vvEPA
United States
Environmental Protection
Agency
Office of Health and
Environmental Assessment
Washington DC 20460
EPA/600/8-85/004A
February 1985
Review Draft
Research and Development
Mutagenicity and
Carcinogenicity
Assessment of
1,3-Butadiene
Review
Draft
(Do Not
Cite or Quote)
Notice
This document is a preliminary draft. It has not been formally
released by EPA and should not at this stage be construed to
represent Agency policy. It is being circulated for comment on its
technical accuracy and policy implications.
-------
DRAFT EPA/600/8-85/004A
DO NOT QUOTE OR CITE February 1985
Review Draft
MUTAGENICITY AND CARCINOGENICITY ASSESSMENT
OF
1,3-BUTADIENE
NOTICE
THIS DOCUMENT IS A PRELIMINARY DRAFT. It has not been formally released by
the U.S. Environmental Protection Agency and should not at this stage be
construed to represent Agency policy. It is being circulated for comments
on its technical accuracy and policy implications.
.'.:', Environmental Protection
• • 'oci 5, Library (5PL-16)
'-,j S. Dearborn Street, Room 167Q
Cljioago,. I& 60604
Office of Health and Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, D.C.
-------
DISCLAIMER
This document is an external draft for review purposes only and does
not constitute Agency policy. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
-------
DRAFT —DO NOT QUOTE OR CITE
CONTENTS (1,3-Butadiene)
Preface. v
Authors3 Contributors, and Reviewers vi
1. SUMMARY AND CONCLUSIONS . 1
1.1. Summary 1
1.2. Conclusions ........ 5
2. INTRODUCTION '..... 7
3. MUTAGENICITY OF 1,3-BUTADIENE AND ITS REACTIVE METABOLITES .... 9
3.1. Mutagenicity of 1,3-Butadiene 9
3.2. Metabolism of 1,3-Butauiene and Reaction of Metabolites
with .DM,,, .,.....--.,..'-..^..1 .... ^s--J< 10
3.3. M'ltagenicity of 3,4-Epoxybutene 13
3.4. Genotoxicity of 1 ,2:3,4-Diepoxybutane 14
3.4.1. Studies in Bacteria 14
3.4.2. Studies in Fungi 15
3.4.3. Studies in Mammalian Cells 17
3.4.4. JJT_ vi_vo Studies 19
3.5. Summary of Mutaoenicity Studies 26
4. ^ CARCINOGENICITY 27
4.1. Toxicology and Pnarmacokinetics 27
4.2. Animal Studies 29
4.2.1. Chronic Toxicity Studies in Mice 29
4.2.2. Chronic Toxicity Studies in Rats 35
4.2.3. Summary of Chronic Toxicity Studies. ....... 39
4.3. Epidemiologic Studies 39
4.3.1. McMicnael et al . (1974, 1976) 41
4.3.2. Andjelkovich et al . (1976, 1977) 44
4.3.3. Checkoway and Williams (1982) 50
4.3.4. Meinhardt et al . (1982) 52
4.3.5 Matanoski et al . (1982) 55
4.3.6. Summary of rpidemiologic Studies 59
4.4. Quantitative Estimation 62
4.4.1. Procedures for Determination of Unit Risk 62
4.4.1.1. Description of the Low-Dose Extrapo-
lation Model 64
4.4.1.2. Selection of Data 67
i i i
-------
DRAFT—DO NOT QUOTE OR CITE
CONTENTS (continued)
4.4.1.3. Calculation of Human Equivalent
Dosages from Animal Data 68
4.4.1.4. Calculation of the Unit Risk from
Animal Studies 70
4.4.1.4.1. Adjustments for Less
Than Lifetime Duration
of Experiment 70
4.4.1.5. Interpretation of Quantitative
Estimates 71
4.4.1.6. Alternative Models 72
4.4.2. Calculation of Quantitative Estimates 73
4.4.3. Comparison of Human and Animal
Inhalation Studies 76
4.4.4. Relative Potency 84
4.4.5. Summary of Quantitative Estimation 90
REFERENCES 92
IV
-------
PREFACE
The Mutagenicity and Carcinogenicity 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 Office of Air Quality Planning and Standards
to support decision-making regarding possible regulation of 1,3-butadiene as a
hazardous air pollutant.
In the development of the assessment document, the scientific literature
has been inventoried, key studies have been evaluated, and the summary and
conclusions have been prepared so that the mutagenicity, carci nogenici ty, m
-------
AUTHORS, CONTRIBUTORS, AND REVIEWERS
The Carcinogen Assessment Group and the Reproductive Effects Assessment
Group within the Office of Health and Environmental Assessment were responsible
for preparing this document.
PRINCIPAL AUTHORS
Steven Bayard, Ph.D. Chapters 1 and 4
Robert P. Beliles, Ph.D. Chapters 1, 2, and 4
Arthur Chiu, Ph.D, M.D. Chapters 1, 2, and 4
Herman J. Gibb, B.S., M.P.H. Chapters 1 and 4
Aparna Koppikar, M.B.B.S.* Chapters 1 and 4
Sneila L. Rosenthal, Ph.D. Chapter 3
•*con$'il tant
Participating Members - CAG Participating Members - REAG
Roy E. Albert, M.D., Chairman Eric Clegg, Ph.D.
David L. Bayliss, M.S. David JacQbson-Kram, Ph.SJ.
Chao W. Chen, Ph.D. Carole Kimmel, Ph.O.
Margaret M.L. Chu, Ph.D. Gary L. Kimmel, Ph.i).
Bernard H. Haberman, D.V.M, M.S. Carol N. Sakai , Ph.D.
Charalingayya B. Hiremath, Ph.D. Ann Huang Sciambi, Ph.D.
James W. Holder, Ph.D. Lawrence Valcovic, Ph.D.
Robert E. McGaughy, Ph.D. Vicki Vaughan-DeIlarci , Ph.D.
Jean C. Parker, Ph.D. Peter E. Voytek, Ph.D.
Wi I Ham E. Pepelko, Ph.D.
Dharm V. Singh, D.V.M., Ph.D.
Todd W. Thorslund, Sc.D.
REVIEWERS
The following individuals provided peer review of the mutagenicity chapter or
this document.
George R. Hoffman, Ph.D.
Department of Biolony
Ho ly Cr oss ;,ol KV-'
Worcester, MA
Stanley Zimmering, Ph.D.
Division of Biology and Medicine
Brawn hnivers i ty
Providence, RI
-------
The following individuals provided peer review of this draft and/or earlier
drafts of this document.
Harriet Ammann, Ph.D.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
Research Triangle Park, NC
Ila Cote, Ph.D.
Pollutant Assessment Branch
Strategies and Air Standards Division
Office of Air Quality Planning and Standards
Research Triangle Park, NC
Daphne Kamely, Ph.D.
Exposure Assessment Group
Office of Health and Environmental Assessment
Washington, DC
Debdas Mukerjee, Ph.D.
Environmental Criteria and Assessment Office
Office of Health and Environmental Assessment
Research Triangle Park, NC
David Patrick
Pollutant Assessment Branch
Strategies and Air Standards Division
Office* of Air Quality Planning and Standards
Research Triangle Park, NC
Ani ta Schmidt
Risk Management Branch
Existing Chemical Assessment Division
Office of Toxic Substances
Washington, DC
Marvin A. Schneiderman, Ph.D.
6503 E. Halbert Road
Bethesda, MD
Jeannette Wiltse, Ph.D.
Risk Management Branch
Existing Chemical Assessment Division
Office of Toxic Substances
Washington, i.«>:
vn
-------
DRAFT—DO NOT QUOTE OR CITE
1. SUMMARY AND CONCLUSIONS
1.1. SUMMARY
1,3-butadiene is a colorless gas with a slight aromatic odor at room
temperature and pressure. It is used mainly by the styrene-butadiene rubber
and polybutadiene rubber industries. No deaths and very few toxic effects have
been reported from acute exposure to the vapor. The symptoms resulting from
acute exposures are lethargy, drowsiness, and irritation to the mucous membranes
of the eyes and the mouth.
The available information on the mutagenicity of 1,3-butadiene is quite
limited in that only two studies have been reported. Both studies, however,
indicate that 1,3-butadiene is a mutagen in Salmonella typhimurium. The
mutagenicity is observed only in the presence of a liver S9 metabolic acti-
vation system. No whole animal studies have been reported. These results
suggest that 1,3-butadiene is a promutagen in bacteria (i.e., its mutageni-
*
city depends on metabolic activation).
There is no information on the metabolism of 1,3-butadiene in humans.
In vitro data suggest that 1,3-butadiene is metabolized to 3,4-epoxybutene
and then to diepoxybutane. Preliminary evidence in rats suggests that 1,3-
butadiene is metabolized to 3,4-epoxybutene in vivo (Bolt et al., 1983),
indicating that the metabolic pathway outlined on the basis of in vitro data
may occur in vivo. More detailed studies by Bolt and his coworkers have been
published recently; the results of these studies will be included in the final
document.
3,4-Epoxybutene is a monofunctional alkylating agent and is a direct-
acting mutagen in bacteria. It has been tested in Klebsiella pneumoniae and
-------
DRAFT--DO NOT QUOTE OR CITE
Escherichia coli. Diepoxybutane is a bifunctional alkylating agent, and as
such it can form cross-links between two strands of DNA. It is mutagenic in
bacteria (K. pneumoniae and S. typhimurium), fungi (yeast and Neurospora
crassa), and the germ cells of Drosophila melanogaster. It also induces DNA
damage in cultured hamster cells and in mice, is clastogenic in fungi and
cultured rat cells, and produces chromosome damage/breakage in D. melanogaster
germ cells. Therefore, there is strong evidence that diepoxybutane is a muta-
gen/clastogen in microbes and animals.
Two lifetime inhalation carcinogenicity studies have been carried out in
mice and rats. There was a marked increase in incidences of primary tumors
among the exposed groups of mice in both sexes. These tumors included lymphomas,
hemangiosarcomas, alveolar/bronchiolar adenomas (and carcinomas), acinar
cell carcinomas, granulosa cell tumors or carcinomas, forestornach papillomas
and carcinomas, and hepatocellular adenomas and carcinomas. The study had
to be-terminated at 60-61 weeks instead of the planned 104 weeks because of
excessive deaths from the neoplasia among the exposed mice.
In female rats (Sprague-Dawl ey) exposed to 1,3-butadiene, increased inci-
dences of mammary tumors, thyroid follicular cell adenomas, and uterine stromal
sarcomas were observed. In the male rats, increases in tumor incidences were
found in the exposed animals in the form of Leydig cell tumors and exocrine
pancreatic adenomas. Zymbal gland tumors were increased in both sexes of ex-
posed rats. The tumor sites involved were different in the mice and rats among
the exposed groups. The severity of the cancers was also widely different; in
the rats, no increase in mortality secondary to neoplasia was observed, and
there was no early termination of the experiments. In addition to the dif-
ferences found in the two sexes, rats were affected less than mice.
-------
DRAFT—DO NOT QUOTE OR CITE
Epidemiclogic studies of the potential health hazards associated with
1,3-butadiene exposure are limited. There are a number of occupational studies
of rubber workers, but only a few of these studies were considered relevant for
this review. These studies concerned workers who were specifically identified
as having worked in the production of synthetic 1,3-butadiene rubber or as
having worked in a synthetic rubber plant. Of the three studies on workers
specifically identified as being exposed to styrene-1,3-butadiene, two of them
also included workplace air sampling. In one study, air samples were analyzed
for 1,3-butadiene, styrene, and benzene. In the other study, the air concen-
trations for 1,3-butadiene, benzene, styrene, and toluene were reported. Both
of these studies found that the time-weighted average for each of the chemical
exposures examined was well below the American Conference of Governmental
Industrial Hygienists (ACGIH) threshold limit values (TLV) at that time for
these chemicals (1,3-butadiene: TLV = 1,000 ppm; styrene: TLV = 100 ppm;
toluene: TLV = 100 ppm; benzene: TLV = 10 ppm). One of the studies was a
cross-sectional investigation designed to look at certain hematologic parameters,
This investigation revealed no evidence of any hematologic abnormality in the
study population.
A case-control study of deaths among rubber plant workers from cancer of
certain sites, diabetes mellitus, and ischemic heart disease found workers in
the synthetic rubber area of the plant to have the highest risk ratio for
deaths from lymphatic and hematopoietic cancer (ICD 200-209). The same authors,
however, had previously found an association of lymphatic leukemia cancer
(:>*{",, wua of gin,c solvent exposures in the rubber industry.
A cohort study of styrene-butadiene rubber (SBR) workers found that the
Standardized Mortality Ratio (SMR) for lymphatic and hematopoietic cancer was
of borderline significance for a subcohort of workers employed at one plant
-------
DRAFT—DO NOT QUOTE OR CITE
during the time when a batch production process was in operation. Here again,
solvent exposure may have been a confounding factor. There is also some limited
evidence that styrene may be a carcinogen, and in particular, a leukemogen.
This factor poses further complications for evaluating the epidemiologic studies
of SBR workers with regard to 1,3-butadiene.
A cohort mortality study of rubber plant workers found that an excess of
lung cancer deaths occurred among workers in the synthetic rubber area of the
plant. This was based on only three deaths, however, and there was no control
for smoking.
A large cohort study of almost 14,000 SBR workers at eight plants found
none of the SMRs for cancer to be significantly elevated. Some bias may have
occurred, however, due to a possible underascertainment of total deaths and a
possible overestimation of deaths among blacks. Given the inconsistency of
results from different studies, the possible confounding due to exposure to
solvents, styrene, and possibly other chemicals, and the potential biases
in some of the studies, the epidemiologic data would have to be considered
inadequate for evaluating whether a causal association exists between
1,3-butadiene exposure and cancer in humans.
Based on the linearized multistage model, a maximum likelihood estimate of
incremental unit risk was calculated for 1,3-butadiene, using the geometric
mean from the pooled male and pooled female significant tumor responses of the
NTP mouse study. The mean value of q^ = 9.0 x 10"3 (ppm) was then used to pre-
dict human responses in several epidemiologic studies, and the predicted and
actual responses were compared. The comparisons were hampered by a scarcity of
information concerning actual exposures, age distributions, and work histories.
In addition, because there was no consistent cancer response across all of the
studies, the most predominant response, cancer of the lymphatic and hemato-
-------
DRAFT--DO NOT QUOTE OR CITE
poietic tissues, was chosen as being the target for 1,3-butadiene. Based on
the comparisons between the predicted and observed human response, the extrapo-
lated unit risk value from the mice data appeared slightly high, but in view of
the uncertainties in the epidemiologic data, no better estimate can be made at
this time.
1.2. CONCLUSIONS
1,3-butadiene has been shown to be an indirect mutagen in bacteria. One
of its potential metabolites, diepoxybutane, is mutagenic in prokaryote as well
as eukaryote test systems. Exposure of rodents to 1,3-butadiene results in
ovarian tumors in mice (Huff et al., 1985) and testicular tumors in rats
(Hazleton Laboratories Europe, Ltd., 1981a). Therefore, these data are sug-
gestive that 1,3-butadiene may present a heritable risk to humans. However,
additional studies in mammalian test systems, as outlined in the Agency's Pro-
posed Guidelines for Mutagenicity Risk Assessment (1984), should be conducted
to further characterize the mutagenic potential of 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,
according to the International Agency for Research on Cancer (IARC) classifi-
cation scheme, as a "probable" human carcinogen, Group 2B. It was placed in
Sub-Group B because of the inadequacy of the available epidemiologic data.
Using the newly proposed EPA classification scheme for carcinogenicity evidence,
the ranking would be B2, meaning that 1,3-butadiene is a "probable" human
carci nogen.
Using a linearized multistage model, a 95% upper-limit incremental unit
risk for 1,3-butadiene is estimated from a NTP mouse study to be q^ = 6.5
x 10~2 (ppm). In addition to a 95% upper-limit incremental unit risk, a mea-
sure of carcinogenic potency was determined for 1,3-butadiene. Among the 54
-------
DRAFT—00 NOT QUOTE OR CITE
chemicals that the CAG has evaluated as suspect carcinogens, 1,3-butadiene
ranks fairly low in potency, placing at the top of the fourth quartile.
-------
OKAFT--UO NOT QUOTE OR CITE
2. INTRODUCTION
1,3-Butadiene (CAS No, 106-99-00) is a colorless gas produced as an ethy-
lene coproduct, by oxidative dehydrogenation of n-butenes, or by dehydrogena-
tion of n-butanes. In 1977, between 2.1 and 7.3 billion pounds of 1,3-butadiene
were produced or imuorced. 1,3-Butaaiene ranked 36th in U.S. domestic chemical
production in 1983. it, > ?. usec as an intermediate in the production of poly-
mers, elastomers, and othtr cnemitals. The major use of 1,3-butadiene is in
the manufacture cf styri-rif-but-ic:1 ene rubber (synthetic rubber). In addition,
1,3-butadiene is uc.ed -s ?>;• viiv mediate to produce a variety of industrial
chemicals, including tr.t fi,jv;!i<. < des. capr.an and captofol . The U.S. Food and
Drug Administration has approved 1,3-butadiene for use in the production of
adhesives used in cental • types of food containers.
Although 1,3-butadiene has '^en found in U.S. drinking water, it is pri-
marily an air contannnant. It has been detected in cigarette smoke, incinera-
tion products of fossil fuels, gasoline vapor, and automotive exhaust. Con-
centrations ranging froip j to 5 pob have been detected in urban air. Higher
concentrations, up to 45 pom, have been reported in air samples and factory
emissions at petrochemical plants.
Approximately 62,000 workers are exposed annually to 1,3-butadiene. Occu-
pational exposure to 1,3-butadiene occurs mainly through inhalation and, to a
lasser extent, by dermal contact. Most occupational exposures occur in plants
manufacturing 1,3-butadiene or using it to produce polymers or elastomers. The
••rrent peri.i \ ss i t: i e exposure limit of 1,000 ppm as an 8-hour time-weighted
average was adopted by tnp Occupational Safety and Health Administration from
the 1968 Threshold Limit Values (TLV) set by the American Conference of Govern-
mental industrial Hygiemsts (AC'^JH). In 1983, the ACGIH, based on the findings
7
-------
DRAFT—DO NOT QUOTE OR CITE
in experimental animals, gave notice of intention to list the material as an
industrial substance suspected of carcinogenic potential for man and to remove
the TLV. The National Institute for Occupational Safety and Health (NIOSH),
considering the same animal information, issued a Current Intelligence Bulletin
recommending that the occupational exposure be reduced to the lowest feasible
level because of the potential for the compound to produce cancer.
The National Toxicology Program is currently undertaking a series of in-
vestigations of 1,3-butadiene. These studies will provide information on the
pharmacokinetics and toxicity of the chemical. The reason for the apparent
differences in sensitivity between the rat and mouse are to be investigated.
In addition, the carcinogenic response at lower airborne concentrations may be
establi shed.
