the epidemiologic study had insufficient power to detect excess
risks at other sites (see below). Second, both the rat and mouse studies suggest that females are
more sensitive to 1,3-butadiene-induced carcinogenicity than males, and the female mammary
gland was the only 1,3-butadiene-related tumor site common to both species.
The mammary tumor unit risk estimated from the female mouse (most sensitive species)
data is just slightly lower (maximum likelihood estimate [MLE] = 0.02/ppm, 95% upper
confidence limit [UCL] = 0.03/ppm) than the human (male) leukemia risk (0.04/ppm based on
linear extrapolation from the LEC01 for leukemia incidence). Thus, EPA decided to apply an
adjustment factor of 2 to the leukemia risk estimate, resulting in a unit risk estimate of 0.08/ppm
intended to cover the combined risks for leukemia and mammary cancer and to provide
additional protection to account for the fact that small increases in risk at other sites, particularly
the lung, cannot be ruled out.
10-21
-------
Table 10-5. Cancer potency (unit risk) estimates for leukemia mortality based on
linear extrapolation from the LEC01 or EC01 calculated from the models presented
by Health Canada (1998).
Model
Log-linear
Power
Linear
Shape
EC01 (ppm)
1.6
0.13
0.78
0.11
Potency estimate
(ppm1)
(0.01/EC01)
6.3 x 1Q-3
7.7 x lO'2
1.3 x ID'2
9.1 x 1C'2
LEC01
(ppm)
0.87
.037
0.375
NA
Potency estimate
(ppm1)
(0.01/LEC01)
1.1 x 1C'2
2.7 x 10'1
2.7 x 1Q-2
NA
The Delzell et al. study was a large cohort study (over 15,000 subjects) with a long
follow-up time (up to 49 years; 25% of the subjects had died by the end of the follow-up), so for
most tumor sites there should be sufficient power to detect an increased risk. The main tumor
site that might be at issue is the lung, which was the most sensitive site in both male and female
mice. Lung cancer is fairly common in humans; therefore, the epidemiology study may have
lacked the statistical power to detect an increase in lung cancer. The crude "power" calculation
presented in Section 10.1.3.1 suggested that if humans were as sensitive as female mice to the
lung cancer effects of 1,3-butadiene, one might have expected to see 26 excess lung cancer cases
in the epidemiology study. In fact, only 2 excess lung cancer cases were observed in the workers
from the 6 plants over a background of 312 expected cases. However, the study has low
statistical power to detect such a small proportional excess (power to detect a statistically
significant increase in risk if the true SMR = 338/312 = 108 is estimated to be 42% according to
the method of Beaumont and Breslow [1981] as presented in Section 10.1.3.1), and a lung cancer
SMR of 107 (319 observed/297 expected) was observed for the ever hourly workers for the 8
plants (although there was no increased lung cancer risk in the overall cohort [SMR=101] or in
the subgroup of ever hourly workers with > 10 years worked and >20 years since hire
[SMR=100]).
The only process group associated with an increased lung cancer SMR was maintenance
(SMR = 141 observed/114 expected = 124). However, 7 mesotheliomas were also observed in
maintenance workers (9 among ever hourly workers in the total cohort), suggesting that these
workers may have been exposed to asbestos, a known lung carcinogen. Furthermore, the
evidence for the association between the increased lung cancers in the maintenance workers and
1,3-butadiene exposure is weakened by the fact that lung cancers are not increased in other
10-22
-------
process groups which exhibited increases in leukemia cases (e.g., 1,3-butadiene production), the
absence of a positive relationship with number of years worked, the absence of a trend with
increasing years since hire, and the fact that the increase was attenuated when state, rather than
national, lung cancer rates were used for comparison (Sathiakumar et al., 1998). Thus, the
overall evidence of an association between lung cancer and 1,3-butadiene exposure is tenuous.
On the other hand, because the background rate of lung cancer is high, the power of the
study to detect small increases in lung cancer risk is low, and, without adjusting for amount of
smoking, it is difficult to make firm conclusions. Workers are not allowed to smoke in the plants
because of the explosive potential of 1,3-butadiene; therefore, the workers may have had lower
cigarette consumption, and this could easily mask a small increase in lung cancer risk. Thus,
while the study does not provide good evidence of an association between lung cancer and 1,3-
butadiene exposure, one cannot rule out a small increase in risk.
Some increases were also observed for laryngeal cancer in the Delzell et al. study,
however, these are based on small numbers (for the overall cohort: 17 observed, 15 expected).
On the other hand, the increases are associated with process groups in which excess leukemias
are observed. No data are provided for duration of exposure or other exposure characteristics.
Thus, while the evidence for an association between laryngeal cancer and 1,3-butadiene exposure
is meager, a small increase in laryngeal cancer cannot be ruled out.
In conclusion, a lifetime excess cancer unit risk estimate of 0.08/ppm for the general
population was obtained by applying an adjustment factor of 2 to the unit risk estimate of
0.04/ppm for leukemia incidence which was derived from the Delzell et al. male worker cohort
data. The adjustment factor of 2 is intended to cover the potential risk of female mammary
cancer, which was observed in both mice and rats, and to provide further protection to account
for the fact that some increases in risk at other sites-particularly the lung, which was the most
sensitive site in female and male mice-cannot be ruled out.
10.2. CANCER RISK ESTIMATES BASED ON RODENT BIOASSAYS
10.2.1. Rat-Based Estimates
Cancer risk estimates based on the 1981 Hazleton rat inhalation study of 1,3-butadiene
were presented in EPA's 1985 1,3-butadiene risk assessment (U.S. EPA, 1985). 95% upper-limit
incremental lifetime unit cancer risk estimates for humans were calculated using the linearized
multistage (LMS) model, after estimating the equivalent human dose assuming 1,3-butadiene
retention based on results of a 1985 NTP absorption study (NTP, 1985; see EPA's 1985 report
for further details). The upper limit based on the male rat tumor incidence data for Leydig cell
tumors, pancreatic exocrine tumors, and/or Zymbal gland carcinomas was 4.2 x 10"3 per ppm 1,3-
butadiene exposure. The upper limit based on the female rat tumor incidence data for mammary
10-23
-------
gland carcinomas, thyroid follicular tumors, and/or Zymbal gland carcinomas was 5.6 x 10"2 per
ppm 1,3-butadiene exposure.
These rat-based estimates are not considered the most appropriate estimates of human
risk; they are merely presented for comparison purposes. EPA believes that the mouse is likely
to represent a better rodent model for human cancer risk from 1,3-butadiene (see below) and that
the cancer risk estimates derived from the epidemiologic data are the best available estimates for
human risk.
10.2.2. Mouse-Based Estimates
Cancer risk estimates based on the 1984 NTP mouse inhalation study were presented in
EPA's 1985 1,3-butadiene risk assessment; however, revisions to these estimates are warranted
because of the new data provided by the 1993 NTP mouse inhalation bioassay, which examined
cancer response from exposure to lower 1,3-butadiene concentrations than those used in the 1984
study (NTP 1984, 1993; see Chapter 6). Groups of male and female B6C3FJ mice were exposed
to 1,3-butadiene concentrations of 0, 6.25, 20, 62.5, 200, or 625 ppm 1,3-butadiene for 6 h/day, 5
days/week, for up to 104 weeks. Significant increases in tumor incidence were observed at
multiple sites: the hematopoietic system (lymphomas, histiocytic sarcomas), heart
(hemangiosarcomas), lung, forestomach, Harderian gland, liver, preputial gland (males), ovary
(females), and mammary gland (females), when adjusted for intercurrent mortality. Significant
increases in lung cancer incidence were observed in female mice at 1,3-butadiene exposure levels
down to 6.25 ppm, the lowest level tested.
10.2.2.1. Quantal
When EPA estimates cancer risks for humans from rodent bioassay data, the risk
estimates are generally calculated from the incidence of rodents of the most sensitive species,
strain, and sex bearing tumors at any of the sites displaying treatment-attributable increases. In
the case of 1,3-butadiene, so many sites demonstrated significant tumor increases attributable to
1,3-butadiene that background levels of tumor-bearing animals obfuscate the effects of 1,3-
butadiene when all these tumor sites are combined. Therefore, risk estimates were derived from
the incidence of female (most sensitive sex) mice with malignant lymphomas, heart
hemangiosarcomas, lung tumors (alveolar/bronchiolar adenomas or carcinomas), mammary
gland tumors (carcinomas, adenocanthomas, or malignant mixed tumors), or benign or malignant
ovary granulosa cell tumors (Table 10-6). These sites were considered the most responsive sites
with low background tumor incidence. Most of the impact on the low-dose linear extrapolation
is from the lung tumor response, because the lung tumor incidences show the largest increases at
the lowest exposures. The 625 ppm exposure group was not included in the dose-response
10-24
-------
analysis because all of the mice were dead by week 65, and the tumor response was already
virtually saturated in the 200 ppm exposure group. Note also that mice that died before the time
of observation of the first tumor were considered to be not at risk and were excluded from the
incidence denominators.
Table 10-6. Dose-response data for linearized multistage model.
Administered exposure
(ppm)
Human equivalent
exposure (ppm)
Number of mice with
tumorsa /Number of mice
at riskb
Control
0
6/50
6.25
1.1
19/49
20
3.6
26/50
62.5
11
31/50
200
36
46/49
aLymphocytic lymphomas, heart hemangiosarcomas, alveolar/bronchiolar adenomas or carcinomas, mammary gland
tumors (carcinomas, adenocanthomas, malignant mixed tumors), or benign or malignant ovary granulosa cell tumors.
bFemale mice surviving to the time of the first significant tumor, which was a lymphocytic lymphoma at day 203.
Human equivalent exposures were based on ppm 1,3-butadiene exposure, adjusted for
continuous daily exposure (e.g., 6.25 ppm x 6/24 x 5/7 =1.12 ppm). No attempt was made to
adjust for internal doses of reactive 1,3-butadiene metabolites because the PBPK data were
inadequate to develop reliable PBPK models (Chapter 9). No adjustments were made for 1,3-
butadiene absorption because there are no adequate human data. Furthermore, there is no reason
to expect nonlinearities in absorption at the lowest exposures (at least < 625 ppm). A 95%
upper-limit incremental lifetime unit cancer risk (extra risk) for humans was calculated from the
incidence data in Table 10-6 using the LMS model. The multistage model has the following
form:
P(d) = 1 - exp[-(q0
q2d2
qkdk)],
where P(d) represents the lifetime risk (probability) of cancer at dose d (human equivalent
exposure concentration, in this case), and parameters q; > 0, for I = 0,1,..., k.
10-25
-------
Extra risk over the background tumor rate is defined as
Point estimates of the dose coefficients (q;s), and consequently the extra risk function, at
any dose d, are calculated by maximizing the likelihood function with respect to tumor incidence
data. The incremental lifetime unit cancer risk for humans (qt*) is defined as the 95% UCL on
the parameter qb which is the linear dose coefficient, for extra risk. This 95% UCL represents a
plausible upper bound for the true risk. The 95% UCL was calculated using the computer
program GLOBAL86 (Howe and Van Landingham, 1986). Both the model and the curve-fitting
methodology used are described in detail by Anderson et al. (1983).
The tumor incidence data in Table 10-6 generated the following results using the LMS
model (GLOBAL86):
MLEs of dose coefficients:
q0 = 0.2629
q1 = 0.07643
q2 = 0.0
q3 = 0.0
q4 = 0.0
p-va\ue for chi-square goodness of fit > 0.01
q!* = 0.10
MLE of risk for an exposure concentration of 1 ppm = 7.4 x 10"2
MLE of risk for an exposure concentration of 1 ppb = 7.6 x 10"5
MLE of exposure concentration for a risk of 0. 10 (EC10) =1.4 ppm
95% UCL on exposure concentration for a risk of 0.10 (LEC10) =1.0 ppm
Thus, the incremental unit cancer risk estimate (95% UCL) for humans calculated from the
mouse 1993 NTP inhalation bioassay results is 0.10 per ppm for continuous lifetime inhalation
exposure to 1,3 -butadiene. The MLE of risk appears to be nearly linear between 1 ppm and
1 ppb and is about 0.075 per ppm 1,3-butadiene exposure.
Under EPA's proposed new cancer risk assessment guidelines (U.S. EPA, 1996, 1999),
unit cancer risk estimates for genotoxic chemicals, such as 1,3-butadiene, would be derived by
straight linear extrapolation to 0 from the LED10 (estimated 95% UCL on the dose corresponding
to a 10% extra cancer risk). Using the LEC10 generated for the LMS model by GLOBAL86
yields a unit cancer risk of 0.10/1.0 ppm = 0.10 per ppm, the same as the qt*. Using the EC10
yields 0.1 0/1. 4 = 7.1 x 10'2 per ppm.
10-26
-------
The unit cancer risk estimate (95% UCL) derived above is intended to depict a plausible
upper limit on the risk of developing any 1,3-butadiene-attributable tumor over a full (70-year)
lifetime. However, using the quantal incidence data for total tumor-bearing mice in each
exposure group does not fully characterize the cancer potency reflected by the mouse bioassay
results. First, the methodology does not account for the fact that many of the mice in the higher
exposure groups had tumors at multiple significant sites. Second, the methodology ignores the
fact that survival was significantly decreased in female mice exposed to 20 ppm or more 1,3-
butadiene as a result of fatal 1,3-butadiene-attributable tumors. Time-to-tumor analyses
conducted for specific tumor sites are presented below and can be used to evaluate the time
component of the cancer risk.
10..2.2.2. Time-to-Tumor
The mouse inhalation bioassay results demonstrate different dose-response relationships
for different tumor sites. To assess the characteristics of the dose-response relationships for
different tumor sites, time-to-tumor analyses were performed to adjust for competing mortality
from cancer at other sites.
Time-to-tumor analyses were conducted from the individual mice data, including the
9-month and 15-month interim sacrifice data, for sites demonstrating increased cancer incidence.
Benign and malignant tumors were combined for sites where appropriate. Thus time-to-tumor
analyses were performed for lung alveolar/bronchiolar adenomas or carcinomas; lymphocytic
lymphomas; histiocytic sarcomas; heart hemangiosarcomas; hepatocellular adenomas or
carcinomas; Harderian gland adenomas or carcinomas; forestomach squamous cell papillomas or
carcinomas; malignant or benign ovary granulosa cell tumors (female); and mammary gland
adenocanthomas, carcinomas, or malignant mixed tumors (female). Preputial gland carcinomas
in male mice were not analyzed because not all the tissues were examined microscopically.
Data from the 625 ppm exposure groups were excluded from analysis because of
excessive early mortality, as in the quantal analysis discussed above. In addition, data from
interim sacrifices for specific sites were excluded for exposure groups for which it appeared that
complete histopathologic examination for that site was not performed on the entire interim
sacrifice group.
Human equivalent exposures were based on ppm 1,3-butadiene exposure, adjusted for
continuous daily exposure, as described above.
The general model used for the time-to-tumor (or time-to-response) analyses was the
multistage Weibull model, which has the form
P(d,t) = 1 - exp[-(q0 + qtd + q2d2 + ... + qkdk)*(t - t0)z],
10-27
-------
where P(d,t) represents the probability of a tumor (or other response) by age t (in bioassay
weeks) for dose d (i.e., human equivalent exposure), and parameters z > 1, t0 > 0, and q; > 0 for I
= 0,1,.. .,k, where k = the number of dose groups -1. The parameter t0 represents the time
between when a potentially fatal tumor becomes observable and when it causes death (see
below). The analyses were conducted using the computer software TOX_RISK version 3.5
(Crump et al., ICF Kaiser International, Ruston, LA), which is based on Weibull models taken
from Krewski et al. (1983). Parameters are estimated using the maximum likelihood method.
Specific n-stage Weibull models were selected for the individual tumor types for each
sex based on the values of the log likelihoods according to the strategy used by NIOSH (1991a).
If twice the difference in log likelihoods was less than a chi-square with degrees of freedom
equal to the difference in the number of stages included in the models being compared, then the
models were considered comparable and the most parsimonious model (i.e., the lowest-stage
model) was selected.
Tumor types were categorized by tumor context as either fatal or incidental tumors.
Incidental tumors are those tumors thought not to have caused the death of an animal, and fatal
tumors are thought to have resulted in animal death. Lymphocytic lymphomas, histiocytic
sarcomas, and heart hemangiosarcomas were treated as fatal tumors, unless observed at an
interim or terminal sacrifice, in which case they were considered incidental. Furthermore, these
fatal tumors were deemed rapidly fatal, and t0 was set equal to 0 (the data were believed
insufficient to reliably estimate t0 in any event). Tumors at all other sites were treated as
incidental. This is basically the same determination as that made by NIOSH (1991a), except the
NIOSH report dealt with preliminary data that did not distinguish histiocytic sarcomas from
lymphomas. NIOSH further cited the work of Portier et al. (1986) analyzing tumor types in NTP
historical controls to lend support to these tumor context assumptions.
Parameter estimates for the time-to-tumor analyses for each tumor type are presented in
Tables 10-7 (based on female mouse data) and 10-8 (male mouse data). For all tumor types
except the heart hemangiosarcomas (both sexes) and the forestomach (male mouse), the one-
stage Weibull model was preferred. For male mice, the heart hemangiosarcomas and
forestomach tumors were best described by the two-stage model, and for female mouse heart
hemangiosarcomas, a three-stage model was preferred.
10-28
-------
Table 10-7. Parameter estimates for multistage Weibull time-to-tumor
model based on female mouse tumor incidence, without 625 ppm group.
Tissue
Lymphocytic
lymphoma
Histiocytic sarcoma
Heart
hemangiosarcoma
Lung
Liver
Forestomach
Harden an gland
Mammary
Ovary
QO
6.23 x 1Q-10
3.68 x 1C'14
0
5.83 x 1C'9
2.12 x 1Q-8
0
1.50 x 1Q-5
2.47 x lO'6
7.83 x 1C'9
Ql
1.67 x 1Q-10
1.23 x lO'14
0
3.40 x lO'9
2.11 x 1C'9
1.29x lO'9
2.06 x 1Q-6
5.42 x lO'5
1.48 x 1Q-8
Q2
—
—
0
Q3
—
—
2.88 x 1Q-17
—
Z
3.92
6.03
6.10
3.69
3.58
3.43
2.03
1.27
3.05
Table 10-8. Parameter estimates for multistage Weibull time-to-tumor
model based on male mouse tumor incidence, without 625 ppm group.
Tissue
Lymphocytic lymphoma
Histiocytic sarcoma
Heart hemangiosarcoma
Lung
Liver
Forestomach
Harden an gland
QO
1.84 x lO'8
0.0
0.0
1.38 x 10'7
1.40 x lO'4
9.68 x lO'10
1.65 x 1Q-7
Ql
1.28 x 1Q-9
1.04x lO'13
0.0
9.53 x lO'8
5.57 x 1Q-6
0.0
7.45 x 1Q-8
Q2
1.14 x ID'23
3.83 x 10'11
—
Z
3.08
5.50
10.0
3.27
1.83
3.39
2.90
10-29
-------
Human unit cancer risk (or potency) estimate results (extra risk) are presented in Tables
10-9 (based on female mouse data) and 10-10 (male mouse data). Mouse lung tumors convey
the greatest amount of extrapolated risk to humans from both the female mouse data (qx* =
0.14/ppm 1,3-butadiene exposure) and the male mouse data (qx* = 0.10/ppm). Note that the unit
risk estimate of 0.14/ppm generated from the female mouse lung tumor data using a time-to-
tumor model is greater than the unit risk estimate of 0.10/ppm generated above from female mice
bearing tumors at any of multiple sites when only the quantal data were used and decreased
survival time was not taken into account.
Although the time-to-tumor modeling does help account for decreased survival times in
the mice, considering the tumor sites individually does not convey the total amount of risk
potentially arising from the sensitivity of multiple sites. To get some indication of the total unit
risk from multiple tumor sites, assuming the multiple sites are mechanistically independent, the
MLEs of the unit potency from the Weibull time-to-tumor models were summed across tumor
sites, and estimates of the 95% upper bound on the summed unit potency were calculated. The
TOX_RISK software provides MLEs and 95% UCLs for human risk at various exposure levels,
allowing for calculation of unit potency estimates at those exposure levels.
When the MLEs of unit potency calculated at 1 ppb from the female mouse data were
summed across the female mouse tumor sites, the MLE of the total unit risk was 0.23/ppm
continuous lifetime 1,3-butadiene exposure. A 95% upper bound for the total potency was
calculated by assuming a normal distribution for the risk estimates, deriving the variance of the
risk estimate for each tumor site from its 95% UCL according to the formula
95% UCL = MLE + 1.645O,
where the standard deviation (SD) o is the square root of the variance, summing the variances
across tumor sites to obtain the variance of the sum of the MLEs, and calculating the 95% UCL
on the sum from the variance of the sum using the same formula. The resulting 95% UCL on the
unit potency for the total unit risk was 0.38/ppm. In comparison, summing the q^s across the
female mouse tumor sites yielded 0.50/ppm.
The unit potencies were also summed using a Monte Carlo analysis and the software
Crystal Ball version 4.0 (Decisioneering, Denver, CO). Normal distributions were assumed for
the unit potency for each tumor site, with the mean equal to the MLE and o as calculated from
the above formula. A distribution around the sum of the MLEs was then generated by simulating
the sum of unit potencies picked from the distributions for each tumor site (according to
probabilities determined by those distributions) 10,000 times. The mean for the sum and the
10-30
-------
Table 10-9. Human unit cancer risk estimates (extra risk, computed for risks of
106) based on female mouse tumor incidences, without 625 ppm group, using
multistage Weibull time-to-tumor model.
Tissue
Lymphocytic lymphoma
Histiocytic sarcoma
Heart hemangiosarcoma
Lung
Liver
Forestomach
Harden an gland
Mammary
Ovary
Ql*
(ppm1)
0.0239
0.1283
4.27 x 1Q-3
0.1404
0.0631
0.0215
0.0443
0.0321
0.0358
MLE
(ppm1)
0.0128
3.36 x lO'3
3.99 x 1Q-6
0.0980
0.0366
0.0112
0.0258
0.0203
0.0218
EC10
(ppm)
8.08
30.8
11.6
1.06
2.82
9.22
4.00
5.09
4.74
LEC10
(ppm)
4.33
0.806
9.24
0.737
1.64
4.80
2.33
3.23
2.89
0.1/LEC10
(ppm1)
0.0231
0.1241
0.0108
0.1357
0.0610
0.0208
0.0429
0.0310
0.0346
Table 10-10. Human unit cancer risk estimates (extra risk, computed for risks of
106) based on male mouse tumor incidences, without 625 ppm group, using
multistage Weibull time-to-tumor model.
Tissue
Lymphocytic lymphoma
Histiocytic sarcoma
Heart hemangiosarcoma
Lung
Liver
Forestomach
Harden an gland
Ql*
(ppm1)
6.437 x lO'3
0.02162
0.01266
0.1023
0.04447
4.258 x 1Q-3
0.07402
MLE
(ppm1)
2.220 x lO'3
0.01394
4.040 x lO'3
0.06998
0.02720
1.660x 1Q-5
0.05398
EC10
(ppm)
46.6
7.42
12.0
1.48
3.80
19.2
1.92
LEC10
(ppm)
16.1
4.78
7.59
1.01
2.33
13.3
1.40
0.1/LEC10
(ppm1)
6.224 x lO'3
0.02090
0.01318
0.09890
0.04300
7.517 x 1Q-3
0.07157
10-31
-------
95th percentile on the distribution were the same as the sum of MLEs and 95% UCL calculated
above, as they should be. However, a sensitivity analysis based on the contribution to variance
revealed that variability associated with the unit potency estimate for the histiocytic sarcomas
was contributing more than 83% of the variance on the sum, and some of the simulated sums
were negative (the distributions for the unit potency estimates were not constrained for the
summation analyses). Excluding the histiocytic sarcomas yielded the same MLE of total risk of
0.23/ppm; however, the 95% UCL decreased to 0.29/ppm. The lung, which then contributes the
most to the sum, contributed about 55% of the variance, followed by the liver with 20%, and no
simulated sums were negative.
The same analyses were performed summing the estimates of unit potency derived from
the male mouse data for the different tumor sites. The resulting MLE for the total unit risk was
0.18/ppm lifetime 1,3-butadiene exposure with a 95% UCL of 0.22/ppm. The lung contributed
about 56% to the variance, followed by the Harderian gland with about 20%. Histiocytic
sarcomas contributed only 3% in this case, and all simulated sums were positive.
Finally, the summation analyses were repeated for unit potency estimates calculated at
1 ppm exposure for comparison with the estimates calculated at 1 ppb. For the female mouse-
based risks (excluding histiocytic sarcomas), the sum of the MLEs was 0.22/ppm (2% lower than
at 1 ppb) and the 95% UCL on the sum was 0.28/ppm (4% lower than at 1 ppb). Thus, the total
unit potency estimates are reasonably linear up to 1 ppm continuous lifetime exposure. Recall
from Table 10-7 that the selected model for the heart hemangiosarcomas in the female mouse
was nonlinear; however, the unit risk estimates based on the heart hemangiosarcomas at these
extrapolated doses are lower than for the other sites and do not affect the total risk summed
across tumor sites. Similarly, the results based on the male mouse (both the sum of the MLEs
and the 95% UCL on the sum) calculated at 1 ppm were 2% lower than those calculated at 1 ppb.
For the male mouse, the selected models for both the heart hemangiosarcomas and the
forestomach tumors were nonlinear (Table 10-8); however, as with the female heart
hemangiosarcomas, the risks from these sites have little impact on the total risk.
The results of these summation analyses are summarized in Table 10-11.
10.2.3. Discussion
Based on the analyses discussed above, the best estimate for an upper bound on human
extra cancer risk from continuous lifetime exposure to 1,3-butadiene derived from animal data is
about 0.3/ppm. This estimate reflects the time-to-tumor response as well as the exposure-
response relationships for the multiple tumor sites (excluding histiocytic sarcomas) in the most
10-32
-------
Table 10-11. Unit potency estimates (extra risk) summed across tumor sites.
Tumor Sites
Female mouse tumor sites
(calculated at 1 ppb)
Female sites excluding
histiocytic sarcomas (at 1 ppb)
Female sites excluding
histiocytic sarcomas (at 1 ppm)
Male mouse tumor sites
(at 1 ppb)
Male mouse tumor sites
(at 1 ppm)
SumofMLEs
(ppm1)
0.23
0.23
0.22
0.18
0.17
95% UCL on
sum (ppm"1)
0.38
0.29
0.28
0.22
0.21
Sum of qt*s
(ppm1)
0.50
0.37
0.36
0.27
0.26
sensitive species and sex (the female mouse). Histiocytic sarcomas were excluded because they
introduced excessive variance into the upper bound and contributed only negligibly to the MLE
of total unit risk.