-------
DRAFT—DO NUT QUOTE OR CITE
3. MUTAGENICITY OF 1,3-BUTADIENE AND ITS REACTIVE METABOLITES
This chapter deals with the mutagenicity of 1,3-butadiene, which is a gas
at room temperature, and also includes a discussion of the metabolism of 1,3-
butadiene and the mutagenicity of its reactive metabolites (3,4-epoxybutene and
l,2:3,4-diepoxybutane). The available evidence suggests that 1,3-butadiene is
mutagenic by virtue of its metabolism to mutagenic intermediates.
3.1. MUTAGENICITY OF 1,3-BUTADIENE
1,3-Butadiene was tested for its mutagenic potential in the Salmonella
typhimurium histidine reversion assay by de Meester et al. (1980). The sample
of 1,3-butadiene studied was 99.5% pure and was obtained from Matheson Gas Pro-
ducts, Belgium. Salmonella strain TA1530 was exposed to 1,3-butadiene for 24
hours at 0, 4, 8, 16, 24, and 32% (vol/vol) in controlled atmospheres in desic-
cators. In the absence of S9 mix or in the presence of S9 prepared from un-
treated, rats, no increase in the revertant frequency was observed. However,
when the bacteria were exposed to 1,3-butadiene in the presence of S9 mix pre-
pared from phenobarbitone or Aroclor 1254-pretreated rats, mutagenic activity
was observed. The number of histidine revertants increased in a dose-related
fashion from 17 per plate in the absence of 1,3-butadiene up to 255 per plate
at 16% (vol/vol) 1,3-butadiene. These results suggest that 1,3-butadiene
itself is not a mutagen, and that it is metabolized into mutagenic metabolic
intermediates that cause base-pair substitutions.
Data suggesting that the mutagenic metabolites are volatile were also
reported by de Meester et al. (1980). When plates containing bacteria and
phenobarbitone or Aroclor-induced S9 mix were coincubated in 4 to 32% 1,3-
butadiene atmospheres with plates containing only the bacteria, mutant colo-
nies appeared on both sets of plates. The number of mutant colonies was pro-
-------
DRAFT—DO NOT QUOTE OR CITE
portional to the level of i,3-butadiene in the desiccators up to 16%.
A small study by Poncelet et al. (1980) supports the conclusion of de
Meester et al. (1980) on the mutayenic potential of 1,3-butadiene in Salmonella
strain TA1530. Mutagenic effects were observed when the assays were performed
in a 16% gaseous atmosphere of 1,3-butadiene in the presence of Aroclor-induced
S9 mix. When the bacteria were exposed to 1,3-butadiene under the conditions
of the plate incorporation method or preincubation in liquid medium, mutageni-
city was not observed. The 1,3-butadiene sample was 99.5% pure and was obtained
from Matheson Gas Products, Belgium.
In summary, the two available studies suggest that 1,3-butadiene is a
promutayen in bacteria; its mutagenicity depends on metabolic activation by a
chemically induced S9 mix. No animal studies have been reported.
3.2. METABOLISM OF 1,3-BUTADIENE AND REACTION OF METABOLITES WITH DNA
As described in the previous section, 1,3-butadiene itself does not appear
to be rautagenic. Mutagenic activity is observed only when 1,3-butadiene has
been metabolized to reactive intermediates. A scheme for the metabolism of 1,3-
butadiene and the DNA binding potential of the probable metabolites are briefly
discussed in this section.
Malvoisin et al. (1979) and Malvoisin and Roberfroid (1982) studied the
metabolism of 1,3-butadiene ij^ vitro using rat liver microsomes. They reported
that the metabolism proceeds via a mixed-function oxidase-catalyzed oxidation
to 3,4-epoxybutene, and they suggest that this compound is subsequently metabo-
lized to l,2:3,4-diepoxybutane (diepoxybutane) and 3-butene-l,2-diol (Figure
1). Both 3,4-epoxybutene and diepoxybutane are probably reactive intermediates,
whereas 3-butene-l,2-diol and its metabolite 3,4-epoxy-l,2-butanediol are pro-
bably detoxification products. Preliminary evidence in rats suggests that 1,3-
butadiene is metaboliz'ed to 3,4-epoxybutene in vivo (Bolt et al., 1983). Both
10
-------
Epoxide
hydrolase
CH2=CH-CHOH-CH2OH
3-Butene-l,2-diol
I
CH2=CH-CH=CH2
1,3-Butadiene
NADPH
02, microsomes
CH2=CH-CH-CH2
\/
0
3,4-Epoxybutene
NADPH
02, microsomes
NADPH
02, microsomes
A
CH2-CH-CH-CH2
0
l,2:3,4-Diepoxybutane
CH2-CH-CHOH-CH2OH
V
3,4-Epoxy-l,2-butanediol
Figure 1. A hypothetical scheme for the metabolism of 1,3-butadiene.
SOURCE: Malvoisin and Roberfroid, 1982.
11
-------
DRAFT—DO NOT QUOTE OR CITE
3,4-epoxybutene and diepoxybutane are mutagenic, as described more fully
below. No information is available on the mutagenicity of 3-butene-l,2-diol
and 3,4-epoxy-l,2-butanediol.
The alkylating ability of the two reactive metabolites of 1,3-butadiene
(3,4-epoxybutene and diepoxybutane) has been investigated, each in a single
study. Hemminki et al. (1980) found that 3,4-epoxybutene alkylated 4-(p-nitro-
benzyl)-pyridine (NBP) and deoxyguanosine, which are nucleophiles that were used
as models for DNA. The NBP reaction was carried out at 37°C using the test
compound at 0.286 pM. The deoxyguanosine reaction was carried out at 37°C
using the test compound at 0.1 M. In both cases, aliquots of the reaction
mixture were assayed for alkylation at 0 minutes, 20 minutes, 1 hour, 3 hours,
and 5 hours. The reaction rates were determined from the initial rates. The
results were calculated using epichlorohydrin as a reference. 3,4-Epoxybutene
alkylated NBP and deoxyguanosine at rates that were 31% and 14%, respectively,
of thai of epichlorohydrin. The alkylation activity correlated with mutageni-
city in Escherichia coli WP2 uvrA, as described in the next section.
Lawley and Brookes (1967) reported that diepoxybutane reacts with DNA in
a manner typical of bifunctional alkylating agents and causes interstrand cross-
linking in DNA. Salmon sperm DNA was dissolved in 0.5 mM sodium citrate at 2
mg/mL (5.4 mM DNA phosphorus) and 25 mL was treated with redistilled diepoxybu-
tane (2.4 mg/mL, 28 mM) at 37°C. Samples were withdrawn after 2, 5, 7, 24, 48,
72, 120, 168, and 193 hours. Ultraviolet spectroscopy was used to measure the
reaction of diepoxybutane with DNA. At 37°C, diepoxybutane reacted with DNA
slowly, as shown by changes in ultraviolet absorption of the reaction mixtures.
The extent of interstrand cross-linking (i.e., covalent linkage of the two
DNA strands by the reaction of diepoxybutane with a nucleotide base in each
DNA strand) was studied by measuring the reversible denaturation (renaturation)
12
-------
DRAFT—DO NOT QUOTE OR CITE
of diepoxybutane-treated DNA. In this experiment, diepoxybutane-treated DNA
was first incubated at 60°C for various time periods and then rapidly cooled.
At 60°C, untreated DNA denatures when it is dissolved in a solution of low
ionic strength (i.e., the two strands separate). When cooled, the DNA rena-
tures (i.e., the two strands rejoin to reform the typical double-stranded DNA
molecule). Diepoxybutane-treated DNA renatured to a greater extent than did
normal untreated DNA, suggesting that diepoxybutane-treated DNA was cross-linked
by a diepoxybutane bridge covalently joining the two DNA strands.
In summary, the above two studies indicate that 3,4-epoxybutene and
diepoxybutane can alkylate DNA and that diepoxybutane causes interstrand cross-
links in DNA.
3.3. MUTAGENICITY OF 3,4-EPOXYBUTENE
The mutagenic potential of 3,4-epoxybutene was studied in the fluctuation
test with Klebsiella pneumoniae as the test organism (Voogd et al., 1981). The
compound was obtained from K and N (ICN, Life Sciences Division, New York), was
analytical grade, and was not further purified. The chemical was dissolved and
diluted in dimethylsulfoxide and subsequently added to broth which was inoculated
with the test organism. The genetic characteristic studied was streptomycin
resistance. The average spontaneous mutation rate for streptomycin resistance
was 0.1676 x 10~9. Triplicate experiments were averaged, and the results were
expressed as the quotient of the observed and spontaneous mutation rates. At 1
and 2 mM 3,4-epoxybutene, the quotients were 1.7 and 2.5, respectively, provi-
ding suggestive evidence of a dose-related positive response.
Besides studying the alkylating activity of 3,4-epoxybutene, Hemminki et
al. (1980) studied its mutagenicity in a tryptophan-requiring strain of £.
coli. The concentrations of 3,4-epoxybutene used were not specified but were
based on toxicity determinations. Bacteria of strain WP2 uvrA were inci/bated
13
-------
DRAFT--DO NOT QUOTE OR CITE
with the chemical for 18 hours at 37°C, after which aliquots were plated onto
minimal ayar to detect reversion to tryptophan prototrophy and onto nutrient
agar containiny tryptophan to determine the total number of cells. Althouyh
the study suygests that 3,4-epoxybutene is mutagenic in E_. coin WP2 uvrA, there
was no indication of a dose-related response; the result for only one dose was
reported and that dose was unspecified. Althouyh this study is of limited use
for risk assessment purposes, it supports the study of Voogd et al. (1981) in
suggesting that 3,4-epoxybutene is mutagenic in bacteria.
3.4. GENOTOXICITY OF 1,2:3,4-DIEPOXYBUTANE
3.4.1. Studies in Bacteria
Voogd et al. (1981) investigated the mutagenic potential of diepoxybutane
in J<. pneumoniae in the same paper cited previously for the mutagenicity of
3,4-epoxybutene. The source of the diepoxybutane was Merck (Darmstadt, F.R.G.);
it was of analytical grade and was not further purified. At an equal chemical
concentration (1 mM), diepoxybutane was approximately 16 times as mutagenic as
3,4-epoxybutene. The quotients of observed and spontaneous rates of mutation
to streptomycin resistance for 0.05, 0.1, 0.2, 0.5, and 1 mM diepoxybutane were
1.7, 3.1, 6.2, 15.7, and 27, respectively. These results clearly indicate that
diepoxybutane is mutagenic in J<. pneumoniae and provide strong evidence of a
dose-related response as well.
Diepoxybutane is also mutagenic in the ^. typhimurium histidine reversion
assay (Wade et al., 1979). Plate incorporation assays were performed with
strains TA98 and TA100, and averages of two to five determinations were reported,
At 0.02, 0.10, and 0.50 mg of diepoxybutane per plate, there were 196, 325, and
663 revertant colonies per plate with strain TA100 and 32, 22, and 29 revertant
colonies per plate with strain TA98. These results suggest that diepoxybutane
is a base-pair substitution mutagen in S_. typhimurium because it produced a
14
-------
DRAFT--DO NOT QUOTE OR CITE
dose-related positive response in strain *TA100. Although strain TA100 is not
specific for mutagens that induce base-pair substitutions, it responds well to
such mutagens, and the result in strain TA98, which detects many frameshift
mutagens, was negative.
3.4.2. Studies in Fungi
The mutagenic potential of diepoxybutane in the yeast Saccharomyces
cerevisiae was studied by Olszewska and Kilbey (1975). They used a diploid
yeast strain that is homozygous for the ilv mutation and therefore requires
isoleucine and valine to grow. Kinetic studies of the induction of revertants
were carried out by treating cells for various times with 0.1 M diepoxybutane at
25°C. About 1.2 x 106 to 1.5 x 106 cells were plated per petri dish. Diepoxy-
butane induced an increase in i1v reversions with increasing time of exposure
up to 20 minutes. At 5, 10, 15, and 20 minutes, the number of ilv+ revertants
per 106 cells was about 3, 10, 20, and 38, respectively. Reversion of the ilv
mutation indicates that diepoxybutane induces point mutations in yeast.
Zaborowska et al. (1983) have shown that diepoxybutane induces mitotic
crossing-over and mitotic gene conversion in the SBTD and D7 strains of _S.
cerevisiae. Stationary-phase cells were suspended in phosphate buffer at 2 x
lO^/mL. The cells were treated with 0.4% (vol/vol) diepoxybutane (Merck, purity
not specified) at 30°C for 15, 30, and 45 minutes. The treatment was terminated
by centrifuging and washing cells twice with buffer.
The results of these experiments are shown in Table 1. The frequency of
mitotic crossing-over in the SBTD strain was dose (exposure time) related.
Dose-dependence of mitotic crossing-over in strain D7 is less clear. Dose-related
increases in mitotic gene conversion were obtained in both strains. Taken
together, these results indicate that diepoxybutane is recombinogenic in yeast,
and recombinogenicity is an indication of DNA damage.
15
-------
DRAFT--DO NOT QUOTE OR CITE
TABLE 1. INDUCTION OF MITOTIC GENE CONVERSION AND MITOTIC CROSSING-OVER IN
SBTD AND D7 STRAINS OF
S. cerevisiae BY DIEPOXYBUTANE
Exposure to 0.4%
diepoxybutane (min)
SBTD strain
0
15
30
45
D7 strain
0
15
30
Survival
100
100
93.2
41.3
100
100
78.4
Mitotic
crossovers (%)
0
1.4
2.8
6.1
0
1.5
1.7
Convertants3
0.7
86.6
167.1
515.2
0
40.4
278.0
aConvertants calculated per 107 survivors in the SBTD strain and per 106
survivors in the D7 strain.
SOURCE: Zaborowska et al., 1983.
16
-------
DRAFT—DO NOT QUOTE OR CITE
Luker and Kilbey (1982) reported that diepoxybutane causes point mutations
and multigenic deletions in Neurospora crassa. They developed a Neurospora
heterokaryon in which both point mutations and deletions can be detected by
the use of selective techniques. Point mutations were scored by reversion to
adenine independence. Deletions were detected by first assaying for resistance •
to £-fluorophenylalanine (pFPA) and then testing for sensitivity to cycloheximide.
These two genes are closely linked on chromosome V.
Information on the source and purity of the sample of diepoxybutane studied
was not provided. Suspensions of Neurospora conidia were treated for various
times with 0.1 M diepoxybutane in 0.067 M phosphate buffer (pH 7.0) at 27°C.
The treatments were terminated by filtering off the mutagenic solution and
washing the cells with 10% sodium thiosulfate solution. Diepoxybutane induced
dose (exposure time) related increases in both adenine reversions and pFPA
resistance, as shown in Table 2. The results shown in Table 3 suggest that
about ope-fourth of the pFPA-resistant mutants were deletions rather than
point mutations because pFPA resistance was associated with sensitivity to
cycloheximide 26.3% of the time. The evidence therefore shows that diepoxybutane
is both mutagenic and clastogenic in Neurospora.
3.4.3. Studies in Mammalian Cells
The results of a study by Dean and Hodson-Walker (1979) suggest that
diepoxybutane is a powerful clastogen in cultured rat liver epithelial cells.
The sample of diepoxybutane studied was obtained from Fluka A.G., Switzerland.
Its purity was not described. The epithelial-like cell line used, designated
RL}, is neardiploid, having a chromosome number of 44 or 45 (compared to the
normal number of 42 in the rat karyotype). The appropriate concentrations of
diepoxybutane for testing were determined from cytotoxicity studies. Because
diepoxybutane is volatile, sealed flask cultures of the rat liver cells were
17
-------
DRAFT—DO NOT QUOTE OR CITE
TABLE 2. INDUCTION OF ADENINE REVERSIONS AND pFPA RESISTANCE IN N. crassa
BY
0.1 M
DIEPOXYBUTANE
Heterokaryotic
conidia screened
(x 106) for
Treatment
(min)
0
10
20
30
Survi
(%)
100
78
60
40
val
.0
.1
.4
.8
ad+
15
6
5
2
.64
.90
.54
.53
pFPA
resistant
4.
1.
1.
0.
17
38
13
51
Number of
mutants
scored
ad +
5
22
34
53
fpr
2
12
21
21
Mutation
frequency
x lO'6
ad
0
3
6
20
-t-
.32
.19
.03
.95
fprr
0.48
8.70
18.58
41.18
SOURCE: Luker and Kilbey, 1982.
TABLE 3. ANALYSIS OF pFPAr MUTANTS AS PUTATIVE DELETIONS
Treatment
with 0.1 M
diepoxybutane
(mi n )
10
15
20
30
45
60
Total
Number of
pFPA-resistant
mutants
tested
12
7
41
94
25
11
190
Number acqui ri ng
sensitivity to
cycloheximide
5
2
10
21
10
2
50
Putative
deletions
(%)
42
29
24
22
40
18
26.3
SOURCE: Luker and Kilbey, 1982.
18
-------
DRAFT —DO NOT QUOTE OR CITE
used. The cell cultures were exposed to diepoxybutane for 24 hours, and colcemid
at 0.3 yg/rnL was added 2 hours before harvesting the cells with 0.25% trypsin.
Distilled water was added to the harvested cells to produce hypotonic conditions.
The cells were fixed in methanol/acetic acid (3:1). Chromosome preparations
were made by air-drying the cells on microscope slides and staining with Giemsa.
The slides were randomly coded, and 100 metaphases from each slide were analyzed
for structural chromosome changes. Diepoxybutane was a powerful clastogenic
agent, producing chromosome damage in a dose-related manner (Table 4).
Perry and Evans (1975) reported that exposure of cultured Chinese hamster
ovary cells to diepoxybutane resulted in a dose-related induction in sister
chromatid exchanges (SCEs). SCE involves the reciprocal exchange of DNA seg-
ments between sister chromatids and is considered an indication of DNA damage.
The cells were treated with 10 yM bromodeoxyuridene and 0.3 yM or 3 yM
diepoxybutane for two cycles of DNA replication before treatment with colcemid
(2 hours at 0.2 pM), collection by mitotic shakeoff, pretreatment with 75 mM
KC1, and fixation in methanol/acetic acid (3:1). Twenty mitotic cells were
scored for each dose. Total SCEs at 0, 0.3, and 3 yM diepoxybutane were 244,
403, and 1818, respectively.
3.4.4. In vivo Studies
Studies of the mutagenic potential of diepoxybutane in whole animals have
been carried out in mice and Drosophila melanogaster (fruit flies).
Conner et al. (1983) studied the ability of diepoxybutane to induce J_n_
vivo SCE in bone marrow, alveolar macrophages, and regenerating liver cells in
mice. The sample of diepoxybutane studied was 97% pure and was obtained from
Aldrich Chemical Co. It was dissolved in phosphate-buffered saline just prior
to injection. The dose-response studies were performed in three intact and
three partially hepatectomized Swiss Webster mice (mean weight of 27 g).
19
-------
DRAFT—DO NOT QUOTE OR CITE
TABLE 4. EFFECTS OF DIEPOXYBUTANE ON CHROMOSOMES OF RAT LIVER CELLS
Diepoxybutane
(yg/mL)
0
0.1
0.5
l.Q
Number
of cells
analyzed
269
117
24
29
Chromatid
gaps
(*)
2.6
11.1
8.3
0
Chromatid
deletions
(*)
0.4
4.3
20.8
0
Chromatid
exchanges
(%)
0
8.b
33.3
3.4
Chromosome
aberrations
(%)
0.4
0.9
0
0
SOURCE: Dean and Hodson-Walker, 1979.
20
-------
DRAFT—DO NOT QUOTE OR CITE
Diepoxybutane (10-291 ymol/kg) was injected intraperitoneally just prior to
bromodeoxyuridine infusion (10 mg/mL; intravenous flow rate of 3.6 mL/24 hours),
Colchicine (3.3 mg/kg) was then injected intraperitoneally. Bone marrow and
alveolar macrophage cells from intact mice and regenerating liver cells from
hepatectomized mice were harvested 4 hours later. As shown in Table 5, diepoxy-
butane produced similar dose-dependent responses for SCE in bone marrow, alveo-
lar macrophages, and regenerating liver cells. Conner et al. (1983) also
reported that the DNA lesions did not persist beyond one cell cycle. These
results show that diepoxybutane is very effective in producing a dose-dependent
SCE response, but that the initially induced lesions disappear in subsequent
division cycles.
A positive response for diepoxybutane in the sex-linked recessive lethal
test in Drosophila was reported in two studies from the same laboratory. The
sex-linked recessive lethal test is a useful assay for testing the mutagenic
potential of chemicals in the germ line of an intact animal. In the first
study, Sankaranarayanan (1983) exposed wild-type Berlin-K adult 3 or 4-day-old
male flies to 2 mM diepoxybutane in 5% sucrose by feeding for 48 hours. The
diepoxybutane sample studied was obtained from Fluka, A.G., Switzerland. Infor-
mation on its purity was not reported. The males were mated to Oster females
to raise three successive 2-day broods (A, B, and C), and the Fj female progeny
were used in the tests for lethals. Brood A tests mature spermatozoa, brood 8
tests late spermatids, and brood C tests early spermatids. The results are
shown in Table 6. A concurrent no-exposure control was not done in this study
or in the subsequent study described below. However, a historical control
value for Drosophila of 0.18% has been established in the same laboratory from
the evaluation of 13,151 chromosomes (Vogel, 1976). The results suggest that
diepoxybutane is a strong inducer of sex-linked recessive lethal mutations
21
-------
DRAFT—DO NOT QUOTE OR CITE
TABLE 5. SCE FREQUENCIES INDUCED IN BONE MARROW, ALVEOLAR MACROPHAGES,
AND REGENERATING LIVER CELLS OF SWISS WEBSTER MICE
FOLLOWING INJECTION OF DIEPOXYBUTANE
Diepoxybutane
(u mol/kg)
Hepatectomized
0
10
39
97
193
291
mice
4
7
8
10
22
32
Bone
.2
.2
.1
.9
.3
.0
Aa
± °
± °
+_ 1
i 1
± 3
± 6
marrow
.6
.9
.4
.5
.5
.6
Bb
66
63
72
63
44
22
3
7
9
13
23
30
Alveolar
macrophages
A B
.5 ^
.9 +_
.8 i
.4 +
.6 +_
.4 +.
0.6
0.4
1.0
3.4
4.5
4.9
77
71
70
75
59
23
Regenerating
liver
A B
3.7 +_ 0.9 80
7.0 +_ 1.4 73
11.9 +_ 2.6 69
14.9 +_ 4.0 69
28.9 +_ 6.1 43
31.1 +_ 5.9 29
Intact mice
0
10
39
97
193
291
3
5
8
9
14
27
.0
.4
.8
.7
.6
.3
± °
1 °
± °
+_ 1
± 2
± 3
.8
.6
.8
.5
.1
.7
64
58
63
63
43
46
3
6
10
12
17
28
.6 +_
.7 +
•2 ±
.8 +_
.1 +.
.6 +_
0.5
1.2
0.3
2.3
2.2
3.8
63
54
57
62
41
42
aMean SCE/cell +_ S.D. of three mice at each dose. Individual animal means
were calculated from SCEs scored in 20 cells of each type.
number of second division cells observed in 100 consecutive metaphases,
SOURCE: Conner et al., 1983.
22
-------
DRAFT—DO NOT QUOTE OR CITE
TABLE 6. FREQUENCIES OF SEX-LINKED RECESSIVE LETHALS INDUCED BY 2 mM
DIEPOXYBUTANE IN POSTMEIOTIC MALE GERM CELLS OF Drosophila
Experimental Number of Lethals
strain Brood3 chromosomes Number %
Berlin-Kb A 800 42 6.5
B 914 32 4.8
C 840 30 4.6
Canton-Sc
Experiment 1 A 934 88 9.4
B 949 68 7.2
C 960 66 6.9
Experiment 2 A 938 64 6.8
B 951 54 5.7
C 350 14 4.0
Ebonyc
Experiment 1 A 932 88 9.4
B 912 65 7.1
C 925 56 6.1
Experiment 2 A 876 57 6.5
B 924 41 4.4
C 885 33 3.7
aBrood A corresponds to treatment of mature spermatozoa; Brood B corresponds
to late spermatids; and Brood C corresponds to early spermatids.
bTaken from Sankaranarayanan, 1983.
GTaken from Sankaranarayanan et al., 1983.
23
-------
DRAFT—DO NOT QUOTE OR CITE
(6.5%, 4.8%, and 4.6% compared to the historical control value of 0.18%). The
results also indicate that mature spermatozoa (brood A) respond with higher fre-
quencies of recessive lethals than late and early spermatids (broods B and C).
Similar positive results (Table 6) were obtained in a later study in Dro-
sophila (Sankaranarayanan et al., 1983). The experimental details in this
study were identical to those described above, except that Canton-S and ebony
males were exposed to diepoxybutane and mated to Muller-5 females. Accordingly,
the strong mutagenic response appears to be independent of the strains employed.
The strongly positive results in the sex-linked recessive lethal test provide
clear evidence that diepoxybutane reaches the gonads and is strongly mutagenic
in germ cells of Drosophila.
There is evidence that diepoxybutane induces chromosome damage in germ
cells of Drosophila (Zimmering, 1983). Treated males carried an X chromosome
in the form of a closed ring and a Y chromosome carrying dominant markers, one
at the end of long arm of the Y and one at the end of the short arm. The males
were permitted to feed on a solution of 1.25 mM diepoxybutane in 5% sucrose for
24 hours and then mated with repair-proficient (ordinary) females or to repair-
deficient females. There was no evidence of toxicity in the treated males.
The F]_ offspring were scored for complete loss of the X or Y chromosomes (in
ring-X males, virtually all complete loss is attributable to ring loss) and for
partial loss of the Y chromosome, indicated by the loss of one but not both of
the Y chromosome markers. Complete loss indicates chromosome breakage and/or
sister chromatid exchange. Partial loss of the Y chromosome is a consequence
of breakage. Results shown in Table 7 provide evidence of a relatively strong
effect on complete loss (5-6%) and a significant increase in partial loss which
is most apparent from matings with the repair-deficient females (approx. 3%).
In summary, the results of the Drosophila experiments assaying for sex-
24
-------
ORAFT--DO NOT QUOTE OR CITE
TABLE 7. CHROMOSOME LOSS IN Drosophlla
FROM MATINGS OF MALES WITH REPAIR-PROFICIENT (RP)
AND REPAIR-DEFICIENT (RD) FEMALES
Series
Female
Percent Induced
Complete Partial Complete Partial
loss loss loss loss
Control RP
Treated
Control RD
Treated
8551
7390
3178
1285
51
515
30
82
0
8 6.37
2
39 5.44
0.11
2.97
All induced frequencies are statistically significant at or below the 0.01 level
The repair-deficient mutant was mei-9a, which is deficient in excision repair.
SOURCE: Zimrnering, 1983.
25
-------
DRAFT—DO NOT QUOTE OR CITE
linked recessive lethals and chromosome loss provide strong evidence that
diepoxybutane is a mutagenic and chromosome damaging agent in germ cells of
Drosophila. In addition, the results of the SCE assay in mice suggest that
diepoxybutane is a DNA damaging agent in mice.
3.5. SUMMARY OF MUTAGENICITY STUDIES
The available information on the mutagenicity of 1,3-butadiene is quite
limited in that only two studies have been reported. Both studies, however,
indicate that 1,3-butadiene is a mutagen in S_. typhimurium. The mutagenicity
is observed only in the presence of a liver S9 metabolic activation system.
No whole animal studies have been reported. These results suggest that 1,3-
butadiene is a promutagen in bacteria (i.e., its mutagenicity depends on
metabolic activation).
Mutagenic metabolites of 1,3-butadiene include 3,4-epoxybutene and diepoxy-
butane. 3,4-Epoxybutene is a monofunctional alkylating agent and is a direct-
acting mutagen in bacteria; it has been tested in J<. pneumoniae and E_. coli.
*
Diepoxybutane is a bifunctional alkylating agent and as such it can form
cross-links between the two strands of DNA. It is mutagenic in bacteria (K_.
pneumoniae and S_. typhimurium), fungi (yeast and Neurospora), and the germ
cells of Drosophila. It also induces DNA damage in cultured hamster cells and
in mice, is clastogenic in fungi and cultured rat cells, and produces chromo-
some damage/breakage in Drosophila germ cells. Therefore, there is strong
evidence that diepoxybutane is a mutagen/clastogen in microbes and animals.
In summary, the weight of the available evidence suggests that 1,3-buta-
diene is a mutagen by virtue of its metabolism to mutagenic intermediates, par-
ticularly diepoxybutane.
26
-------
DRAFT—DO NOT QUOTE OR CITE
4. CARCINOGENICITY
The purpose of this chapter is to provide an evaluation of the likelihood
that 1,3-butadiene is a human carcinogen and, on the assumption that it i^s a
human carcinogen, to provide a basis for estimating its public health impact
and evaluating its potency in relation to other carcinogens. The evaluation of
carcinogenicity depends heavily on animal bioassays and epidemiologic evidence.
However, additional factors, including mutagenicity, pharmacokinetics, and
other toxicological characteristics have an important bearing on both the
qualitative and quantitative assessment of carcinogenicity. This section pre-
sents an evaluation of the animal bioassays, the epidemiologic evidence, and
the quantitative aspects of risk assessment.
4.1. TOXICOLOGY AND PHARMACOKINETICS
Information on the toxicity of 1,3-butadiene resulting from acute exposure
is limited. The median lethal concentrations for mice and rats of 1,3-buta-
diene for periods of exposure ranging from 2 to 4 hours are above 100,000 ppm.
The oral LD^Q values for rats and mice are 5.48 g/kg and 3.21 g/kg, respectively.
The major acute toxic effects are irritation of the respiratory tract, mucous
membranes, and eyes, and narcosis (NTP, 1984).
A teratogenicity study sponsored by the International Institute of Synthe-
tic Rubber Producers, Inc., was conducted at Hazleton Laboratories Europe, Ltd.
(1981b) in England. Pregnant female Charles River CD rats (Sprague-Dawley),
obtained from Charles River Ltd., were exposed to concentrations of 200, 1,000,
and 8,000 ppm of 1,3-butadiene, 6 hours/day, on days 6-15 of gestation. The
rats were observed daily and weighed at intervals during the study. On day
20 of gestation the females were killed, necropsied, and the uterine contents
inspected. One-third of each litter was examined for visceral abnormalities,
27
-------
DRAFT—DO NOT QUOTE OR CITE
and the remainder were examined for skeletal anomalies. There were 24 female
rats assigned to each treatment group, and 40 were assigned to the control
group which was exposed to filtered air.
The maternal animals were not affected by exposure to the 1,3-butadiene
except for reduced weight gain in those exposed to 8,000 ppm. The authors
concluded, as a result of the maternal toxicity, that there was embryonic
growth retardation and slight embryolethality among all dose groups, and that
the magnitude of the effect was dose-related. The relationship between mater-
nal toxicity and the fetal effects was a subjective judgment and not experimen-
tally established. The authors further concluded that at the highest dose
there was an indication of teratogenicity based on the presence of major fetal
defects. There was a significant increase in "minor" defects at the lowest
airborne concentration tested (200 ppm). Whether this represents a qualitative
change in the dose-response or is due to other factors, such as maternal
toxicity, cannot be determined from the information available. In an earlier
but inadequately reported study (Carpenter et al., 1944), decreases in litter
size were reported in rats exposed to 6,700 and 2,300 ppm, but not at 600 ppm.
As detailed in the preceding section, in vitro studies have indicated
that 1,3-butadiene is metabolized in liver microsomes by P-450-dependent mixed-
function oxidases. The biotransformation, although only partially confirmed by
in vivo experiments in rats (Bolt et al., 1983), could lead to the formation of
3,4-epoxybutene, then to l,2:3,4-diepoxybutane (diepoxybutane) or to 3-butene-
1,2-diol, then to 3,4-epoxy-l,2-butanediol, followed by the formation of ery-
thritol (Malvoisin and Roberfroid, 1982). In addition to being mutagenic,
diepoxybutane is classified, in the Third Annual Report on Carcinogens, as a
substance that may be reasonably anticipated to be a carcinogen. This deter-
mination was in turn based on the International Agency for Research on Cancer
28
-------
DRAFT—DO NOT QUOTE OR CITE
(IARC) finding of sufficient evidence for carcinogenicity in experimental
animals. The studies underlying these determinations show a carcinogenic
response following skin application in mice and the production of local cancer
in mice and rats by subcutaneous injection.
Information concerning the pharmacokinetics of 1,3-butadiene is limited.
Carpenter et al. (1944) determined that blood concentrations in the femoral
artery and in the femoral vein of rabbits at nine minutes after exposure to an
airborne concentration of 250,000 ppm were 0.26 mg/mL and 0.18 mg/mL, respec-
tively. Rats exposed to 1,3-butadiene for 2 hours at an airborne concentration
of 130,000 ppm had the highest concentrations of the chemical in peri renal fat
(152 mg%); lower concentrations (36-51 mg%) were found in the liver, brain,
spleen, and kidney. Ninety minutes after exposure to 130,000 ppm for one hour
the tissue concentrations were minimal (Shugaev, 1969).
Preliminary results from further investigations by NTP (personal communica-
tion, 1S85) indicated that following intraperitoneal injection of radiolabeled
1,3-butadiene, male B6C3F1 mice exhale most of the dose unchanged. Exhaled
carbon dioxide was the next largest pool for the 14C label. Lesser amounts
were detected in the urine and feces. Little remained in the carcass 65 hours
after intraperitoneal administration. Nose-only exposure of male mice at con-
centrations of 7.5, 97, 119, and 770 ppm for 6 hours did not alter the minute
volume significantly. Although pulmonary absorption was lower than expected
(3-11%), it was linearly related to the inhaled concentration. The urine was
the major route of excretion of the absorbed 1,3-butadiene. The excretion of
the metabolized products is linearly related to the inhaled concentration.
4.2. ANIMAL STUDIES
4.2.1. Chronic Toxicity Studies in Mice
A chronic toxicity and carcinogencity inhalation study of 1,3-butadiene in
29
-------
DRAFT—DO NOT QUOTE OR CITE
B6C3F1 mice, sponsored by the NTP, was conducted at Battelle Pacific Northwest
Laboratories. Preliminary inhalation toxicity studies in mice were used as a
basis for dose selection for a chronic study. A 15-day study and a 14-week
study were conducted at International Bio-Test Laboratories. In the 15-day
study, weight loss at airborne concentrations of 1,250 ppm was observed. The
mice exposed to 8,000 ppm, the highest airborne concentration, survived the
exposure period. In the 14-week study, reduced body weight and death were
observed among mice treated at 2,500 ppm or more. Necropsy findings were not
reported (NTP, 1984).
The mice used in the chronic study were obtained from Charles River
Laboratories and were exposed to graded concentrations of 625 and 1,250 ppm
of 1,3-butadiene for 6 hours/day, 5 days/week. The exposures were conducted in
dynamic negative-pressure exposure chambers, and the chamber concentrations
were generated by mixing the test gas with filtered air. The chamber concen-
trations were measured 7 to 12 times a day for the first 150 days with a photo-
ionization detector, and thereafter by gas chromatography. It was intended
that the dimer (4-vinyl-1-cyclohexene) concentration in the test material be-
fore use should be controlled to less than 100 ppm. However, three cylinders
with slightly more than 100 ppm of dimer were used because replacements were
not available. The mice were 8 to 9 weeks of age when the exposures began, and
were housed individually throughout the study. There were 50 mice per sex per
dose group.
The mice were weighed weekly for the first 12 weeks of the study, and
monthly thereafter. They were examined for subcutaneous masses beginning after
the 12th week. Clinical signs were recorded weekly. Histopathological evalua-
tion (32 tissues) was performed on all mice.
While the original plan was for this to be a 2-year study, all surviving
30
-------
DRAFT—DO NOT QUOTE OR CITE
mice were killed after week 60-61 because of excessive deaths among the treated
mice. Many of these deaths were caused primarily by the developing neoplasia.
The survival (mice at risk; corrected for mice that were missing or were acci-
dentally killed) at this early termination was as follows:
Airborne concentration (ppm)
0 625 1,250
Males 49/49 11/50 7/46
Females 46/46 14/47 30/48
There were no increases in clinical signs that could be associated with
exposure to 1,3-butadiene except those related to tumor development and death.
The body weights were not affected by inhalation exposure to the test chemical.
There was a marked increase in the overall frequency of mice with primary
tumors, as indicated below:
Airborne concentration (ppm)
0 625 1.250
Males 10/50 44/50 40/50
Females 6/50 40/50 46/49
In addition to a marked increase in the number of animals with primary
tumors, there was also an increase in the number of animals with multiple
primary tumors. Among the tumor-bearing male mice, there were 11, 73, and 61
such tumors in the control, low-, and high-exposure groups, respectively. In
the females, there were 6, 66, and 100 tumors in the tumor-bearing animals of
the control, low-, and high-exposure groups, respectively.
The histopathologic evaluation indicated significant increases in tumors
31
-------
DRAFT—DO NOT QUOTE OR CITE
of various types, as shown in Table 8. These tumors began to appear remark-
ably early in the course of the study. Lymphomas were diagnosed in mice dead
at 22 and 20 weeks of exposure for males and females, respectively, of the high-
exposure group. The first tumors of this type were found in low-dose mice at 24
and 29 weeks, respectively. Of the survivors, two males in the low-dose group
and one in the high-dose group had lymphomas. In the females, one lymphoma
was found among the surviving control mice, three in the low-dose group, and
one in the high-dose group. Many of the early deaths were judged to be caused
by this type of tumor.
The heart was the principal organ in which hemangiosarcomas occurred. The
first hemangiosarcomas were diagnosed at 32 and 42 weeks in the low- and high-
dose males and at 41 and 43 weeks among the females. The cardiac hemangiosarco-
mas may have caused some of the mice to die early. Atypical cardiac endothelial
hyperplasia, a likely preneoplastic lesion, was not observed among the controls
but was. present in treated males (625 ppm - 10%, 1,250 ppm - 4%) and females
(625 ppm - 10%, 1,250 ppm - 16%).