The greatest source of uncertainty in this estimate is from the interspecies extrapolation
of risk from mouse to human. The two rodent species for which bioassay data were available—
mouse and rat—vary significantly in their carcinogenic responses to 1,3-butadiene, in terms of
both site specificity and magnitude of response (Chapter 6). The mouse and rat also exhibit
substantial quantitative differences in their metabolism of 1,3-butadiene to potentially reactive
metabolites (Chapter 3). Unfortunately, existing pharmacokinetic models cannot yet adequately
explain the species differences in carcinogenic response (Chapter 9), and it is possible that there
are pharmacodynamic as well as pharmacokinetic differences between the mouse and rat with
respect to their sensitivities to 1,3-butadiene.
The mouse was the more sensitive species to the carcinogenic effects of 1,3-butadiene
exposure and, hence, the more conservative (i.e., public health protective) for extrapolation of
risk to humans. In addition, the mouse appears to be the more relevant species for extrapolation
to humans in terms of site specificity, in that 1,3-butadiene induces tumors of the
lymphohematopoietic system in both mice and humans. Melnick and Kohn (1995) further
10-33
-------
suggest that the genetic mutations observed in 1,3-butadiene-induced mouse tumors are
analogous to genetic alterations frequently observed in human tumors.
In addition to uncertainties pertaining to the relevance of the rodent models to human
risk, there is uncertainty in quantitatively scaling the animal risks to humans. Ideally, a PBPK
model for the internal dose(s) of the reactive metabolite(s) would decrease some of the
quantitative uncertainty in interspecies extrapolation; however, current PBPK models are
inadequate for this purpose (Chapter 9). In vitro metabolism data for humans suggest that
interhuman variability in the capacity to metabolically activate 1,3-butadiene nearly spans the
range between rats and mice (Chapter 3). Overall, based on current PBPK and genotoxicity data,
one would generally expect humans to have lower cancer risks than mice from 1,3-butadiene
exposure, but further advancements are needed in the PBPK models before these differences can
be reliably quantified.
Another major source of uncertainty in the unit potency estimate of 0.3/ppm is the
extrapolation of high-dose risks observed in the mouse bioassay to lower doses that would be of
concern from human environmental exposures. A multistage Weibull time-to-tumor model was
the preferred model because it can account for differences in mortality between the exposure
groups in the mouse bioassay; however, it is unknown how well this model predicts low-dose
extrapolated risks for 1,3-butadiene.
Uncertainties also pertain to the specific assumptions used in conducting these multistage
Weibull time-to-tumor analyses. Alternative analyses were performed to consider the sensitivity
of the results to some of these assumptions. For example, for each of the tumor types assumed to
be fatal, alternative analyses were conducted in which the modeling software estimated t0. In
each case, the resulting q^s, EC10s, and LEC10s were identical to those generated when t0 was set
equal to 0 a priori.
In addition, analyses were performed on the lymphocytic lymphoma data including the
625 ppm group, because this was the exposure group most affected by lymphocytic lymphomas
and relatively few animals in this group survived to develop tumors at other sites. From the
female mouse data, the resulting qx* was 0.515/ppm, or roughly twice that calculated when the
625 ppm group was excluded. From male mice, the qx* was 0.0215/ppm, or roughly three times
higher than that obtained when the 625 ppm group was excluded.
NIOSH (1991a) examined the sensitivity of its results for each tumor type to (a) model
selection (i.e., stage of Weibull model) from among models deemed to be comparable, (b) tumor
context assumptions, and (c) exclusion/inclusion of the 625 ppm exposure group, and generally
found only small discrepancies in the results. Moreover, uncertainties in some of the model
assumptions are trivial compared with the major uncertainties introduced by the interspecies and
high- to low-dose extrapolations.
10-34
-------
In conclusion, because of the high uncertainty in extrapolating 1,3-butadiene cancer risks
from rodents to humans and the existence of good-quality occupational epidemiology data with
exposure measures, the epidemiology-based risk estimates presented at the beginning of this
chapter are the preferred human risk estimates. The rodent-based estimates are presented
primarily for comparison purposes. Given that different quantitative methodologies and
assumptions were used to calculate the various risk estimates, the estimated upper bound (95%
UCL) on human incremental lifetime unit cancer risk from continuous 1,3-butadiene exposure
was 6 x 10"2/ppm based on the female rat tumors, 3 x lO'Vppm based on the female mouse
tumors, and 2 x 10"2/ppm and 6 x 10"3/ppm based on lymphocytic lymphomas in female and male
mice, respectively (lymphocytic lymphomas being the tumor site that most closely resembles the
lymphohematopoietic cancers observed in male workers exposed to 1,3-butadiene). The human
incremental lifetime unit cancer (incidence) risk estimate based on extrapolation from the
leukemias observed in occupational epidemiology studies was 4 x 10"2/ppm. A twofold
adjustment to the epidemiology-based estimate was then applied to reflect evidence from the
rodent bioassays suggesting that the epidemiology-based estimate may underestimate total cancer
risk from 1,3-butadiene exposure in the general population (see Section 10.1.4), resulting in a
lifetime excess cancer unit risk estimate of 8 x 10"2/ppm.
10.3. DOSE-RESPONSE MODELING OF REPRODUCTIVE AND DEVELOPMENTAL
TOXICITY
10.3.1. Introduction
The reproductive and developmental effects of 1,3-butadiene are among the health effects
(both cancer and noncancer) observed at the lowest exposure levels following short-term or
chronic inhalation exposure. Data on reproductive and developmental effects were available
from three types of studies for modeling and calculation of a benchmark concentration (BMC)
and the lower confidence limit on the BMC (BMCL). The three types of studies were (a)
prenatal developmental toxicity studies, (b) male-mediated dominant lethal studies, and (c)
chronic studies. Data on mice were used in these analyses because there were few if any effects
in the rat studies done. The studies are briefly described here, and pertinent endpoints used in the
BMC and BMCL calculations are shown in the tables. For more detailed evaluation of these and
other studies on 1,3-butadiene toxicity, the reader is referred to Chapters 4 and 5.
10.3.2. Prenatal Developmental Toxicity Studies
In the first type of study, a prenatal developmental toxicity, the effects of 1,3-butadiene
were evaluated in mice exposed 6 h/day for 10-days via inhalation at 0, 40, 200, and 1,000 ppm
10-35
-------
on gestation days (gd) 6-15 (Hackett et al., 1987b). Reduced fetal weights were seen in mice at
all exposure levels (Table 10-12). Thus, 40 ppm was considered a LOAEL for mice.
Fetal weight data were modeled using three approaches. In the first approach, data were
fit using a log-logistic model for developmental toxicity, as described by Allen et al. (1994a).
The TERALOG model software (ICF Kaiser International, KS Crump Group) was used for this
purpose. This analysis considers the within-dam grouping of fetal observations (using a beta-
binomial approach to account for intralitter correlation) as well as litter size. To apply this
model, the individual fetal weights were converted to dichotomous data using two different
values as the cutoff for defining an adverse level of response:
• A decrease below the 5th percentile of the control distribution, and derivation of the
EC053 and LEC054 associated with a 5% additional risk of obtaining a fetal weight
below the 5th percentile of the controls, or
• A decrease below the 10th percentile of the control distribution, and derivation of the
EC10 and LEC10 associated with a 10% additional risk of obtaining a fetal weight
below the 10th percentile of controls.
These cutoffs were based on Kavlock et al. (1995) who showed that, using this approach, the
BMDL estimates for a large database of similarly designed studies were close (on average) to the
corresponding NOAELs. Using the "percentiles" function within EXCEL, the 5th and 10th
percentiles of the control fetal weights were determined to be 1.13 and 1.169, respectively. The
model can be expressed as:
P(d, s} = a+ 6,s + [1 - a - 6ls]l{ 1 + exp[/?+ 62s - y\og (d - d0}}},
where P(d, s) is the probability of a low-weight fetus at dose d and litter size s, and the
parameters a, /?, y, $,, and 62 are estimated by maximum likelihood methods. The intercept
parameter (d0; sometimes referred to as a threshold parameter, i.e., the point at which the model
can no longer distinguish from background) was set equal to 0 in the model. Other parameter
constraints were: y> 1; 0 < a - d^s < 1.
The EC is the effective (exposure) concentration associated with a given level of risk, 5% in this case. The EC is also known as the
benchmark concentration (BMC).
The LEG is the 95% lower confidence limit on the effective concentration associated with a given level of risk. The LEG is also known as the
lower confidence limit on the benchmark concentration (BMCL).
10-36
-------
Table 10-12. Fetal weight data from prenatal (developmental) toxicity studya.
Exposure level
0
40 ppm
200 ppm
1,000 ppm
No. litters
18
19
21
20
Mean fetal weight/litter + SD
1.341 + .1068
1.282 + .05806
1.126b + . 09849
1.038b+.1118
aSpecies/strain: Pregnant CD-I mice. Exposure time: Gestational day (GD) 6-15. Exposure regimen: 6 h/day.
Exposure levels: 0, 40, 200, or 1,000 ppm.
V0.05.
Source: Hackett et al, 1987b.
In the second approach, fetal weight also was modeled as the average of mean fetal
weights per litter using the continuous power model (Allen et al., 1994b). The THWC model
software (ICF Kaiser International, KS Crump Group) was used for this purpose. Several cutoff
values were used, again based on those suggested by Kavlock et al. (1995):
• A 5% reduction in mean fetus weight/litter from the control mean,
• A reduction in mean fetus weight/litter to the 25th percentile of the control
distribution, and
• A reduction in mean fetus weight/litter to 0.5 SD of the control mean.
Again using the "percentiles" function within EXCEL, the 25th percentile of the mean fetal
weights among the control fetuses was estimated to be 1.295.
The continuous power model can be expressed as
m(d)=a+ (MY,
where m(d) is the mean of the mean fetus weight/litter for dose d, and a, /?, and y are parameters
estimated by maximum likelihood methods. The parameter constraints were: a > 0; y ^ 1-
Goodness of fit was determined by a £ test for the log-logistic model, and by an F test
for the continuous power model (U.S. EPA, 1995, Appendix A). The model was considered to
provide an acceptable fit if the/? value was greater than 0.05 and a graphical display of the data
showed a good fit of the model.
10-37
-------
The third approach used to model fetal weight data was the hybrid approach proposed by
Gaylor and Slikker (1990) and further developed by Crump (1995). The BENCH_C model
software (ICF Kaiser International, KS Crump Group) was used for this purpose. This approach
uses all of the information contained in the original observations by modeling changes in mean
response as a function of exposure concentration, but defines ECs and LECs in terms of
probability of response. In order to derive probabilities of response from changes in average
values, one must assume some distribution of mean fetal weights around the averages (an
assumption of normality, with a constant SD for all concentrations, in this case). The continuous
data are fit using a model that incorporates parameters from the quantal model. Several models
are possible within the software for both continuous data and quantal risk estimates.
For this study, the log-logistic model (not including litter size) was used for the quantal risk
estimates and the following model for the continuous portion of the hybrid model:
m(d) = m(0) + QtN-'O-n) - ^{(1 - PJ[~ l/[l+(/^)]]}],
where N is the standard normal distribution function, m(d) is the mean response at dose d, <7is
the SD of the response fixed for all dose groups, and ft and k are the log-logistic model
parameters estimated by the maximum likelihood method. The parameter constraints were
k> !;/?> 0.
Crump (1995) indicated that a background rate (P0) of 5% and an EC for 10% additional
risk corresponds to a change from the control mean of 0.61 SD. Because a change in mean fetal
weight/litter of 0.5 SD corresponded on average to aNOAEL in studies by Kavlock et al. (1995),
a P0 of 0.05 and an EC10 (10% additional risk) were used here.
Results of the modeling for fetal weight are shown in Table 10-13. The values were
converted from a 6 h/day exposure to continuous exposure (6/24). Although a dose-duration
adjustment is not typically applied for developmental toxicity endpoints, it was judged more
appropriate, especially in the case of fetal weight, because of the many possible times during
gestation when weight may be affected. In fact, exposure after the period of major organogenesis
(gestation days 6-15 in the mouse) may result in further decrements in fetal body weight (RTI,
1999). Because the animals in this study were killed on gestation day 18, an additional
adjustment from 10 days to 13 days could be considered. However, the adjustment for 6 h/day to
24 h/day is likely to be conservative, and any further adjustment for additional days of exposure
may be overly conservative. The application of dose-duration adjustments to developmental
toxicity data was explored and reported by Weller et al. (1999).
The log-logistic model resulted in an adequate fit of the data. Because the log-logistic
model requires converting continuous data to quantal responses, the continuous power model
also was applied but did not give an adequate fit with all four exposure levels. When it was fit to
10-38
-------
Table 10-13. Fetal weight modeling.
Model
Log-logistic (l-4)a
Continuous power (l-3)a
Hybrid model* ( 1-4)
Response
Individual fetal
weight
Mean fetal
weight/litter
Mean fetal
weight/litter
Cutoff
5th percentile
lOthpercentile
5% relative reduction
25th percentile
0.5 SD absolute
reduction
P0 = 0.05
EC" (ppm)
EC05 = 6.9
EC10 = 10
16.275
11.28
13.09
EC10 = 7.07
LEC" (ppm)
LEC05 =2.9
LEC10=4.7
13.38
9.17
10.64
LEC10 = 3.325
"Exposure levels modeled in each case are shown in parentheses.
bValues were adjusted to 24-h exposures, e.g., (40) (6/24)= 10 ppm.
the first three exposure levels, an adequate fit was obtained. The continuous power model gave
similar ECs and LECs but these were somewhat larger than those obtained with the log-logistic
model except for the one based on a cutoff using the 25th percentile. The hybrid approach
resulted in a quantal estimate of dose at the LEC10 that was lower than that for either the log-
logistic or continuous power model.
All three models have strengths and limitations that must be considered. For example,
neither the continuous power model nor the hybrid model are currently structured to account for
intralitter correlation or litter size, whereas the log-logistic model accounts for both. The hybrid
and the continuous power modeling approaches are based on mean fetal weight responses. With
the hybrid approach, the EC10 and LEC10 estimates derived correspond to the concentration (and
its lower bound) estimated to give a 10% risk of having a mean fetal weight below the cutoff.
This is not the same as having a 10% risk of an individual fetus with a weight less than the
cutoff. It is important to note that the SDs associated with average responses are smaller than
those corresponding to individual observations, and so the cutoffs defined by the spread around
the background mean (e.g., defined by assuming a 5% background response) tend to be closer to
the background means than would be the case for individual responses. As a consequence, the
models predict lower concentrations associated with smaller changes and the ECs therefore tend
to be smaller than would be the case when the cutoff is based on individual responses. This
may in part account for the fact that the hybrid approach gives a smaller EC10 than the log-
10-39
-------
logistic model. In addition, the version of the hybrid model used here does not allow use of the
SD (a) for individual dose groups, so the o at dose d0 was used for all dose groups. The log-
logistic model, on the other hand, makes use of the individual fetal weight data, although it does
require conversion of continuous data to quantal responses.
Unlike the log-logistic approach or the continuous modeling approach, the hybrid
approach has not been "calibrated" to find the optimal choices of P0 and risk level to result in
LECs (BMCLs) that on average are close to the NOAEL. Even though the LECs derived from
the continuous power approach do not correspond directly to the probability of individual
response, the options presented (e.g., a 5% relative reduction) have been shown to correspond on
average to NOAELs (Kavlock et al., 1995) and hence there is some basis for selecting those
options. In this case, it was not possible using the continuous power approach to obtain an
adequate fit of the data except when the highest concentration was ignored. Because the EC and
LEG values estimated from the continuous model were within the range of the remaining
concentrations (0, 40, and 200 ppm), the modeling approach was acceptable, but it would be
preferable to include the information contained in the 1,000 ppm data, if possible.
Based on the above considerations, the log-logistic model was considered the preferred
approach for modeling fetal weight data from this study, i.e., the log-logistic model fit all four
exposure levels adequately. Figure 10-6 shows the log-logistic model fit of the fetal weight
data. The combinations of the 5th percentile cutoff for a 5% additional risk or 10th percentile
cutoff for a 10% additional risk were shown by Kavlock et al. (1995) to result in BMDLs similar
to the NOAEL. The LEG (BMCL) value for the 5th percentile cutoff for a 5% additional risk of
2.9 ppm is the more health protective choice to use for the point of departure for the RfC.
10.3.3. Dominant Lethal Studies
In the second type of study used for BMCL calculations, the male-mediated effects of
1,3-butadiene were evaluated in three dominant lethal studies in which male CD-I mice were
exposed (Anderson et al., 1993, 1995, 1996, 1998; Brinkworth et al., 1998)5 and subsequently
mated with unexposed females. All three studies were conducted in the same laboratories under
essentially the same conditions and with the same personnel.
10.3.3.1. Anderson et al (1993,1995,1996)
In the first Anderson et al. study (reported in 1993,1995, 1996), males were exposed to 0,
12.5, or 1,250 ppm for 6 h/day, 5 days/week, for 10 weeks, then mated immediately after
exposure ended. Two females were mated with each male. One group of females at each
Drs. Anderson and Brinkworth very kindly supplied the individual animal data from all three of these studies after obtaining agreement from the
sponsors for use in the modeling of data for this effort.
10-40
-------
50 100 150 200
Human Equivalent, Continous Concentrations (ppm)
- 0.95
Mean Observed Response Rates
Model Predictions
-0.05
250
Figure 10-6. Log-logistics model for fetal weight data, with data points and standard error
bars.
10-41
-------
exposure level was killed on gestation day 17 (reported in Anderson et al., 1993), while another
was allowed to litter (reported in Anderson et al., 1996). Data from the final study report
(Anderson et al., 1995) were used to obtain individual animal data for modeling. The data used
for modeling are summarized in Tables 10-14a and 10-14b. In the prenatal component of the
study, there was a statistically significant decrease in the mean number of implantations, and a
significant increase in incidence and percentage of early deaths, late deaths, and abnormal fetuses
at 1,250 ppm. At 12.5 ppm, a statistically significant increase in incidence of late deaths (with
and without dead fetuses) and abnormal fetuses (including runts) was noted. However, incidence
of late deaths and abnormal fetuses was lower at 1,250 ppm than at 12.5 ppm, perhaps because of
the large increase in early deaths at the higher exposure level. For the purposes of modeling the
data, early and late deaths (including dead fetuses) are combined into "all deaths," resulting in a
consistently increasing response with increasing exposure level. In addition, the number of
abnormal fetuses was changed from that reported by Anderson et al. (1993) to include only those
with actual malformations, whereas the runts (those with <75% the body weight of the rest of the
litter) were considered in an analysis of fetal body weights. Skeletal evaluations were conducted
on a small subset of normal and abnormal fetuses (each abnormal fetus and a normal littermate as
well as one control corresponding to each abnormal fetus were examined). There was no clear
indication of any increase in skeletal defects among the animals examined. The mean fetal body
weights from the analysis of individual data are shown in Table 10-14a and indicate a slight but
insignificant decrease at 1,250 ppm.
In the postnatal component of the study, there was a significant decrease in the number of
implantations and litter size at birth, as well as an increase in postimplantation loss at 1,250 ppm.
Litter size at weaning was similar to that at birth, because very few pups died between birth and
weaning. No changes were detected at 12.5 ppm. Total litter weight at birth was significantly
decreased at 1,250 ppm, but when mean pup weight per litter was calculated, there was no
difference between the offspring of control males and offspring of treated males; in fact, there
was a slight increase in the 1,250 ppm pup weights (data not shown). This is consistent with data
from Anderson et al. (1995) indicating that male and female body weights of Fl offspring from
treated males were increased above controls at 8 through 71 weeks of age. No further analysis of
these data was conducted.
Table 10-15 shows the EC05, LEC05, EC10, and LEC10 values for various endpoints from
the Anderson et al. study (1993, 1995, 1996). Implant data from both the prenatal and postnatal
10-42
-------
Table 10-14A. Male-mediated developmental toxicity, prenatal component".
Exposure
level
0
12.5 ppm
1250 ppm
No. males
exposed
(no. mated)
25
(23)
25
(25)
48
(48)
No. preg.
females'"
23
24
38
Total
no. implants
(mean ± SD)
278
(12.09 ± 1.276)
306
(12.75 ±2.507)
406
(10.68f±3.103)
No. early deaths
(meanc + SD)
13
(0.050 ± .0597)
16
(0.053 ± .0581)
87
(0.204h±.1611)
No.
late deaths
(meanc + SD)
0
7
(0.023f±.038)
6
(0.014h±.0324)
No. late deaths
incl. dead
fetuses
(meanc + SD)
2
(0.007 ± .022)
8
(0.026g ± .0424)
7
(0.016 ±.0339)
No.
abnormal
fetuses
(no. litters)"
0
4(2)
1(1)
Fetal weight
mean6 + SD
(no. litters)
1.142 ±.1055
(23)
1.116±.1030
(24)
1.109 ±.0935
(37)
aSpecies/strain: CD-I mice, adult males. Exposure regimen: 6 h/day, 5 days/week. Exposure time: 10 weeks. Exposure levels: 0, 12.5 ppm, 1250 ppm.
bUsed in prenatal component.
°Based on number of implants/litter.
dNot including runts.
eBased on number of live fetuses/litter.
Statistically significantly different from controls,p<.Q\.
Statistically significantly different from controls,p<-05.
hStatistically significantly different from controls, /K.001.
Source: Anderson et al, 1993, 1995, 1996.
-------
Table 10-14B. Male-mediated developmental toxicity, postnatal componenta.
Exposure
level
0
12.5 ppm
1,250 ppm
Litter size at
birth mean ±
SD
(no. litters)
12.22 ±2.16
(18)
11.14± 1.71
(21)
8.31c+3.48
(36)d
No. implants
mean + SD
(no. litters)
205
12.81±2.07(16)
210
12.35±1.69(17)
335
10.47C±2.90 (32)
No.
postimplantation
losses
(mean" + SD)
10
(0.625±.500)
19
(0.790±.676)
80
(2.50C±1.78)
Litter size at
weaning
mean + SD
(no. litters)
12.17±2.07
(18)
10.95±1.99
(20)
9.03C±2.53
(33)
aSpecies/strain: CD-I mice, adult males. Exposure regimen: 6 h/day, 5 days/week. Exposure time: 10 weeks.
Exposure levels: 0, 12.5 ppm, 1250 ppm.
bPer pregnant female.
"Statistically significantly different from controls, p<00l.
dThree litters completely resorbed.
Source: Anderson et al., 1993, 1995, 1996.
components of the study were fit using a linear model because no corpora lutea data were
available for normalizing the implant data. The model used was
y(d) = a + Pd,
where y(d) is the mean response for the endpoint of interest at dose (concentration) d.
For all other endpoints, the log-logistic model (TERALOG software), as discussed above
for fetal weight modeling, was used because the individual animal data for the dominant lethal
studies were available. Adjustment for duration of exposure was applied after the modeling (i.e.,
the model outputs for EC and LEG were multiplied by the factor (6/24)(5/7)). The endpoints
included in the modeling were those that appeared significantly related to 1,3-butadiene
exposure, and included early prenatal deaths, late prenatal deaths, all prenatal deaths, abnormal
fetuses in the prenatal component of the study, and postimplantation loss in the postnatal
component. Dead fetuses were not included in late prenatal deaths but were included in all
prenatal deaths. Abnormalities included only external effects detected in fetuses and did not
include skeletal or visceral effects (only a small subset of fetuses was examined for skeletal
effects), and also did not include runts or one fetus with blood in the amniotic sac. The litter size
used in the model was the number of live fetuses for each litter because only live fetuses were
10-44
-------
examined for abnormalities. Fetal weights were analyzed separately and did not appear to be
changed significantly with treatment. Fertility, numbers of dead fetuses, and postnatal deaths did
not appear to be significantly related to treatment and were not included in the analyses. Litter
size at weaning was not included because of the relatively small number of postnatal deaths, and
litter size at birth was captured by an analysis of all prenatal deaths.
As shown in Table 10-15, the EC and LEG estimates for early deaths and all deaths were
similar because early deaths dominated the "all deaths" values. The values for postimplantation
loss from the postnatal portion of the study also were similar even though this value may include
some deaths that occurred after prenatal day 17 (when fetuses in the prenatal component were
killed) up to postnatal day 0 when pups were delivered and examined.
Table 10-15. ECs and LECs for male-mediated developmental toxicitya.
Prenatal component
Estimate
(ppm)
EC05
LEC05
FT
-CA_x10
T Fr
J-JJA^IO
No.
implants
89
56
178
112
No.
early
deaths
55
40
116
85
No.
late
deaths'5
oo
oo
oo
oo
No.
all
deaths
53
39
112
83
No.
abnormal
fetuses0
oo
oo
oo
oo
Postnatal component
No.
implants
67
46
135
93
No. post-
implantation
losses
98
54
206
114
Exposures were adjusted to 24-h daily exposures, e.g., (12.5) (6/24) (5/7) = 2.2 ppm. Entries with the "°°" symbol
indicate essentially flat dose-response curves.
bNot including dead fetuses.
°Not including runts.
Source: Anderson et al., 1993, 1995, 1996.
10.3.3.2. Anderson et al (1998)
The second dominant lethal study, by Anderson et al. (1998), involved exposure of male
CD-I mice to 0, 12.5, 65, or 130 ppm for 6 hr/day, 5 days/week, for 4 weeks. Mating of males
began after 1 day of rest at the end of exposure with two females per male. Both females, if
pregnant, were included in the examination of effects reported in the paper. Data are shown in
Table 10-16. The ECs and LECs are shown in Table 10-17. Implant data were fit using a linear
10-45
-------
Table 10-16. Male-mediated developmental toxicitya.
Exposure
level
0
12.5 ppm
65 ppm
130 ppm
No.
fertile
males
25
24
23
24
No. implants
Mean + SDb
(no. preg. females)
572
12.43 ±2.60
(46)
528
12.00 ±2.37
(44)
559
12.42 ±2.58
(45)
562
11.96 ±2.52
(47)
No. early
deaths
(mean + SD)C
17
(0.039 ±0.10)
12
(0.022 ±0.05)
44
(0.083e±0.11)
49
(0.082e±0.10)
No. late
deaths
(mean + SD)C
7
(0.011 ±0.04)
3
(0.005 ± 0.02)
10
(0.017 ±0.04)
7
(0.013 ±0.03)
No. late deaths
incl. dead fetuses
(mean + SD)C
13
(0.023 ± 0.05)
12
(0.022 ±0.05)
14
(0.024 ±0.05)
16
(0.030 ±0.05)
No.
abnormal
fetus esd
0
0
0
0
Fetal weight
mean + SD
(no. litters)
1.178 ±.1962
(11)
1.107±.1941
(14)
1.049 ±.0830
(13)
1.161 ±.1801
(16)
aSpecies/strain: CD-I mice, adult males. Exposure regimen: 6 h/day, 5 days/week. Exposure time: 4 weeks. Exposure levels: 0, 12.5, 65, 130 ppm.
bMean per litter.
°Mean per implant per litter.
^Excludes runts.
Statistically significantly different from controls, p<.01.
Source: Anderson et al., 1998.
-------
Table 10-17. ECs and LECs for male-mediated developmental toxicitya.
Estimate
(ppm)
FT
-c,^05
T Fr
-L-J-A-'05
EC10
LEL10
No.
implants
- [12]
oo [7.3]
oo [24]
- [15]
No. early
deaths
14 [23]
8.9 [11]
30 [49]
19 [23]
No. late deaths"
oo [86]
oo [29]
oo [CF]
oo [CF]
No. all deaths
17 [18]
9.5 [10]
37 [38]
20 [21]
Exposures were adjusted to 24-h daily exposures, e.g., (12.5) (6/24) (5/7) = 2.2 ppm. Entries with the "°°" symbol
indicate essentially flat dose-response curves, i.e., no estimates for the EC or LEG values could be derived. Values in
brackets are the estimates from the alternative analyses checking sensitivity of results. All fits described the data
satisfactorily (P-values all greater than 0.10). CF indicates that convergence failed, likely due to the flat dose-response
curve.
bNot including dead fetuses.
Source: Anderson et al., 1998.
model, as for the Anderson et al. (1993, 1995, 1996) study. In the 1998 study, there were often,
but not always, two pregnant females per male. Because, to our knowledge, no dose-response
models exist that incorporate the added layer of nesting (litter within male) present with these
data sets, we have adopted the following procedure. When there were two litters for any
experimental unit (male mouse), one of the two was randomly chosen for inclusion in the dose-
response analysis. When there was only one litter for a particular male, that one litter was
included in the analysis.
In order to assess the sensitivity of the results to that random assignment of litters to the
dose-response analysis, the same analysis was conducted using the "leftover" litters (the litters
not randomly chosen for inclusion in the original analysis). When there was only one litter, it
was used in this alternative analysis as well as in the primary analysis. The results presented in
Table 10-17 include both sets of EC and LEG estimates, one for the primary analysis (using the
chosen litters) and one for the alternative analysis (using the leftover litters). However, when
combinations of studies were considered, only the chosen litters were used (see below).
As in the first Anderson et al. study, this study reported significant effects of treatment
with 1,3-butadiene on incidence of early deaths (Anderson et al., 1998). In this study, exposure
levels of 65 and 130 ppm resulted in increased incidence of early deaths, but 12.5 ppm did not
show a statistically significant increase in early deaths as in the first Anderson et al. (1993) study.
There were no effects at any exposure level on late deaths, nor on incidence of abnormal fetuses
10-47
-------
including runts. All the abnormal fetuses reported by Anderson et al. (1998) were runts (<75%
of the weight of littermates), and no fetuses had structural abnormalities. Analysis of individual
fetal weight data also showed no effect of exposure to 1,3-butadiene; however, data were
available from only about one-half the litters because there apparently were problems in
detecting pregnancies in a number of animals. In this analysis, fetal weight data were not
included from "unplugged" females (those in which a vaginal plug was not detected) that
eventually were found to be pregnant, nor from the "plugged" females that were considered to
have been necropsied too early, i.e., those litters in which the fetuses were found to be
underdeveloped for the presumed gestational age, because gestational age could not be accurately
determined. Data from these litters were not combined with those from pregnancies of known
gestational age, because fetuses grow rapidly during late gestation and as little as 1 day can make
a significant difference in fetal weight. The lower sample size and other experimental
considerations may explain the somewhat larger standard deviations for data from this study
compared with those from the Anderson et al. (1993) study.
In Table 10-17, ECs and LECs for implants and late deaths could not be estimated in the
primary analysis because they did not show a dose-response relationship, although there were
cases in the alternative analyses where these values could be estimated. The ECs and LECs for
early deaths and all deaths from the primary analysis were very similar to those from the
alternative analysis. The values for early deaths and all deaths were similar to each other as well,
again reflecting the fact that early deaths made up the majority of all deaths. The EC05 for both
endpoints was between the two higher exposure levels tested, and the EC10 was above the highest
exposure level. Given that the data are nested (i.e., fetuses within litters; two litters/male), and
thus are similar to the developmental toxicity data evaluated by Allen et al. (1994b), the EC05 and
LEC05 could easily be estimated. The LEC05 for both endpoints (early deaths and all deaths) was
below the highest exposure level tested and near the 65 ppm exposure level (11.6 ppm adjusted
value).
10.3.3.3. Brinkworth et al. (1998)
The third dominant lethal study, by Brinkworth et al. (1998), involved exposure of male
CD-I mice to 0, 12.5 ppm, and 125 ppm for 6 h/day, 5 days/week for 10 weeks. Mating of males
began 4 days after the end of exposure with two females/male. Data are shown in Table 10-18.
The ECs and LECs are shown in Table 10-19. The analysis was identical to that described above
for the Anderson et al. (1998) study, in that a primary and an alternative analysis were done for
the endpoints modeled. In this study, early deaths were significantly increased after exposure of
males to 125 ppm, but not after 12.5 ppm. Incidence of late deaths, dead fetuses, and abnormal
fetuses was not affected by treatment. Fetal weight also was not affected by exposure, but data
10-48
-------
Table 10-18. Male-mediated developmental toxicitya.
Exposure
level
0
12.5 ppm
125 ppm
No. of
fertile males
23
22
25
No. implants
Mean + SDb
(no. preg.
females)
576
13.09±2.604
(44)
502
13.21±1.862
(38)
602
13.09±1.532
(46)
No. early deaths
(mean + SD)C
32
(0.055±0.0691)
28
(0.055±0.0694)
59
(0.097f±0.0835)
No. late deaths
(mean + SD)C
9
(0.014±0.0365)
13
(0.028±0.0647)
14
(0.024 ±0.05 14)
No. dead fetuses
(mean + SD)C
1
(0.002±0.0099)
3
(0.006±0.0202)
4
(0.008±0.0235)
Abnormal
fetuses'1
(no. litters)
2
(2)
1
(1)
0
Fetal weight6
Mean + SD
(no. litters)
1.084±.1380
(9)
1.004±.0719
(4)
1.068±.1338
(9)
o
VO
aSpecies/strain: CD-I mice, adult males. Exposure regimen: 6 h/day, 5 days/week. Exposure time: 10 weeks. Exposure levels: 0, 12.5, 125 ppm.
bMean per litter.
°Mean per implant per litter.
^Excludes runts.
eOnly runts and a single littermate as well as one control were weighed.
Statistically significantly different from controls, p<01
Source: Brinkworth et al, 1998.
-------
Table 10-19. ECs and LECs for male-mediated developmental toxicitya.
Estimate
(ppm)
FC
-c,^05
T Fr
-L-J-A-'05
EC10
LEC10
No. implants
oo [50]
- [13]
oo [100]
oo [26]
Early deaths
26 [20]
18 [11]
31 [41]
23 [24]
Late deaths'5
42C [oo]
13C [oo]
89C [oo]
28C [oo]
All deaths
19 [22]
9.9 [11]
41 [25]
21 [23]
^Exposures were adjusted to 24-h daily exposures, e.g., (12.5) (6/24) (5/7) = 2.2 ppm. Entries with the "°°" symbol
indicate essentially flat dose-response curves, i.e. no estimates for the EC or LEG values could be derived. Values in
brackets are the estimates from the alternative analyses checking sensitivity of results.
''Not including dead fetuses.
°The fit of the log-logistic model to these data was relatively poor (/>=0.04). All other fits described the data
satisfactorily (/^-values all greater than 0.10).
Source: Brinkworth et al, 1998.
are sparse and are likely biased in this case, because only those litters with fetuses considered to
be runts (<75% or <80% the weight of the rest of the litter) were weighed, along with "matching"
control litters. This likely explains the lower mean fetal weights in this study versus those seen
in either the Anderson et al. (1993) or the Anderson et al. (1998) study. In addition, the SD was
somewhat larger for these data than those from the Anderson et al. (1993) study, probably a
reflection of the smaller sample size and other experimental considerations.
The ECs and LECs (Table 10-20) could not be calculated in the primary analysis for the
number of implants, nor in the alternative analysis for late deaths. It was possible to calculate
values for implants in the alternative analyses, but the fit of the model for late deaths was poor,
so the estimates are not used. Values in brackets are the estimates from the alternative analyses
checking sensitivity of results. All fits described the data satisfactorily (p-values all greater than
0.10). CF indicates that convergence failed, likely due to the flat dose-response curve; i.e., no
estimates for the EC or LEG values could be derived. The ECs and LECs for early deaths were
similar to those for all deaths, again reflecting the fact that early deaths make up the largest
component of all deaths. The EC05 for both was similar to the highest exposure level tested (i.e.,
22.3 ppm adjusted for continuous exposure), and, along with the LEC05, could be estimated
easily.
10-50
-------
Table 10-20. ECs and LECs for male-mediated developmental toxicity from the combined studies".
Endpoint
Implants
Early deaths
Late deaths"
All deaths0
Estimate
(ppm)
EC05
J-^^05
EC10
-LJiCjo
EC05
J-^^05
EC10
-LJiCjo
EC05
J-^^05
EC10
-LJiCjQ
EC05
J-^^05
EC10
-LJiCjQ
Anderson et
al. (1993)
89
56
178
112
55
40
116
85
CO
CO
CO
CO
53
39
112
83
Anderson et
al. (1998)
- [12]
- [7.3]
= [24]
- [15]
14 [23]
8.9 [11]
30 [49]
19 [23]
co [86]
co [29]
-[CF]
co[CF]
17 [18]
9.5 [10]
37 [38]
20 [21]
Brinkworth
et al. (1998)
co [50]
- [13]
co [100]
co [26]
26 [20]
18 [11]
31 [41]
23 [24]
42eM
13eM
89e[co]
28e[co]
19 [22]
9.9 [11]
41 [25]
21 [23]
Combined Anderson et al.
(1993) and Brinkworth et
al. (1998)"
72
50
144
100
50
37
106
77
ND
ND
ND
ND
52
38
109
80
Combined 3
studies'1
79
53
158
106
44
32
92
67
ND
ND
ND
ND
47
34
99
72
Combined 3 studies
except 1,250 ppm
group*1
—
—
—
—
24
15
50
28
—
—
—
—
15
10
31
21
a Exposures were adjusted to 24-h daily exposures, e.g., (12.5) (6/24) (5/7) = 2.2 ppm. Values in brackets are the estimates from the alternative analyses checking sensitivity of results. Entries with
the "°°" symbol indicate essentially flat dose-response curves, i.e., no estimates for the EC or LEG values could be derived. CF indicates that convergence failed, likely because of the flat dose-
response curve. ND indicates "not done" — the failure of individual studies to estimate EC and LEC values suggested that analysis of combined data would be futile. "—" indicates that the
combined analysis of data excluding the 1,250 ppm group was not done; only the analysis for early deaths and all deaths were done, since those endpoints showed the clearest dose-response
relationship.
b Not including dead fetuses.
0 Including dead fetuses.
d Only data on litters used in the primary analysis were used in combining studies for modeling.
" The fit of the log-logistic model to these data was relatively poor (p=0.04). All other fits described the data satisfactorily (p-values all greater than 0.10).
-------
10.3.3.4. Analysis of Combined Dominant Lethal Studies
As stated earlier, the three dominant lethal studies were conducted in the same laboratory,
under essentially the same conditions, and by the same investigators. The only notable
difference was the number of weeks of exposure (4 weeks in the Anderson et al. [1998] study
versus 10 weeks in the other two studies). Because Adler and Anderson (1994) showed that the
longer period of exposure did not increase the dominant lethal effects of 1,3-butadiene, the data
were combined. Table 10-20 presents the results of the analyses for the combinations of studies.
For completeness, results of analysis of the individual studies are also included.
Initially, only the data from Anderson et al. (1993) and Brinkworth et al. (1998) were
combined because they were similar in terms of duration of exposure. Ultimately, all three
studies were combined because the results indicated that the difference in duration of exposure
was not crucial to estimation of the EC and LEG values.
Note that the random assignment of litters for inclusion in the dose-response analysis of
the Anderson et al. (1998) study and the Brinkworth et al. (1998) study did not appear to affect
the results to any great degree, with the exception of the estimates for implantations. Except for
that case, the alternative estimates (in brackets in Table 10-20) that would have been obtained
had the other litters been included in the analysis are very similar to those obtained from the
primary analysis.
For implantations, the choice of primary or alternative data subsets did have a relatively
large impact, because for the primary subset no downward trend in number of implants was
detected, whereas for the alternative subset such a downward trend was in evidence. The
apparent sensitivity of the implantation EC and LEG estimates to the choice of the data subsets
suggests the need for additional investigation of this endpoint (including, perhaps, examination
of whether 5% or 10% reductions in number of implantations are suitable reference points—
response levels—for defining RfCs).
Both the Anderson et al. (1998) and the Brinkworth et al. (1998) studies, which had
similar maximum exposure levels of 130 and 125 ppm, respectively, estimated very similar EC
and LEG values, especially for all deaths. In fact, the EC and LEG estimates from both of those
studies were less than the corresponding estimates from Anderson et al. (1993). The reason for
this is as follows. In Anderson et al. (1998) and Brinkworth et al. (1998), the maximal observed
increases in mean response, for all deaths as an example, were 0.058 and 0.066, respectively.
Consistent with those observations, the models estimated EC05s of slightly less than 22 or 23
ppm (the continuous exposure equivalents of 125 and 130 ppm, respectively). Given the rates of
change in response observed in these two studies, the projections out to 1,250 ppm (the
maximum experimental concentration in Anderson et al., 1993) would overestimate the response
observed there. In fact, the observed average increase for the high-exposure group from the first
10-52
-------
study was about 0.16; the projections from the two 1998 studies for the concentration
corresponding to a 0.16 increase in response are about 350 ppm. When all the data were
combined for modeling, the rate of increase of the dose-response curve was somewhat less and
the fit at the high dose was better. However, the fit at the lower doses was not as good. Because
we are most concerned with fitting the data in the range of the EC and LEG, it was decided to run
the model for all three studies combined without the highest exposure level (1,250 ppm). When
this was done, the ECs and LECs for early deaths and all deaths from the combined studies were
very similar to those for the Anderson et al. (1998) and the Brinkworth et al. (1998) studies.
Figure 10-7 shows the fit of the data for all prenatal deaths.
Combined analysis of the three studies provides many more data points suitable for dose-
response analysis and enhances the credibility of the estimates precisely because of the general
agreement observed among the results. We believe that the most suitable combination for
estimating the EC05 and LEC05 is without the 1,250 ppm exposure level from the Anderson et al.
(1993) study because this data point results in a higher estimate and a visual fit of the model to
the data that is not as good at the EC05. Therefore, a suitable basis for derivation of an RfC for
1,3-butadiene based on these dominant lethal results would be the value of 10 ppm (adjusted for
continuous exposure) obtained when all of the studies are combined except for the 1,250 ppm
exposure level and "all deaths" is the endpoint used (see Table 10-20). This value, while being
the most health-protective, is consistent with values derived with other endpoints ("early deaths"
with an LEC05 of 15 ppm), and with the estimates from either the Anderson et al. (1998) study or
the Brinkworth et al. (1998) study estimates for early deaths or all deaths.
10.3.4. Chronic Studies
In the third type of study, reproductive effects of 1,3-butadiene were seen in lifetime
studies in mice after chronic inhalation exposure to 6.25, 20, 62.5, 200, and 625 ppm for 6 h/day,
5 days/week (NTP, 1993). The lowest exposure level studied in mice (6.25 ppm) showed
increased ovarian atrophy and was considered a LOAEL (Table 10-21). Minimal data from
studies on rats suggested their lesser sensitivity to chronic exposure than for mice in that effects
on fertility were noted only at high exposure levels (600 ppm and above).
In conclusion, each of these three types of studies indicated the potential of 1,3-butadiene
to affect reproduction and development in mice at low levels of exposure.
The quantal Weibull model was used initially to model all data. In cases where this
model did not provide a good fit of the data, a log-logistic model was used. The 15-month and
chronic ovarian atrophy data could not be fit adequately using the quantal Weibull model. A log-
logistic model similar to that used for fetal weight (setting OjS and 02s to zero) was found to fit
the data well. The model was run to determine the probability of additional risk and extra risk.
10-53
-------
0.17-
0.15-
D)
^
i
$
cc
0.13-
0.11 -
0.09-
0.07-
0.05
0.12
• Mean Observed Response Rates
Model Predictions
0.00
5.00 10.00 15.00 20.00 25.00
Human Equivalent, Continuous Concentrations (ppm)
30.00
0.08
- 0.06
LEC10=21.3 EC10=31.2
- 0.04
0.02
-0.02
35.00
x
LU
Figure 10-7. Log-logistic model for data on all prenatal deaths from the three dominant lethal studies, excluding the 1,250
ppm exposure group, with data points and standard error bars".
aThe EC05, LEC05, EC10, and LEC10 shown correspond to those in the last column of Table 10-20.
-------
Table 10-21. NTP chronic study (1993)a.
Incidence Data—Ovarian Atrophy
Exposure level
0
6.25 ppm
20.00 ppm
62.50 ppm
200.00 ppm
625.00 ppm
Ovarian atrophy — 9 mo
No. examined
10
—
—
10
10
8
% affected
0
—
—
0
90
100
Ovarian atrophy — 15 mo
No. examined
10
10
10
10
10
2
% affected
0
0
10
90
70
100
Ovarian atrophy — 2 years
No. examined
49
49
48
50
50
79
% affected
8.16
38.78
66.67
84.00
86.00
87.34
Ovarian Atrophy — Lesion Distribution
Number (%)
Exposure level
0.00
6.25 ppm
20.00 ppm
62.50 ppm
200.00 ppm
625.00 ppm
Ovarian atrophy — 9 mo
Minimal
0
—
—
0
0
0
Mild
0
—
—
0
0
0
Moderate
0
—
—
0
9(90)
8 (100)
Ovarian atrophy — 15 mo
Minimal
0
0
1(10)
1(10)
0
0
Mild
0
0
0
7(70)
1(10)
0
Moderate
0
0
0
1(10)
6(60)
2 (100)
Ovarian atrophy — 2 years
Minimal
1(2)
0
1(2)
3(6)
0
0
Mild
2(4)
15(31)
23 (48)
18 (36)
9(18)
19 (24)
Moderate
1(2)
4(8)
8(17)
21 (42)
34 (18)
47 (59)
Marked
0
0
0
0
0
3(4)
-------
Table 10-21. NTP chronic study (1993) (continued).
Incidence Data—Uterine Atrophy
Exposure level
0
6.25 ppm
20ppm
62.5 ppm
200 ppm
625 ppm
Uterine atrophy — 9 mo
No. examined
10
—
—
10
10
8
No. (%) Affected
0
—
—
0
3(30)
6(75)
Uterine atrophy — 15 mo
No. examined
10
1
10
10
10
2
No. (%) Affected
0
0
0
0
0
2 (100)
Uterine atrophy — 2 years
No. examined
50
49
50
49
50
78
No. (%) Affected
1(2)
0
1(2)
1(2)
8(16)
41 (53)
Incidence Data—Testicular Atrophy
Exposure level
0
6.25 ppm
20 ppm
62.5 ppm
200 ppm
625 ppm
Testicular atrophy — 9 mo
No. examined
10
—
—
—
10
10
No. (%) Affected
0
—
—
—
0
6(60)
Testicular atrophy — 15 mo
No. examined
10
—
1
—
10
7
No. (%) Affected
0
—
0
—
0
4(57)
Testicular atrophy — 2 years
No. examined
50
50
50
48
49
72
No. (%) Affected
1(2)
3(6)
4(8)
2(4)
6(12)
53 (74)
aSpecies/strain: Male and female B6C3F[ mice. Exposure regimen: 6 h/day, 5 days/week for 2 years. Exposure levels: 0, 6.25, 20, 62.5, 200, or 625 ppm.
-------
Goodness of fit was determined by a %2 test. The model was considered to give a good fit if the
p value was greater than 0.05 and a graphical display of the data showed a good fit of the model.
An attempt was made to model various levels of severity in the lesions seen, based on the
data shown in Table 10-21. The data for moderate lesions were fit using the quantal Weibull
model (Allen et al., 1994b) for dichotomous data. This model can be expressed as:
P(d) = 1 - exp[- (a + 0dY)],
where P(d) is the probability of response at exposure level d and a, P, and y are parameters
estimated from the observed dose-response data. Parameter constraints were a > 0; P > 0; y >
0. The model was run to determine the probability of additional risk. Goodness of fit was
determined by a %2 test. The model was considered to provide an acceptable fit if the/? value
was greater than 0.05 and a graphical display of the data showed a good fit of the model.
Table 10-22 gives the results of fitting the log-logistic model to the 2-year ovarian
atrophy data for exposure groups 1-5 and 1-4. The model gave a poor fit for all six exposure
groups, because of leveling off of the response at exposures above 62.5 ppm (36 ppm adjusted
for continuous exposure). The best fit of the model was for exposure groups 1-4, although the
model also fit exposure groups 1-5 well (Figure 10-8; exposures adjusted for continuous
exposure), and the EC10s and LEC10s obtained for groups 1-4 and 1-5 were similar. As expected,
LEC10s were lowest for ovarian atrophy at 2 years. Moderate ovarian atrophy at 2 years also was
modeled using the quantal Weibull model with exposure groups 1-5 or 1-4. The EC10 and LEC10
were higher than those for all lesions. Ovarian atrophy data for all six exposure groups at 9 and
15 months were fit using the quantal Weibull or log-logistic model.
Uterine and testicular atrophy data also were modeled using the quantal Weibull model.
The quantal Weibull model resulted in an acceptable fit of the 2-year uterine atrophy and
testicular atrophy data (Table 10-22 and Figures 10-9 and 10-10; exposures adjusted for
continuous exposure). However, the EC10s and LEC10s were much higher for these endpoints
than for 9-month, 15-month, or 2-year ovarian atrophy data. LEC10s were estimated because it
has been shown that, for quantal responses, the LEC10 is near or below the range of detectable
responses (Allen et al. 1994b). Also, the Proposed Guidelines for Carcinogen Risk Assessment
propose use of an LEC10 as the default point of departure for low-dose extrapolation, and use of
an LEC10 as a default for noncancer estimation of an RfC would be consistent with this approach
(U.S. EPA, 1996, 1999).
10-57
-------
Table 10-22. ECs and LECs for ovarian, uterine, and testicular atrophy using the
quantal Weibull and log-logistic models"
Endpoint
Ovarian atrophy —
2 years
Ovarian atrophy —
2 year
Moderate lesions only
Ovarian atrophy —
15 mos
Ovarian atrophy —
9 mos
Uterine atrophy
Testicular atrophy
Model
Log-logistic (l-5)b
Log-logistic (1-4)
Quantal Weibull (1-
5)
Quantal Weibull (1-
4)
Log-logistic (1-6)
Quantal Weibull (1-
6)
Quantal Weibull (1-
6)
Quantal Weibull (1-
6)
NOAEL/LOAEL
l.lppm(LOAEL)
1.1 ppm
1.1 ppm
11 ppm
11 ppm
36 ppm
EC10
(ppm)
0.32
0.29C
0.27
0.24C
3.02
2.31
2.10
20.04
29.37
40.59
LEC10
(ppm)
0.22
0.21C
0.18
0.17C
2.35
1.67
0.72
9.95
18.43
25.64
P-
Value
0.11
0.96
0.55
0.96
0.66
0.83
0.66
0.55
^Exposures were adjusted for continuous exposure, e.g., (6.25) (6/24) (5/7) = 1.1
bExposure levels included in the model.
°Extra risk. All other values are estimates of additional risk.
ppm.
10-58
-------
100
Q
UJ
UJ
u.
u.
<
UJ
o
QC
UJ
Q.
0
Observed
Predicted
50
200
100 150
EXPOSURE (ppm)
Figure 10-8. Ovarian atrophy (groups 1-5) using log-logistic model.
250
l_l
E
u
u.
u.
UJ
UJ
Q.
40 60 80
EXPOSURE (ppm)
Figure 10-9. Uterine atrophy (groups 1-5) using quantal Weibull model.
Observed
Predicted
10-59
-------
Q
UJ
UJ
UJ
UJ
Q.
Observed
Predicted
20
40 60 80
EXPOSURE (ppm)
100
120
Figure 10-10. Testicular atrophy (groups 1-6) using quantal Weibull model.
Although some 9- and 12-month interim sacrifice data were available for ovarian, uterine,
and testicular atrophy (Table 10-21), these were less than ideal for modeling because smaller
numbers of animals were killed and not all dose groups were represented. In addition, some
animals died or became moribund and were killed before the 2-year death time point (see
Chapter 6). To include the interim data and to account for the variability in time of death, time-
to-response analyses were done using the multistage Weibull model as discussed in Section
10.2.2.2. Exposures were adjusted to the equivalent continuous lifetime exposures. An EC10 and
an LEC10 were calculated in each case. All the reproductive responses were treated as incidental,
not fatal. Parameter estimates for each reproductive endpoint are presented in Table 10-23.
Results of the Weibull time-to-response model are shown in Table 10-24. The ECs and
LECs were similar to those from other models used for ovarian atrophy, uterine atrophy, and
testicular atrophy, with the exception of those from the modeling of testicular atrophy including
the highest exposure group, for which the Weibull time-to-response model yields results roughly
five times lower than the quantal Weibull model. The quantal Weibull model results for uterine
and testicular atrophy were for additional risk, whereas the Weibull time-to-response results were
for extra risk; however, because of the low background rates of both uterine and testicular
atrophy, additional risk and extra risk should be nearly the same. The results of the time-to-
response model are used in the calculation of potential RfCs.
The time-to-response model also allows for calculation of extra risks at ages less than full
lifetime. For uterine and testicular atrophy, full 70-year lifetimes were used. For ovarian
10-60
-------
Table 10-23. Parameters for Weibull time-to-response model used to model
reproductive effects observed in mice based on ppm butadiene exposure*
Response
Ovarian atrophy
Uterine atrophy
Testicular
atrophy
625 ppm
group
included
no
yes
no
yes
no
yes
QO
4.86 x lO'6
9.01 x ID'7
6.73 x 10'5
9.08 x lO'5
4.28 x 10'4
1.60 x lO'4
Qi
7.06 x 10'6
1.32 x lO'6
5.28 x 10'5
9.74 x lO'6
2.24 x 10'5
1.52 x lO'4
Q2
—
—
—
1.31 x lO'6
—
—
Z
2.21
2.58
1.0
1.0
1.0
1.0
aEach response was considered to be incidental with induction time, T0=0. See Section 10.2.2.2 on time-to-tumor
modeling of the mouse carcinogenicity data for a discussion of the Weibull model structure and selection.
Table 10-24. Human benchmark 1,3-butadiene exposure concentrations
calculated for reproductive effects observed in mice using a Weibull
time-to-response model (extra risk).