Alveolar/bronchiolar adenomas and carcinomas occurred (both separately and
combined) at increased frequency in both male and female mice. In the high-dose
groups the first such lesions appeared at week 42 for males and week 50 for
females. Neoplastic changes in the lungs of the controls were not detected
until the termination of the study. While neoplastic lesions of the nasal
cavity were not found at any dose, there was an increase in nonneoplastic
changes at the high dose. At 1,250 ppm, chronic inflammation of the nasal cavity
(male, 33/50; female, 2/49), fibrosis (male, 35/50; female, 2/44), cartilaginous
metaplasia (male, 16/50; female 1/49), osseous metaplasia (male, 11/50; female,
2/49), and atrophy of the sensory epithelium (male, 32/50) were observed. No
nonneoplastic lesions of the nasal cavity were found in the controls. Huff et
32
-------
DRAFT—DO NOT QUOTE OR CITE
TABLE 8. SUMMARY OF THE STATISTICALLY SIGNIFICANT INCIDENCE OF TUMORS
IN MICE EXPOSED FOR 60-61 WEEKS TO 1,3-BUTADIENE
Tumor type and site
Hemangiosarcomas
(heart)
Malignant lymphomas
(hematopoietic system)
Al veolar/bronchiolar
adenoma
adenoma/carci noma
»
Acinar cell carcinoma
(mammary)
Granulosa cell tumor
or carcinoma (ovary)
Forestomach
(All papilloma
and carcinoma)
Hepatocellular
adenoma
adenoma/carci noma
Sex
M
F
M
F
M
F
M
F
F
F
M
F
F
F
Airborne
0
0/50
p=0.032a
0/49
p=0.001a
0/50
p<0.001a
1/50
p<0.006a
2/50
p=0.10a
3/49
p<0.001a
2/50
p<0.001a
3/49
p<0.001a
0/50
p=0.007a
0/49
p<0.001a
0/49
p=0.354a
0/49
p<0.001a
0/50
p=0.025a
0/50
p=0.016a
concentration
625
16/49
p<0.001b
11/48
p<0.001b
23/50
p=0.001b
10/49
p=0.003b
12/49
p=0.003b
9/48
p=0.056b
14/49
p<0.001b
12/48
p=0.010b
2/49
p=0.242b
6/45
p=0.010b
7/40
p=0.037b
5/42
p=0.018b
1/47
p=0.485b
2/47
p=0.232b
(ppm)
1,250
7/49
p=0.006b
18/49
p<0.001b
29/50
p=0.001b
10/49
p=0.003b
11/49
p=0.007b
20/49
p<0.001b
15/49
p<0.001b
23/49
p<0.001b
6/49
p=0.012b
13/48
p<0.001b
1/44
p=0.473b
10/49
p<0.001b
4/49
p=0.056b
5/49
p=0.015b
aCochran-Armitage Trend Test.
bFisher Exact Test.
33
-------
DRAFT—00 NOT QUOTE OR CITE
al. (1985) have suggested that the lack of neoplasms in the nasal cavity may
reflect a requirement for biotransformation of 1,3-butadiene to a reactive
epoxide intermediate.
Among the 10 control male mice with primary tumors, eight had hepatocell-
ular adenomas and/or carcinomas. This type of tumor is normally observed among
male mice of this strain in 2-year bioassays. It may be that the inclusion of
mice with this type of tumor in considering the number of tumor-bearing animals
tends to deemphasize the frequency of compound-induced neoplasia. On the other
hand, among the females, the frequency of hepatocellular adenomas and/or carci-
nomas was increased. The occurrence of this type of tumor among females of
this strain is more suggestive of adverse chemical-related effects. Among the
male mice, there was a significant increase in liver necrosis at both doses.
In the female mice, liver necrosis was significantly elevated only at the
higher airborne concentration.
In^addition to the neoplastic changes in the ovary and forestomach,
ovarian atrophy and forestomach epithelial hyperplasia were elevated among
the mice at both doses. Since Zymbal gland tumors have been reported in the
chronic rat study to be discussed, it is worth noting that in this study, one
also occurred in high-dose female mice and two occurred in high-dose male mice.
This tumor is not normally found in control mice, even at the termination of a
2-year study. Testicular atrophy was observed in male mice at both dose levels;
however, the increase in tumors of the testes that had occurred in the rats did
not occur in the mice.
An audit of this chronic study was conducted by the NTP. Potential dis-
crepancies that might have significantly influenced the interpretation of this
study were resolved. The NTP considered that this study provided clear evi-
dence of carcinogenicity, which is the highest classification in their system
34
-------
DRAFT—DO NOT QUOTE OR CITE
of categorizing evidence of carcinogenicity.
4.2.2. Chronic Toxic, ity Studies in Rats
A 2-year chronic inhalation toxicity study using rats as the experimental
animals was conducted by Hazleton Laboratories Europe, Ltd. (HLE, 1981a) in
England. The study was sponsored by the International Institute of Synthetic
Rubber Producers, Inc. The chronic study was preceded by a 3-month toxicity
study. The airborne concentrations of 1,3-butadiene used in the 3-month study
were 1,000, 2,000, 4,000, and 8,000 ppm. A control group was exposed to fil-
tered air. No effects considered by the authors to be attributable to exposure
to the test chemical were seen in growth rate, food consumption, hematology,
blood biochemical investigations, or pathological evaluation. The only effect
the investigators observed that might be related to 1,3-butadiene exposure was
a moderate increase in salivation, particularly in female rats during the last
6 to 8 weeks of exposure at the higher airborne concentrations (Crouch et al.,
1979)..
For the chronic investigation (HLE, 1981a), Charles River CD rats (Sprague-
Dawley rats obtained from Charles River Ltd.) were exposed to graded concen-
trations of 1,000 and 8,000 ppm of 1,3-butadiene. The exposures (6 hours/day,
5 days/week) for 111 and 105 weeks for males and females, respectively, were
conducted in a dynamic negative-pressure exposure chamber. The chamber concen-
trations were generated by mixing the test gas with filtered air. The concen-
centrations were measured with an infrared gas analyzer. The dimer (4-vinyl-
1-cyclohexene) concentrations in the test material before use were less than
1,000 ppm, but samples in the 700 to 800 ppm range were used, and the dimer
concentration of these averaged 413 ppm. The rats were 4 1/2 weeks of age
when the exposures began, and were housed five to a cage throughout the study.
There were 110 rats per sex per dose group, and a similar number of rats exposed
35
-------
DRAFT—DO NOT QUOTE OR CITE
to filtered air served as a control group.
The rats were weighed weekly and examined for subcutaneous masses and
other clinical signs. Blood chemistries, hemograms, and urine analyses were
evaluated at 3, 6, and 12 months. Neuromuscular function was evaluated period-
ically through week 77 of the study. Ten rats per sex per dose were killed and
necropsied at week 52. Histopathological evaluations were performed on all rats
from the high-dose group and the control group. Tissues from the low-dose
group that were deemed to be of toxicological significance were also examined.
Variations in mean body weight suggested no consistent adverse effect.
Review of the hemograms, blood chemistry, urine analysis, and behavioral testing
was likewise not indicative of an adverse effect.
The clinical signs in the high-dose group, consisting of excessive secre-
tion of the eyes and nose plus a slight ataxia, were observed between months 2
and 5. In addition, in the females of the treated groups subcutaneous masses
appeared earlier and at a higher incidence than in the control group. A dose-
related increase in liver weights was observed at the necropsy performed at 52
weeks and at the termination of the study. This could indicate that the chemi-
cal induces liver enzymes. Otherwise, no significant changes were noted at the
52-week kill.
In the control group, 45% of the males and 46% of the females survived until
the end of the study (note: corrected for interim kill). In the high-dose group,
survival was 32% and 24%, while in the low-dose group 50% and 32% survived.
The decreased survival in the high-dose group was statistically significant.
Increased alveolar metaplasia and nephropathy were observed among males of
the 8,000-ppm treatment groups at the termination of the study. Marked or se-
vere nephropathy occurred in 27% of the male rats in the high-dose group, as
compared with 9 to 10% in the control and the low-dose groups. The authors con-
36
-------
DRAFT—DO NOT QUOTE OR CITE
sidered nephropathy to be the cause of some of the early deaths in this study.
The frequency of metaplasia was 5/44 in the surviving male rats (8,000 ppm) as
compared to 5/45 in the controls.
With regard to the carcinogenic potential of 1,3-butadiene, the authors
of this study concluded that exposure of male and female rats under the
conditions of this investigation was associated with significant increases in
both common and uncommon tumors. Furthermore, they stated that the results of
this 2-year inhalation study supported the premise that 1,3-butadiene is a
suspect weak oncogen.
The incidence of selected neoplasms is shown in Table 9. In the females
there was an increase in mammary carcinoma tumors (control - 8%, 1,000 ppm - 42%,
8,000 ppm - 38%). Also, in the females follicular thyroid adenomas were encoun-
tered more frequently among the treated females than among the controls (control
- 0%, 1,000 ppm - 2%, 8,000 ppm - 8%). An increase in uterine/vaginal tumors
(control - 0%, 1,000 ppm - 2%, 8,000 ppm - 8%) was also observed in the females.
One female of the 8,000 - ppm exposure group had an uterine stromal tumor
(unspecified).
In the males there was an increase in Leydig cell adenomas (control - 0%,
1,000 ppm - 2%, 8,000 ppm - 7%). A single Leydig cell tumor (unspecified) was
observed in one male of each exposed group. Exocrine pancreatic adenomas were
increased in the male rats of the high-dose group (control - 3%, 1,000 ppm - 1%,
8,000 ppm - 10%). One carcinoma was observed in this tissue in the males of the
high-dose group.
Zymbal gland tumors were increased in the high-dose group, when male and
female rats were combined. Except for these, the increase in tumors in this
investigation was limited to tumors that developed in hormonal-dependent
tissues.
37
-------
ORAFT--DO NOT QUOTE OR CITE
TABLE 9. SUMMARY OF THE INCIDENCE OF TUMORS IN RATS
EXPOSED TO 1,3-BUTADIENE (100 RATS PER SEX PER DOSE GROUP)
Tumor type and site
Multiple mammary
gland tumors
Thyroid follicular
(adenoma and carcinoma)
Uterine cervical/
stromal sarcoma
Leydig cell
(adenoma and carcinoma)
Pancreatic exocrine
Carcinoma
Adenoma
Zymbal 3} and
(carcinoma)
Sex
F
F
F
M
M
M
M
F
Airborne
0
50
p<0.001a
0
p<0.001a
1
p=0.115a
0
p<0.003a
0
3
p=0.019a
0
p=0.384a
0
p=0.037a
concentration
1,000
79
p<0.001b
4
p=0.06b
4
p=0.184b
3
p=0.12b
0
1
p=0.879b
1
p=0.5b
0
(ppm)
8,000
84
p<0.001b
11
p<0.001b
5
p=0.106b
8
p<0.001b
1
10
p=0.041b
1
p=0.5b
4
p=0.061b
aCochran-Armitage Trend Test.
bFisher Exact Test.
38
-------
DRAFT—DO NOT QUOTE OR CITE
4.2.3. Summary of Chronic Toxicity Studies
The two chronic studies available at this time are compared in Table 10.
The obvious difference is the marked sensitivity of the mice compared to the
rats with regard to carcinogenic response. Some difference in this direction
might be expected. For example, if the carcinogenic response is elicited by a
metabolite, as has been suggested (de Meester et al., 1978, 1980), mice, because
of their higher rate of metabolism, might be expected to yield a greater response
than rats. The variation in housing during exposure (individually in the case
of the mice versus in groups in the case of the rats), and the higher dimer
content of the material to which the rats were exposed, might also be expected
to contribute to the difference in the sensitivity of the mice. Among the
other factors that might have led to this rather remarkable species differences
are intralaboratory variations in dimer formation in the test atmospheres and
variations in metabolite formation.
Tlie tumors in the rats are largely characterized as occurring in hormonal -
dependent tissues. Some suggestion of this is observed in the mice, but not to
so great an extent. The occurrence of a similar array of tumors in the mice
could have been prevented by the early deaths from more rapidly developing neo-
plasia. It is worth noting the Zymbal gland tumors did develop in both species,
but were not as marked in the mice.
In the NTP mouse study, hemangiosarcomas of the heart, a very rare tumor,
were markedly elevated in both groups exposed to 1,3-butadiene. The possible
cardiac selectivity of 1,3-butadiene is further suggested by the presence of
cardiac disease in two of the epidemiologic studies, as well as the occurrence
of cardiac anomalies in the teratogenic investigations.
4.3. EPIDEMIOLOGIC STUDIES
The manufacture of styrene-butadiene rubber (SBR) involves the use of, and
39
-------
TABLE 10.
SIGNIFICANT EFFECTS OF EXPOSURE TO 1,3-BUTADIENt ON SPRAliUb-DAWLEY RATS
AND B6C3F1 MICE IN INHALATION STUDIES
Rats8 (Hazleton Laboratories Europe, Ltd., 1981a)
1,000 ppm 8,000 ppm
Neoplasms:
Male
Female
Leydig cell adenoma''
Mammary gland:
fibroadenoma/carci nomab
Thyroid: follicular cell
adenomab
Uterus: stromal sarcomab
Nonneoplastic lesions:
Males
Leydig cell adenomab
Pancreas: exocrine tumors'3
Brain: glioma
Mammary gland:fibroadenoma/
carcinoma"
Thyroid:follicular cell
adenomab
Uterus: stromal sarcomab
Zymbal gland: carcinomab
Increased focal alveolar
epithelialization
Nephropathy
Hicec (National Toxicology Program, 1984)
625 ppm
1,250 ppm
Neoplasms:
Males
Females
Heart: hemangiosarcomab
Malignant lymphomab
Lung: alveolar/bronchiolar
adenoma and carcinoma*3
Forestomach: papillomab
Preputial gland:
squamous cell carcinomad
Brain: glioma''
Heart: hemangiosarcomab
Malignant lymphomab
Lung: alveolar/bronchiolar
adenoma and carcinoma*5
Forestomach: papillomab
Ovary: granulosa cell
tumorb
Nonneoplastic lesions:
Male Forestomach: epithelial
hyperplasia*3
Liver necrosis'5
Testicular atrophyb
Female
Liver necrosis'3
Forestomach: epithelial
hyperplasiab
Ovary, atrophyb
Uterus: involution13
Heart: hemangiosarcomab
Malignant lymphomab
Lung: alveolar/bronchiolar
adenoma and carcinoma*3
Preputial gland:
squamous cell carcinoma''
Zymbal gland: carcinomad
Brain: gliomad
Heart: hemangiosarcomab
Malignant lymphoma*3
Lung: alveolar/bronchiolar
adenoma and carcinoma*3
Forestomach: papilloma*3
Mammary gland: acinar cell
carcinoma*3
Ovary granulosa cell tumorb
Liver: hepatocellular adenoma
or carcinoma (combined)*3
Forestomach: epithelial
hyperplasia*3
Liver necrosis'3
Nasal cavity lesions (chronic
inflammation, fibrosis, car-
tilaginous metaplasia, osse-
ous metaplasia, atrophy of
sensory epithelium*3
Testicular atrophy*3
Forestomach: epithelial
hyperplasiab
Ovary: atrophy*3
Uterus: involution*3
aGroups of 100 male and female Sprague-Dawley rats were exposed to air contain-
ing 0, 1,000 or 8,000 ppm 1,3-butadiene 6 hours/day, 5 days/week for
105 weeks (female), or 111 weeks (male); survival in dosed groups decreased.
bStatistically significant (p < 0.05)
cGroups of 50 male and female B6C3F1 mice were exposed to air containing 0, 625
or 1,250 ppm 1,3-butadiene 6 hour/day, 5 days/week for 60 week (male) or 61
weeks (female); survival in dosed groups decreased and was the reason for early
termination.
^Considered uncommon at 60 weeks.
SOURCE: National Toxicology Program, 1984.
40
-------
DRAFT—DO NOT QUOTE OR CITE
hence exposures to, several different chemicals. The two major components of
SBR polymers are styrene and butadiene. In a typical recipe for the production
of SBR, butadiene and styrene account for 26% and 9%, respectively, of the
total ingredients. It should be pointed out that water accounts for 63% of the
volume. At room temperature styrene is a clear, colorless liquid, while buta-
diene is a gas.
Two other agents, toluene and benzene, need to be considered, although they
are not used directly in the manufacture of SBR. Toluene exposures result from
its periodic use as a tank-cleaning agent; it may also exist as an impurity of
styrene. Benzene exposures may occur as an impurity of styrene or toluene.
j
Occupational epidemiologic studies investigating the potential health ha-
zards associated with the production of synthetic rubber have been very limited
in number. Because styrene and butadiene are the two basic materials used in
the manufacture of SBR, with benzene and toluene as byproducts, it is at best
difficult to assess the singular contribution of each. Benzene exposure has
been identified with excessive risk, particularly acute leukemia (Linet, 1985).
Styrene also may be a leukemogen in humans (Ott et al., 1980; Lilis and Nichol-
son, 1976). Although many studies of rubber production workers have been con-
ducted, only a few of those studies are relevant to butadiene exposure. Those
studies, which are reviewed here, include studies of workers specifically iden-
tified as working in styrene-butadiene production or the manufacture of synthe-
tic rubber. A study was also included if it was a preliminary study to one in
which the workers were identified as SBR or synthetic rubber workers, or if it
added to the interpretation of one of the studies of SBR or synthetic rubber
workers.
4.3.1. McHichael et al. (1974, 1976)
In 1974, McMichael et al. identified, through company records, a historic
41
-------
DRAFT—DO NOT QUOTE OR CITE
prospective cohort of 6,678 hourly paid male workers in a rubber tire manufac-
turing plant in Akron, Ohio. The cohort was composed of all active and retired
male employees aged 40 to 84 years as of January 1, 1964.
During the 9-year follow-up period from 1964 through 1972, 1,783 workers
died. Death certificates were obtained for 99.5% of these workers, and the
causes of death were coded according to the 8th revision of the International
Classification of Diseases (ICD), by a National Center for Health Statistics
nosologist.
SMRs were calculated for all males using the 1968 U.S. male population as
a standard population. SMRs for all causes for the active age range of 40-64
and the full age range of 40-84 were 93 and 99, respectively. In cause-speci-
fic SMRs, statistically significant excesses were observed for stomach cancer
(SMR = 219, observed = 12, expected = 5.5, p < 0.01), lymphosarcoma (SMR = 251,
observed = 6, expected = 2.4, p < 0.05), and leukemia (SMR = 315, observed =
11, expected = 3.5, p < 0.001), in the active age range of 40-64. For the full
age range of 40-84, significant SMR increases were observed for cancers of the
stomach (SMR = 187, observed = 39, expected = 20.9, p < 0.001), prostate, (SMR
= 142, observed = 49, expected = 34.4, p < 0.05), lymphosarcoma (SMR = 226,
observed = 14, expected = 6.2, p < 0.01), diabetes mellitus (SMR = 143, observed
= 43, expected = 30, p < 0.05), and arteriosclerosis (SMR = 154, observed =
34, expected = 22.1, p < 0.05).