Response
Ovarian
atrophyb
Uterine
atrophy
Testicular
atrophy
625 ppm group
included3
No
Yes
No
Yes
No
Yes
Based on ppm continuous butadiene exposure
EC10
1.05
1.13
18.8
24.0
44.3
6.54
LEC10
0.878
0.958
12.0
15.6
15.9
5.39
a Because of the high early mortality in the 625 ppm exposure group, results from the models excluding this group will be
used.
bResults for ovarian atrophy reflect extra risks only until 50 years of age (see text).
atrophy, however, extra risks were calculated for 50 years to reflect only the time before average
age at menopause when follicles are no longer present and available for ovulation, because in the
mouse studies of ovarian atrophy, the atrophy occurs as a result of follicular failure (see Chapter
5). For comparison purposes, calculations for the full 70 years yield EC10 = 0.50 ppm and LEC10
= 0.38 ppm for ovarian atrophy (625 ppm dose group excluded). Prenatal exposures may also
10-61
-------
affect ovarian atrophy; however, there are no specific data for such exposures and they were not
included explicitly in the analysis.
10.3.5. Summary and Conclusions
ECs and LECs were estimated for three types of exposure scenarios to 1,3-butadiene
based on different endpoints:
1. Short-term exposure (10 days)—fetal weight reduction
2. Subchronic exposure (10 weeks)—male-mediated developmental toxicity
3. Chronic exposure—ovarian, uterine, and testicular atrophy
These analyses demonstrate approaches for estimation of ECs and LECs based on continuous
and quantal data.
Results of the fetal weight analysis illustrate how both continuous and quantal modeling
approaches can be used for continuous data. All of the LECs calculated were below the LOAEL
of 40 ppm (i.e., 10 ppm adjusted for continuous exposure), except for two LECs calculated using
the continuous power model, which were near this value. Since the log-logistic modeling
approach is considered the preferred method for modeling continuous data, the LEC05 of 2.9 ppm
from this model will be used for calculating the reference concentration for acute exposure.
Results of the combined analysis of subchronic studies of male-mediated developmental
toxicity (dominant lethal effects) gave an EC05 of 15 ppm for all deaths and an LEC05 of 10 ppm.
However, since the value for fetal weight is lower, it will also be used as the basis for the
subchronic RfC.
Note that the LEC05 is selected as the appropriate "point of departure" for the dominant
lethal effects and fetal weight reductions, whereas, immediately below, the LEC10 is used for
ovarian atrophy. This is because the dominant lethal and fetal weight effects occur within litters.
Studies with data nested within litters typically have sufficient power to observe levels of
response of 5%. [Fetal weight is actually a continuous measure; however, for the log-logistic
model recommended here, the data were dichotomized and the LEC05 corresponds with a 5%
additional risk of obtaining a fetal weight below the 5th percentile of the controls.] The ovarian
atrophy data are from a cancer bioassay study, which typically has the power to detect 10%
response levels. The "point of departure" is selected to be at the low end of the observable range;
thus, a 5% response level is used for the nested data, and a 10% level for the data from the cancer
bioassay (see also Section 10.4.2 below).
Modeling of the 2-year ovarian atrophy data, the effect occurring at the lowest chronic
exposure level, gave a good fit with the log-logistic model, but only when the highest exposure
10-62
-------
level was dropped. This approach was justified because the responses leveled off for the top
three exposure groups. The LECs derived for a 10% increase in additional risk or extra risk were
five- to sixfold below the LOAEL of 6.25 ppm (i.e., 1.1 ppm adjusted for continuous exposure).
When the time-to-response model was applied to account for interim sacrifice data and early
mortality, an LEC10 of 0.88 ppm (extra risk, up to age 50) was calculated, a value slightly higher
than that using the log-logistic model. For the chronic effects of uterine and testicular atrophy,
higher LEC10s were obtained: 12 and 16 ppm, respectively.
There is strong evidence suggesting that the diepoxide metabolite (DEB) is responsible
for 1,3-butadiene-induced ovarian atrophy (Doerr et al., 1996). Uterine atrophy may be
secondary to ovarian atrophy, and thus may also be related to DEB levels. However, an adequate
pharmacokinetic model was not available to estimate levels of DEB for dose-response modeling
(Chapter 9).
Chronic RfC calculations will be made for both ovarian atrophy, the reproductive effect
occurring at the lowest chronic exposure level, and testicular atrophy, the reproductive effect
observed in male mice following chronic exposure.
10.4. REFERENCE CONCENTRATIONS FOR REPRODUCTIVE AND
DEVELOPMENTAL EFFECTS
10.4.1. Introduction
As discussed in Chapter 5 and Section 10.3, a variety of reproductive and developmental
effects have been observed in mice exposed to 1,3-butadiene by inhalation. (No human
reproductive or developmental data are available for 1,3-butadiene.) Some of these reproductive
and developmental effects represent the most sensitive noncancer endpoints that have been
observed in available toxicity studies. In this section, sample chronic reference concentrations
(RfCs) are calculated for the most sensitive reproductive and developmental endpoints, that is,
those effects exhibiting responses at the lowest exposure concentrations from various exposure
scenarios, using both the traditional NOAEL/LOAEL approach and the "benchmark dose"
approach (Crump, 1984). A reference concentration (or dose) is an estimate of a daily exposure
to humans that is "likely to be without an appreciable risk of deleterious [noncancer] effects
during a lifetime" (Barnes et al., 1988). The final reported chronic RfC will be based on the
endpoint resulting in the lowest calculated RfC level. An RfC for acute exposure and an RfC for
subchronic exposure will also be presented.
10.4.2. Calculation of RfCs
The most sensitive short-term developmental effect was decreased fetal weight in a
prenatal developmental toxicity study in the mouse. The most sensitive reproductive effect
10-63
-------
observed in subchronic exposure studies was increased prenatal deaths ("all deaths") in dominant
lethal studies of mice (i.e., male mice were exposed to 1,3-butadiene and effects on litters were
measured after mating to unexposed females). Dominant lethal effects in humans would likely
be manifested as infertility (due to reduced fertility or very early deaths) or spontaneous
abortions. From chronic exposure studies (2-year bioassays), the most sensitive reproductive
endpoints were ovarian atrophy in female mice and testicular atrophy in male mice.
Table 10-25 summarizes the EC10, and in some cases the EC05 (i.e., the exposure
concentration resulting in a 10% [or 5%] increase in risk based on modeling the exposure-
response data in the observable range), the LEC10, and in some cases the LEC05 (i.e., the 95%
lower confidence limit on the exposure concentration estimated to result in a 10% [or 5%]
increase in risk), and the NOAEL (i.e., no observed adverse effect level) or LOAEL (i.e., lowest
observed adverse effect level; reported when no NOAEL was observed) for these 1,3-butadiene-
induced effects. Table 10-25 also provides sample calculations of RfCs using the NOAEL (or
LOAEL), as well as the LEC10 (or LEC05), as "points of departure." Uncertainty factors are
applied to the "points of departure" to calculate the sample RfCs.
A 10% response level is the standard "benchmark response" for dichotomous data (or
continuous data that have been dichotomized for analysis), and the corresponding LEC10 is the
standard point of departure for such data. This response level is chosen because it is typically at
the low end of the range of observation for increased responses in studies of experimental
animals. However, in certain cases, other response levels may be more appropriate. For
example, studies of developmental effects often have more power to detect effects at lower
response levels because of their nested (fetuses within litters) study design, and a response level
of 5% is typically within the range of observation (Allen et al., 1994b). Therefore, the LEC05 is
used here for the point of departure for decreased fetal weight. Similarly, the dominant lethal
data have a nested design, and an LEC05 is used as the point of departure for this endpoint as
well.
For interspecies uncertainty, a factor of 10 is generally used when the "point of departure"
is based on nonhuman data. However, when ppm equivalence across species is assumed for
inhalation studies, as was done here, a factor of 3 is used instead. (Note that the assumption of
ppm equivalence is identical to using EPA's inhalation dosimetry methodology with RGDR,.=1
[U.S. EPA, 1994c]). This convention assumes that pharmacokinetic differences have been
accounted for but not pharmacodynamic differences. Thus, in Table 10-25 an interspecies
uncertainty factor of 3 was used for all endpoints. For ovarian atrophy, evidence is convincing
that the diepoxide metabolite (l,2:3,4-diepoxybutane, DEB) is required to elicit the effect (see
Chapter 5), and it is expected, based on pharmacokinetic data, that humans produce less DEB
than mice (see Chapter 3). However, DEB levels cannot be quantified without an adequate
10-64
-------
Table 10-25. Points of departure and sample chronic RfC calculations for reproductive and developmental effects of
1,3-butadiene.
Effect
Decreased
fetal
weight
All
prenatal
deaths
(dominant
lethal
effect)
Ovarian
atrophy
Testicular
atrophy
NOAEI/
(or LOAEL)
(ppm)
10 (LOAEL)
(GD6-15)
2.2°
(10 week)
1.1 (LOAEL)
(2 year)
36
(2 year)
EC10"
(ppm)
EC05=6.9
(EC10=10)
EC05=15
(EC10=31)
1.0s
44
LEC10"
(ppm)
LEC05=2.9
(LEC10=4.7)
LEC05=10
(LEC10=21)
0.888
16
Interspecies
uncertainty
factor
3
3
3
3
3
3
3
3
Intraspecies
uncertainty
factor
10
10
10
10
10
10
10
10
Acute/
subchronic-
to-chronic
uncertainty
factor
ld
ld
lf
lf
1
1
1
1
LOAEL-to-
NOAEL
uncertainty
factor
10
1
10
1
Effect level
extrapolation
factor"
4
5
10
10
Modifying
factor for
incomplete
database1
3
3
3
3
3
3
3
3
RfC based
on NOAEL
(ppm)
0.01
0.02
0.001
0.4
RfC based
on LEC10
(or LEC05)
(ppm)
0.007
0.02
0.0009
0.02
a Adjusted to 24-h daily exposure for the exposure period. Ppm equivalence across species was assumed; this is the same as using U.S. EPA's inhalation dosimetry
methodology withRGDR^l (U.S. EPA, 1994c).
b For benchmark concentrations - to decrease risk to below the benchmark response level; analogous conceptually to the LOAEL-to-NOAEL uncertainty factor.
0 Reflects primarily the absence of a multigenerational reproductive study and neurodevelopmental studies.
d Although from acute study, only exposure during gestation is assumed to be relevant to fetal weight.
e Although this exposure level was a LOAEL in one study, it was a NOAEL in the other two, and the combined analysis of the three studies did not find the increase in
prenatal deaths at this level to be statistically significant.
f Although from subchronic studies, exposure appears to affect only post-meiotic stages of spermatogenesis, thus dominant lethal effects are not expected to be
cumulative with exposure.
8Extra risk has been calculated for a 50-year exposure up to the time of menopause, after which ovarian atrophy from follicular loss is not expected to be an adverse
health effect for humans.
-------
physiologically based pharmacokinetic (PBPK) model, thus default dosimetry (i.e., 1,3-butadiene
exposure concentration) was used for dose-response modeling and the default value of 1 for the
pharmacokinetic portion of the interspecies uncertainty factor was retained.
A large degree of human variability has been observed in metabolic activities that could
affect 1,3-butadiene toxicity. For example, Seaton et al. (1995) measured a 60-fold variation in
the initial rate of oxidation of l,2-epoxy-3-butene (EB) to DEB in microsomes from 10 different
human livers. However, overall variability in total metabolism and susceptibility is unknown,
thus the conventional intraspecies uncertainty factor of 10 for human variability was used for
each endpoint in Table 10-25.
With respect to the acute/subchronic-to-chronic uncertainty factor, none was needed for
ovarian or testicular atrophy because these effects were based on chronic studies. No acute-to-
chronic uncertainty factor was used for fetal weight either, because only exposures during
gestation are relevant. Similarly, the dominant lethal effects appear to occur with exposure
during a specific time period of spermatogenesis (i.e., only postmeiotic stages of developing
sperm appear susceptible), and results after a 4-week exposure were similar to those from a 10-
week exposure. Thus, even though these effects were observed in subchronic studies, no
subchronic-to-chronic uncertainty factor is used for the dominant lethal effects either, since the
effects are not expected to be cumulative with exposure.
Under the NOAEL/LOAEL approach, the NOAEL is defined as the exposure level for
which there is no observed adverse effect, although it is circumscribed by the detection limit of
the study. For endpoints for which there is no NOAEL, an uncertainty factor of 10 is typically
used to attempt to extrapolate from the LOAEL to a level at which there are presumed to be no
detectable effects. In the benchmark dose approach, the "points of departure" used here
correspond to 5% and 10% increased response levels, which are explicitly not no-effect levels,
and an analogous uncertainty factor is needed. EPA is planning to develop guidance for
calculating an "effect level extrapolation factor" to be used as an uncertainly factor to decrease
the risk below the effect level at the point of departure. Pending final guidance on this
uncertainty factor, in this risk assessment we use a formula that accounts for the slope of the
dose-response curve and generates a factor between 3 and 10 for use with 10% response levels
and between 3 and 5 for a 5% response level. The formula is as follows: uncertainty factor = x x
(slope of the line from the ECX to 0)/(slope of the dose-response curve at the ECX), where x% is
the response level. Results of the formula are confined within a minimum value of 3 and a
maximum value of x.
A modifying factor of 3 was applied to all the sample calculations to reflect the fact that
the database for calculating the RfCs is incomplete. In particular, a multigeneration reproductive
study and neurodevelopmental studies are lacking.
10-66
-------
The sample chronic RfCs in Table 10-25 suggest that ovarian atrophy is the most sensitive
reproductive/developmental endpoint (i.e., the "critical" effect), and thus, should be the basis for
the final chronic RfC. Based on the critical effect of ovarian atrophy and the benchmark dose
methodology, a final chronic RfC of 0.0009 ppm, or 0.9 ppb, is adopted.
In addition to the chronic RfC, two "RfC's were calculated for acute and subchronic
exposures. (EPA is currently developing guidance for deriving reference values for acute and
subchronic exposure scenarios. These values are not technically RfCs, because RfCs are by
definition for chronic exposure scenarios.) An "RfC" for acute exposure of 0.007 ppm, or 7 ppb,
was calculated for decreased fetal weight, based on the benchmark dose methodology and using
the same factors as those depicted in Table 10-25. This acute "RfC" is identical to the sample
chronic RfC calculated for decreased fetal weight because no subchronic-to-chronic uncertainty
factor was used in that calculation. A multigenerational reproductive study would not be
considered essential for a complete database for an acute reference value; however, an
uncertainty factor of 3 for incomplete database is still warranted because of the absence of
neurodevelopmental studies. Effects observed in such studies may result from acute exposures
during a sensitive period of development. Finally, an "RfC" for subchronic exposure was
calculated. As shown in Table 10-25, the sample chronic RfC for the prenatal death endpoint
from the subchronic dominant lethal studies, using the benchmark dose methodology, was 0.02
ppm. Because no subchronic-to-chronic uncertainty factor was used here either, this value is can
be considered a sample "RfC" for subchronic exposure, without adjustment. However, this value
is higher than the acute "RfC" of 7 ppb from the fetal weight data. Therefore, 7 ppb is used for
the "RfC" for subchronic exposure as well.
10.4.3. Discussion
There are substantial uncertainties in estimating low-exposure human risks for noncancer
effects observed in animals exposed to high concentrations of an agent. It is generally believed
that there is a sublinear low-dose exposure-response relationship for noncancer effects, and
perhaps a threshold, although this is difficult to demonstrate empirically. The shape of this low-
dose exposure-response relationship is unclear, however, and thus RfCs are calculated for most
noncancer effects rather than exposure-based risk estimates. Uncertainties related to the shape of
the exposure-response curve still pertain to RfCs, however. It is anticipated that by using
NOAELs or LECs coupled with an effect level extrapolation factor, as well as the other
uncertainty factors, the ultimate chronic RfC represents an exposure level "likely to be without
appreciable risk of deleterious effects during a lifetime." Note, however, that in situations in
which most of the uncertainty factors do not apply, e.g., if the chronic RfC is based on good-
quality human data from a diverse population, the resulting value may not actually be far enough
10-67
-------
down the sublinear part of the exposure-response curve to ensure that it is "likely to be without
appreciable risk."
Major uncertainties considered in deriving a chronic RfC include extrapolation of effects
observed in laboratory animals to humans (interspecies extrapolation), potential existence of
sensitive human subpopulations resulting from human (intraspecies) variability, extrapolation of
effects observed from acute or subchronic exposures to chronic exposure scenarios, and various
deficiencies in the database. These areas of uncertainty are addressed to some extent by the
uncertainty factors. Other methodological uncertainties arise in determination of the "point of
departure" and in selection of the relevant exposure metric for equating experimental animal
exposure-response relationships to humans.
There are a number of limitations in using the NOAEL/LOAEL approach for obtaining a
"point of departure"; these have inspired development of an alternative "benchmark dose" (or
concentration) methodology. First, the NOAEL/LOAEL approach relies on one exposure level
and ignores the rest of the exposure-response data. Second, the NOAEL/LOAELs depend
explicitly on the specific exposure levels selected for the study. They are also a function of study
power because a LOAEL is the lowest exposure level with a statistically significant increase in
an adverse effect, whereas a NOAEL could represent an increase that lies below the study's limit
of detection. Finally, NOAEL/LOAELs are not readily comparable across endpoints or studies
because they can refer to different response levels.
The alternative benchmark concentration approach involves modeling the full exposure-
response curve in the observable range and calculating an effective concentration (EC)
corresponding to some level of response (e.g., 10%) that can be used as a point of comparison
across endpoints and studies (the 10% effect level is typically at the low end of the observable
range, although sometimes a lower level of response can be estimated). The 95% lower bound
on the EC is used as the "point of departure" to take into account statistical variability around the
EC estimate. While the benchmark concentration approach alleviates some of the limitations of
the NOAEL/LOAEL approach, there are still uncertainties regarding the appropriate exposure-
response model to use. It is generally expected that models that provide a good fit to the data in
the observable range should yield reasonably similar ECxs, as shown for quantal models by Allen
etal. (1994b).
As shown in Table 10-25, these two approaches yielded nearly identical sample RfCs for
decreased fetal weight, prenatal deaths (dominant lethal effect), and ovarian atrophy. For
testicular atrophy, the NOAEL-based sample RfC is about 20 times higher than the LEC10-based
sample RfC. At least part of this discrepancy is likely attributable to the fact that the time-to-
response modeling conducted to derive the LEC10 took into account the decreased survival times
in the higher exposure groups in the chronic study, thus increasing the effective percent with
10-68
-------
testicular atrophy in the midrange of the exposure-response curve, which otherwise is fairly flat.
This assessment adopts the use of the benchmark dose/concentration approach as the superior
methodology in general.
Uncertainties also exist in the choice of exposure metric. Ideally, NOAELs or LOAELs
and LECxs should be converted to appropriate human equivalent exposures before using these
exposure levels as "points of departure." Theoretically, this is best accomplished by using a
PBPK model to convert animal exposures to biologically effective doses to the target organ and
then to convert these tissue concentrations back to human exposures to the parent compound.
Unfortunately, current PBPK data and models are inadequate for use in risk assessment;
therefore, exposure concentrations of 1,3-butadiene are used as the default exposure metric (this
risk assessment assumes equivalence of effects from equivalent ppm exposures across species).
For the lifetime chronic exposure study, demonstrating ovarian and testicular atrophy, mouse
exposure concentrations were adjusted to human equivalent continuous chronic exposures.
For the subchronic and acute studies, however, the appropriate timeframe for exposure
averaging is less clear. In this assessment, daily exposures have been adjusted to an equivalent
24-h exposure for the exposure period (e.g., exposure to 10 ppm, 6 h/day, 5 days/week, would be
adjusted to 10 ppm (6/24)(5/7) =1.8 ppm continuous daily exposure). However, the exposure
concentrations for these subchronic and acute effects have not been adjusted to reflect total
duration of exposure because the critical timeframes are unknown. Thus, for example, a
1-day exposure is treated equivalently to a 10-week exposure to the same daily level.
10.4.4. Conclusions
A chronic RfC of 0.9 ppb was adopted for the critical endpoint of ovarian atrophy, based
on mouse data. This reference concentration, the uncertainties discussed above notwithstanding,
is presumed to represent a daily exposure to humans that is likely to be without an appreciable
risk of deleterious noncancer effects during a lifetime. In addition, an "RfC" for acute exposure
of 7 ppb and an "RfC" for subchronic exposure of 7 ppb were calculated based on fetal weight
data for mice. Each of these RfCs was obtained using benchmark concentration methodology.
10.5. CONCLUSIONS ON QUANTITATIVE RISK ESTIMATES
In this chapter, a lifetime extra unit cancer risk of 0.027 per ppm of continuous 1,3-
butadiene exposure was calculated based on linear modeling and extrapolation of the excess
leukemia mortality reported in a high-quality occupational epidemiology study of male workers.
Applying the same linear relative rate model and actuarial program to background leukemia
incidence rates resulted in an excess cancer unit risk estimate of 0.04 per ppm for leukemia
incidence. However, results in experimental animal studies suggest that 1,3-butadiene is a
10-69
-------
multisite carcinogen and that females are more sensitive to 1,3-butadiene-induced
carcinogenicity than males; therefore, the leukemia results based on male workers may
underestimate the total cancer risk to the general population, in particular females. Thus, an
adjustment factor of 2 was used, yielding a lifetime extra cancer unit risk estimate of 0.08/ppm.
Using this cancer potency estimate, the chronic exposure level resulting in an increased cancer
risk of 10"6 can be estimated as follows:
(I(r6)/(0.08/ppm) = i x lO'5 ppm = 0.01 ppb.
A range of human cancer potency estimates from 4 x 10"3/ppm to 0.29/ppm was also
calculated based on a variety of tumors observed in mice and rats exposed to 1,3-butadiene.
These risk estimates are considered inferior to those based on the epidemiologic data, primarily
because of the large uncertainties in extrapolating 1,3-butadiene cancer risks across species in
light of the large unexplained differences in responses of rats and mice.
In addition, benchmark concentrations and reference concentrations were calculated for an
assortment of reproductive and developmental effects observed in mice exposed to 1,3-butadiene.
These effects represent the most sensitive noncancer endpoints that have been observed in
available toxicity studies. A chronic RfC of 0.9ppb was adopted for the critical effect of
ovarian atrophy observed in mice, using a benchmark concentration approach to obtain the "point
of departure." The RfC is presumed to be a chronic exposure level without "appreciable risk" of
deleterious noncancer effects. This chronic RfC of 0.9 ppb is about 90 times higher than the
chronic exposure level estimated for a 10~6 cancer risk.
Finally, an "RfC"for acute exposure of 7ppb and an "RfC"for subchronic exposure of
7ppb were calculated from mouse fetal weight data, each using benchmark concentration
methodology.
10-70
-------
11. DOSE-RESPONSE CHARACTERIZATION
Cancer potency estimates of lifetime extra cancer risk in humans were derived from both
human and animal data. In addition, after benchmark dose modeling of a variety of reproductive
and developmental effects, which are the most sensitive noncancer endpoints that have been
observed in available toxicity studies, a chronic reference concentration (RfC) was calculated
based on the critical effect of ovarian atrophy, observed in female mice. "RfC"s for acute and
subchronic exposures were also calculated, based on reduced fetal weight in mice.
Several physiologically based pharmacokinetic (PBPK) models have been developed to
attempt to explain the interspecies differences in the potency and site specificity of the
carcinogenic response between mice and rats and to provide a dosimetric basis for quantitatively
extrapolating carcinogenic (or reproductive) risks from rodents to humans. While progress has
been made in recent years, our understanding of the bases for the species differences in 1,3-
butadiene carcinogenicity is still inadequate, and major uncertainties in the existing PBPK
models preclude their use in human risk assessment at this time (see Chapter 9).
11.1. QUANTITATIVE RISK ESTIMATION FOR CANCER
Lifetime extra cancer risk is estimated to be about 0.08perppm continuous
1,3-butadiene exposure, based primarily on human data. When adequate human data are
available, as is the case with 1,3-butadiene, it is generally preferable to base cancer risk estimates
on the human data rather than on data from experimental animals because of the inherent
uncertainties associated with interspecies extrapolation. Cancer risk estimates were calculated
for increased leukemia risk from the Delzell et al. (1995) study of 1,3-butadiene polymer
workers. There were insufficient exposure-response data to calculate a lymphoma risk estimate
from the monomer cohorts. Cancer risk estimates based on the experimental animal data are also
presented for comparison purposes (see below).
The Delzell et al. (1995) retrospective cohort study of more than 15,000 male
styrene-butadiene rubber production workers provides high-quality epidemiologic data on
leukemia mortality risk from 1,3-butadiene exposure. In this study, 1,3-butadiene exposure was
estimated for each job and work area for each study year, and these estimates were linked to
workers' work histories to derive cumulative exposure estimates for each individual worker.
Subsequent to the Poisson regression analyses by Delzell et al. (1995), which used four different
mathematical models to fit the exposure-response data, Health Canada obtained the data on this
cohort and performed their own analyses. The Health Canada analyses were similar to those of
Delzell et al., but involved some minor refinements, and it is the Health Canada analyses that are
used for this risk assessment (Health Canada, 1998).
11-1
-------
The results were adjusted for age, calendar period, years since hire, and cumulative
styrene exposure. Benzene exposure was also estimated for each worker but was not found to be
a confounder and, hence, was not included in the models. Risk estimates were made using the
relative rate models and an actuarial program that accounts for the effects of competing causes of
death. U.S. age-specific mortality rates for all race and gender groups combined (NCHS, 1996)
were used to specify the leukemia and all-cause background rates. Excess leukemia mortality
risks were computed up to age 85 for continuous 1,3-butadiene exposures. The occupational
exposures in the epidemiology study were converted to continuous exposures by adjusting for the
differences in the number of days exposed per year (240/365 days) and differences in the amount
of (contaminated) air inhaled per day (10/20 m3).
Consistent with EPA's 1986 Guidelines for Carcinogen Risk Assessment (U.S. EPA,
1986) and evidence of the genotoxicity of 1,3-butadiene, the linear relative rate
exposure-response model was used to calculate a maximum likelihood estimate (MLE) of 1.3 x
10"2/ppm (or 1 x 10"2/ppm, rounded to one significant figure) for lifetime extra risk of leukemia
mortality from continuous environmental 1,3-butadiene exposure. The corresponding 95% upper
limit on unit risk is 0.03/ppm.
Alternatively, interpreting the proposed new carcinogen risk assessment guidelines (U.S.
EPA, 1996, 1999), linear extrapolation from the LEC01 (i.e., the 95% lower confidence limit of
the exposure concentration associated with a 1% increased risk) is warranted given the clear
genotoxicity of 1,3-butadiene and the fact that a 1% increase in risk is within the range of the
epidemiologic data. The models presented by Health Canada yield LEC01 values ranging from
0.037 to 0.87 ppm. Using a linear extrapolation and the LEC01 as the "point of departure," the
corresponding cancer potency estimates range from 0.011/ppm to 0.27/ppm. While various
models adequately fit the data in the range of observation, there is no compelling reason to
deviate from the linear model historically used for human data. Therefore, the potency estimate
of 0.027/ppm from the linear model (LEC01 = 0.375 ppm) was used to represent the extra risk of
leukemia mortality from the occupational data. Applying the same linear relative rate model and
actuarial program to background leukemia incidence rates resulted in an excess cancer unit risk
estimate of 0.04 per ppm for leukemia incidence.
An adjustment factor of 2 was then applied to this unit risk estimate to reflect evidence
from rodent bioassays suggesting that extrapolating the excess risk of leukemia in a male-only
occupational cohort may underestimate the total cancer risk from 1,3-butadiene exposure in the
general population. First, studies in both rats and mice indicate that 1,3-butadiene is a multisite
carcinogen. It is possible that humans exposed to 1,3-butadiene may also be at risk of cancers
other than leukemia and that the epidemiologic study had insufficient power to detect excess
11-2
-------
Table 11-1. Estimates of upper bounds on human extra unit cancer risk
(potency) from continuous lifetime exposure to 1,3-butadiene based on animal
inhalation bioassays.
Species
Raf
Mouseb
Sex
M
F
M
F
M
F
Tumor sites/types
Leydig cell, pancreatic exocrine cell, Zymbal gland
Mammary gland, thyroid follicular cell, Zymbal gland
Lymphocytic lymphomas, histiocytic sarcomas, heart
hemangiosarcomas, lung, forestomach, Harderian gland,
liver, preputial gland
Lymphocytic lymphomas, heart hemangiosarcomas, lung,
forestomach, Harderian gland, liver, ovary, mammary gland
Lymphocytic lymphomas
Lymphocytic lymphomas
Upper bound on potency
(ppm1)
4.2 x 1Q-3
5.6 x icr2
0.22
0.29
6.4 x IQ-3
2.4 x IQ-2
Trom U.S. EPA's 1985 assessment, based on the Hazleton (1981b) study; linearized multistage model.
bBased on 1993 NTP study; Weibull multistage time-to-tumor model.
risks at other sites. Second, both the rat and mouse studies suggest that females are more
sensitive to 1,3-butadiene-induced carcinogenicity than males (see, e.g., Table 11-1), and the
female mammary gland was the only 1,3-butadiene-related tumor site common to both species.
For the female mouse, which is the most sensitive sex and species tested, the MLE of cancer
potency for the mammary gland is 0.02/ppm and the 95% upper confidence limit is 0.03/ppm
(see Section 10.2.2.2). Applying a twofold adjustment to the potency estimate of 0.04/ppm
derived for leukemia incidence from the occupational epidemiologic study yields a cancer
potency estimate of 0.08/ppm, which roughly corresponds to a combination of the human
leukemia and mouse mammary gland tumor risk estimates, addressing the concerns that the
leukemia risk estimated from the occupational data may underestimate total cancer risk to the
general population, in particular females. It is this excess cancer unit risk estimate of 0.08/ppm,
derived using the 1996 and 1999 proposed guidelines linear default methodology and a twofold
adjustment factor, that is adopted in this assessment. (See Section 10.1 for additional details
about the derivation of this risk estimate.)
For comparison purposes, human unit cancer risk estimates based on extrapolation from
the results of lifetime animal inhalation studies are summarized in Table 11-1. These potency
estimates are 95% upper confidence limits on unit extra cancer risk calculated from incidence
data on all significantly elevated tumor sites using a linearized low-dose extrapolation model.
Such estimates are generally considered by EPA to represent plausible upper bounds on the extra
11-3
-------
unit cancer risk to humans. The rat-based estimates are from EPA's 1985 assessment and use the
linearized multistage model, consistent with the 1986 Guidelines. The mouse-based estimates
use a Weibull multistage time-to-tumor model with the low-dose extrapolation methodology of
the 1986 Guidelines because it is more amenable to the methodology used for combining risk
estimates from different tumor sites (see Section 10.2.2.2). Table 11-1 also includes unit risk
estimates based only on the lymphocytic lymphomas in mice, because this was the tumor type in
rodents most analogous to the lymphohematopoietic cancers observed in workers exposed to
1,3-butadiene.
Human health risk estimates based on extrapolation from high-quality epidemiologic
results are preferable to those based on rodent data because the uncertainties inherent in
extrapolating across species are avoided and, typically, the human exposures in epidemiologic
studies are closer to anticipated environmental exposures than the high exposures used in animal
studies, thus the extent of low-dose extrapolation is reduced. In the case of 1,3-butadiene, while
the rat exposures were at least 100-fold higher than human exposures, the lowest exposure in the
1993 NTP mouse study (4.7 ppm, 8 h time-weighted average [TWA]) is within the range of
occupational exposures (0.7-1.7 ppm median and 39-64 ppm max 8 h TWAs for work-area
groups). However, interspecies differences in tumor sites and susceptibilities between rats and
mice are especially pronounced, and the biological bases for these differences are unresolved. A
review of available pharmacokinetic data and models revealed that the state of the science is
currently inadequate for explaining interspecies differences or improving on default dosimetry
assumptions. Therefore, the quantitative extrapolation of rodent risks to humans is highly
uncertain for 1,3-butadiene.
Even though high-quality human data were used for the quantitative cancer risk
estimation for 1,3-butadiene, there are inevitable uncertainties in the calculated risk estimate.
First, there are uncertainties inherent in the epidemiologic study itself. In particular, there are
uncertainties in the retrospective estimation of 1,3-butadiene exposures, which could have
resulted in exposure misclassification. Nondifferential exposure misclassification would tend to
bias estimates of effect toward the null, resulting in an underestimate of risk. Differential
misclassification could bias results in either direction.
Nonetheless, a great deal of effort and detail went into estimating exposures in the Delzell
et al. (1995) study. The investigators reviewed relevant plant records and conducted plant walk-
throughs and interviews with plant experts and long-term employees (at least 2 employees with
work experience for each work area and time period) (Macaluso et al., 1996). They also assessed
determinants of exposure for specific component tasks (e.g., frequency and duration of task,
work practice, distance from emission source) in each job at different time periods. Then they
developed exposure models to estimate exposure intensities associated with specific tasks in
11-4
-------
different time periods. There was some validation of these exposure models in one plant.
Finally, the job-exposure matrices were linked to each worker's work history to calculate
cumulative exposure estimates for each individual worker. Furthermore, these individual
cumulative exposure estimates were derived with the investigators "blind" to the mortality data
for the individuals.
However, as discussed in Section 10.1.3, some concerns were later raised by the
investigators about the accuracy of the exposure estimates (letters from Elizabeth Moran, CMA,
March 25, 1996, and Delzell and Macaluso, University of Alabama, April 2, 1996, to Aparna
Koppikar). In 2000, Delzell et al. completed a re-assessment of exposure estimates. Analyses
performed by Robert Sielken, a consultant to the American Chemistry Council who had access to
the exposure-response data, suggest that the revised exposure estimates result in about a 70%
decrease in the leukemia risk estimate (Sielken and Valdez-Flores, 2001). It should be noted,
though, that Sielken's analyses differ somewhat from those of EPA (e.g., Sielken based his linear
extrapolation on the EC01 rather than on the LEC01), thus the Agency's results with the revised
exposure estimates may differ slightly from the 70% decrease.
EPA is considering conducting a critical review of both the revised exposure estimates as
well as the a priori estimates. If EPA concludes that the exposure re-assessment yields more
credible exposure estimates, the Agency will consider modifying the cancer risk estimate,
provided that the revised exposure estimates have a substantial impact. EPA does not at this time
have the data needed to conduct a comparison cancer risk analysis based on the revised exposure
estimates. If the revised exposure estimates are an improvement over the a priori estimates and
EPA's analysis yielded the same results as Sielken's, then the leukemia risk estimate would
decrease about 70%. However, EPA would still apply an adjustment factor that would
effectively increase the leukemia risk estimate by 0.03-0.04/ppm to account for possible cancer
risk at other sites, in particular the female mammary gland. Thus, under the above assumptions,
the decrease to the total cancer unit risk estimate would be closer to 40%.
Second, there are uncertainties regarding the appropriate dose metric for dose-response
analysis. Although the dose surrogate of cumulative exposure (i.e., ppm x years) yielded highly
statistically significant exposure-response relationships, cumulative exposure is strongly
correlated with other possible exposure measures, and there may be a dose-rate effect (e.g., risk
at high exposures may be more than proportionately greater than at lower exposures) obscured in
the analysis, or operative at exposures below the observable range but relevant to low-dose
extrapolation.
Third, there are uncertainties pertaining to the model to use for the epidemiologic data.
Several mathematical models adequately fit the exposure-response data from the epidemiology
study, and because the specific mechanisms of 1,3-butadiene carcinogenesis are unknown, there
11-5
-------
is no biological basis for choosing one model over another. The linear model was chosen in this
risk assessment to derive the "point of departure" for low-dose extrapolation because there was
no compelling reason to deviate from historical approaches.
Fourth, it is uncertain which potential modifying or confounding factors should be
included in the model. The linear model of Health Canada, which is used in this risk assessment,
adjusted for age, calendar year, years since hire, race, and exposure to styrene. Plant and benzene
exposure were ruled out as potential confounders. However, there may be other relevant factors
that were not included in the models.
Fifth, there are uncertainties in the parameter estimates used in the models. The study of
Delzell et al. (1995, 1996) is large, providing some degree of reliability in the parameter
estimates; however, especially given the large human variability that has been observed in
metabolic activities that could affect cancer risk from 1,3-butadiene exposure, the
generalizability of the occupational results is unclear.
Sixth, there are uncertainties in extending the relative rate models from the epidemiology
study to derive lifetime excess leukemia incidence unit risk estimates for the U.S. population.
Notwithstanding, the actuarial-type analysis that was used is a well-established methodology, and
the background leukemia incidence rates and mortality rates used in the analysis are from large
national databases.
Seventh, the precise model for low-dose extrapolation is unknown. The linear default
extrapolation procedure in the 1996 proposed guidelines (U.S. EPA, 1996) was used in this
assessment because of the well-established genotoxicity of 1,3-butadiene, via its metabolites.
In addition, important concerns are raised by comparison with the rodent data. First, the
rodent studies suggest that 1,3-butadiene is a multisite carcinogen (see Chapter 6). It is possible
that humans may also be at risk of 1,3-butadiene-induced carcinogenicity at other sites and that
the epidemiologic study had insufficient power to detect the other excess risks. In the mouse, for
example, the lung is the most sensitive tumor site. Significant excesses of lung cancer may not
have been detectable in the epidemiologic study because of the high background rates of lung
cancer in humans. A crude "power" calculation presented in Section 10.1.3 suggested that if
humans were as sensitive as female mice to the lung cancer effects of 1,3-butadiene, one might
have expected to see 26 excess lung cancer cases in the epidemiology study. However, the
statistical power of the study to detect a corresponding SMR of 108 is estimated to be only 42%.
The epidemiology-based excess cancer risk estimate of 0.04/ppm, which is based only on
leukemias, may be an underestimate if other sites are also at risk.
Second, both the rat and mouse studies suggest that females are more sensitive to
1,3-butadiene-induced carcinogenicity than males, and the mammary gland in females was the
only tumor site common to both species. If female humans are also more sensitive than males,
11-6
-------
then the male-based risk estimates calculated from the epidemiology study would underestimate
risks to females. Because of these concerns, an adjustment factor of 2 is used, as discussed
above, yielding a cancer potency estimate of 0.08/ppm.
Despite these uncertainties, confidence in the excess cancer risk estimate of 0.08/ppm
is moderate. First, the estimate is based primarily on human data. Furthermore, these data are
from a large, high-quality epidemiologic study in which 1,3-butadiene exposures were estimated
for each individual a priori to conducting the exposure-response analysis. Although there are
uncertainties in the exposure estimation, a serious attempt was made to reconstruct historical
exposures for specific tasks and work areas. It is virtually unprecedented to have such a
comprehensive exposure assessment for individual workers in such a large occupational
epidemiologic study. In addition, the assumption of linearity for low-dose extrapolation is
reasonable given the clear evidence of genotoxicity by 1,3-butadiene metabolites.
Using the cancer potency estimate of 0.08/ppm, the lifetime average exposure level
resulting in an increased cancer risk of 10"6 (i.e., one in a million) can be estimated as follows:
(10-6)/(0.08/ppm) = 1 x 1Q-5 ppm = 0.01 ppb.
11.2. QUANTITATIVE ESTIMATION (RfC) FOR NONCANCER EFFECTS
A chronic RfC of 0.9 ppb was obtained for the critical effect of ovarian atrophy, from a
chronic lifetime bioassay of female mice. In addition, an "RfC" for acute exposures of 7 ppb
and an "RfC" for subchronic exposures of 7 ppb were calculated based on fetal weight data
for mice. A chronic RfC is an estimate of the daily exposure to humans that is "likely to be
without appreciable risk of deleterious (noncancer) effects during a lifetime." The RfC is
calculated for the "critical (noncancer) effect," that is, the effect for which an increased response
is observed at the lowest concentration used in the study, or for which benchmark concentration
modeling yields the lowest EC10, or the effect which yields the lowest concentration after
application of the relevant uncertainty factors. In this assessment, sample RfC calculations were
performed for several reproductive and developmental effects, because these represent the most
sensitive noncancer endpoints observed in the available toxicity studies.
Of the 1,3-butadiene noncancer effects, the critical effect was ovarian atrophy, from a
chronic lifetime bioassay of female mice. The RfC calculation was based on the LEC10 (0.88
ppm), which was derived from data from the 1993 NTP 2-year bioassay using benchmark
concentration methodology (Weibull time-to-response model), and uncertainty factors for
interspecies extrapolation (3), intraspecies variability (10), "effect level extrapolation" to
extrapolate to a level below the 10% effect level (analogous to the LOAEL-to-NOAEL
uncertainty factor) (10), and incomplete database (3). The LEC10 was derived from a chronic
bioassay; therefore, an acute/subchronic-to-chronic factor was not required. A factor of 3 (the
11-7
-------
standard factor for the pharmacodynamic portion) was used for the interspecies uncertainty factor
because the use of "ppm equivalence" in determining the human equivalent exposures is
assumed to account for the pharmacokinetic portion. Note that the assumption of ppm
equivalence is the same as using EPA's inhalation dosimetry methodology with RGDR^l (U.S.
EPA, 1994c). There is strong evidence that the ovarian atrophy is caused by the diepoxide
metabolite of 1,3-butadiene (DEB; see Chapter 5), and it is expected, based on pharmacokinetic
data, that humans produce less DEB than mice (see Chapter 3). However, PBPK models are
inadequate to quantify levels of DEB (see Chapter 9), thus default dosimetry (i.e., 1,3-butadiene
exposure concentration) was used for dose-response modeling and the default value of 1 for the
pharmacokinetic portion of the interspecies uncertainty factor was retained. The effect level
extrapolation factor was derived from a formula that takes into account the slope of the exposure-
response curve at the "point of departure." However, because the slope was supralinear at the
LEC10, a maximum factor of 10 for the 10% level was used (see Section 10.4). The default factor
of 10 was used for intraspecies variability, and a factor of 3 was used to reflect an incomplete
database, in particular the absence of a multigenerational reproductive study and a developmental
neurotoxicity study. The resulting chronic RfC is 0.9 ppb (i.e., 0.88 ppm/(3 x 10 x 10 x 3)).
Thus, a chronic RfC of 0.9 ppb is adopted for the critical effect of ovarian atrophy. The
actual risks at low exposure levels are unknown; the chronic RfC merely provides a bound on
chronic exposure below which no "appreciable risk" of noncancer effects is expected.
In addition, an "RfC" for acute exposures of 7 ppb and an "RfC" for subchronic
exposures of 7 ppb were calculated from the mouse fetal weight data (Hackett et al., 1987b),
using benchmark concentration methodology (log-logistic model) to obtain the "point of
departure" for applying uncertainty/modifying factors. For the fetal weight data, an LEC05 = 2.9
ppm was obtained, and uncertainty/modifying factors for interspecies extrapolation (3),
intraspecies variability (10), "effect level extrapolation" to extrapolate to a level below the 5%
effect level (analogous to the LOAEL-to-NOAEL uncertainty factor) (4), and incomplete
database (3) (e.g., the absence of neurodevelopmental studies), were applied, yielding 7 ppb (i.e.,
2.9 ppm/(3x 10x4x3)). These RfC values are summarized in Table 11-2.
11.3. SUMMARY AND CONCLUSIONS
The estimate of human lifetime extra cancer risk from chronic exposure to 1,3-butadiene
is 0.08 per ppm based primarily on linear modeling and extrapolation of the increased leukemia
risks observed in occupationally exposed workers. Although there is uncertainty in extrapolating
from occupational exposures to lower environmental exposures, this risk estimate has the
advantage of being based on a large, high-quality human study, and linear extrapolation is
warranted by the known genotoxicity of 1,3-butadiene metabolites. This cancer potency estimate
11-8
-------
Table 11-2. Reference concentrations (RfCs) for 1,3-butadiene, based on
mouse data.
TypeofRfC
Chronic
Subchronic
Acute
Critical effect
Ovarian atrophy
Fetal weight
Fetal weight
Point of departure3
LEC10 = 0.88ppm
LEC05=2.9ppm
LEC05=2.9ppm
Uncertainty
facto rsb
3 x 10 x 1 x 10 x 3 (1000)
3xlQxNAx4x3 (400)
3xlQxNAx4x3 (400)
RfC (ppb)
0.9
7
7
aThe point of departure (or "benchmark dose") is the 95% lower bound on the exposure concentration corresponding to
an effect level at the low end of the range of observations (in this case a 5% or 10% effect level).
blnterspecies x intraspecies x acute/subchronic-to-chronic x effect level extrapolation x incomplete database (total); see
text and Section 10.4 for explanations.
NA: not applicable.
incorporates an adjustment factor of 2 to address concerns that the leukemia risk estimate from a
male worker population may underestimate total cancer risk to the general population. The
corresponding estimate of the chronic exposure level of 1,3-butadiene resulting in an extra cancer
risk of 10"6 (i.e., one in a million) is 0.01 ppb.
1,3-Butadiene also causes a variety of reproductive and developmental effects in mice; no
human data on these effects are available. The most sensitive effect was ovarian atrophy
observed in a lifetime bioassay of female mice (LEC10= 0.88 ppm). Based on this critical effect
and using benchmark concentration methodology, an RFC (i.e., a chronic exposure level
presumed to be "without appreciable risk" for noncancer effects) of 0.9 ppb was calculated. The
actual risks at low exposure levels are unknown; this RfC merely provides a bound on chronic
exposure below which no "appreciable risk" of noncancer effects is expected. An acute "RfC"
and a subchronic "RfC" were also calculated (7 ppb) based on reduced fetal weight.
In summary, the primary changes in EPA's conclusions about the health effects of 1,3-
butadiene from the 1985 document are:
• The cancer characterization has been changed from probable human carcinogen to
carcinogenic to humans by inhalation.
• The unit cancer risk estimate has been changed from 0.64/ppm (upper bound based on
mouse data) to 0.08/ppm (based primarily on linear modeling and extrapolation of
human data).
• For the first time, a (chronic) RfC (0.9 ppb), an acute "RfC" (7 ppb), and a subchronic
"RfC" (7 ppb) are calculated for noncancer effects.
11-9
-------
REFERENCES
Acquavella, JF. (1989) The paradox of butadiene epidemiology. Exp Pathol 37:114.
Adler, ID; Anderson, D. (1994) Dominant lethal effects after inhalation exposure to 1,3-butadiene. Mutat Res
309:295-297.
Adler, ID; Cao, J; Filser, JG; et al. (1994) Mutagenicity of 1,3-butadiene inhalation in somatic and germinal cells of
mice. Mutat Res 309:307-314.
Adler, ID; Filser, JG; Gassner, P; et al. (1995) Heritable translocations induced by inhalation exposure of male mice
to 1,3-butadiene. Mutat Res 347:121-127.
Adler, ID; Kliesch, U; Nylund, L; et al. (1997) In vitro and in vivo mutagenicity of the butadiene metabolites
butadiene diolepoxide, butadiene monoepoxide and diepoxybutane. Mutagenesis 12:339-345.
Adler, ID; Filser, J; Gonda, H; et al. (1998) Dose response study for 1,3-butadiene-induced dominant lethal
mutations and heritable translocations in germ cells of male mice. Mutat Res 397:85-92.
Agency for Toxic Substances and Disease Registry (ATSDR). (1991) Toxicological profile for 1,3-butadiene. TP-
91/07, PB93-11690. Public Health Service, U.S. Department of Health and Human Services, Atlanta, GA.
Aiman, J; Forney, JP; Parker, CR, Jr. (1986) Secretion of androgens and estrogens by normal and neoplastic ovaries
in postmenopausal women. Obstet Gynecol 68:1-5.
Alberts, B; Bray, D; Lewis, J; et al. (1983) Molecular biology of the cell. New York: Garland Academic Press, pp.
319-384.
Albrecht, OE; Filser, JG; Neumann, HG. (1993) Biological monitoring of 1,3-butadiene: species differences in
haemoglobin binding in rat and mouse. In: IARC Scientific Publications Vol. 127. Sorsa, M; Peltonen, K; Vainio, H,
eds. Lyon, France: International Agency for Research on Cancer, pp. 135-142.
Allen, BC; Kavlock, RJ; Kimmel, CA; et al. (1994a) Dose-response assessment for development toxicity: III.
Statistical models. Fundam Appl Toxicol 23:496-509.
Allen, BC; Kavlock, RJ; Kimmel, CA; et al. (1994b) Dose-response assessment for developmental toxicity: II.
Comparison of generic benchmark dose estimates with NOAELs. Fundam Appl Toxicol 23:487-495.
Altaian, PL; Dittmer, DS. (1971) Respiration and circulation. Bethesda, MD: Federation of American Societies for
Experimental Biology, pp. 384-385.
Anderson, EL; U.S. EPA Carcinogen Assessment Group. (1983) Quantitative approaches in use to assess cancer
risk. Risk Anal 3:277-295.
Anderson, ME; Clewell, HJ; Gargas, ML; et al. (1987) Physiologically based pharmacokinetics and the risk
assessment process for methylene chloride. Toxicol Appl Pharmacol 87:185-205.
Anderson, D; Edwards, AJ; Brinkworth, MH. (1993) Male-mediated ¥{ effects in mice exposed to 1,3-butadiene. In:
Butadiene and styrene: assessment of health hazards. IARC Scientific Publications. Vol. 127. Sorsa, M; Peltonen, K;
Vainio, H; et al., eds. Lyon, France: International Agency for Research on Cancer, pp. 171-181.
Anderson, D; Edwards, AJ; Brinkworth, MH. (1995) The detection of dominant lethal mutations and foetal
malformations, tumours and chromosome damage in the offspring of male mice treated sub-chronically with
butadiene. BIBRA Report No. 1060/5/95.
R-l
-------
Anderson D; Edwards AJ; Brinkworth MH; et al. (1996) Male-mediated Fl effects in mice exposed to 1,3-butadiene.
Toxicology 113(1-3): 120-127.
Anderson, D; Dobrzynka, MM; Jackson, LI; et al. (1997) Somatic and germ cell effects in rats and mice after
treatment with 1,3-butadiene and its metabolites, 1,2-epoxybutene and 1,2,3, 4- diepoxybutane. Mutat Res
391:233-242.
Anderson, D; Hughes, JA; Edwards, AJ; et al. (1998) A comparison of male-mediated effects in rats and mice
exposed to 1,3-butadiene. Mutat Res 397(l):77-84.
Andjelkovich, D; Taulbee J; Symons, M. (1976) Mortality experience of a cohort of rubber workers, 1964-1973. J
OccupMed 18:387-394.
Andjelkovich, D; Taulbee, J; Symons, M. (1977) Mortality of rubber workers with reference to work experience. J
OccupMed 19:397-405.
Anonymous. (1991) Chemical profile: butadiene. Chem Mark Rep 239:50.
Arce, GT; Vincent, DR; Cunningham, MJ; et al. (1990) In vitro and in vivo genotoxicity of 1,3-butadiene. Environ
Health Perspect 86:75-78.
Armitage, P. (1971) Statistical methods in medical research. New York, NY: John Wiley and Sons, pp. 362-365.
Arms, A; Travis, C. (1988) Reference physiological parameters in pharmacokinetic modelling. U.S. Environmental
Protection Agency. EPA/600/6-88/004.
Astwood, EB. (1938) A six-hour assay for the quantitative determination of estrogen. Endocrinology 23:25.
Au, WW; Bechtold, WE; Wharton, EB, Jr.; et al. (1995) Chromosome aberrations and response to gamma-ray
challenge in lymphocytes of workers exposed to 1,3-butadiene. Mutat Res 334:125-130.
Bailer, AJ; Portier, CJ. (1988) Effects of treatment-induced mortality and tumor-induced mortality on tests for
carcinogenicity in small samples. Biometrics 44:417-431.
Barnes, DG; Dourson, M; U.S. EPA Reference Dose Work Group. (1988) Reference dose (RfD): Description and
use in health risk assessments. Reg Toxicol Pharmacol 8:471-86.
Bartsch, H; Malaveille, C; Barbin, A; et al. (1979) Mutagenic and alkylating metabolites of halo-ethylenes,
chlorobutadienes and dichlorobutenes produced by rodent or human liver tissues. Evidence for oxirane formation by
P450-linked microsomal mono-oxygenases. Arch Toxicol 41:249-277.
Beaumont, JJ; Breslow, NE. (1981) Power considerations in epidemiologic studies of vinyl chloride workers. Am J
Epidemiol 114:725-734.
Bechtold, WE; Strunk, MR; Chang, I-Y; et al. (1994) Species differences in urinary metabolite ratios between mice,
rats, and humans. Toxicol Appl Pharmacol 127:44-49.
Bechtold, WE; Strunk, MR; Thornton-Manning, JR; et al. (1995) Analysis of butadiene, butadiene monoxide and
butadiene diepoxide in blood by gas chromatography/gas chromatography/mass spectroscopy. Chem Res Toxicol
8:182-187.
Berard, CW; Hutchison, RE. (1997) The problem of classifying lymphomas: an orderly prescription for progress.
Ann Oncol (Suppl 2):S3-9.
R-2
-------
Bernardini, S; Pelin, K; Peltonen, K; et al. (1996) Induction of sister chromatid exchange by 3,4-epoxybutane-l,2-
diol in cultured human lymphocytes of different GSTT1 and GSTM1 genotypes. MutatRes 361:121-127.
Bernardini, S; Hirvonen, A; Pelin, K; et al. (1998) Induction of sister chromatid exchange by l,2-epoxy-3-butene in
cultured human lymphocytes: influence of GSTT1 genotype. Carcinogenesis 19:377-380.
Beaumont, JJ; Breslow, NE. (1981) Power considerations in epidemiologic studies of vinyl chloride workers. Am J
Epidemiol 114:752-34.
Bevan, C; Stadler, JC; Elliott, GS; et al. (1996) Subchronic toxicity of 4-vinylcyclohexene in rats and mice by
inhalation exposure. Fundam Appl Toxicol 32:1-10.