McMichael et al. (1976) attempted to evaluate the relationship of these
mortality excesses to specific jobs within this plant by designing a nested
case-control study. Out of a total of 1,983 deaths observed during the 10-year
follow-up period of 1964 through 1973, 455 individuals who had died from certain
specific causes were selected as cases. The specific causes of death included
stomach, colorectal, respiratory, prostate, and bladder cancers; cancers of the
42
-------
DRAFT—DO NOT QUOTE OR CITE
lymphatic and hematopoietic systems; lymphatic leukemias; ischemic heart dis-
ease; and diabetes mellitus. Out of these 455 cases, 353 deaths were attri-
buted to cancers and 102 to noncancer causes. From the remainder of the plant
population of male workers, an age-stratified random sample of 1,500 individuals,
with 500 individuals in each age group of 40-54, 55-64 and 65-84 was obtained.
Complete work histories were obtained for 99% (1,482) of this age-stratified
sample drawn as a control group.
As of January 1, 1964, the plant population of male workers had a racial
composition of 86% white and 14% black. Thirty-eight percent, 30%, and 32% were
in the 40-54, 55-64, and 65-84 age ranges, respectively. Forty-eight percent
had begun work in the plant at least 25 years prior to 1964, and 99% had worked
for at least 10 years by 1964.
Work exposure histories were restricted to the period from 1940 through
1960. Cumulative job exposures of less than 2 years were excluded from the
analyses. Because follow-up extended to 1972, the period between first exposure
and death could range from 12 to 32 years, which should allow for the observation
of occupationally induced cancers in adults.
For each of the cause-specific mortality groups, as well as the group of
controls, rates of exposure for minimum duration of 2 years and 5 years were
calculated for each of 16 occupational title groups (OTGs) in order to ascer-
tain any dose-response relationships. The exposure rates in each case group
were age-adjusted by the direct method of adjustment to the age distribution of
the controls. For the nine cause-specific mortality groups, the ratios of
their age-adjusted job classification exposure rates to the rates within the
sample of controls were calculated in order to provide an approximation for
the more conventional odds ratios.
For all of the causes of death under investigation, there were statisti-
43
-------
DRAFT—DO NOT QUOTE OR CITE
cally significant (p < 0.001) associations with many of the work areas in which
workers had had at least 5 years of exposure. In the synthetic plant area, the
significant (p < 0.001) risk ratios were 6.2 for lymphatic and hematopoietic
cancer, 3.9 for lymphatic leukemia, 3.0 for ischemic heart disease, and 2.2 for
stomach cancer. Among the various work areas, the risk ratios for lymphatic
leukemia and for lymphatic and hematopoietic cancer were the highest in the
synthetic plant. Spirtas (1976) reported, however, that the risk ratio for
lymphatic and hematopoietic cancer dropped from 6.2 to 2.4 when a smaller
matched control group was used; the statistical significance of the 2.4 risk
ratio was not indicated.
McMichael et al. (1976) reported that a case-control study (McMichael et
al., 1975) had found an association between lymphatic leukemia and solvent expo-
sure in the rubber industry. Many of the lymphatic leukemia deaths were the
same as those reported in the McMichael et al. (1976) study. Further analysis
by Checkoway et al. (1984) of 11 of the lymphatic leukemia cases studied by Mc-
Michael et al. also found an association between lymphatic leukemia and solvent
exposure. Spirtas (1976) reported that of the six deaths from neoplasms of
lymphatic and hematopoietic tissue of individuals who had worked in the synthetic
rubber area of the plant in the McMichael et al. (1976) study, three were due
to leukemia. Of these three individuals, two had had experience with solvents
other than in the synthetic plant. Thus the role of the transfer of individuals
from one work area into another needs to be investigated. Also, racial factors
could not be accounted for in exposure calculations because data on race were
not available for much of the study population at the time of sampling.
4.3.2. Andjelkovich et al. (1976, 1977)
The mortality experience during the period from January 1, 1964 through
December 31, 1973 of a historic prospective cohort of 8,938 male rubber workers
44
-------
DRAFT—DO NOT QUOTE OR CITE
(known as the "1964 cohort") from a single plant located in Akron, Ohio, was
observed by Andjelkovich et al. in 1976. Any person who was 40 years of age
or more on January 1, 1964, and was an active or living retired hourly worker
from the plant under study, was included in the 1964 cohort.
Data were collected from company records, life insurance death claims,
and bureaus of vital statistics of several states. A trained nosologist coded
the causes of death according to the 8th revision of the ICD. Follow-up was
achieved for 96.7% of the cohort. Out of 8,938 males, 94% (8,418) were white
males and were equally distributed in the age groups 40-54, 55-64, and 65-84.
Although 6% of the cohort consisted of black males, the major analyses were
done on white males. During the 10-year observation period 2,373 (28%) of the
white males died. SMRs were calculated using the age-, race-, and sex-specific
rates of the 1968 U.S. population. SMRs for deaths from all causes in the
40-64, 65-84, and 40-84 age groups were, respectively, 92 (observed = 619), 95
(observed = 1,754, p < 0.05) and 94, (observed = 2,373, p < 0.01).
Many cause-specific SMRs showed increases, but only two disease SMRs,
those for neoplasms of lymphatic and hematopoietic tissue (SMR = 138, observed
= 40) and chronic rheumatic heart disease (SMR = 170, observed = 16) for the
age group 65-84 were statistically significant (p < 0.05). The only statisti-
cally significant (p < 0.05) SMR for the age group 40-64 was for cerebrovascular
disease (SMR = 138, observed = 48). On further detailed examination of neoplasms
of lymphatic and hematopoietic tissue, statistically significant excesses were
found for monocytic leukemia (SMR = 441, observed = 3, p < 0.01) and other
neoplasms of lymphatic and hematopoietic tissue (SMR = 276, observed = 10, p <
0.001) in the age group 65-84. There were no deaths by either of these causes
in the age group 40-64.
An important finding of this study is that it found a high mortality rate
45
-------
DRAFT—DO NOT QUOTE OR CITE
in workers who had retired between the ages of 40 and 64, the mandatory retire-
ment age being 65. The SMR for all causes for this group was 202, which is
highly statistically significant (observed = 299, p < 0.001). The SMR for
almost every cause analyzed was elevated, and out of 26 categories, 13 of them
were statistically significant. For malignant diseases, the authors found
significant elevations in SMRs for malignant neoplasms of the prostate (SMR =
278, observed = 4, p < 0.05); large intestine (SMR = 231, observed = 6, p <
0.05); trachea, bronchus, and lung (SMR = 241, observed = 28, p < 0.001); and
brain and central nervous system (SMR = 323, observed = 3, p < 0.05). In
non-malignant diseases, SMRs were statistically significant for 1) benign neo-
plasms and neoplasms of unspecified nature (SMR = 541, observed = 2, p < 0.01);
2) endocrine, nutritional, and metabolic diseases (SMR = 396, observed = 11,
p < 0.001); 3) diseases of the nervous system and sense organs (SMR = 577,
observed = 6, p < 0.001); 4) chronic rheumatic heart disease (SMR = 440, ob-
served^ 7, p < 0.001); 5) ischemic heart disease (SMR = 180, observed = 112,
p < 0.001); 6) cerebrovascular disease (SMR = 258, observed = 22, p < 0.001);
7) other respiratory disease (SMR = 309, observed = 5, p < 0.01); 8) diseases
of the digestive system (SMR = 357, observed = 16, p < 0.01); and 9) symptoms
and ill-defined conditions (SMR = 263, observed = 4, p < 0.05).
As opposed to these increases, the SMR for deaths from all causes for ac-
tive workers in the 40-64 age group was 61 (observed = 320), substantially
lower than the SMR of 202 for retired workers in this age group. The overall
SMR for active and retired workers combined was 92 (observed = 619), showing a
dilution effect by the active workers and confirming the "healthy worker"
effect. Some cause-specific SMRs were elevated slightly in this active group,
but none of them were statistically significant.
In an attempt to evaluate the relationship between the mortality increases
46
-------
DRAFT—DO NOT QUOTE OR CITE
and occupational exposures, Andjelkovich et al. (1977) re-analyzed the same
data for 28 work areas within the plant under study in 1977. The Occupational
Title Group (OTG) of each person was decided on the basis of the most represen-
tative department (obtained from personnel folders) in which the individual had
worked. The closing date for the active workers was December 31, 1973, while
for the retired or terminated workers the period of study was from the date of
hire to the last date worked.
All causes of death and cause-specific SMRs were calculated by using the
experience of the entire cohort as a reference group. Marginal increases in
SMRs for all causes were observed for many OTGs for all three of the age groups
considered: 40-64, 65-84, and 40-84 years. The only statistically significant
excess observed was for cast film manufacture (SMR = 230, observed = 7, p <
0.05) in age group 65-84. Statistically significant (p < 0.05) SMR deficits
from all causes were observed for OTGs: 1) product fabrication (tire and beads),
2) product fabrication (valves, tubes, and flaps), and 3) bulk chemicals and
*
metal products, for at least one or more of the three age groups.
SMR increases were statistically significant for all neoplasms in the fol-
lowing four OTGs: 1) cast film manufacture for age group 40-84, 2) special
products manufacture for age group 40-84, 3) milling for age groups 65-84 and
40-84, and 4) miscellaneous for age groups 40-64 and 40-84. Out of these four
OTGs, the first three departments dealt with the manufacture of industrial
products.
For selected cancers, the SMRs were significantly (p < 0.05) elevated in
age group 40-84 in certain OTGs, namely, cancer of the stomach in compounding
and mixing (SMR = 479, observed = 3), and milling (SMR = 369, observed = 6);
cancer of the large intestine in special products (SMR = 629, observed = 4);
cancers of the trachea, bronchus, and lung in synthetic latex (SMR = 434,
47
-------
DRAFT—DO NOT QUOTE OR CITE
observed = 3); cancer of the prostate (SMR = 212, observed = 10); and leukemia
(SMR = 246, observed = 6) in general services. All SMRs, except the one for
the compounding and mixing department, showed statistically significant excesses
of deaths in more than one age group.
For non-malignant diseases, Andjelkovich et al. (1977) calculated signifi-
cantly (p < 0.05) elevated SMRs for diabetes mellitus, acute myocardial infarc-
tion, arteriosclerosis, and suicide in various OTGs for various age groups.
Significant excesses of deaths in the general service department and arterio-
sclerosis in the shipping and receiving department were observed in more than
one age group. Statistically significant deficits were observed for ischemic
heart disease in the industrial chemicals department and acute myocardial
infarction in the stock preparation department, both in the 40-84 age group.
Since the authors were aware that the job transfer patterns of the deceased
workers were not necessarily representative of the job transfer patterns of the
entire-cohort, they used the available information on deceased workers to esti-
mate the length of time spent in the most representative department. A simple
random sample of 50 deceased workers was chosen, and detailed work histories
were reviewed for them. The 50 workers had spent an average of 28.3 years in
the industry. On an average, each worker had spent 50% of his work time in
his most representative OTG. However, the fraction of time spent in the most
representative OTG ranged from less than 10% up to 100% of total employment
duration.
The Andjelkovich et al. studies have a number of limitations. First,
their use of the most representative departments should be questioned in view
of the fact that the people under study could have worked in these departments
from 10% to 100% of their total employment duration. The only elevated SMR in
synthetic latex was for cancer of the trachea, bronchus, and lung, based on only
48
-------
DRAFT—DO NOT QUOTE OR CITE
three deaths, while cigarette smoking, which is a potential confounder, was not
controlled for. Another limitation with regard to the observed elevation of
trachea, bronchus, and lung cancer was the use of 1968 mortality data to calcu-
late the expected number of deaths. Mortality for trachea, bronchus, and lung
cancer was rising sharply during the period 1964 to 1983, the follow-up period
of this study. The use of 1968 data may have underestimated the expected num-
ber of deaths and thus overestimated the SMR. Although a statistically signi-
ficant excess of deaths for cancers of lymphatic and hematopoietic tissue was
observed for persons whose most representative department was general services,
this job category does not necessarily involve contact with SBR production.
With regard to the questionable exposure, Taulbee et al. (1976) reported that
in an analysis of the work histories of 37 leukemia cases and four matched
controls per case from the 1964 cohort of Andjelkovich et al., none of the
cases were found to have worked in the OTG "synthetic plant." Some of the
cases Had worked in departments in which there may have been exposure to the
synthetic process, but this association was not statistically significant
(p < 0.05), nor was the association found to increase by duration of exposure.
One positive aspect of the Andjelkovich et al. (1977) study is that the
entire cohort is used as a reference group, which should reduce the "healthy
worker effect" and allow for a more unbiased evaluation. It would have been
interesting, however, to see the comparison of SMRs calculated from the U.S.
population (1968), which was used as a standard population in the 1976 study,
with the SMRs calculated from the internal cohort as a reference group.
Both the McMichael et al. and the Andjelkovich et al. studies are sugges-
tive of some health problems in the synthetic plant, indicating that specific
exposure investigations should be undertaken.
49
-------
DRAFT--DO NOT QUOTE OR CITE
4.3.3. Checkoway and Williams (1982)
Since the study «;, McMichael et al. (1976) indicated the potential pre-
sence of carcinogens in SBR plants, Checkoway and Williams conducted a combined
industrial hygiene and hematology cross-sectional survey at the same plant
studied by McMichael et al. The objectives of the Checkoway and Williams study
were to quantify workplace exposures to styrene, butadiene, benzene, and toluene,
and to relate exposure levels to hematologic measurements.
During the week of May 15-19, 1979, personal breathing-zone air samples
were collected with both a charcoal tube and a passive diffusion dosimeter
during the day and evening shifts for seven different departments. The depart-
ments were: 1) tank farm, 2) reactor and recovery, 3) latex blending and
solution, 4) shipping and receiving, 5) storeroom, 6) factory service, and 7)
maintenance areas. Sampling periods ranged from 4 to 6 hours. Charcoal tubes
were changed at intervals of 1 to 2 hours during the sampling period to avoid
overloading.
Blood samples of male hourly production workers for the same departments
were obtained on 4 separate days, May 15-18. Of the 163 workers (26-65 years
of age with a median age of 45 years; 144 whites and 19 blacks) who participa-
ted in the industrial hygiene survey, 154 (135 whites and 19 blacks) also par-
ticipated in the blood survey. Because of work scheduling demands, blood
samples were collected from participants from each of the departments on all 4
days, thereby minimizing any bias due to day effect. The hematological parame-
ters measured included red cell count (RBC), hemoglobin concentration, hemato-
crit, mean corpuscular volume (MCV), mean corpuscular hemoglobin concentration,
reticulocyte count, platelets, total leukocytes (WBC), and differential distri-
butions of neutrophils, neutrophil band forms, eosinophils, basophils, monocytes,
and lymphocytes. Medical histories were obtained by means of questionnaires.
50
-------
DRAFT—DO NOT QUOTE OR CITE
Data from persons who reported positive histories of either malignant disease,
radiation therapy, or current anemia of known etiology were excluded from the
analyses. Only one individual, who reported a history of leukemia, was excluded
from the study.
The mean 8-hour time-weighted averages and ranges show that all four che-
mical exposures were well below the American Conference of Governmental Indus-
trial Hygienists (ACGIH) Threshold Limit Values (TLVs) recommended at that
time. The TLVs in parts per million (ppm) for butadiene, styrene, benzene, and
toluene are 1,000, 100, 10, and 100, respectively. With the exception of
butadiene and styrene, for which time-weighted averages of 20.03 ppm and 13.67
ppm, respectively, were observed in the tank farm area, the mean levels for the
four chemicals in all other departments were less than 2 ppm. Even the maximum
concentration of benzene, the most strongly suspected leukemogen of the four
chemicals analyzed, was less than 1 ppm in all plant departments.
Wi_th regard to the hematologic survey, there were generally no associations
(p > 0.05) of hematologic values with chemical exposures. Red blood cell count
was negatively associated with butadiene and styrene exposure, while basophiI
count was positively related to the aforementioned chemicals, as measured by
Pearson product moment analysis. The negative association of styrene with
erythrocyte counts and the positive association of the basophil proportions
with butadiene persisted after controlling for age and medical status in step-
wise multiple linear regression analyses. However, there were curious opposing
findings for mean corpuscular hemoglobin concentration—a positive relationship
to butadiene and a negative association with styrene.
Mean hematologic mesurements adjusted for age and medical status were com-
pared for tank farm area workers and all other workers. The tank farm workers
had slightly lower levels of circulating erythrocytes, hemoglobin, platelets,
51
-------
DRAFT—DO NOT QUOTE OR CITE
and neutrophils, in addition to mean corpuscular red cell volumes and neutro-
phil band circuits that were slightly higher than those of the other workers.
This study was undertaken to quantify exposure levels and to find out if
there is any evidence of hematopoietic toxicity in relation to these exposure
levels. With the exception of the tank farm area, the average exposures to the
four chemicals assayed were uniformly less than 2 ppm; even in the tank farm
area, the styrene and butadiene concentrations were considerably lower than the
recommended ACGIH TLVs, although they were considerably higher than in other
work places studied.
Because this study is cross-sectional in design, it is very limited with
regard to determining whether styrene-butadiene exposure is carcinogenic.
Individuals in the plant who may have developed cancer probably left the work
force and hence were not available for blood sampling. The industrial hygiene
survey findings cannot be generalized to the past, since the concentrations may
have differed quantitatively as well as qualitatively. It can be concluded
that there was no pronounced evidence of hematologic abnormality in this study
population.
4.3.4. Meinhardt et al. (1982)
Meinhardt et al. (1982) reported on a retrospective cohort mortality study
conducted by NIOSH at two adjacent SBR facilities in eastern Texas. This study
was motivated by the report of two men who had worked at both plants, and who
had died of leukemia in January 1976.
Personnel employment records documenting the employment of 3,494 workers
from plant A and 2,015 from plant B were available since January 1943 and
January 1950, respectively, to the study cut-off date of March 31, 1976. The
study cohorts from plants A and B consisted of 1,662 and 1,094 white males who
had had at least 6 months of non-management and non-administrative employment,
52
-------
DRAFT—DO NOT QUOTE OR CITE
respectively. The average lengths of employment for the study cohort in plants
A and B were, respectively, 9.48 and 10.78 years.
At the time of the study, environmental samples were obtained at each
plant. At plant A, time-weighted average exposures of styrene, butadiene, and
benzene were 0.94 ppm (0.03-6.46), 1.24 ppm (0.11-4.17) and 0.10 ppm (0.08-0.14)
respectively. For plant B, time-weighted average exposures for styrene and
butadiene were 1.99 ppm (0.05-12.3) and 13.5 ppm (0.34-174.0), respectively
(benzene was not measured). No historical monitoring data were available for
either plant.
The study cohorts from plants A and B accounted for 34,187 and 19,742
person-years at risk of dying. It is also important to note that the survival
status of 54 individuals (3.25%) from cohort A and 34 individuals (3.11%)
from cohort B was unable to be determined. In subsequent analyses, these
individuals were considered to be alive.
Thje age, race, sex, calendar time, and cause-specific mortality rates of
the U.S. population were applied to the appropriate strata of person-years at
risk in order to obtain the expected number of cause-specific deaths in the
study populations. Differences in observed and expected numbers of deaths were
evaluated by a test statistic based on the Poisson distribution.
In cohorts from plants A and B, observed and expected numbers of deaths
were compared for the following cause-specific categories: tuberculosis;
malignant neoplams (including cancers of the lymphatic and hematopoietic tis-
sues); all other cancers; diseases of the nervous, circulatory, respiratory,
and digestive systems; accidents; and all other causes. With the exception of
mortalities from cancers of the lymphatic and hematopoietic tissues, there were
deficits (in some instances, striking deficits) in the cause-specific SMRs for
the study cohorts in both the plants.
53
-------
DRAFT--DO NOT QUOTE OR CITE
With reyard to the total number of deaths due to all causes, cohorts A and
B had observed numbers of deaths of 252 (SMR = 80) and 80 (SMR = 66), respec-
tively. Although it is possible that the "healthy worker" effect may, in part,
explain these deficits, the relative magnitudes of the deficits, particularly
for plant B, suggest that there may have been an underreporting of deaths or that
selection factors in the choice of the study groups reduced the mortality rate.
Meinhardt et al. observed that all five of the individuals from plant A
whose underlying cause of death was leukemia began employment before the end of
December 1945. This date corresponds to the time when the batch process for
SBR production was converted to a continuous-feed operation. The decision was
made, therefore, to evaluate the mortality experience of those 600 white male
employees who had had at least 6 months of employment in plant A between Janu-
ary 1, 1943 and December 31, 1945. This subgroup was employed for an average
of 11.9 years, and had an accumulation of 17,086 person-years at risk of dying.
The survival status of 34 individuals (5.7%) was unknown. As in the previous
mortality analyses, with the exception of deaths from cancers of the lymphatic
and hematopoietic tissues, there generally were large deficits in the cause-
specific observed number of deaths. For malignant neoplasms of lymphatic and
hematopoietic tissues, the SMR was 212 (9 observed, 4.25 expected, 0.05 < p
< 0.1); for lymphosarcoma and reticulosarcoma the SMR was 224 (3 cases, 1.34
expected, p > 0.05); for Hodgkin's disease the SMR was 213 (1 case, 0.47 ex-
pected, p > 0.05); and for leukemia and aleukemia the SMR was 278 (5 cases,
1.80 expected, 0.05 < p < 0.1). The total number of observed deaths due to all
causes was 201 (SMR = 83, 242.09 expected, p < 0.05). For cohort B, there
were no significant (p > 0.05) excesses of mortality from any cause. Deaths
from all malignant neoplasms (SMR = 53, observed = 11, expected = 20.78, p <
0.05) and "all other causes" (SMR = 54, observed = 9, expected = 16.80, p <
54
-------
DRAFT—DO NOT QUOTE OR CITE
0.05) were significantly decreased, however.
The authors calculated the likelihood of detecting a doubling and a qua-
drupling of the expected occurrence of leukemia for cohort A, subcohort A, and
cohort B. These likelihoods were 26% and 88% for cohort A, 20% and 77% for
subcohort A, and 13% and 51% for cohort B. Thus, this study suggests that some
component of the SBR manufacturing process may be a leukemogen.
4.3.5. Matanoski et al. (1982)
This study was performed to determine whether there are health risks
associated with the production of synthetic rubber, specifically styrene-buta-
diene rubber (SBR). The populations studied were obtained from seven U.S. and
one Canadian rubber plants. The study population consisted of males who had
worked for more than one year, and whose records contained birth dates and
employment dates. In addition, workers selected had been employed by the
facility from the date of the facility's first SBR production to December 1976.
The total number of people available from the eight plants surveyed was 29,179,
of whom 13,920 (48%) met the criteria for selection.
Data obtained from the personnel records of each facility included employee
name, Social Security number, job history, date(s) of employment, birth infor-
mation, death information, and limited data on retirees. Individual workers'
jobs were coded according to first job, last job, and job held longest during
the period of employment. For the analyses, jobs were categorized in four
general work areas: production, utilities, maintenance, and other. In three
plants (plants 3, 4, and 5), race classification was unknown for 176 (85%),
329 (50%), and 4,540 (98%) of the cohort populations, respectively. Plants 3
and 4 were expected to have employed black workers; plant 5 had employed few or
no blacks at any time in its history. In all, 7,209 (52%) of the study popula-
tion were unable to be classified racially. If race was not specified, indivi-
55
-------
DRAFT--DO NOT QUOTE OR CITE
duals were assumed to be white males.
Follow-up activities to determine vital status of the study population
in the seven U.S. plants included searching by the Social Security Administra-
tion, tracing through motor vehicle administration records, and contacts by
telephone. Through these follow-up activities, 42% of the study population
were traced. For the Canadian plant, follow-up was performed by searching the
company insurance plan records for death benefit information. Determination
of vital status for the study populations revealed 10,899 workers alive, 2,097
known dead, and 924 lost to follow-up. It was determined from a 10% sample in
each plant that about 4% of the study population assumed to be living was
actually dead. Thus, as many as 440 deaths, or 17% of the total possible
deaths, might have been missed. For the populations who were known dead, 90%
of the death certificates requested were received. Death certificates were
coded according to the 8th revision of the ICD.
Mast of the statistical analyses were done using the worker records from
the time when record-keeping systems became complete, or any time thereafter
through December 1976. Females, workers employed less than one year, and those
with unknown birth dates or employment dates were omitted. The total eligible
population numbered 13,608. SMRs for the workers as compared to the general
population were calculated by using a modified, version of U.S. Death Rates Pro-
grams (Monson, 1979). SMRs were calculated separately for the white population
and the black population, and the ratios were combined to correct for differen-
ces in age, race, and calendar time. The total number of deaths occurring
among the eligible study population was 1,995, with an SMR of 81. The study
cohort accounted for 250,000 person-years.
Power calculations were performed to test the ability of the data to deter-
mine increases in risk of 0.1, 0.25, 0.5, and 2 times greater then the U.S.
56
-------
DRAFT—DO NOT QUOTE OR CITE
population. The calculations showed a 100% probability of detecting a twofold
increase in all causes of death and in all cancers, and an 89% probability of
detecting a 25% increase in lung cancer.
The average period of follow-up was 19 to 20 years. The average age at
death was 61 years. The overall SMR for all causes of death was 81, with SMRs
of 98 for blacks and 78 for whites. The low mortality among members of this
population is, in part, a reflection of the "healthy worker" effect. However,
a question must also be raised as to the effect of the 440 possible deaths that
may have been missed. Furthermore, the large difference in SMRs for blacks and
whites may be due to an undercounting of blacks and an overcounting of whites.
This is further exemplified by similar patterns in SMRs for all accidents,
motor vehicle accidents, and suicides. The SMRs for all causes of death do not
appear to increase with duration of employment but do appear to increase with
increasing follow-up period.
The leukemias found in the population included nine acute, five chronic,
and three unspecified on the death certificate, with a median latent period
of 17 years from time of first employment. The types found were not considered
to have a distribution remarkably different from that found in the general
population. The study population as a whole did not demonstrate leukemia in
excess.
None of the analyses demonstrated significant increases in SMRs for other
specific causes in the total study population. Black males appear to have a
significantly elevated risk of arteriosclerotic heart disease (p < 0.05), but
this value may be artificially inflated due to undercounting of the blacks,
resulting in a smaller denominator. Vascular lesions of the central nervous
system were also in excess in black males, but they were not statistically
significant.
57
-------
DRAFT—DO NOT QUOTE OR CITE
Increases in mortality according to job classifications are noteworthy,
but only two of them are statistically significant (p < 0.05). The SMR for
testicular cancer among maintenance workers is 294 (observed = 3, p < 0.05).
SMRs for esophagus, stomach and large intestine, and larynx cancers are
elevated in the utilities and maintenance work areas. The SMR for larynx can-
cer in utilities workers is statistically significant (SMR = 476, observed =
4, p < 0.05). Hodgkin's disease is associated with high SMRs in all work
areas. Very few high SMRs are found for the production area.
It should be further pointed out that, with the exceptions of Hodgkin's
disease and stomach cancer, all of the cancers had relatively long latent peri-
ods. For all cancers, half of the individuals had had 12 years of employment
or more.
In addition to complete ascertainment of deaths and racial distribution,
further investigation is needed in order to obtain information specific to
various jobs and the associated exposures of individuals in those jobs. It
would also be useful to separate the number of years employed within each job
category, in order to determine the periods of possible exposure in the work-
place. Moreover, changes in the SBR process, in plant design, and in worker
practices should be given greater attention in the evaluation of mortality for
the four worker categories studied.
Other methodologic limitations of this study include the fact that less
than 50% of the total population of eight plants was studied. This raises
questions about the population that was excluded due to lack of birth dates or
employment dates. This may have been an older population, which probably had
longer exposure and was therefore more likely to suffer from occupational dis-
eases. Out of eight cohorts, only 50% were followed from 1943, whereas in
the rest of the plants, follow-up starting dates ranged from 1953 to 1970. It
58
-------
DRAFT--DO NOT QUOTE OR CITE
is probable that the employees from the latter plants were not followed long
enough for malignancies to develop.
4.3.6. Summary of Epidemlologic Studies
McMichael et al. (1974) found significant (p < 0.05) excess mortality from
cancers of the stomach, prostate, and lymphatic and hematopoietic system as
well as from diabetes mellitus, arteriosclerosis, and ischemic heart disease in
their historic prospective cohort study. To evaluate these excesses, McMichael
et al. (1976), in a case-control study, investigated exposure rates for various
jobs within the same rubber plant for several cause-specific deaths. The data
indicated that the most probable health risks were of prostate cancer in jani-
toring and trucking; bladder cancer in milling and reclaiming operations; lym-
phatic and hematopoietic cancers in the synthetic plant; and lymphatic leukemia
in the synthetic plant, inspection-finishing-repair, and tread cementing. Non-
malignant mortality excesses included ischemic heart disease in the synthetic
plant and tread cementing and diabetes mellitus in janitoring and trucking and
inspection-finishing-repair.
Andjelkovich et al. (1976) carried out a similar kind of study (historic
prospective cohort) in white males in a single rubber plant, and observed sig-
nificantly (p < 0.05) increased SMRs for malignant neoplasms of lymphatic and
hematopoietic tissues (monocytic leukemia and other neoplasms of lymphatic and
hematopoietic tissue), chronic rheumatic heart disease, and cerebrovascular
disease. They also observed high SMRs (p < 0.001) for all causes and for most
of the cause-specific deaths for a group of workers who had retired between the
ages of 40 and 64. Andjelkovich et al. (1977) evaluated these excesses in
mortality ratios in relation to various work areas by using the entire cohort
as a reference group, and found that only malignant neoplasms of the trachea,
bronchus, and lung were associated with the synthetic latex department. This
59
-------
DRAFT—DO NOT QUOTE OR CITE
finding was based on only three observed deaths, however, and no smoking data
were taken.
Checkoway and Williams (1982) carried out an industrial hygiene and hema-
tology cross-sectional survey at the same plant in which McMichael et al. had
conducted their case-control study. With the exception of the tank farm area,
in which 8-hour time-weighted averages for butadiene and styrene were observed
to be 20.3 ppm and 13.67 ppm, respectively, all other departments had mean
exposure levels of less than 2.0 ppm. No evidence of hematologic abnormality
was noted. Because of its cross-sectional design, however, this study could
not be expected to identify an excess cancer risk.
Meinhardt et al. (1982), conducted a retrospective cohort mortality study
at two plants in Texas. Male rubber workers in one of the plants (plant A)
were followed from January 1, 1943 through March 31, 1975; employees in the
other plant (plant B) were followed from January 1, 1950 through March 31,
1976. These intervals of observation are important to note because the two
plants changed from a batch process for SBR production to a continuous-feed
operation in 1946. With regard to cancers of the lymphatic and hematopoietic
system and lymphatic leukemia, plant A exhibited excess mortalities, although
these were not statistically significant (p > 0.05); plant B did not show any
mortality excesses. When the mortality experience in plant A was analyzed
further for those workers who had had at least 6 months of employment between
January 1, 1943, and December 31, 1945 (the interval for which the batch pro-
cess was used), excess mortalities for the above-mentioned cancers were shown
to be of borderline statistical significance (0.05 < p < 0.01, two-sided).
It is also of interest to note that all of the employees from the total cohort
of plant A whose causes of death were cancers of the lymphatic and hematopoietic
tissues had been employed between 1943 and 1945. Had the analysis in plant A
60
-------
DRAFT--DO NOT QUOTE OR CITE
commenced with the date of first employment in 1946, the SMRs in question would
have been reduced to zoro, and the lack of excess mortality would have been
similar to plant B.
Matanoski et al. (1982) also conducted a retrospective cohort mortality
study in which eight SBR plants were involved. As with the cohort in plant B
investigated by Meinhardt et al., there was a general lack of excess mortali-
ties. It should also be noted that half of the cohort was followed from 1943
to 1979. The date of entry for the remaining half of the cohort ranged from
1953 to 1976, with follow-up terminating in 1979.
Although both the McMichael et al. (1976) and the Meinhardt et al. (1982)
studies found some evidence of an association between styrene-butadiene exposure
and lymphatic and hernatopoietic cancer, confounding due to exposures to solvents
cannot be ruled out for either study. In addition, the results from the Mein-
hardt et al. study for the subcohort employed during the batch process of pro-
duction were only of borderline (0.05 < p < 0.1, two-sided) significance, and
a possibly serious underascertainment of deaths, and/or selection factors in
the choice of the study group may have biased the results.
The Andjelkovich et al . (1977) study found an association between employ-
ment in the synthetic part of the plant with mortality from cancer of the
trachea, bronchus, and lung. The association was based on only three deaths,
however, and there was no control for smoking.
The study by Matanoski et al. of almost 14,000 styrene-butadiene produc-
tion workers found no excesses of cancer mortality that were statistically
significant (p < 0.05). Again, however, a possibly serious underascertainment
of deaths may have biased the results. An under-counting of blacks in the study
population may also have resulted in a potential bias.
The epidemiologic evaluation of SBR workers with regard to the carcinoge-
61
-------
DRAFT--DO NOT QUOTE OR CITE
nicity of 1,3-butadiene is particularly difficult because styrene may also be
a carcinogen and, in particular, a leukemogen (Ott et al., 1980; Lilis and
Nicholson, 1976). Because of the inconsistency of the results from different
studies, the possible confounding due to solvent and styrene exposures, and the
potential for bias in some of the studies, the epidemiologic data are considered
inadequate for determining a causal association between 1,3-butadiene exposure
and cancer in humans.
4.4. QUANTITATIVE ESTIMATION
This section deals with the incremental unit risk for 1,3-butadiene in
air, and the potency of 1,3-butadiene relative to other carcinogens that the
GAG has evaluated. The incremental unit risk estimate for an air pollutant is
defined as the excess lifetime cancer risk occurring in a hypothetical popula-
tion in which all individuals are exposed continuously from birth throughout
their lifetimes to a concentration of 1 g/m^ of the agent in the air they
breathe. This calculation estimates in quantitative terms the impact of the
agent as a carcinogen. Unit risk estimates are used for two purposes: 1) to
compare the carcinogenic potency of several agents with each other, and 2) to
give a crude indication of the population risk that might be associated with
air exposure to these agents if the actual exposures were known. Hereafter,
the term "unit risk" will refer to incremental unit risk.
4.4.1. Procedures for Determination of Unit Risk
The data used for quantitative estimation are taken from one or both of
the following: 1) lifetime animal studies, and 2) human studies where excess
cancer risk has been associated with exposure to the agent. In animal studies
it is assumed, unless evidence exists to the contrary, that if a carcinogenic
response occurs at the dose levels used in the study, then responses will also
occur at all lower doses, with an incidence determined by an extrapolation model.
62
-------
DRAFT--DO NOT QUOTE OR CITE
There is no solid scientific basis for any mathematical extrapolation
model that relates carcinogen exposure to cancer risks at the extremely low
concentrations that must be dealt with in evaluating environmental hazards.
For practical reasons, such low levels of risk cannot be measured directly
either by animal experiments or by epidemiologic studies. Ue must, therefore,
depend on our current understanding of the mechanisms of carcinogenesis for
guidance as to which risk model to use. At the present time the dominant view
of the carcinogenic process involves the concept that most agents that cause
cancer also cause irreversible damage to DNA. This position is reflected by
the fact that a very large proportion of agents that cause cancer are also
mutagenic.
There is reason to expect that the quanta! type of biological response,
which is characteristic of mutagenesis, is associated with a linear nonthreshold
dose-response relationship. Indeed, there is substantial evidence from mutage-
nicity studies with both ionizing radiation and a wide variety of chemicals
that this type of dose-response model is the appropriate one to use. This is
particularly true at the lower end of the dose-response curve; at higher doses,
there can be an upward curvature, probably reflecting the effects of multistage
processes on the mutagenic response. This linear nonthreshold dose-response
relationship is also consistent with the relatively few epidemiologic studies
of cancer responses to specific agents that contain enough information to make
the evaluation possible (e.g., radiation-induced leukemia, breast and thyroid
cancer, skin cancer induced by arsenic in drinking water, liver cancer induced
by aflatoxin in the diet). There is also some evidence from animal experiments
that is consistent with the linear non-threshold model (e.g., liver tumors
induced in mice by 2-acetylaminofluorene in the large-scale EDgi study at the
National Center for Toxicological Research, and the initiation stage of the
63
-------
DRAFT — DO NOT QUOTE OR CITE
two-stage carcinogenesis model in rat liver and mouse skin).
Based on the above evidence of low-dose linearity, and because very few
compounds exhibit low-dose responses that are super-linear, the linear non-
threshold model has been adopted as the primary basis for risk extrapolation to
low levels of the dose-response relationship. The incremental risk estimates
made with this model and the corresponding 95% upper-limit incremental unit
risks, should be regarded as conservative, representing the most plausible
upper limits for the risk, i.e., the true risk is not likely to be higher than
the estimates, but it could be lower.
The mathematical formulation chosen to describe the linear nonthreshold
dose-response relationship at low doses is the linearized multistage model.
This model employs enough arbitrary constants to fit almost any monotonically
increasing dose-response data, and it incorporates a procedure for estimating
the largest possible linear slope (in the 95% confidence limit sense) at low
extrapolated doses that is consistent with the data at all dose levels of the
experiment. The multistage model, described below, is fit to the data in the
observational or experimental range. The fit of the curve allows for a linear
term which dominates the risk estimate at low doses. The 95% upper limit,
described below, is technically an upper-limit estimate on the linear term,
but, practically, functions as the upper-limit low dose-response function.
4.4.1.1. Description of the Low-Dose Extrapolation Model
Let P(d) represent the lifetime risk (probability) of cancer at dose d.
The multistage model has the form
P(d) = 1 - exp [-(q0 + qjd + qxd2 + ...+ qkdk)]
where
64
-------
DRAFT—DO NOT QUOTE OR CITE
q-j >_ 0, 1 =0, 1, 2, ..., k
Equivalently,
Pt(d) = 1 - exp [-(q-^d + q2d2 + ... + qkdk)]
where
P,(d) = P(<0 - P(0)
t 1 - P(0)
is the extra risk over background rate at dose d.