Bittersohl, G. (1974) Epidemiologic investigation on cancer incidence in workers contacted by acetaldol and other
aliphatic aldehydes. Arch Geschwulstforsch 43:172-176.
Bogaards, JJ; Venekamp, JC; van Bladeren, PJ. (1996) The biotransformation of isoprene and the two isoprene
monoepoxides by human cytochrome P450 enzymes, compared to mouse and rat liver microsomes. Chem Biol
Interact 102(3): 169-182.
Bogaards, JJ; Venekamp, JC; Salmon, FGC; et al. (1999) Conjugation of isoprene monoepoxides with glutathione,
catalyzed by alpha, mu, pi, and theta-class glutathione S-transferases of rat and man. Chem Biol Interact 117(1): 1-
14.
Bolt, HM; Filser, JG; Buchter, A. (1981) Inhalation pharmacokinetics based on gas uptake studies. III. A
pharmacokinetic assessment in man of "peak concentrations" of vinyl chloride. Arch Toxicol 48:213-228.
Bolt, HM; Schmeidel, G; Filser, JG; et al. (1983) Biological activation of 1,3-butadiene to vinyl oxirane by rat liver
microsomes and expiration of the reactive metabolite by exposed rats. J Cancer Res ClinOncol 106:112-116.
Bolt, HM; Filser, JG; Stormer, F. (1984) Inhalation pharmacokinetics based on gas uptake studies. Arch Toxicol
55:213-218.
Bond, JA; Dahl, AR; Henderson, RF; et al. (1986) Species differences in the disposition of inhaled butadiene.
Toxicol Appl Pharmacol 84:617-627.
Bond, JA; Dahl, AR; Henderson, RF; et al. (1987) Species differences in the distribution of butadiene in tissues. Am
Ind Hyg Assoc J 48:867-872.
Bond, JA; Martin, OS; Birnbaum, LS; et al. (1988) Metabolism of 1,3-butadiene by lung and liver microsomes of
rats and mice repeatedly exposed by inhalation to 1,3-butadiene. Toxicol Lett 44:143-151.
Bond, JA; Bechtold, WE; Birnbaum, LS; et al. (1991) Disposition of inhaled isoprene inB6C3Fl mice. Toxicol
Appl Pharmacol 107(3):494-503.
Bond, GG; Bodner, KM; Olsen, GW; et al. (1992) Mortality among workers engaged in the development or
manufacture of styrene-based products-an update. Scand J Work Environ Health 18(3): 145-154.
Boogaard, PJ; Bond, JA. (1996) The role of hydrolysis in the detoxication of l,2:3,4-diepoxybutane by human, rat,
and mouse liver and lung in vitro. Toxicol Appl Pharmacol 141:617.
Boogaard, PJ; Sumner, SCJ; Bond, JA. (1996) Glutathione conjugation of l,2:3,4-diepoxybutane in human liver and
rat and mouse liver and lung in vitro. Toxicol Appl Pharmacol 136:307-316.
Borman, SM; VanDePol, BJ; Kao, S; et al. (1999) A single dose of the ovotoxicant 4-vinylcyclohexene diepoxide is
protective in rat primary ovarian follicles. Toxicol Appl Pharmacol 158:244-252.
R-3
-------
Brinkworth, MH; Anderson, D; Hughes, JA; et al. (1998) Genetic effects of 1,3-butadiene on the mouse testis. Mutat
Res 397:67-75.
Brunnemann, KD; Kagan, MR; Cox, JE; et al. (1990) Analysis of 1,3-butadiene and other selected gas-phase
components in cigarette mainstream and sidestream smoke by gas chromatography-mass selective detection,
Carcinogenesis 11:1863.
Buckley, LA; Coleman, DP; Burgess, JP; et al. (1999) Identification of urinary metabolites of isoprene in rats and
comparison with mouse urinary metabolites. Drug Metab Dispos 27(7):848-854.
Budavari, S, ed. (1989) The Merck index, 11th ed. Rahway, NJ: Merck & Co., pp. 230-231.
California Air Resources Board. (1992) Proposed identification of 1,3-butadiene as a toxic air contaminant. Part A.
Exposure assessment. Stationary Source Division, Sacramento, CA.
Campbell, DL; Mangino, J. (1994) Evaluation and improvement of the Puget Sound toxic air contaminants emissions
inventory. Prepared by Radian Corporation for the Office of Research and Development, U.S. Environmental
Protection Agency, Washington, DC, under EPA contract no. 68-D1-0031.
Capen, CC; Beamer, WG; Tennent, BJ; et al. (1995) Mechanisms of hormone-mediated carcinogenesis of the ovary
in mice. Mutat Res 333:143-151.
Carpenter, CP; Shaffer, CB; Weil, CS; et al. (1944) Studies on the inhalation of 1,3-butadiene; with a comparison of
its narcotic effect with benzol, toluol and styrene, and a note on the elimination of styrene by the human. J Ind Hyg
Toxicol 26:69-78.
Chakravarti, S; Collins, WP; Forecast, JD; et al. (1976) Hormonal profiles after the menopause. Br Med J
2:784-787.
Checkoway, H; Williams, TM. (1982) A hematology survey of workers at a styrene-butadiene synthetic rubber
manufacturing plant. Am Ind Hyg Assoc J 43:164-169.
Cheng, X; Ruth, JA. (1993) A simplified methodology for quantitation of butadiene metabolites. Application of the
study of butadiene metabolism by rat liver microsomes. Drug Metab Dispos 21:121-124.
Cheng, X; Maniglier-Poulet, C; Ross, D; et al. (1994) Solvolytic formation of l,2-dichloro-3,4-epoxybutane from
butadiene monoxide under physiological conditions. Drug Metab Dispos 22:206-210.
Christian, MS. (1996) Review of reproductive and developmental toxicity of 1,3-butadiene. Toxicology 113:137-
143.
Cochrane, JE; Skopek, TR. (1994a) Mutagenicity of butadiene and its epoxide metabolites: I. Mutagenic potential
of 1,2-epoxybutene, 1,2,3,4-diepoxybutane and 3,4-epoxy-l,2-butanediol in cultured human lymphoblasts.
Carcinogenesis 15:713-717.
Cochrane, JE; Skopek, TR. (1994b) Mutagenicity of butadiene and its epoxide metabolites: II. Mutational spectra of
butadiene, 1,2-epoxybutene, and diepoxybutane at the hprt locus in splenic T cells from exposed B6C3Fj mice.
Carcinogenesis 15:719-723.
Cole, P; Delzell, E; Acquavella, J. (1993) Exposure to butadiene and lymphatic and hematopoietic cancer.
Epidemiology 4:86.
Concawe. (1987) A survey of exposures to gasoline vapour. Report No. 4/87. de Hague: Concawe.
R-4
-------
Coughtrie, MWH; Burchell, B; Bend, JR. (1987) Purification and properties of rat kidney UDP-
glucuronosyltransferase. Biochem Pharmacol 36:245-251.
Cowles, SR; Tsai, SP; Snyder, PJ; et al. (1994) Mortality, morbidity, and haematologic results from a cohort of long-
term workers involved in 1,3-butadiene monomer production. Occup Environ Med 51:323-329.
Cox, DR. (1972) Regression models and life tables. J R Stat Soc 634:187-220.
Crump, KS. (1984) A new method for determining allowable daily intakes. Fundam Appl Toxicol 4:854-871.
Crump, KS. (1995) Calculation of benchmark doses from continuous data. Risk Anal 15:79-89.
Csanady, GA; Bond, JA. (1991) Species differences in the biotransformation of 1,3-butadiene to DNA-reactive
epoxides: role in cancer risk assessment. Chemical Industry Institute of Toxicology, CUT Activities 11:1-8.
Csanady, GA; Guengerich, FP; Bond, JA. (1992) Comparison of the biotransformation of 1,3-butadiene and its
metabolite, butadiene monoepoxide, by hepatic and pulmonary tissue from humans, rats, and mice. Carcinogenesis
13:1143-1153.
Csanady, GA; Guengerich, FP; Bond, JA. (1993) Comparison of the biotransformation of 1,3-butadiene and its
metabolite, butadiene monoepoxide, by hepatic and pulmonary tissues from humans, rats and mice [published
erratum]. Carcinogenesis 14:784.
Csanady, GA; Kreuzer, PE; Baur, C; et al. (1996) A physiological toxicokinetic model for 1,3-butadiene in rodents
and man: blood concentrations of 1,3-butadiene, its metabolically formed epoxides, and of haemoglobin adducts -
relevance of glutathione depletion. Toxicology 113:300-305.
D'Souza, RW; Francis, WR; Anderson, ME. et al. (1988) Physiological model for tissue glutathione depletion and
increased resynthesis after ethylene dichloride exposure. J PharmExp Ther 245:563-56.
Dahl, AR; Birnbaum, LS; Bond, JA; et al. (1987) The fate of isoprene inhaled by rats: comparison to butadiene.
Toxicol Appl Pharmacol 89:237-242.
Dahl, AR; Bechtold, WE; Bond, JA; et al. (1990) Species differences in the metabolism and disposition of inhaled
1,3-butadiene and isoprene. Environ Health Perspect 86:65-69.
Dahl, AR; Sun, DJ; Birnbaum, LS; etal. (1991) Toxicokinetics of inhaled 1,3-butadiene in monkeys: comparison to
toxicokinetics in rats and mice. Toxicol Appl Pharmacol 110:9-19.
de Meester, C. (1988) Genotoxic properties of 1,3-butadiene. MutatRes 195:273-281.
Del Monte, M; Citti, L; Gervasi, PG. (1985) Isoprene metabolism by liver microsomal monooxygenases.
Xenobiotica 15:591-597.
Delzell, E; Sathiakumar, N; Macaluso, M. (1995) A follow-up study of synthetic rubber workers. Final report
prepared under contract to International Institute of Synthetic Rubber Producers.
Delzell, E; Sathiakumar, N; Hovinga, M. (1996) A follow-up study of synthetic rubber workers. Toxicology
113:182-189.
Deutsche Forschungsgemeinschaft. (1999) List of MAK and BAT Values. Commission for the Investigation of
Health Hazards of Chemical Compounds in the Work Area. Report No. 35.
Deutschmann, S. (1988) Vergleichende Untersuchungen zur Depletion von zellularem Glutathion durch 1,3-
Butadienbei Maus und Ratte. Thesis. Ruhr-Universitat, Bochum, Germany.
R-5
-------
Deutschmann, S; Laib, RJ. (1989) Concentration-dependent depletion of non-protein sulfhydryl (NPSH) content in
lung, heart and liver tissue of rats and mice after acute inhalation exposure to butadiene. Toxicol Lett 45:175-183.
Divine, BJ. (1990) An update on mortality among workers at a 1,3-butadiene facility— preliminary results. Environ
Health Perspect 86:119-128.
Divine, BJ; Hartman, CM. (1996) Mortality update of butadiene production workers. Toxicology 113:169-181.
Divine, BJ; Wendt, JK; Hartman, CM. (1993) Cancer mortality among workers at a butadiene facility. In: Butadiene
and styrene: assessment of health hazards. IARC Scientific Publications. Vol. 127. Sorsa, M; Peltonen, K; Vainio, H;
et al., eds. Lyon, France: International Agency for Research on Cancer, pp. 345-362.
Doerr, JK; Hooser, SB; Smith, BJ; et al. (1995) Ovarian toxicity of 4-vinylcyclohexene and related olefins in
B6C3F! mice: role of diepoxides. ChemRes Toxicol 8:963-969.
Doerr, JK; Hollis, EA; Sipes, IG. (1996) Species difference in the ovarian toxicity of 1,3-butadiene epoxides in
B6C3Fj mice and Sprague-Dawley rats. Toxicology 113:128-136.
Downs, TD; Crane, MM; Kim, KW. (1987) Mortality among workers at a butadiene facility. Am J Ind Med
12:311-329.
Downs, T; Pier, S; Crane, M. (1992) Cause-specific mortality in a cohort of 1,000 ABS workers. Abstr Symp.
Butadiene and styrene: assessment of health hazards. Espoo, Finland.
Duescher, RJ; Elfarra, AA. (1992) 1,3-Butadiene oxidation by human myeloperoxidase: role of chloride ion in
catalysis of divergent pathways. J Biol Chem 267:19859.
Duescher, RJ; Elfarra, AA. (1993) Chloroperoxidase-mediated oxidation of 1,3-butadiene to 3-butenal, a
crotonaldehyde precursor. Chem Res Toxicol 6:669-673.
Duescher, RJ; Elfarra, AA. (1994) Human liver microsomes are efficient catalysts for 1,3-butadiene oxidation:
evidence for major roles by cytochrome P450 2A6 and 2E1. Arch Biochem Biophys 311:342-349.
Dunnick, JK; Eustis, SL; Piegorsch, WW; et al. (1988) Respiratory tract lesions in F344/N rats and B6C3F! mice
after inhalation exposure to 1,2-epoxybutane. Toxicology 50:69-82.
Elfarra, AA; Duescher, RJ; Pasch, CM. (1991) Mechanisms of 1,3-butadiene oxidations to butadiene monoxide and
crotonaldehyde by mouse liver microsomes and chloroperoxidase. Arch Biochem Biophys 286:244-251.
Elfarra, AA; Sharer, JE; Duescher, RJ. (1995) Synthesis and characterization of N-acetyl-L-cysteine S-conjugates of
butadiene monoxide and their detection and quantitation in urine of rats and mice given butadiene monoxide. Chem
Res Toxicol 8:68-76.
Evelo, CTA; Oostendorp, JGM; ten Berge, WF; et al. (1993) Physiologically based toxicokinetic modeling of 1,3-
butadiene lung metabolism in mice becomes more important at low doses. Environ Health Perspect 101:496-502.
Federal Register. (2001, November 29) Notice of opportunity to provide additional information and comment.
Washington, DC; U.S. GPO, 66(23):59593-59594.
Filser, JG. (1992) The closed chamber technique—uptake, endogenous production, excretion, steady-state kinetics
and rates of metabolism of gases and vapors. Arch Toxicol 66:1-10.
Filser, JG; Bolt, HM. (1979) Pharmacokinetics of halogenated ethylenes in rats. Arch Toxicol 42:123-136.
R-6
-------
Filser, JG; Bolt, HM. (1981) Inhalation pharmacokinetics based on gas uptake studies. I. Improvement of kinetic
models. ArchToxicol 47:279-292.
Filser, JG; Bolt, HM. (1983) Exhalation of ethylene oxide by rats on exposure to ethylene. Mutat Res 120:57-60.
Filser, JG; Bolt, HM. (1984) Inhalation pharmacokinetics based on gas uptake studies. VI. Comparative evaluation
of ethylene oxide and butadiene monoxide as exhaled reactive metabolites of ethylene and 1,3-butadiene in rats.
ArchToxicol 55:219-223.
Filser, JG; Althaler, B; Welter, HF; et al. (1992) Metabolism of 1,3-butadiene in microsomes from livers of mouse,
rat, and man. Naunyn-Schmiedeberg's ArchToxicol Suppl 345:R31.
Fiserova-Bergerova, V; Diaz, ML. (1986) Determination and prediction of tissue gas partition coefficients. Int Arch
Occup Environ Health 58:75-87.
Flaws, JA; Doerr, JK; Sipes, IG; et al. (1994a) Destruction of preantral follicles in adult rats by 4-vinyl-l-
cyclohexene diepoxide. Reprod Toxicol 8:509-514.
Flaws, JA; Salyers, KL; Sipes, IG; et al. (1994b) Reduced ability of rat preantral ovarian follicles to metabolize 4-
vinyl-1-cyclohexene diepoxide in vitro. Toxicol Appl Pharmacol 126:286-294.
Frank, PM. (1978) Introduction to system sensitivity theory. New York: Academic Press.
Frentzel-Beyme, R; Thiess, AM; Wieland, R. (1978) Survey of mortality among employees engaged in the
manufacture of styrene and polystyrene at the BASF Ludwigshafen works. Scan J Environ Health 4 Suppl 2:231-
239.
Frome, EL; Checkoway, H. (1985) Use of Poisson regression models in estimating incidence rates and ratios. Am J
Epidemiol 121(2):309-322.
Gargas, ML; Burgess, RJ; Voisard, DE; Cason, GH; Andersen, ME. (1989) Partition coefficients of low-molecular-
weight volatile chemicals in various liquids and tissues. Toxicol Appl Pharmacol 98(l):87-99.
Gart, JJ; Chu, KC; Tarone, RE. (1979) Statistical issues in interpretation of chronic bioassay for carcinogenicity. J
Natl Cancer Inst 62:957-974.
Gaylor, DW; Slikker, W. (1990) Risk assessment for neurotoxic effects. Neurotoxicology 11:211-218.
Center, MB; Recio, L. (1994) Absence of detectable P450 2E1 in bone marrow of B6C3F1 mice: relevance to
butadiene-induced bone marrow toxicity. Fundam Appl Toxicol 22:469-473.
Gervasi, PG; Longo, V. (1990) Metabolism and mutagenicity of isoprene. Environ Health Perspect 86:85-87.
Gibson, RW; Bross, IDJ; Lilienfeld, AM; et al. (1968) Leukemia in children exposed to multiple risk factors. New
EnglJMed 279:906-909.
Goodrow, T; Reynolds, S; Maronpot, R; et al. (1990) Activation of K-ras by codon 13 mutations in C57BL/6 x
CSHFj mouse tumors induced by exposure to 1,3-butadiene. Cancer Res 50:4818-4823.
Creep, RO; Weis, L. (1977) Histology. New York: McGraw-Hill.
Grizzle, TB; George, JD; Fail, PA; et al. (1994) Reproductive effects of 4-vinylcyclohexene in Swiss mice assessed
by a continuous breeding protocol. Fundam Appl Toxicol 22:122-129.
R-7
-------
Hackett, PL; Sikov, MR; Mast, TJ; et al. (1987a) Inhalation developmental toxicology studies of 1,3-butadiene in the
rat (final report). Richland, WA: Pacific Northwest Laboratory; PNL Report No. PNL-6414 UC-48; NIH Report No.
NIH-401-ES-41031; 104 p. Prepared for NIEHS, NTP, under a Related Services Agreement with the U.S.
Department of Energy under contract DE-AC06-76RLO-1830.
Hackett, PL; Sikov, MR; Mast, TJ; et al. (1987b) Inhalation developmental toxicology studies: teratology study of
1,3-butadiene in mice (final report). Richland, WA: Pacific Northwest Laboratory; PNL Report No. PNL-6412 UC-
48; NIH Report No. NIH-401-ES-40131; 92 p. Prepared for NIEHS, NTP, under a Related Services Agreement with
the U.S. Department of Energy under contract DE-AC06-76RLO-1830.
Hackett, PL; McClanahan, BJ; Brown, MG; et al. (1988a). Sperm-head morphology study inB6C3F! mice following
inhalation exposure to 1,3-butadiene (final technical report). Richland, WA: Pacific Northwest Laboratory; PNL
Report No. PNL-6459 UC-48; NIH Report No. NIH-Y01-ES-70153; 51 p. Prepared for NIEHS, NTP, under a
Related Services Agreement with the U.S. Department of Energy under contract DE-AC06-76RLO-1830.
Hackett, PL; McClanahan, BJ; Mast, TJ; et al. (1988b). Dominant lethal study in CD-I mice following inhalation
exposure to 1,3-butadiene (final technical report). Richland, WA: Pacific Northwest Laboratory; PNL no. PNL-6545
UC-408; NIH no. NIH-Y01-ES-70153; 85 p. Prepared for NIEHS, NTP, under a Related Services Agreement with
the U.S. Department of Energy under contract DE-AC06-76RLO-1830.
Haley, TJ, Jr. (1978) Chloroprene (2-chloro-l,3-butadiene)—what is the evidence for its carcinogenicity? Clin
Toxicol 13:153-170.
Hallberg, LM; Bechtold, WE; Grady, J; et al. (1997) Abnormal DNA repair activities in lymphocytes of workers
exposed to 1,3-butadiene. MutatRes 383:213-221.
Hallenbeck, W. (1992) Cancer risk assessment for the inhalation of 1,3-butadiene using physiologically based
pharmacokinetic modelling. Bull Environ Contain Toxicol 49:66-70.
Hanks DK; Townsley, WE. (2000) A letter submitted to the editors about lymphatic and haematopoietic cancer
mortality in a population attending school adjacent to styrene-butadiene facilities, 1963-1993. J Epidemiol
Community Health 54:480.
Hattis, D; Wasson, J. (1987) Pharmacokinetic/mechanism-based analysis of the carcinogenic risk of butadiene.
PB88-20817. Rockville, MD: National Institute for Occupational Safety and Health, pp. 1-122.
Hayes, RB; Xi, L; Bechtel, W; et al. (1996) hprt Mutation frequency among workers exposed to 1,3-butadiene in
China. Toxicology 113:100-105.
Hayes, RB; Yin, SN; Dosemeci, M; et al. (1997) Benzene and the dose-related incidence of hematologic neoplasms
in China. J Natl Cancer Inst 89:1065-1071.
Hazleton Laboratories Europe, Ltd. (HLE). (1981a) 1,3-Butadiene: inhalation teratogenicity study in the rat (final
report). Harrowgate, England: Hazleton Labs; addendum no. 2788-522/3.
Hazleton Laboratories Europe, Ltd. (HLE). (1981b) The toxicity and carcinogenicity of butadiene gas administered
to rats by inhalation for approximately 24 months. Prepared for the International Institute of Synthetic Rubber
Producers, New York, NY. Unpublished.
Health Canada. (1998) Canadian Environmental Protection Act Priority Substances List Health Assessment: 1,3-
Butadiene. Draft for second stage peer review. Ottawa. March, 1998.
Heindel, JJ. (1999) Oocyte quantitation and ovarian histology. In: An evaluation and interpretation of reproductive
endpoints for human health risk assessment. Washington, DC: ILSI Press, pp. 57-74.
R-8
-------
Himmelstein, MW; Turner, MJ; Asgharian, B; et al. (1994) Comparison of blood concentrations of 1,3-butadiene
and butadiene epoxides in mice and rats exposed to 1,3-butadiene by inhalation. Carcinogenesis 15:1479-1486.
Himmelstein, MW; Asgharian, B; Bond, JA. (1995) High concentrations of butadiene epoxides in livers and lungs of
mice compared to rats exposed to 1,3-butadiene. Toxicol Appl Pharmacol 132:281-288.
Himmelstein, MW; Acquavella, JF; Recio, L; et al. (1997) Toxicology and epidemiology of 1,3-butadiene. Crit Rev
Toxicol 27(1): 1-108.
Hooser, SB; Parola, LR; VanErt, MD; et al. (1993) Differential ovotoxicity of 4-vinylcyclohexene and its analog, 4-
phenylcyclohexene. Toxicol Appl Pharmacol 119:302-305.
Hooser, SB; Douds, DP; DeMerell, DG; et al. (1994) Long-term ovarian and gonadotropin changes in mice exposed
to 4-vinylcyclohexene. Reprod Toxicol 8:315-323.
Hooser, SB; DeMerell, DG; Douds, DP; et al. (1995) Testicular germ cell toxicity caused by vinylcyclohexene
diepoxide in mice. Reprod Toxicol 9:359-367.
Howe, RB; Van Landingham, C. (1986) GLOBAL86: a computer program to extrapolate quantal animal toxicity
data to low doses. EPA Contract No. 68-01-6826. KS Crump and Company, Ruston, LA. Unpublished.
Hoyer, PB; Sipes, IG. (1996) Assessment of follicle destruction in chemical-induced ovarian toxicity. Annu Rev
Pharmacol Toxicol 36:307-331.
Hughes, CL, Jr; Wall, LL; Creasman, WT. (1991) Reproductive hormone levels in gynecologic oncology patients
undergoing surgical castration after spontaneous menopause. Gynecol Oncol 40:42-45.
Hunt, WF, Jr.; Faoro, R; Hudischewskyj, AB; et al. (1988) The EPA interim database for air toxics volatile organic
chemicals. Systems Applications, Inc., and U.S. Environmental Protection Agency, Office of Air Quality Planning
and Standards, Research Triangle Park, NC. EPA/450/4-88-014.
International Institute of Synthetic Rubber Producers (IISRP). (1982) 1,3-Butadiene: inhalation teratogenicity in the
rat (final report with cover letter dated 08/11/82). Report no. 2788-522/3; submission 8EHQ-0382-0441.
Harrowgate, England: Hazleton Laboratories Europe, Ltd.
International Agency for Research on Cancer (I ARC). (1976) I ARC monographs on the evaluation of carcinogenic
risk of chemicals to humans. Vol. 12. Some carbamates, thiocarbamates, and carbazides. Lyon, France: International
Agency for Research on Cancer, pp 85-95.
International Agency for Research on Cancer (IARC). (1986) IARC monographs on the evaluation of carcinogenic
risk of chemicals to humans. Vol. 39. Some chemicals used in plastics and elastomers butadiene. Lyon, France:
International Agency for Research on Cancer, pp. 155-179.
International Agency for Research on Cancer (IARC). (1991) IARC monographs on the evaluation of carcinogenic
risks to humans. Vol. 53. Occupational exposures in insecticide application, and some pesticides. Lyon, France:
International Agency for Research on Cancer, pp. 403-422.
International Agency for Research on Cancer (IARC). (1999) IARC monographs on the evaluation of carcinogenic
risks to humans. Vol. 71. 1,3-Butadiene. Re-evaluation of some organic chemicals, hydrazine and hydrogen peroxide
(part one). Lyon, France: International Agency for Research on Cancer, pp. 109-225.
Irons, RD. (1990) Studies on the mechanism of 1,3-butadiene-induced leukemogenesis: the potential role of
endogenous murine leukemia virus. Environ Health Perspect 86:49-55.
R-9
-------
Irons, RD; Pyatt, DW. (1998) Commentary: dithiocarbamates as potential confounders in butadiene epidemiology.
Carcinogenesis 19(4):539-542.
Irons, RD; Smith, CN; Stillman, WS; et al. (1986a) Macrocytic-megaloblastic anemia in male B6C3F! Swiss mice
following chronic exposure to 1,3-butadiene. Toxicol Appl Pharmacol 83:95-100.
Irons, RD; Smith, CN; Stillman, WS; et al. (1986b) Macrocytic-megaloblastic anemia in male NIH Swiss mice
following repeated exposure to 1,3-butadiene. Toxicol Appl Pharmacol 85:450-455.
Irons, RD; Cathro, HP; Stillman, WS; et al. (1989) Susceptibility to 1,3-butadiene-induced leukemogenesis
correlates with endogenous ecotropic retroviral background in the mouse. Toxicol Appl Pharmacol 101:170-176.
Jacobson-Kram, D; Rosenthal, S. (1995) Molecular and genetic toxicology of 1,3-butadiene. Mutat Res 339:121-
130.
Jelitto, B; Vangala, RR; Laib, RJ. (1989) Species differences inDNA damage by butadiene: role of diepoxybutane.
Arch Toxicol 13 (suppl.):246-249.