The point estimate of the coefficients q-j, i = 0, 1, 2, ..., k, and con-
sequently, the extra risk function, Pt(d), at any given dose d, is calculated
by maximizing the likelihood function of the data.
The point estimate and the 95% upper confidence limit of the extra risk,
Pt(d), are calculated by using the computer program GLOBAL83, originally
developed by Crump and Watson (1979). At low doses, upper 95% confidence
limits on the extra risk and lower 95% confidence limits on the dose producing
a given risk are determined from a 95% upper confidence limit, q^, on parameter
qj. Whenever q^ > 0, at low doses the extra risk Pt(d) has approximately the
form Pt(d) = qj x d. Therefore, q^ x d is a 95% upper confidence limit on
the extra risk, and R/q^ is a 95% lower confidence limit on the dose, pro-
ducing an extra risk of R. Let LQ be the maximum value of the log-likelihood
function. The upper-limit, q^, is calculated by increasing q1 to a value
q1 such that when the log-likelihood is remaximized subject to this fixed
value qj for the linear coefficient, the resulting maximum value of the
65
-------
DRAFT--DO NOT QUOTE OR CITE
log-likelihood LI satisfies the equation
2 (L0 - LI) = 2.70554
where 2.70554 is the cumulative 90% point of the chi-square distribution with
one degree of freedom, which corresponds to a 95% upper limit (one-sided). This
approach of computing the upper confidence limit for the extra risk Pt(d) is an
improvement on the Crump et al . (1977) model. The upper confidence limit for
the extra risk calculated at low doses is always linear. This is conceptually
consistent with the linear nonthreshold concept discussed earlier. The slope,
q-p is taken as an upper bound of the potency of the chemical in inducing
cancer at low doses.
In fitting the dose-response model, the number of terms in the polynomial
is chosen equal to (h-1), where h is the number of dose groups in the experiment,
including the control group.
Whenever the multistage model does not fit the data sufficiently well,
data at the highest dose is deleted and the model is refit to the rest of the
data. This is continued until an acceptable fit to the data is obtained. To
determine whether or not a fit is acceptable, the chi-square statistic
X2=
.=1 NiPi (1-Pi)
is calculated where N^ is the number of animals in the i^1 dose group, Xj is
the number of animals in the itn dose group with a tumor response, P^ is the
probability of a response in the itn dose group estimated by fitting the
66
-------
DRAFT—DO NOT QUOTE OR CITE
multistage model to the data, and h is the number of remaining groups. The
fit is determined to be unacceptable whenever X2 is larger than the cumulative
99% point of the chi-square distribution with f degrees of freedom, where f
equals the number of dose groups minus the number of non-zero multistage co-
efficients.
When all the higher order terms in the multistage model are zero except
the linear term, the multistage model reduces to the one-hit model, which is a
true low-dose linear nonthreshold model. As will be seen with the animal data,
this is the case with 1,3-butadiene.
For cases of partial lifetime exposure where time-to-tumor or time-to-
tumor death is known, Crump and Howe (1983a) have developed the multistage
model to include a time-dependent dose pattern. The form of this model is
one which is linear in dose and in which time has a power and form determined
by both the number of assumed stages and the stage affected by the carcinogen.
A best fit is determined by the method of maximum likelihood in the ADOLL183
computer program (Crump and Howe, 1983b). Application of this program to the
NTP data was unsuccessful because of lack of convergence.
4.4.1.2. Selection of Data—For some chemicals, several studies in different
animal species, strains, and sexes, each run at several doses and different
routes of exposure, are available. A choice must be made as to which of the
data sets from several studies to use in the model. It may also be appropriate
to correct for metabolism differences between species and for absorption factors
via different routes of administration. The procedures used in evaluating these
data are consistent with the approach of making a maximum-likely risk estimate.
They are listed below:
1. The tumor incidence data are separated according to organ sites or
tumor types. The set of data (i.e., dose and tumor incidence) used in the
67
-------
DRAFT—DO NOT QUOTE OR CITE
model is the set in which the incidence is statistically significantly higher
than for the control for at least one test dose level, or in which the tumor
incidence rate shows a statistically significant trend with respect to dose
level, or in which a specific tumor appears unusually early for that site. The
data set that gives the highest estimate of the lifetime carcinogenic risk,
q-p is selected in most cases. However, efforts are made to exclude data sets
that produce spuriously high risk estimates because of a small number of animals.
That is, if two sets of data show a similar dose-response relationship, and one
has a very small sample size, the set of data having the larger sample size is
selected for calculating the carcinogenic potency.
2. If there are two or more data sets of comparable size that are iden-
tical with respect to species, strain, sex, and tumor sites, the geometric mean
of qp estimated from each of these data sets, is used for risk assessment.
If, as is the case with the 1,3-butadiene NTP (1984) study, males and females of
the species show a basically similar response, the usual procedure would be to
use the geometric mean of their individual q-^'s for the incremental 95% upper-
limit unit risk.
3. If two or more significant tumor sites are observed in the same study,
and if the data are available, the number of animals with at least one of the
specific tumor sites under consideration is used as incidence data in the model.
4.4.1.3. Calculation of Human Equivalent Dosages from Animal Data--Fo1lowing
the suggestion of Mantel and Schneiderman (1975), it is assumed that mg/surface
area/day is an equivalent dose between species. Since, to a close approximation,
surface area is proportional to the two-thirds power of weight, as would be the
case for a perfect sphere, the exposure in mg/day per two-thirds power of the
weight is also considered to be equivalent exposure. In an animal experiment,
this equivalent dose is computed in the following manner:
68
-------
DRAFT—DO NOT QUOTE OR CITE
Let
Le = duration of experiment
le = duration of exposure
m = average dose per day in mg during administration of the agent (i.e.,
during le), and
W = average weight of the experimental animal
The lifetime exposure is then
2/3
Le x W
1,3-Butadiene is slightly soluble in water and can be considered a par-
tially soluble vapor. The dose in m = mg/day of partially soluble vapors is
proportional to the 02 consumption, which in turn is proportional to W2/3 and
is also proportional to the solubility of the gas in body fluids, which can be
expressed as an absorption coefficient, r, for the gas. Therefore, expressing
the Q£ consumption as 02 = k W2/3, where k is a constant independent of species
and V = mg/m3 of the agent in air, it follows that:
m = k VK/'* x v x r
or
d = m = kvr
w2/3
In the absence of experimental information or a sound theoretical argument
to the contrary, the absorption fraction, r, is assumed to be the same for all
69
-------
DRAFT--DO NOT QUOTE OR CITE
species. Therefore, for these substances (e.g., 1,3-butadiene) a certain con-
centration in ppm orp g/m3 in experimental animals is assumed equivalent to
the same concentration in humans. This is supported by the observation that
the minimum alveolar concentration necessary to produce a given "stage" of
anesthesia is similar in man and animals (Dripps et al., 1977). When the
animals are exposed via the oral route and human exposure is via inhalation or
vice versa, the assumption is made, unless there is pharmacokinetic evidence to
the contrary, that absorption is equal by either exposure route.
4.4.1.4. Calculation of the Unit Risk from Animal Studies—The risk associated
with d mg/kg2/3/day is obtained from GLOBAL83 and, for most cases of interest to
risk assessment, can be adequately approximated by P(d) = 1 - exp (-q*d). An
incremental "unit risk" in units X is simply the risk corresponding to an expo-
sure of X = 1. To estimate this value, the number of mg/kg2/3/day corresponding
to one unit of X is determined and substituted into the above relationship.
Thus, for example, if X is in units of ppm orp g/m3 in the air, then for 1,3-
butadiene, d = 1, when ppm oru g/m3 is the unit used to compute parameters in
animal experiments.
If exposures are given in terms of ppm in air, then the conversion factor
to mg/m3 is
ppm = 1.2 x mo]ecular weight (gas) mg/nr
molecular weight (air)
4.4.1.4.1. Adjustments for less than lifetime duration of experiment—When
analyzing quantal data, if the duration of experiment Le is less than the
natural lifespan of the test animal L, the slope q*, or more generally the
exponent g(d), is increased by multiplying by a factor (L/LP)3. We assume that
70
-------
DKAFT--DO NOT QUOTE OR CITE
if the average dose d had been continued, the age-specific rate of cancer would
have continued to increase as a constant function of the background rate. The
age-specific rates for humans increase at least by the second power of the age
and often by a considerably higher power, as demonstrated by Doll (1971). Thus,
it is expected that the cumulative tumor rate would increase by at least the
third power of age. Using this fact, it is assumed that the slope q1, or
more generally the exponent g(d), would also increase by at least the third
power of age. As a result, if the slope q-^ [or g(d)] is calculated at age
Le, we expect that if the experiment had been continued for the full lifespan,
L, at the given average exposure, the slope q^ [or g(d)] would have been
o
increased by at least (L/l_e) .
For time-to-tumor data, this adjustment is also conceptually consistent
with the proportional hazard model proposed by Cox (1972) and the time-to-tumor
model considered by Crump (1979), where the probabiity of cancer by age t and
at dose d is given by
P(d,t) = 1 - exp[-f(t) x
It is also consistent with the partial lifetime exposure extension of the
multistage model developed by Crump and Howe (1983a), which as discussed above
is linear in dose, but has a power function of time.
4.4.1.5. Interpretation of Quantitative Estimates—For several reasons, the
unit risk estimate based on animal bioassays is only an approximate indication
of the absolute risk in populations exposed to known carcinogen concentrations.
First, there are important species differences in uptake, metabolism, and organ
distribution of carcinogens, as well as species differences in target site
susceptibility, immunological responses, hormone function, dietary factors, and
71
-------
DRAFT—DO NOT QUOTE OR CITE
disease. Second, the concept of equivalent doses for humans compared to animals
on a mg/surface area basis is virtually without experimental verification
regarding carcinogenic response. Finally, human populations are variable with
respect to genetic constitution and diet, living environment, activity patterns,
and other cultural factors.
The unit risk estimate can give a rough indication of the relative potency
of a given agent as compared with other carcinogens. The comparative potency
of different agents is more reliable when the comparison is based on studies
in the same test species, strain, and sex, and by the same route of exposure,
preferably inhalation.
The quantitative aspect of carcinogen risk assessment is included here
because it may be of use in the regulatory decision-making process, e.g.,
setting regulatory priorities, evaluating the adequacy of technology-based
controls, etc. However, it should be recognized that the estimation of cancer
risks to humans at low levels of exposure is uncertain. Because of the limited
data available from animal bioassays, especially at the high dose levels re-
quired for testing, almost nothing is known about the true shape of the dose-
response curve at low environmental levels. At best, the linear extrapolation
model used here provides a rough but plausible estimate of the upper limit of
risk; i.e., it is not likely that the true risk would be much more than the
estimated risk, but it could be considerably lower. The risk estimates pre-
sented in subsequent sections should not be regarded as accurate representations
of the true cancer risk even when the exposures are accurately defined. The
estimates presented may, however, be factored into regulatory decisions to the
extent that the concept of upper risk limits is found to be useful.
4.4.1.6. Alternative Models--The methods used by the CAG for quantitative ex-
trapolation from animal to man are generally conservative, i.e., tending toward
72
-------
DRAFT--DO NOT QUOTE OR CITE
high estimates of risk. The most important part of the methodology contribu-
ting to this conservatism is the CAG's use of the linearized multistage non-
threshold extrapolation model. There are a variety of other extrapolation
models that could be used, most of which would give lower risk estimates.
Among these alternative models, two which are currently popular and which often
tend to give different low-dose extrapolations than the multistage model are
the log-probit and Weibull models. These models have not been used in the fol-
lowing analysis of the mouse data because their fits to the 1,3-butadiene data
base were considered poorer than that of the multistage model. As discussed
below, all models are of limited value for predicting low-dose risks for 1,3-
butadiene because mouse responses were greater than 60% at the lowest dose
tested, and rat responses were far less sensitive than those of the mouse.
4.4.2. Calculation of Quantitative Estimates
Human studies have provided inadequate evidence for the carcinogenicity
of 1,3-butadiene. Furthermore, concurrent exposure to several other possible
carcinogens also limits the use of epidemiologic studies as primary sources for
calculating quantitative risk estimates. For animal-to-human extrapolation,
there are two suitable animal bioassays, both showing significant carcinogenic
response. The first, a rat inhalation study, showed statistically significant
increases in several tumor sites in both males and females (Hazleton Laborato-
ries, Ltd., 1981a). The second study was the NTP (1984) mouse inhalation
study, which showed high statistically significant increases (p < 0.01) both in
hemangiosarcomas of the heart and malignant lymphomas and in both males and
females at 625 ppm and 1,250 ppm. Both of these tumor types are life-threaten-
ing and appeared quite early in the study. As discussed at length in the
qualitative section and shown in Table 8, several other tumor sites were also
significantly increased in the study, which was stopped at 60-61 weeks due to
73
-------
DRAFT—DO NOT QUOTE OR CITE
high mortality from tumors in the treated groups.
The mice were used for risk estimation, since they were far more sensitive
than the rats. Incremental 95% upper-limit unit risk estimates were calculated
from both the male and female mouse data. For the male mice, the numbers of
animals either with tumors at significantly increased sites, or with tumors
considered unusual for 60 weeks (preputial gland squamous cell carcinomas and
Zymbal gland carcinomas (see Table 10) were 2/50, 43/50, and 40/50 for the
control, 625-ppm, and 1,250-ppm groups, respectively. For the females, the
numbers of animals with significantly increased tumors or brain gliomas were
4/50, 31/49, and 45/49. These results are presented in Table 11. Also presented
in Table 11 are the maximum likelihood estimates (MLE) and the 95% upper-limit
unit risk estimates (q^) based on these data. The initial upper-limit esti-
mates based on the 60-week (male) and 61-week (female) studies are then adjusted
to project for natural lifetime risk (see section 4.4.1.4.1.). The final
estimates are q^ = 7.1 x 10~2 (ppm)"1 for the males and q^ = 5.9 x 10~2
(ppm)~l for the females. Since the data sets are so comparable, the geometric
if 01
mean, q^ = 6.5 x 10~^ (ppm)"1, was chosen as the final 95% upper-limit unit
risk estimate.
This estimate can also be presented in terms of yg/m3. The conversion
factor is
M.W. 1,3-butadiene 54.1
1 ppb = 1.2 x M.W. air= 1-2 * 28.8
or 1 ppb = 2.25 ug/m3
or 1 ppm = 2.25 x 103 yg/m3
74
-------
DRAFT—DO NOT QUOTE OR CITE
TABLE 11. NUMBERS OF MICE WITH AT LEAST ONE OF THE
STATISTICALLY SIGNIFICANT INCREASED TUMORS, OR
TUMORS CONSIDERED UNUSUAL AT TIME OF TERMINAL SACRIFICE.
ALSO, MLE AND 95%-UPPER-LIMIT INCREMENTAL UNIT RISK ESTIMATES3
Nominal
exposure
(ppm)
0
625
1,250
Equivalent
continuous
exposure^
(ppm)
0
111.6
223.2
No. males
with tumors/
no. examined0
2/50 (4%)
43/50 (86%)
40/50 (80%)
No. females
with tumors/
no. examined0
4/50 (8%)
31/49 (63%)
45/49 (92%)
dSee Tables 8 and 10 for tumor sites.
^Continuous equivalent dose = animal exposure x 6/24 x 5/7.
cExamined for either hemangiosarcomas or lymphomas.
Males
Females
Initial maximum likelihood estimates q-^ = 8.1 x 10~3 (ppm)"1 q-^ = 1.0 x 10"2 (ppm)"1
Initial estimates of 95% upper limit q*
1.36 x ID'2 (ppm)-1 q* = 1.18 x 10~2 (ppm)'1
Adjustment factor for early sacrifice /104\3 c
V~~'
Final estimate of 95% upper limit
Geometric mean of 95% upper limit
= 4.96
7.1 x 10-2 (ppm)-1 5.9 x 10~2 (ppm)'1
qi = 6.5 x ID"2 (ppm)"1
75
-------
DRAFT—DO NOT QUOTE OR CITE
then q* = 6.5 x 10'2 (pptn)'1 x 1 = 2,9 x 10'5 (yg/m3)
2.25 xlO3 ug/m3
4.4.3. Comparison of Human and Animal Inhalation Studies
The purpose of this section is to evaluate whether or not the animal-to-
man extrapolated estimate of 1,3-butadiene-caused cancer is reasonably borne
out by human data. The section considers the limited data base and determines
to what extent extrapolation from the positive animal data might overestimate
the human response.
While mouse exposures of 625 ppm and 1,250 ppm of 1,3-butadiene (6 hours/
day, 5 days/week for 60-61 weeks) caused a broad spectrum of cancers, human
response associated with the SBR process was not consistent across studies.
Various cohorts displayed excess mortality from cancers of the stomach or
intestine, prostate, and/or respiratory system. The most consistent excesses
(and, therefore, the focus of this section) appear to be restricted to cancers
of the lymphatic and hematopoietic systems, cancers which include leukemias,
Hodgkin's disease, and lymphosarcomas.
It must be emphasized that exposure to 1,3-butadiene alone cannot be iso-
lated from exposure to several other potential carcinogens. Especially associ-
ated with the SBR process is concurrent exposure to styrene, a compound for
which there is limited evidence of carcinogenicity in animals and inadequate
evidence in humans (IARC, 1982). The small amount of human evidence associated
with styrene exposure and cancer suggests an association with leukemia and,
possibly, lymphomas. Styrene, like 1,3-butadiene, metabolizes to an epoxide;
both epoxides are the suspected carcinogens. (Ethylene oxide, also an epoxide,
is also associated with leukemias). In addition to styrene, the SBR process
76
-------
DRAFT—DO NOT QUOTE OR CITE
involves numerous other exposures concurrent with 1,3-butadiene. These concur-
rent exposures will not be dealt with in the following analysis, because if the
animal risk extrapolation based on 1,3-butadiene alone overestimates the human
risk, then the animal risk extrapolation will most likely be too high.
Probably the strongest evidence for human cancer associated with the SBR
process is that of Meinhardt et al. (1982), in which workers exposed to the
high-temperature batch polymerization process from 1943 through 1945 showed a
marginally significant increase in cancers of the lymphatic and hernatopoietic
tissues, with an SMR of 212 from 9 deaths out of 600 study members. For workers
first exposed after the process was changed to continuous feed in 1946, with
correspondingly less exposure, no deaths from lymphopoietic system cancers
occurred among more than 1,000 study members. Unfortunately, no exposure
estimates are available for the pre-1946 cohort. For the cohort exposed after
1946, only 1,3-butadiene measurements taken after 1975 are available. They
show an 8-hour time-weighted average mean concentration of 1.24 ppm butadiene
fc 1.20 standard deviation [SD]), 0.10 ppm benzene (t 0.035 SO), and 0.94
ppm styrene (t 1.23 SD). (Benzene is not used in SBR manufacturing, but may
be present as an impurity of styrene or toluene.) The Meinhardt et al. (1982)
study also contained an analysis from a second plant whose workers were first
exposed in 1950. Based on a cohort of 1,094, the SMR for cancers of the lym-
phatic and hematopoietic tissues was 78, slightly higher than the overall SMR
of 66, the latter being significantly (p < 0.05) less than that of the compa-
rable general population. Meinhardt et al. reported average 1,3-butadiene
exposure levels in 1977 of 13.5 ppm.