Johanson, G; Filser, JG. (1992) Experimental data from closed chamber gas uptake studies in rodents suggest lower
uptake rate of chemical than calculated from literature values on alveolar ventilation. Arch Toxicol 66:291-295.
Johanson, G; Filser, JG. (1993) A physiologically based pharmacokinetic model for butadiene and its metabolite
butadiene monoxide in rat and mouse and its significance for risk extrapolation. Arch Toxicol 67:151-163.
Johanson, G; Filser, JG. (1996) PBPK model for butadiene metabolism to epoxides: quantitative species differences
in metabolism. Toxicology 113:40-47.
John, EM; Savitz, DA; Sandier, DP. (1991) Prenatal exposure to parents' smoking and childhood cancer. Am J
Epidemiol 133:123-132.
Judd, HL; Lucas, WE; Yen, SS. (1976) Serum 17 beta-estradiol and estrone levels in postmenopausal women with
and without endometrial cancer. J Clin Endocrinol Metab 43:272-278.
Kao, S-W; Sipes, IG; Hoyer, PB. (1999) Early effects of ovotoxicity induced by 4-vinylcyclohexene diepoxide in
rats and mice. Reprod Toxicol 13:67-75.
Kaplan, EL; Meyer, P. (1958) Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-
481.
Kavlock, R; Allen, BC; Kimmel, CA; et al. (1995) Dose-response assessment for developmental toxicity: IV.
Benchmark doses for fetal weight changes. Fundam Appl Toxicol 26:211-222.
Keller, DA; Carpenter, SC; Cagen, SZ; et al. (1997) In vitro metabolism of 4-vinylhexene in rat and mouse liver,
lung and ovary. Toxicol Appl Pharmacol 144:36-44.
Kelsey, KT; Wiencke, JK; Ward, J; et al. (1995) Sister-chromatid exchanges, glutathione S-transferase theta deletion
and cytogenetic sensitivity to diepoxybutane in lymphocytes from butadiene monomer production workers. Mutat
Res 335:267-273.
Kemper, RA; Elfarra, AA. (1996) Oxidation of 3-butene-l,2-diol by alcohol dehydrogenase. ChemRes Toxicol
9:1127-1134.
Kemper, RA; Elfarra, AA; Myers, SR. et al. (1998) Metabolism of 3-butene-l,2-diol in B6C3F1 mice: evidence for
involvement of alcohol dehydrogenase and cytochrome P450. Drug Metab Dispos 26:914-920.
R-10
-------
Kirshenbaum, I. (1978) Butadiene. In: Kirk-Othmer encyclopedia of chemical technology, 3rd ed., vol. 4. Mark, HF;
Othmer, DF; Overberger, CG; et al., eds. New York, NY: John Wiley and Sons, pp. 313-337.
Kligerman, AD; DeMarini, DM; Doerr, CL; etal. (1999) Comparison of cytogenetic effects of 3,4-epoxy-l-butene
and 1,2 : 3,4- diepoxybutane in mouse, rat and human lymphocytes following in vitro G(0) exposures. Mutat Res
439:13-23.
Koc, H; Tretyakova, NY; Walker, VE; et al. (1999) Molecular dosimetry of n-7 guanine adduct formation in mice
and rats exposed to 1,3-butadiene. ChemRes Toxicol 12:566-574.
Kogevinas, M; Ferro, G; Saracci, R; et al. (1993) Cancer mortality in an international cohort of workers exposed to
syrene. IARC SciPub 127:289-300.
Kohn, MC. (1997) The importance of anatomical realism for validation of physiological models of disposition of
inhaled toxicants. Toxicol Appl Pharmacol 147:448-458.
Kohn, MC; Melnick, RL. (1993) Species differences in the production and clearance of 1,3-butadiene metabolites: a
mechanistic model indicates predominantly physiological, not biochemical, control. Carcinogenesis 14:619-628.
Kohn, MC; Melnick, RL. (1996) Effects of the structure of a toxicokinetic model of butadiene inhalation exposure
on computed production of carcinogenic intermediates. Toxicology 113:31-39.
Koivisto, P; Sorsa, M; Pacchierotti, F; et al. (1997) 32P-postlabelling/HPLC assay reveals an enantioselective adduct
formation in N7 guanine residues in vivo after 1,3-butadiene inhalation exposure. Carcinogenesis 18:439-443.
Koivisto, P; Adler, ID; Pacchierotti, F; et al. (1998) DNA adducts in mouse testis and lung after inhalation exposure
to 1,3-butadiene. Mutat Res 397:3-10.
Korach, KS; Couse, JF; Curtis, SW; et al. (1996) Estrogen receptor gene disruption: molecular characterization and
experimental and clinical phenotypes. Recent Prog Horm Res 51:159-186, discussion 186-188.
Kosaric, N; Duvnjak, Z; Farkas, A; et al. (1987) Ethanol. In: Ullmann's encyclopedia of industrial chemistry, 5th
Rev. Ed., Vol. A9. Gerhartz, W, ed. New York, NY:VCH Publishers, p. 590.
Krause, PJ; Elfarra, AA. (1997) Oxidation of butadiene monoide to meso- and (+/-)-diepoxybutane by cDNA-
expessed human cytochrome P450s and by mouse, rat, and human liver microsomes: evidence for preferential
hydration of meso-diepoxybutane in rat and human liver microsomes. Arch Biochem Biophys 337:176-184.
Krause, PJ; Sharer, JE; Elfarra, AA. (1997) Epoxide hydrolase-dependent metabolism of butadiene monoxide to 3-
butene-l,2-diol in mouse, rat, and human liver. Drug Metab Dispos 25:1013-1015.
Kraybill, HF. (1980) Evaluation of public health aspects of carcinogenic biorefractories in drinking water. Prev Med
9:212-218.
Kreiling, R. (1987) Formation of 7-(l-hydroxy-3-butene-2-yl)guanine in liver DNA of mice exposed to 14C-
butadiene. Naunyn-Schmiedebergs Arch Pharmacol 335 (suppl):31.
Kreiling, R; Laib, RJ; Bolt, HM. (1986a) Alkylation of nuclear proteins and DNA after exposure of rats and mice to
[1,4-14C] 1,3-butadiene. Toxicol Lett 30:131-136.
Kreiling, R; Laib, RJ; Filser, JG; et al. (1986b) Species differences in butadiene metabolism between mice and rats
evaluated by inhalation pharmacokinetics. Arch Toxicol 58:235-238.
R-ll
-------
Kreiling, R; Laib, RJ; Filser, JG; et al. (1987) Inhalation pharmacokinetics of l,2-epoxybutene-3 reveal species
differences between rats and mice sensitive to butadiene induced carcinogenesis. Arch Toxicol 61:7-11.
Kreiling, R; Laib, RJ; Bolt, HM. (1988) Depletion of hepatic non-protein sulfhydryl content during exposure of rats
and mice to butadiene. Toxicol Lett 41:209-214.
Kreuzer, P; Kessler, W; Welter, HF; et al. (1991) Enzyme specific kinetics of l,2-epoxybutene-3 in microsomes and
cytosol from livers of mouse, rat, and man. Arch Toxicol 65:59-67.
Lahdetie, J; Peltonen, K; Sjoblom, T. (1997) Germ cell mutagenicity of three metabolites of 1,3-butadiene in the rat:
induction of spermatid micronuclei by butadiene mono-, di-, and diolepoxides in vivo. Environ Mol Mutagen
29:230-239.
Laib, RJ; Filser, JG; Kreiling, R; et al. (1990) Inhalation pharmacokinetics of 1,3-butadiene and l,2-epoxybutene-3
in rats and mice. Environ Health Perspect 86:57-63.
Laib, RJ; Tucholski, M; Filser, JG; et al. (1992) Pharmacokinetic interaction between 1,3-butadiene and styrene in
Sprague-Dawley rats. Arch Toxicol 66:310-314.
Landis, SH; Murray, T; Bolden, S; et al. (1999) Cancer statistics, 1999. CA Cancer J Clin 49:8-31.
Leavens, TL; Moss, OR; Turner, MJ; et al. (1996) Metabolic interactions of 1,3-butadiene and styrene in male
B6C3F1 mice. Toxicol Appl Pharmacol 141:628-636.
Leavens, TL; Farris, GM; James, RA; et al. (1997) Genotoxicity and cytotoxicity in male B6C3F1 mice following
exposure to mixtures of 1,3-butadiene and styrene. Environ Mol Mutagen 29:335-345.
Legator, MS; Au, WW; Ammenheuser, M; et al. (1993) Elevated somatic cell mutant frequencies and altered DNA
repair responses in nonsmoking workers exposed to 1,3-butadiene. In: Butadiene and styrene: assessment of health
hazards. IARC Scientific Publications. Vol. 127. Sorsa, M; Peltonen, K; Vainio, H; et al., eds. Lyon, France:
International Agency for Research on Cancer, pp. 253-263.
Lemen, RA; Meinhardt, TJ; Crandall, MS; et al. (1990) Environmental epidemiologic investigations in the
styrene-butadiene rubber production industry. Environ Health Perspect 86:103-106.
Lemieux, PM; DeMarini, DM. (1992). Mutagenicity of emissions from the simulated open burning and scrap rubber
tires. EPA-600/R-92-127. Washington, DC: U.S. Environmental Protection Agency.
Leukemia Society of America. (1999) Leukemia, lymphomas, and myeloma [on line]. Available from:
http://leukemia.org/docs/leuk_rel.html (1999, April 15).
Leuratti, C; Jones, NJ; Marafante, E; et al. (1994) DNA damage induced by the environmental carcinogen butadiene:
identification of a diepoxybutane-adenine adduct and its detection by 32P-postlabelling. Carcinogenesis
15:1903-1910.
Lewis, DF; Bird, MG; Parke, DV. (1997) Molecular modelling of CYP2E1 enzymes from rat, mouse and man: an
explanation for species differences in butadiene metabolism and potential carcinogenicity, and rationalization of
CYP2E substrate specificity. Toxicology 118:93-113.
Lieser. (1983) Tierexperimentelle Pharmakokinetik von 1,3-Butadiene. Thesis. Johannes-Gutenberg-Universitat,
Mainz, Germany.
Ligocki, MP. (1993) Review of sources and emission estimates for 1,3-butadiene. SYSAPP-93/074. Prepared by
Systems Applications International, Inc., San Rafael, CA.
R-12
-------
Linet, MS. (1985) The Leukemias: epidemiologic aspects. Monographs in epidemiology and biostatistics. Vol. 6.
New York: Oxford University Press, p. 16.
Lofgren, L; Berglund, P; Nordlinder, R; etal. (1991) Selective assessment of C2-C0 alkenes in air by adsorption
sampling and gas chromatography. Int J Environ Anal Chem 45:39-44.
Lofroth, G; Burton, RM, Forehand, L; et al. (1989) Characterization of environmental tobacco smoke. Environ Sci
Technol23:610.
Loughlin, JE; Rothman, KJ; Dreyer, NA. (1999) Lymphatic and haematopoietic cancer mortality in a population
attending school adjacent to styrene-butadiene facilities, 1963-1993. J Epidemiol Community Health 53:283-287.
Lyman, WJ; Reehl, WF; Rosenblatt, DH. (1990) Handbook of chemical property estimation methods. Washington,
DC: American Chemical Society.
Macaluso, M; Larson, R; Delzell, E. (1996) Leukemia and cumulative exposure to butadiene, styrene and benzene
among workers in the synthetic rubber industry. Toxicology 113:190-202.
Maldotti, A; Chiorboli, C; Bignozzi, CA; et al. (1980) Photooxidation of 1,3-butadiene containing systems: rate
constant determination for the reaction of acrolein with -OH radicals. Int J ChemKinet 12:905-913.
Malvoisin, E; Roberfroid, M. (1982) Hepatic microsomal metabolism of 1,3-butadiene. Xenobiotica 12:137-144.
Malvoisin, E; Lhoest, G; Poncelet, F; et al. (1979a) Identification and quantitation of l,2-epoxybutene-3 as the
primary metabolite of 1,3-butadiene. J Chromatogr 178:419-425.
Malvoisin, E; Evrard, E; Roberfroid, M; et al. (1979b) Determination of Kovats retention indices with a capillary
column and electron capture detection: application to the assay of the enzymatic conversion of 3,4-epoxy-l-butene
into diepoxybutane. J Chromatogr 179:81.
Maniglier-Poulet, C; Cheng, X; Ruth, JA; et al. (1995) Metabolism of 1,3-butadiene to butadiene monoxide in
mouse and human bone marrow cells. Chem-Biol Interact 97:119-129.
Mannervik, B. (1985) The isoenzymes of glutathione transferase. Adv Enzymol 57:357-417.
Maronpot, RR. (1987) Ovarian toxicity and carcinogenicity in eight recent National Toxicology Program studies.
Environ Health Perspect 73:125-130.
Matanoski, GM; Schwartz, L. (1987) Mortality of workers in styrene-butadiene polymer production. J Occup Med
29:675-680.
Matanoski, GM; Santos-Burgoa, C. (1994) Butadiene and lymphatic and hematopoietic cancer (letter to the editor).
Epidemiology 5:261.
Matanoski, GM; Schwartz, L; Sperrazza, J; et al. (1982) Mortality of workers in the styrene-butadiene rubber
polymer manufacturing industry. A report prepared under contract to International Institute of Synthetic Rubber
Producers.
Matanoski, GM; Santos-Burgoa, C; Zeger, SL; et al. (1989) Epidemiologic data related to health effects of
1,3-butadiene. In: Assessment of inhalation hazards. Mohr, U; Bates, DV; Dungworth, DL; et al., eds. New York,
NY: Springer-Verlag, pp. 201-214.
Matanoski, GM; Santos-Burgoa, C; Schwartz, L. (1990) Mortality of a cohort of workers in the styrene-butadiene
polymer manufacturing industry (1943-1982). Environ Health Perspect 86:107-117.
R-13
-------
Matanoski, GM; Francis, M; Correa-Villasenor, A. (1993) Cancer epidemiology among styrene-butadiene rubber
workers. In: Butadiene and styrene: assessment of health hazards. Scientific Publications. Vol. 127. Sorsa, M;
Peltonen, K; Vainio, H; et al, eds. Lyon, France: International Agency for Research on Cancer, pp. 363-374.
Matanoski, G; Elliott, E; Tao, X; et al. (1997) Lymphohematopoietic cancers and butadiene and styrene exposure in
synthetic rubber manufacture. Ann N Y Acad Sci 837:157-169.
May, PC; Finch, CE. (1987-88) Aging and responses to toxins in female reproductive functions. Reprod Toxicol
1:223-238.
McAlister, RA; Moore, WH; Rice, J; et al. (1989) Nonmethane organic compound monitoring program final report
volume II: urban air toxics monitoring program, 1988. EPA-450/4-89-005. Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency.
McAlister, RA; Moore, WH; Rice, J; etal. (1990) Urban air toxics monitoring program, 1989. EPA-450/4-91-001.
Research Triangle Park, NC: Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency.
McAlister, RA; Epperson, DL; Jongleux, RF. (1991) Urban air toxics monitoring program aldehyde results, 1989.
EPA-450/4-91-006. Research Triangle Park, NC: Radian Corp. and Office of Air Quality Planning and Standards,
U.S. Environmental Protection Agency.
McCabe, RW; Siegl, O; Chun, W; et al. (1992) Speciated hydrocarbon emissions from the combustion of single
component fuels. II. Catalyst effects. J Air Waste Manage Assoc 42:1071-1077.
McMichael, AJ; Spirtas, R; Kupper, LL. (1974) An epidemiologic study of mortality within a cohort of rubber
workers, 1964-1972. JOccupMed 16:458-464.
McMichael, AJ; Spirtas, R; Gamble, JF; et al. (1976) Mortality among rubber workers: relationship to specific jobs.
JOccupMed 18:178-185.
Medinsky, MA; Leavens, TL; Csanady, GA; et al. (1994) In vivo metabolism of butadiene by mice and rats: a
comparison of physiological model predictions and experimental data. Carcinogenesis 15:1329-1340.
Meinhardt, TJ; Lemen, RA; Crandall, MS; et al. (1982) Environmental epidemiologic investigations of the
styrene-butadiene rubber industry: mortality patterns with discussion of the hematopoietic and lymphatic
malignancies. Scand J Work Environ Health 8:250-259.
Melnick, RL; Huff, JE. (1992) 1,3-Butadiene: toxicity and carcinogenicity in laboratory animals and in humans. Rev
Environ Contain Toxicol 124:111-144.
Melnick, RL; Huff, JE. (1993) 1,3-Butadiene induces cancer in experimental animals at all concentrations from 6.25
to 8,000 parts per million. In: Butadiene and styrene: assessment of health hazards. IARC Scientific Publications.
Vol. 127. Sorsa, M; Peltonen, K; Vainio, H; et al., eds. Lyon, France: International Agency for Research on Cancer,
pp. 309-322.
Melnick, RL; Kohn, MC. (1995) Mechanistic data indicate that 1,3-butadiene is a human carcinogen. Carcinogenesis
16:157-163.
Melnick, RL; Huff, JE; Chou, BJ; et al. (1990a) Carcinogenicity of 1,3-butadiene in C57BL/6 x C3RFl mice at low
exposure concentrations. Cancer Res 50:6592-6599.
Melnick, RL; Huff, JE; Bird, MG; et al. (1990b) Symposium overview: toxicology, Carcinogenesis, and human
health aspects of 1,3-butadiene. Environ Health Perspect 86:3-5.
R-14
-------
Melnick, RL; Huff, JE; Roycroft, JH; et al. (1990c) Inhalation toxicology and carcinogenicity of 1,3-butadiene in
B6C3F! mice following 65 weeks of exposure. Environ Health Perspect 86:27-36.
Melnick, RL; Sills, RC; Roycroft, JH; et al. (1994) Isoprene, an endogenous hydrocarbon and industrial chemical,
induces multiple organ neoplasia in rodents after 26 weeks of inhalation exposure. Cancer Res 54:5333-5339.
Melnick, RL; Sills, RC; Roycroft, JH; et al., (1996) Inhalation toxicity and carcinogenicity of isoprene in rats and
mice: comparisons with 1,3-butadiene. Toxicology 113:247-252.
Melnick, RL; Sills, RC; Portier, CJ; et al., (1999) Multiple organ carcinogenicity of inhaled chloroprene (2-chloro-
1,3-butadiene) in F344/N rats and B6C3F! mice and comparison of dose-response with 1,3-butadiene in mice.
Carcinogenesis 20(5): 867-878.
Meng, QX; Recio, L; Reilly, AA; et al. (1998) Comparison of the mutagenic potency of 1,3-butadiene at the hprt
locus of T-lymphocytes following inhalation exposure of female B6C3F1 mice and F344 rats. Carcinogenesis
19:1019-1027.
Meng, QX; Henderson, RF; Chen, T; et al. (1999a) Mutagenicity of 1,3-butadiene at the Hprt locus of T-
lymphocytes following inhalation exposures of female mice and rats. Mutat Res 429:107-125.
Meng, QX; Henderson, RF; Walker, DM; et al. (1999b) Mutagenicity of the racemic mixtures of butadiene
monoepoxide and butadiene diepoxide at the Hprt locus of T-lymphocytes following inhalation exposures of female
mice and rats. Mutat Res 429:127-140.
Menzel, DB; Wolpert, RL; Boger, JR; et al. (1987) Resources available for simulation in toxicology: specialized
computers, generalized software, and communication networks. In: Pharmacokinetics in risk assessment, Vol. 8,
Drinking water and health, Washington, DC: National Academy Press, pp. 229-250.
Miller, LM. (1978) Investigations of selected potential environmental contaminants: butadiene and its oligomers.
U.S. Environmental Protection Agency, Washington, DC. EPA-560/2-78-008.
Miller, RA; Boorman, GA. (1990) Morphology of neoplastic lesions by 1,3-butadiene inB6C3F! mice. Environ
Health Perspect 86:37-48.
Miller, BA; Ries, LAG; Hankey, BF; et al., eds. (1993) SEER cancer statistics review: 1973-1990. Cincinnati:
National Cancer Institute. 18:8 (NIH Publ No 93-2789).
Monson, RR. (1974) Analysis of relative survival and proportional mortality. Comput Biomed Res 7:325-332.
Morrow, ML. (1990) The industrial production and use of 1,3-butadiene. Environ Health Perspect 86:7-8.
Moutschen, J; Moutschen-Dahmen, M; Loppes, R. (1963) Differential mutagenic activity of L(+) and D(-)
diepoxybutane. Nature (London) 199:406-407.
Nagamani, M; Hannigan, EV; Dillard, EA, Jr; et al. (1986) Ovarian steroid secretion in postmenopausal women
with and without endometrial cancer. J Clin Endocrinol Metab 62:508-512.
National Toxicology Program (NTP), U.S. Public Health Service, U.S. Department of Health and Human Services.
(1984) Toxicology and Carcinogenesis studies of 1,3-butadiene (CAS No. 106-99-0) inB6C3F! mice (inhalation
studies). NTP TR 288, NIH Pub. No. 84-2544. Research Triangle Park, NC.
National Toxicology Program (NTP), U.S. Public Health Service, U.S. Department of Health and Human Services.
(1985) Quarterly report from Lovelace Research Institute, January 1 through March 31, 1985. Interagency agreement
22-Y01-ES-0092. (L. Birnbaum, NTP Project Officer).
R-15
-------
National Toxicology Program (NTP), U.S. Public Health Service, U.S. Department of Health and Human Services.
(1986) Toxicology and carcinogenesis studies of 4-vinylcyclohexene (CAS No. 100-40-3) inF344/N rats and
B6C3F; mice (gavage studies). NTP TR 303. Research Triangle Park, NC.
National Toxicology Program (NTP), U.S. Public Health Service, U.S. Department of Health and Human Services.
(1989) Toxicology and carcinogenesis studies of 4-vinyl-l-cyclohexene diepoxide (CAS No. 106-87-6) inF344/N
rats and B6C3FJ mice (dermal studies). NTP TR 362, NIH Pub. No. 90-2817. Research Triangle Park, NC.
National Toxicology Program (NTP), U.S. Public Health Service, U.S. Department of Health and Human Services.
(1993) Toxicology and carcinogenesis studies of 1,3-butadiene (CAS No. 106-99-0) inB6C3F! mice (inhalation
studies). NTP TR 434. Research Triangle Park, NC.
National Toxicology Program (NTP) (1997). NTP Technical Report on the toxicology and carcinogenesis studies of
isoprene in F344/N rats. Technical Report No. 486 (Draft), U.S. DHHS.
National Toxicology Program (NTP) (1998) Toxicology and carcinogenesis studies of chloroprene inF344/N rats
and B6C3FJ mice. Technical Report No. 467, U.S. DHHS.
National Toxicology Program (NTP), U.S. Public Health Service, U.S. Department of Health and Human Services.
(2000) 9th Report on Carcinogens. NTP, Research Triangle Park, NC. http://ntp-server.niehs.nih.gov.res.
Nauhaus, SK; Fennell, TR; Asgharian, B; et al. (1996) Characterization of urinary metabolites from Sprague-Dawley
rats andB6C3F! mice exposed to [l,2,3,4-C-13]butadiene. ChemRes Technol 9:764-773.
Neal, RA; Halpert, J. (1982) Toxicology of thiono-sulfur compounds. Ann Rev Pharmacol Toxicol 22:321-339.
Nelson, JF; Karelus, K; Bergman, MD; et al. (1995) Neuroendocrine involvement in aging: evidence from studies of
reproductive aging and caloric restriction. Neurobiol Aging 16(5):837-843, discussion 855-856.
Nieusma, JL; Claffey, DJ; Maniglier-Poulet, C; et al. (1997) Stereochemical aspects of 1,3-butadiene metabolism
and toxicity in rat and mouse liver microsomes and freshly isolated rat hepatocytes. Chem Res Toxicol 10:450-456.
Nieusma, JL; Claffey, DJ; Koop, DR; et al. (1998a) Oxidation of 1,3-butadiene to (R)- and (S)-butadiene monoxide
by purified recombinant cytochrome P450 2E1 from rabbit, rat and human. Toxicol Lett 95:123-129.
Nieusma, JL; Claffey, DJ; Ruth, JA; et al. (1998b) Stereochemical aspects of the conjugation of epoxide metabolites
of butadiene with glutathione in rat liver cytosol and freshly isolated rat hepatocytes. Toxicol Sci 43:102-109.
Nordlinder, RG; Nilsson, RI; Buskhe, AB. (1996) 1,3-butadiene in gasoline: an analytical conclusion. J Occup
Environ Med 38:463.
Norppa, H; Sorsa, M. (1993) Genetic toxicology of 1,3-butadiene and styrene. In: Butadiene and styrene: assessment
of health hazards. IARC Scientific Publications. Vol. 127. Sorsa, M; Peltonen, K; Vainio, H; et al., eds. Lyon,
France: International Agency for Research on Cancer, pp. 185-193.
Norppa, H; Hirvonen, A; Jarventaus, H; et al. (1995) Role of GSTT1 and GSTM1 genotypes in determining
individual sensitivity to sister chromatid exchange induction by diepoxybutane in cultured human lymphocytes.
Carcinogenesis 16:1261-1264.
Occupational Safety and Health Administration (OSHA), U.S. Department of Labor. (1990, August 10)
Occupational exposure to 1,3-butadiene; proposed rule and notice of hearing. Federal Register 55:32736-32776.
Occupational Safety and Health Administration (OSHA), U.S. Department of Labor. (1996, November 4)
Occupational exposure to 1,3-butadiene; final rule. Federal Register 61(214):56745-56795.
R-16
-------
Oe, T; Kambouris, SJ; Walker, VE; et al. (1999) Persistence of N7-(2,3,4-trihydroxybutyl)guanine adducts in the
livers of mice and rats exposed to 1,3-butadiene. ChemRes Toxicol 12:247-257.
Okun, AH; Beaumont, JJ; Meinhardt, TJ; et al. (1985) Mortality patterns among styrene exposed boatbuilders. Am J
IndMed 8(3): 193-205.
Osterman-Golkar, SM; Kautianen, A; Bergmark, E; et al. (1991) Hemoglobin adducts and urinary mercapturic acids
in rats as biological indicators of butadiene exposure. Chem Biol Interact 80:291-302.
Osterman-Golkar, SM; Bond, JA; Ward, JB, Jr; et al. (1993) Use of hemoglobin adducts for biomonitoring exposure
of 1,3-butadiene. In: Butadiene and styrene: assessment of health hazards. IARC Scientific Publications. Vol. 127.
Sorsa, M; Peltonen, K; Vainio, H; et al., eds. Lyon, France: International Agency for Research on Cancer, pp. 127-
134.
Osterman-Golkar, SM; Peltonen, K; Anttinen-Klementi, T; et al. (1996) Haemoglobin adducts as biomarkers of
occupational exposure to 1,3-butadiene. Mutagenesis 11:145-149.
Osterman-Golkar, SM; Moss, O; James, A; et al. (1998) Epoxybutene-hemoglobin adducts in rats and mice: dose
response for formation and persistence during and following long-term low-level exposure to butadiene. Toxicol
Appl Pharmacol 150:166-173.
Owen, PE; Glaister, JR. (1990) Inhalation toxicity and carcinogenicity of 1,3-butadiene in Sprague-Dawley rats.
Environ Health Perspect 86:19-25.
Owen, PE; Glaister, JR; Gaunt, IF; et al. (1987) Inhalation toxicity studies with 1,3-butadiene. 3. Two-year
toxicity/carcinogenicity study in rats. Am Ind Hyg Assoc J 48(5):407-413.
Pacchierotti, F; Adler, ID; Anderson, D; et al. (1998a) Genetic effects of 1,3-butadiene and associated risk for
heritable damage. MutatRes 397:93-115.
Pacchierotti, F; Tiveron, C; Ranaldi, R; et al. (1998b) Reproductive toxicity of 1,3-butadiene in the mouse:
cytogenetic analysis of chromosome aberrations in first-cleavage embryos and flow cytometric evaluation of
spermatogonial cell killing. Mutation Res 396:55-66.
PDR - Physicians Desk Reference, 52nd ed. (1998) Antabuse. Montvale, NJ: Medical Company, Inc., pp. 3008-
3009.
Percy, C; Stanek, E; GloecklerL. (1981) Accuracy of cancer death certificates and its effects on cancer mortality
statistics. Am J Public Health 71:242-250.
Perez, HL; Lahdetie, J; Landin, HH; et al. (1997) Haemoglobin adducts of epoxybutanediol from exposure to 1,3-
butadiene epoxides. ChemBiol Interact 105:181-198.
Peter, H; Wiegand, HJ; Bolt, HM; et al. (1987) Pharmacokinetics of isoprene in mice and rats. Toxicol Lett 36:9-14.
Peterson, J; Ward, D. (1989) An inventory of paniculate matter and air toxic emissions from prescribed fires in the
United States for 1989. Final Report IAG #DW12934736-01-0-1989. U.S. Department of Agriculture, Forest
Service.
Placke, ME; Griffis, L; Bird M; et al. (1996) Chronic inhalation oncogenicity study of isoprene in B6C3F! mice.
Toxicology 113:253-262.
Plugge, H; Jaeger, PJ. (1979) Acute inhalation toxicity of 2-chloro-1,3-butadiene (chloroprene): effects onliverand
lung. Toxicol Appl Pharmacol 50:565-572.
R-17
-------
Portier, CJ; Bailer, AJ. (1989) Testing for increased carcinogenicity using a survival-adjusted quanta! response test.
Fundam Appl Toxicol 12:731-737.
Portier, CJ; Hedges, JC; Hoel, DG. (1986) Age-specific models of mortality and tumor onset for historical control
animals in the National Toxicology Program's carcinogenicity experiments. Cancer Res 46:4372-4378.
Pyatt, DW; Gruntmeir, J; Stillman, WS. (1998) Dimethyldithiocarbamate inhibits in vitro activation of human CD4+
T lymphocytes. Toxicology 128(2):83-90.
Ramna's, O; Ostermark, U; Pettersson, G. (1994) Characterization of sixty alkenes in a cat-cracked gasoline naphtha
by gas chromatography. Chromatographia 38:222-223.
Recio, L; Goldsworthy, TL. (1995) The use of transgenic mice for studying mutagenicity induced by 1,3-butadiene.
Toxicol Lett 82:607-612.
Recio, L; Meyer, KG; Pluta, LJ; et al. (1996a) Assessment of 1,3-butadiene mutagenicity in the bone marrow of
B6C3F1 lacl transgenic mice (Big Blue(R)): a review of mutational spectrum and lacl mutant frequency after a 5-day
625 ppm 1,3-butadiene exposure. Environ Mol Mutagen 28:424-429.
Recio, L; Sisk, SC; Meyer, K; et al. (1996b) Mutagenicity and mutational spectra of 1,3-butadiene in the bone
marrow of B6C3F1 lacl transgenic mice. Toxicology 113:106-111.
Recio, L; Pluta, LJ; Meyer, KG. (1998) The in vivo mutagenicity and mutational spectrum at the lacl transgene
recovered from the spleens of B6C3F1 lacl transgenic mice following a 4-week inhalation exposure to 1,3-butadiene.
Mutat Res 401:99-110.
Reitz, RH; Gargas, ML; Hays, SM; et al. (1996) Expansion of the CUT PBPK model for butadiene (description of
diepoxide kinetics). Report prepared for Chemical Manufacturers Association by McLaren/Hart, ChemRisk Division
and RHR Toxicology Consulting, 28 May 1996.
Reynolds, SH; Stowers, SJ; Patterson, RM; et al. (1987) Activated oncogenes in B6C3F! mouse liver tumors:
implications for risk assessment. Science 237:1309-1316.
Richardson, KA; Peters, MMCG; Megens, RHJJJ; et al. (1998) Identification of novel metabolites of butadiene
monoepoxide in rats and mice. ChemRes Toxicol 11:1543-1555.
Robinson, JP; Wiley, JA; Piazza, T; et al. (1989) Activity patterns of California residents and their implications for
potential exposure to pollution. Draft final report. California Air Resources Board, Sacramento, CA.
CARB-46-177-33.
Rolzhauser, HP. (1985) Pharmakokinetik von Butadienmonoxid. Thesis, Johannes-Gutenberg-Universitat, Mainz,
Germany.
Rosenthal, SL. (1985) The reproductive effects assessment group's report on the mutagenicity of 1,3-butadiene and
its reactive metabolites. Environ Mutagen 7:933-945.
Rothman, K; Boice, J. (1982) Epidemiologic analysis with a programmable calculator. Boston, MA: Epidemiology
Resources, Inc., pp. 11-17.
Rubber Manufacturers Association (RMA). (1998) Comments of the Rubber Manufacturers Association on health
risk assessment of 1,3-butadiene - external review draft 63, Federal Register 7167, April 17. Washington, DC.
Russo, A; Nogara, C; Renzi, L; et al. (1997) Micronucleus induction in germ and somatic cells of the mouse after
exposure to the butadiene metabolites diepoxybutane and epoxybutene. Mutat Res 390:129-139.
R-18
-------
Sabourin, PJ; Burka, LT; Bechtold, WE; et al. (1992) Species differences in urinary butadiene metabolites;
identification of l,2-dihydroxy-4-(A^-acetylcysteinyl) butane, a novel metabolite of butadiene. Carcinogenesis
13:1633-1638.
Sandberg, DV; Pickford, SG; Darley, EF. (1975) Emissions from slash burning and the influence of flame retardant
chemicals. J Air Pollut Control Assoc 25:278-281.
Santodonato, J. (1985) Monograph on human exposure to chemicals in the workplace: 1,3-butadiene (PB86-
147261). Washington, DC: U.S. National Technical Information Service.
Santos-Burgoa, C; Matanoski, GM; Zeger, S; et al. (1992) Lymphohematopoietic cancer in styrene-butadiene
polymerization workers. AmJEpidemiol 136:843-854.
Sathiakumar, N; Delzell, E; Hovinga, M; et al. (1998) Mortality from cancer and other causes of death among
synthetic rubber workers. Occup Environ Med 55:230-235.
Sax, NI; Lewis, RJ, Sr. (1987) Hawley's condensed chemical dictionary, llthed. New York, NY: VanNostrand
Reinhold, p. 177.
Schachter, A; Segal, A; Avram, E; et al. (1979) Cytologic evaluation of hormonal dynamic in the postmenopause.
Maturitas 1:269-277.
Schmidt, U; Loeser, E. (1985) Species differences in the formation of butadiene monoxide from 1,3-butadiene. Arch
Toxicol 57:222-225.
Schottenfeld, D; Fraumeni, JF, Jr. (1996) Cancer epidemiology and prevention. 2nd ed. New York: Oxford Univ
Press, p. 856.
Seaton, MJ; Follansbee, MH; Bond, JA. (1995) Oxidation of l,2-epoxy-3-butene to l,2:3,4-diepoxybutane by cDNA
expressed human cytochromes P450 2E1 and 3 A4 and human, mouse and rat liver microsomes. Carcinogenesis
16:2287-2293.
Seaton, MJ; Plopper, CG; Bond, JA. (1996) 1,3-Butadiene metabolism by lung airways isolated from mice and rats.
Toxicology 113:314-317.
Selzer, RR; Elfarra, AA. (1996) Synthesis and biochemical characterization of N1-, N2-, and N7- guanosine adducts
of butadiene monoxide. ChemRes Toxicol 9:126-132.
Shah, JJ; Heyerdahl, EK. (1988) National ambient volatile organic compounds database update (project report). U.S.
Environmental Protection Agency, Atmospheric Sciences Research Laboratory, Research Triangle Park, NC.
EPA/600/3-88/010(a).
Sharer, JE; Elfarra, AA. (1992) S-(2-Hydroxy-3-buten-l-yl)glutathione and S-(l-hydroxy-3-buten-yl-)glutathione are
in vivo metabolites of 1,3-butadiene monoxide: detection and quantitation in bile. ChemRes Toxicol 5:787-790.
Sharer, JE; Duescher, RJ; Elfarra, AA. (1992) Species and tissue differences in the microsomal oxidation of 1,3-
butadiene and the glutathione conjugation of butadiene monoxide in mice and rats. Drug Metab Dispos 20:658-664.
Shields, PG; Xu, GX; Blot, WJ; et al. (1995) Mutagens from heated Chinese and U.S. cooking oils. J Natl Cancer
Inst 87:836-841 [abstract from Toxline 95:258350].
Sielken, RL; Valdez-Flores, C. (2001) Dose-response implications of the University of Alabama study of
lymphohematopoietic cancer among workers exposed to 1,3-butadiene and styrene in the synthetic rubber industry.
ChemBiol Interact 135-136:637-51.
R-19
-------
Sjoblom, T; Lahdetie, J. (1996) Micronuclei are induced in rat spermatids in vitro by 1,2,3,4-diepoxybutane but not
by l,2-epoxy-3-butene and l,2-dihydroxy-3,4-epoxybutane. Mutagenesis 11:525-528.
Smith, BR; Bend, JR. (1980) Prediction of pulmonary benzo(a)pyrene 4,5-oxide clearance: a pharmacokinetic
analysis of epoxide-metabolizing enzymes in rabbit lung. J Pharmacol Exp Ther 214:478-482.
Sobels, FH. (1977) Extrapolation from experimental test systems for evaluation of genetic risks in man. In: Scott, D;
Bridges, BA; Sobels, FH, eds. Progress in genetic toxicology. Amsterdam: Elsevier/North Holland Biomedical
Press; pp. 175-182.
Soderkvist, P; Goodrow, T; Cochran, C; et al. (1992) Characterization of p53 mutations and allelic losses on
chromosome 4, 11, and 14 in butadiene-induced tumors of B6C3F! mice. J Cell Biochem 16B:138.
Sorsa, M; Autio, K; Demopoulos, NA; et al. (1994) Human cytogenetic biomonitoring of occupational exposure to
1,3-butadiene. MutatRes 309:321-326.
Sorsa, M; Osterman-Golkar, S; Peltonen, K; et al. (1996) Assessment of exposure to butadiene in the process
industry. Toxicology 113:77-83.
Springer, LN; McAsey, ME; Flaws, JA; et al. (1996a) Involvement of apoptosis in 4-vinylcyclohexene diepoxide-
induced ovotoxicity in rats. Toxicol Appl Pharmacol 139:394-401.
Springer, LN; Tilly, JL; Sipes, IG; et al. (1996b) Enhanced expression of bax in small preantral follicles during 4-
vinylcyclohexene diepoxide-induced ovotoxicity in the rat. Toxicol Appl Pharmacol 139:402-410.
Springer LN, Flaws JA, Sipes IG, et al. (1996c) Follicular mechanisms associated with 4-vinylcyclohexene
diepoxide-induced ovotoxicity in rats. Reprod Toxicol 10:137-143.
Sram, RJ; Rossner, P; Peltonen, K; et al. (1998) Chromosomal aberrations, sister-chromatid exchanges, cells with
high frequency of SCE, micronuclei and comet assay parameters in 1,3- butadiene-exposed workers. Mutat Res
419:145-154.
SRI International. (1993) 1993 directory of chemical producers-USA. Menlo Park,CA: SRI International.
Startin, JR; Gilbert, J. (1984) Single ion monitoring of butadiene in plastics and foods by coupled mass
spectrometry-automatic headspace gas chromatography. J Chromatogr 294:427-430.
Steen, AM; Meyer, KG; Recio, L. (1997a) Characterization of hprt mutations following l,2-epoxy-3-butene
exposure of human TK6 cells. Mutagenesis 12:359-364.
Steen, AM; Meyer, KG; Recio, L. (1997b) Analysis of hprt mutations occurring in human TK6 lymphoblastoid cells
following exposure to 1,2,3,4-diepoxybutane. Mutagenesis 12:61-67.
Stephanou, G; Andrianopoulos, C; Vlastos, D; et al. (1997) Induction of micronuclei and sister chromatid exchange
in mouse splenocytes after exposure to the butadiene metabolite 3,4-epoxy-l-butene. Mutagenesis 12:425-429.
Stephanou, G; Russo, A; Vlastos, D; et al. (1998) Micronucleus induction in somatic cells of mice as evaluated after
1,3-butadiene inhalation. MutatRes 397:11-20.
Stjernfeldt, M; Berglund, K; Lindsten, J; et al. (1986) Maternal smoking during pregnancy and risk of childhood
cancer. Lancet 1(8494): 1350-1352.
Sudman, PD; Rutledge, JC; Bishop, JB; et al. (1992) Bleomycin: female-specific dominant lethal effects in mice.
Mutat Res 296:143-156.
R-20
-------
Sun, JD; Dahl, AR; Bond, JA; et al. (1989a) Characterization of hemoglobin adduct formation in mice and rats after
administration of [14C]butadiene or [14C]isoprene. Toxicol Appl Pharmacol 100:86-95.
Sun, JD; Dahl, AR; Bond, JA; et al. (1989b) Metabolism of inhaled butadiene to monkeys: comparison to rodents.
ExpPathol 37:133-135.
Sweeney, LM; Himmelstein, MW; Schlosser, PM; et al. (1996) Physiologically based pharmacokinetic modeling of
blood and tissue epoxide measurements for butadiene. Toxicology 113:318-321.
Sweeney, LM; Schlosser, PM; Medinsky, MA; et al. (1997) Physiologically based pharmacokinetic modeling of 1,3-
butadiene, l,2-epoxy-3-butene, and l,2:3,4-diepoxybutane toxicokinetics in mice and rats. Carcinogenesis 18:611-
625.
Swenberg, JA; Christova-Gueorguieva, NI; Upton, PB; et al. (1999) Hemoglobin adducts as biomarkers of
butadiene exposure and metabolism. In: 1,3-Butadiene: cancer, mutations, and adducts. HEI research report number
92.
Swenberg, JA; Christova-Gueorguieva, NI; Upton, PB; et al. (2000) 1,3-butadiene: cancer, mutations, and adducts.
Part V: Hemoglobin adducts as biomarkers of 1,3-butadiene exposure and metabolism. Res Rep Health Eff Inst, pp.
191-210.
Tarone, RE. (1975) Tests for trend in life table analysis. Biometrika 62:679-682.
Tates, AD; van Dam, FJ; de Zwart, FA; et al. (1994) Development of a cloning assay with high cloning efficiency to
detect induction of 6-thioguanine-resistant lymphocytes in spleen of adult mice following in vivo inhalation exposure
to 1,3-butadiene. MutatRes 309:299-306.
Tates, AD; van Dam, FJ; de Zwart, FA; et al. (1996) Biological effect monitoring in industrial workers from the
Czech Republic exposed to low levels of butadiene. Toxicology 113:91-99.
Tates, AD; van Dam, FJ; van Teylingen, CMM; et al. (1998) Comparison of induction of hprt mutations by 1,3-
butadiene and/or its metabolites 1,2-epoxybutene and 1,2,3,4-diepoxybutane in lymphocytes from spleen of adult
male mice and rats in vivo. Mutat Res 397:21-36.
Thornton-Manning, JR; Dahl, AR; Bechtold, WE; et al. (1995a) Disposition of butadiene monoepoxide and
butadiene diepoxide in various tissues of rats and mice following a low-level inhalation exposure to 1,3-butadiene.
Carcinogenesis 16:1723-1731.
Thornton-Manning, JR; Dahl, AR; Bechtold, WE; et al. (1995b) Gender differences in the metabolism of 1,3-
butadiene in Sprague-Dawley rats following low-level inhalation exposure. Carcinogenesis 16:2875-2878.
Thornton-Manning, JR; Dahl, AR; Bechtold, WE; et al. (1997) Comparison of the disposition of butadiene epoxides
in Sprague-Dawley rats and B6C3F1 mice following a single and repeated exposures to 1,3- butadiene via inhalation.
Toxicology 123:125-134.
Thornton-Manning, JR; Dahl, AR; Allen, ML; et al. (1998) Disposition of butadiene epoxides in Sprague-Dawley
rats following exposures to 8000 ppm 1,3-butadiene: comparisons with tissue epoxide concentrations following low-
level exposures. Toxicol Sci 41:167-173.
Thurmond, LM; Lauer, LD; House, RV; et al. (1986) Effect of short-term inhalation exposure to 1,3-butadiene on
murine immune functions. Toxicol Appl Pharmacol 86:170-179.
Tiveron, C; Ranaldi, R; Bassani, B; et al. (1997) Induction and transmission of chromosome aberrations in mouse
oocytes after treatment with butadiene diepoxide. Environ Mol Mutagen 30:403-409.
R-21
-------
Tommasi, AM; deConti, S; Dobrzynska, MM; et al. (1998) Evaluation and characterization of micronuclei in early
spermatids of mice exposed to 1,3-butadiene. MutatRes 397:45-54.
Travis, CC. (1988) Carcinogen risk assessment: contemporary issues in risk analysis. Vol. 3. New York, NY:
Plenum Press.
Tretyakova, N; Lin, YP; Upton, PB; etal. (1996) Macromolecular adducts of butadiene. Toxicology 113:70-76.
Tretyakova, N; Chiang, SY; Walker, VE; et al. (1998) Quantitative analysis of 1,3-butadiene-induced DNA adducts
in vivo and in vitro using liquid chromatography electrospray ionization tandem mass spectrometry. J Mass
Spectrom 33:363-376.
Trochimowicz, HJ; Loser, E; Feron, VJ; et al. (1998) Chronic inhalation toxicity and carcinogenicity studies on P-
chloroprene in rats and hamsters. Inhal Toxicol 10:443-472.
U.S. Department of Health and Human Services (U.S. DHHS), Public Health Service. (1992) Toxicological profile
for 1,3-butadiene.
U.S. Environmental Protection Agency. (1978) Interim primary drinking water regulations. Federal Register
43:20135-29150.
U.S. Environmental Protection Agency. (1985) Mutagenicity and carcinogenicity assessment of 1,3-butadiene.
Office of Health and Environmental Assessment, Office of Research and Development, Washington, DC.
EPA/600/8-85/004F.
U.S. Environmental Protection Agency. (1986, Sept. 24) Guidelines for carcinogen risk assessment. Federal Register
51(185):33992-34003.
U.S. Environmental Protection Agency. (1987) Health assessment document for acetaldehyde. EPA/600/8-86/015A,
external review draft.
U.S. Environmental Protection Agency. (1989a) AIRS user's guide, volumes I-VII. Office of Air Quality Planning
and Standards, Research Triangle Park, NC.
U.S. Environmental Protection Agency. (1989b) 1989 Urban Air Toxics Monitoring Program. Office of Air Quality
Planning and Standards, Research Triangle Park, NC. EPA/450/4-91-001.
U.S. Environmental Protection Agency. (1989c) 1989 Urban Air Toxics Monitoring Program; aldehyde results.
Office of Air Quality Planning and Standards, Research Triangle Park, NC. EPA/450/4-91-006.
U.S. Environmental Protection Agency. (1990a) 1989 Urban Air Toxics Monitoring Program. Office of Air Quality
Planning and Standards, Research Triangle Park, NC. EPA/450/4-91-024.
U.S. Environmental Protection Agency (1990b) 1990 Urban Air Toxics Monitoring Program carbonyl results. Office
of Air Quality Planning and Standards, Research Triangle Park, NC. EPA/450/4-91-025.
U.S. Environmental Protection Agency. (1991) Non-road engine and vehicle emission study. Office of Mobile
Sources, Ann Arbor, MI. EPA 21A-2001.
U.S. Environmental Protection Agency. (1992) Compilation of air pollutant emission factors (AP-42). Supplement
E, October 1992. U.S. Environmental Protection Agency.
U.S. Environmental Protection Agency. (1993a) Motor vehicle-related air toxics study. Office of Mobile Sources,
Ann Arbor, MI. EPA/420-R-93-005.
R-22
-------
U.S. Environmental Protection Agency. (1993b) The Integrated Air Cancer Project, Boise Study; research to
improve risk assessment of area sources, woodstoves and mobile sources. Office of Research and Development.
Presentation to the Office of Mobile Sources, August 1993.
U.S. Environmental Protection Agency. (1994a) Locating and estimating air emissions from sources of 1,3-
butadiene. (September 1994 draft.) Office of Air Quality Planning and Standards, Research Triangle Park, NC.
U.S. Environmental Protection Agency. (1994b) Memo from Rich Cook, Office of Mobile Sources to Dennis
Beauregard, Office of Air Quality Planning and Standards. November 9, 1994.
U.S. Environmental Protection Agency. (1994c) Methods for Derivation of Inhalation Reference Concentrations and
Application of Inhalation Dosimetry. Office of Research and Development, Washington, DC. EPA/600/8-90/066F.
U.S. Environmental Protection Agency. (1995) The use of the benchmark-dose approach in health risk assessment.
Office of Research and Development, Washington, DC. EPA/630/R-94/007.
U.S. Environmental Protection Agency. (1996, April 23) Proposed guidelines for carcinogen risk assessment.
Federal Register 61(79):17960-18011.
U.S. Environmental Protection Agency. (1999a) Estimation of motor vehicle toxic emissions and exposure in
selected urban areas. Office of Mobile Sources, Ann Arbor, MI. EPA420-D-99-002. March 1999.
U.S. Environmental Protection Agency. (1999b, July) Guidelines for carcinogen risk assessment. Review Draft,
NCEA-F-0644, July 1999. Risk Assessment Forum, http://www.epa.gov/ncea/raf/cancer.htm.
U.S. Environmental Protection Agency. (2001) Toxics release inventory. Office of Toxic Substances, Washington,
DC. http://www.epa.gov/tri/chemical.htm.
U.S. International Trade Commission (USITC). (1990) Synthetic organic chemicals; United States production and
sales. 1989. Pub. 2338. Washington, DC: Government Printing Office.
Uuskula, M; Jarventaus, H; Hirvonen, A; et al. (1995) Influence of GSTM1 genotype on sister chromatid exchange
induction by styrene-7,8-oxide and l,2-epoxy-3-butene in cultured human lymphocytes. Carcinogenesis 16:947-950.
Vagenakis, AG. (1989) Endocrine aspects of menopause. Clin Rheumatol Suppl 2:48-51.
Valentine, JL; Boogaard, PJ; Sweeney, LM; et al. (1997) Disposition of butadiene epoxides in Sprague-Dawley rats.
ChemBiol Interact 104:103-115.
VanDuuren, BL; Nelson, N; Orris, L; etal. (1963) Carcinogenicity of epoxides, lactones, andperoxy compounds. J
Natl Cancer Inst 31:41-55.
Van Duuren, BL; Langseth, L; Orris, L; et al. (1966) Carcinogenicity of epoxides, lactones, and peroxy compounds.
IV. Tumor response in epithelial and connective tissue in mice and rats. J Natl Cancer Inst 37:825-838.
Vermeulen, A. (1976) The hormonal activity of the postmenopausal ovary. J Clin Endocrinol Metab 42:247-253.
Verschueren, K. (1983) Handbook of environmental data on organic chemicals, 2nd ed. New York, NY: Van
Nostrand Reinhold, pp. 295-297.
Villani, P; Andreoli, C; Crebelli, R; et al. (1998) Analysis of micronuclei and DNA single-stand breaks in mouse
splenocytes and peripheral lymphocytes after oral administration of tetramethylthiuram disulfide (thiram). Food
ChemToxicol 36(3): 155-164.
R-23
-------
Ward, DE; Hao, WM. (1992) Air toxic emissions from burning of biomass globally—preliminary estimates. Paper
92-98.03 Presented at the 85th Annual Meeting of the Air and Waste Management Association, Kansas City, MO.
Ward, JB, Jr; Au, WW; Ammenheuser, MM; et al. (1994) hprt Mutant lymphocyte frequencies in workers at a 1,3-
butadiene production plant. Environ Health Perspect 102:79-85.
Ward, EM; Fajen, JM; Ruder, A. (1995) Mortality study of workers in 1,3-butadiene production units identified from
a chemical workers cohort. Environ Health Perspect 103:598-603.
Ward, JB, Jr.; Ammenheuser, MM; Wharton, EB, Jr.; et al. (1996a) Biological monitoring for mutagenic effects of
occupational exposure to butadiene. Toxicology 113:84-90.
Ward, EM; Fajen, JM; Ruder, A. (1996b) Mortality study of workers employed in 1,3-butadiene production units
identified from a large chemical workers cohort. Toxicology 113:157-168.
Weast, RC, ed. (1989) CRC handbook of chemistry and physics, 70th ed. Boca Raton, FL:CRC Press, p. C-160.
Weller, E; Long, N; Smith, H; et al. (1999) Dose-rate effects of ethylene oxide exposure on developmental toxicity.
Toxicol Sci 50(2):259-270.
Wijnberg, L; Faoro, R. (1989) Urban air toxics monitoring program—results of aldehyde monitoring fiscal year
1988. PEI Associates and the U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards.
Research Triangle Park, NC. September 1989.
Wise, PM; Kashon, ML; Krajnak, KM; et al. (1997) Aging of the female reproductive system: a window brain
aging. Recent Prog Horm Res 52:279-303, discussion 303-305.
Wistuba, D; Weigand, K; Peter, H. (1994) Stereoselectivity of in vitro isoprene metabolism. ChemRes Toxicol
7(3):336-343.
Wong, O. (1990) A cohort mortality study and a case-control study of workers potentially exposed to styrene in the
reinforced plastics and composites industry. Br J Ind Med 47(ll):753-762.
Xiao, Y; Tates, AD (1995) Clastogenic effects of 1,3-butadiene and its metabolites 1,2-epoxybutene and 1,2,3,4-
diepoxybutane in splenocytes and germ cells of rats and mice in vivo. Environ Mol Mutagen 26:97-108.
Xiao, Y; de Stoppelaar, JM; Hoebee, B; et al. (1996) Analysis of micronuclei induced by 1,3-butadiene and its
metabolites using fluoresecence in situ hybridization. Mutat Res 354:49-57.
R-24
------- |