The next strongest evidence for cancer associated with the SBR process is
based on the case-control study of McMichael et al. (1976). These authors
estimated an age-standardized risk ratio of 6.2 for lymphatic and hematopoietic
77
-------
DRAFT—00 NOT QUOTE OR CITE
cancers among workers with at least 5 years of exposure in the synthetic plant,
relative to all other workers as controls. (This ratio decreased to 2.4 when
a matched control analysis was used.) The synthetic plant is where the SBR
process is located. McMichael et al. also found a dose-related risk ratio in
the synthetic plant by number of years worked there.
Estimates of exposures in the McMichael et al. study are based upon a
later paper. Checkoway and Williams (1982) measured 1,3-butadiene, styrene,
benzene, and toluene levels at the same synthetic plant in which McMichael et
al. (1976) found that "leukemia and lymphoma (cases) among hourly paid rubber
workers from one company were 6 times more likely than controls to have worked
at jobs in the SBR plant." Exposure levels of 1,3-butadiene typically averaged
below 1 ppm, but exposure levels in the tank farm area averaged 20 ppm.
The most extensive investigation specifically designed to study the health
effects of the SBR process shows very little association of 1,3-butadiene with
lymphatic and hematopoietic tissue cancer (ICD 200-207). The Matanoski et al.
study of one Canadian and seven U.S. synthetic rubber plants showed, possibly,
a trend with more exposure as defined by production, maintenance, utilities,
or other jobs, but none of the SMRs (Table 12) are statistically significant.
Only Hodgkin's disease (ICD 201), shows a consistently high SMR in all three
cohorts, but the numbers of cases were small. No exposure estimates were
presented in the Matanoski et al. report, but the plants studied were of the
same type as those studied by Meinhardt et al. Some workers in seven of the
eight plants might have started as early as 1943, and Matanoski states (per-
sonal communication) that the open batch process in several of these plants was
continued into the early 1970s. Based on these observations, the estimates of
20 ppm for production workers, 10 ppm for maintenance workers, and 5 ppm for
utility workers have been used for calculation purposes.
78
-------
TABLE 12. SUMMARY OF CANCER OF LYMPHATIC AND HEMATOPOIETIC TISSUES ASSOCIATED WITH THE
STYRENE-BUTADIENE SYNTHETIC RUBBER PROCESS. CONFIDENCE LIMITS ON ACTUAL SMRs
PREDICTED EXCESS DEATHS BASED ON INITIAL MLE INCREMENTAL UNIT RISKS OF THE MOUSE DATA
Study
Meinhardt
et al .
(1982)
Plant A
1943-1945
Plant A
1946-1976
Plant B
1950-1976
Matanoski
et al.
(1982)
(by job
last held)
Production
Maintenance
Utilities
McMichael
et al.
TT976)
Sample Cancer
size type Observed
Lymphatic and
hematopoietic
tissues ICD
600 (200-205) 9
1,062 Same as above 0
1,094 Same as above 2
All lympho-
poietic ICD
3,269 (200-207) 11
3,683 Same as above 13
550 Same as above 4
All lym-
phatic and
hematopoietic
(200-209)
?Test for observed cancers not significantly
a = 0.05, power determined by formula Z-, a
cMore deaths
*p < 0.05.
**p < 0.01.
predicted than observed.
Excess deaths
from exposure
predicted from
95% mouse data
Confidence (observed-
Expected SMR intervals expected)8
4.25 2.12 0.96-4.02 1.7 (4.75)*c
1.54 0 0-2.34 0.2 (-1.54)
2.55 0.78 0.09-2.83 2.3 (-0.55)
10.6 1.04 0.52-1.86 18.0 (0.4)**
16.2 0.80 0.47-1.37 9.3 (-3.2)**
6.5 0.62 0.17-1.58 0.5 (-2.5)
6.2 No
(risk further
ratio in estimates
synthetic possible
plants
2.4
in matched
controls
different from predicted (Poisson test or normal
= Z - 2 (SMR0-5 - 1)E°-5 (Beaumont and Breslow,
Power
to detect
predicted
deaths
(a-. k
0.05)D Comments
Estimates
from 1943-
1945: 20 ppm
from 1946-
12% 1952: 10 ppm
Average
3% exposure =
1.24
23% Average
exposure =
13.5
Same types of
plants as those
studied by
Meinhardt et al .
99% Assume: 20 ppm
(estimate)
53% 10 ppm (estimate)
4% 5 ppm (estimate)
Increases asso-
ciated with com-
pounding, mixing,
cement mixing,
inspection-
finishing-repai r,
synthetic plant.
Average exposure
< 1 ppm except
for tank farm
area (20 ppm)
approximation to the Poisson).
1981).
-------
DRAFT—DO NOT QUOTE OR CITE
Is the unit risk estimate based on the mouse tumor data a reasonable
extrapolation? To answer this question we must be able to estimate the predic-
ted effect based on actual (or estimated) exposure. (We must also assume the
effect on these cancers from other exposures to be nil.) For illustration, we
choose the Meinhardt et al. (1982) study plant B, with estimated exposure of
13.5 ppm. Although this exposure was measured in 1977, we will consider it to
be representative of exposures from 1950 through 1976. We must also know that
the 1,094 study members converted to 19,742 person-years at risk for an average
of 18 years per person. Since average employment or exposure is given as 10.78
years, we can estimate the expected contribution as follows:
1. The continuous lifetime equivalent exposure based on 10 working
years out of about 50 possible remaining years is:
13.5 ppm x 10.78 years x 240 day x 8_ hours = 0.64 ppm
50 365 24
2. The inital best estimate (MLE) of risk based on the mouse data is Q]_ =
8.1 x 10~3 (ppm)"1 for the males, qx = 1.0 x 10"2 (ppm)"1 for the fe-
males, and a geometric mean of q-^ = 9.0 x 10"3 (ppm)"1, based on human
equivalent continuous exposure. This geometric mean MLE incremental
risk estimate will be used to predict human excess deaths for the
present purpose of deciding whether the number of deaths among indus-
trial workers is consistent with the expected deaths derived using the
animal extrapolation procedure. The initial estimate (Table 11) is
used since it projects the limited animal observation period (60-61
weeks out of a 2-year lifetime) to the the limited human observation
period (average of up to 29 years out of 50 years remaining lifetime).
80
-------
DRAFT--DO NOT QUOTE OR CITE
3. Based on the unit risk factor q^ = 9.0 x 10~3 (ppm) , each of the
1,094 workers would be expected to have an additional lifetime risk of
R = 9.0 x lO-3 (ppm)-1 x 0.64 = 5.8 x 10-3
4. Based on 1,094 workers at risk for 18 of their 50 remaining years,
this converts to the following expected excess number of cancers:
5.8 x ID'3 x i,094 x 18 = 2.3
50
5. In addition to the 2.55 cancer deaths expected based on no exposure
(Table 12), we could expect, with the exposure, to observe 4.8
deaths from cancers of the lymphatic and hematopoietic tissues. The
probability of observing two or fewer deaths (Table 12) with 4.8
expected is
2 xx
P (deaths <_ 2|x = 4.8) = I e~ x = 0.14
x=o xT
The statistical power to detect a predicted SMR of (4.8/2.55) or 1.9 is
given by Beaumont and Breslow (1981) as Z1-g = Za - 2 (SMR0-5 - 1)E°-5. For
plant B this is
Zj_g = 1.96 - 2 (1.9°-5 - 1) (2.55)0-5 = 0.75
which corresponds to a power of 0.23, or 23% at a = 0.05.
81
-------
DRAFT--DO NOT QUOTE OR CITE
Results based on similar calculations are presented in Table 12 for the
other two Meinhardt cohorts and for three Matanoski cohorts. They show incon-
sistent results. For the 1943-1945 plant A Meinhardt cohort, the deaths pre-
icted from animal extrapolation actually underpredict the observed human
response (p < 0.5). For the other two Meinhardt cohorts and the Matanoski
utilities cohort, the predicted and observed results are not significantly dif-
ferent, although the power to detect the predicted difference in these three
cases is low--23% or less. For the two larger Matanoski cohorts, the extrapo-
lated deaths do significantly (p < 0.01 in both cases) overpredict response.
Furthermore, the power to detect the predicted response is 99% in the mainte-
nance cohort.
The interpretation of these results, if we dismiss momentarily the large
uncertainties in the exposure estimates, is that no predicted results will
satisfy the observed results in all six cohorts. Were we to lower the risk
estimates to try to better accommodate the Matanoski maintenance and production
cohorts, we would further underestimate the observed results for the Meinhardt
and early plant A cohort. Based on the information we have, no single extrapo-
lated value can predict the human response. It seems probable that the potency
value we have extrapolated will overpredict human response by a factor of about
3. Considering the uncertainties in the human exposure data, however, the
animal extrapolation is the best that can be achieved at present.
Finally, the same analysis as computed for lymphatic and hernatopoietic
cancers in Table 12, can be done for all cancers on the theory that 1,3-buta-
diene might be a broad-spectrum carcinogen in humans as it is in mice. This
analysis is presented in Table 13. Since we have used the same extrapolation
from the mouse data, the same number of excess deaths as predicted in Table 12
will result, but these excess deaths are spread out over all cancers. It should
82
-------
TABLE 13. SUMMARY OF ALL CANCERS (ICD 140-205) ASSOCIATED WITH THE
STYRENE-BUTADIENE SYNTHETIC RUBBER PROCESS. CONFIDENCE LIMITS ON ACTUAL SMRs
PREDICTED EXCESS DEATHS BASED ON INITIAL MLE INCREMENTAL UNIT RISKS OF THE MOUSE DATA
oo
CO
All cancers
Sample (ICD 140-205)
Study size Observed Expected
Meinhardt
et al .
(1982)
Plant A
1943-1945 600 39 45.14
Plant A
1946-1976 1,062 10 12.19
Plant B
1950-1976 1,094 11 20.78
Matanoski
et al.
(1982)
(by job
last held)
Production 3,269 94 105.6
Maintenance 3,683 168 176.8
Utilities 550 25 26.3
Excess deaths
from exposure
predicted from
95% mouse data
confidence (observed-
SMR intervals expected)3
0.86 0.63-1.18 1.7 (-6.14)
0.82 0.45-1.51 0.2 (-2.19)
0.58 30-0.95* 2.3 (-9.78)**
0.89 0.73-1.09 18.0 (-9.6)**
0.95 0.82-1.11 9.3 (-8.8)
0.95 0.58-2.07 0.5 (-1.3)
Power
to detect
predicted
deaths
u=
0.05)b
5%
3%
7%
39%
10%
3%
Comments
Estimates
from 1943-
1945: 20 ppm
from 1946-
1952: 10 ppm
Average
exposure =
1.24
Average
exposure =
13.5
Same types of
plants as those
studied by
Meinhardt et al .
Assume: 20 ppm
(estimate)
10 ppm (estimate)
5 ppm (estimate)
?Test for observed cancers not significantly different from predicted (Poisson test or normal approximation to the Poisson),
Da = 0.05, power determined by formula ll_& =1-2 (SMR0-5 - 1)E°'5 (Beaumont and Breslow, 1981).
cMore deaths predicted than observed.
*p < 0.05.
**p < 0.01.
-------
URAFT--DO NOT QUOTE OR CITE
be noted that only one of the SMRs is statistically significant and that all,
in fact, are less than one. Based on the predicted excess deaths extrapolated
from animal data and estimated exposures, two of the cohorts experienced signi-
ficantly fewer deaths than predicted. For one of these cohorts, the Meinhardt
plant 8, the deficit in observed versus expected deaths was significant even
if there were no predicted deaths. For the other, the Matanoski production
workers, a reduction of predicted deaths from 18 to 6 would eliminate statis-
tical significance at the p = 0.05 level (one-sided). This corresponds to a
reduction of the extrapolated unit risk estimate by a factor of 3. This is the
only cohort with a power greater than 10%--it has a power of 39%--of detecting
the predicted results.
Comparing Tables 12 and 13, we see fairly similar results: weak, if any,
evidence of a human carcinogenic risk from 1,3-butadiene, but also no strong
evidence that the unit risk extrapolation from animal to human results is
unreasonable, or that it seriously overpredicts a potential risk.
4.4.4. Relative Potency
One of the uses of quantitative estimation is to compare the relative
potencies of different carcinogens. To estimate relative potency, the unit
risk slope factor is multiplied by the molecular weight, and the resulting
number is expressed in terms of (mmol/kg/day)-1. This is called the relative
potency index.
Figure 2 is a histogram representing the frequency distribution of potency
indices of 54 suspect carcinogens evaluated by the CAG. The actual data summa-
rized by the histogram are presented in Table 14. Where positive human data are
available for a compound, they have been used to calculate the index. Where no
human data are available, animal oral studies and animal inhalation studies have
been used, in that order. In the present case, only the animal inhalation studies
84
-------
FREQUENCY
00
in
c
-s
(D
CTT3 DC
^ O -••
c-t- CO rt
:3- 3 O
fD o to.
•< -s
o 01
Cu —*• 3
o cx -s
_i. -J. n>
3 O ~O
O fD ~S
CQ l/> fD
fD to
3 O fD
on en -".
CO j^ 3
n> «£j
l/> I/)
fD ~O n>
r+ O -h
r+ -S
£T3 fD
"S O -Cl
O O) £=
C T fD
TD O 3
. -J. O
3 '<
O
to O-
fD -••
3 to
to c+
-s
fD ->•
< cr
—' r+
c: -'•
Qj O
n- 3
fD
Q- O
c-t
^~
n>
O
D
to
O
-i
g -^
O
_ 01
2
O
m
X O)
00
MJ
t I I I i l i i i l I i I i i i i i I I
X
~A
o
ro
to
x
O
c
m
O
c
>
m
O
c
m
O
c
m
-------
TABLE 14. RELATIVE CARCINOGENIC POTENCIES AMONG 54 CHEMICALS EVALUATED BY THE CARCINOGEN ASSESSMENT GROUP
AS SUSPECT HUMAN CARCINOGENS
Level
of evidence3
Compounds
Acrylonitrile
Aflatoxin B^
Aldri n
Allyl chloride
gj Arseni c
B[a]P
Benzene
Benzidene
Beryl 1 ium
1,3-Butadiene
Cadmi um
Carbon tetrachloride
Chi ordane
CAS Number
107-13-1
1162-65-8
309-00-2
107-05-1
7440-38-2
50-32-8
71-43-2
92-87-5
7440-41-7
106-99-0
7440-43-9
56-23-5
57-74-9
Humans
L
L
I
S
I
S
S
L
I
L
I
I
Animals
S
S
L
I
S
S
S
S
S
S
S
L
Grouping
based on
IARC
criteria
2A
2A
2B
1
28
1
1
2A
2B
2A
2B
3
SI ope
(mg/kg/day)-l
0.24(W)
2900
11.4
1.19x10-2
15(H)
11.5
2.9xlO-2(W)
234(W)
2.6
1.0x10-1(1)
7.8(W)
1.30x10-1
1.61
Molecul ar
weight
53.1
312.3
369.4
76.5
149.8
252.3
78
184.2
9
54.1
112.4
153.8
409.8
Potency
i ndex
1x10+1
9xlO+5
4x10+3
9x10-1
2x10+3
3x10+3
2x100
4x10+4
2x10+1
5x100
9x10+2
2x10+1
7x10+2
Order of
magnitude
(Iog10
i ndex)
+1
+6
+4
0
+3
+3
0
+5
+1
+1
+3
+1
+3
= Sufficient evidence; L = Limited evidence; I = Inadequate evidence.
(continued on the following page)
-------
TABLE 14. (continued)
CO
Level
of evidence3
Compounds CAS Number Humans
Chlorinated ethanes
1 ,2-dichl oroethane
hexachl oroethane
1 ,1,2,2-tetrachl oroethane
1 ,1 ,2-trichloroethane
Chloroform
107-06-2
67-72-1
79-34-5
79-00-5
67-66-3
Chromium 7440-47-3
DOT
Dichlorobenzidine
1 ,1-Di chl oroethyl ene
(Vinylidene chloride)
Di chl oromethane
(Methylene chloride)
Diel dri n
2,4-Dinitrotoluene
Diphenylhydrazi ne
Epichlorohydri n
Bis(2-chloroethyl)ether
50-29-3
91-94-1
75-35-4
75-09-2
60-57-1
121-14-2
122-66-7
106-89-8
111-44-4
I
I
I
I
I
S
I
I
I
I
I
I
I
I
I
Animal s
S
L
L
L
S
S
S
S
L
L
S
S
S
s
s
Groupi ng
based on
IARC
criteria
2B
3
3
3
2B
1
2B
28
3
3
2B
2B
2B
2B
28
Slope
(mg/kg/day)-l
6.9x10-2
1.42x10-2
0.20
5.73x10-2
7x10-2
41(W)
0.34
1.69
1.47x10-1(1)
6.3x10-4(1)
30.4
0.31
0.77
9.9x10-3
1.14
Molecular
weight
98.9
236.7
167.9
133.4
119.4
100
354.5
253.1
97
84.9
380.9
182
180
92.5
143
Potency
i ndex
7x100
3x100
3x10+1
8x100
8x100
4x10+3
1x10+2
4x10+2
1x10+1
5x10-2
1x10+4
6x10+1
1x10+2
9x10-1
2x10+2
Order of
magnitude
(Iog10
i ndex)
+1
0
+ 1
+ 1
+ 1
+4
+2
+3
+1
-1
+4
+2
+2
0
+2
= Sufficient evidence; L = Limited evidence; I = Inadequate evidence.
(continued on the following page)
-------
TABLE 14. (continued)
Level
of evidence3
Compounds
Bi s(chl oromethyl )ether
Ethylene dibromide (EDB)
Ethylene oxide
Heptachlor
Hexachl orobenzene
CO
00 Hexachlorobutadiene
Hexachl orocycl ohexane
technical grade
alpha isomer
beta isomer
gamma isomer
Hexachlorodibenzodioxin
Nickel
Nitrosami nes
Dimethyl nitrosami ne
Di ethyl nitrosami ne
Di butyl nitrosami ne
CAS Number
542-88-1
106-93-4
75-21-8
76-44-8
118-74-1
87-68-3
319-84-6
319-85-7
58-89-9
34465-46-8
7440-02-0
62-75-9
55-18-5
924-16-3
Humans
S
I
L
I
I
I
I
I
I
I
L
I
I
I
Animals
S
S
S
S
S
L
S
L
L
S
S
S
S
S
Grouping
based on
IARC
criteria
1
2B
2A
28
2B
3
2B
3
28
2B
2A
2B
2B
2B
Slope
(mg/kg/day)-l
9300(1)
41
3.5x10-1(1)
3.37
1.67
7.75x10-2
4.75
11.12
1.84
1.33
6.2x10+3
1.15(W)
25.9(not by ql
43.5(not by q^
5.43 J
Molecular
weight
115
187.9
44.1
373.3
284.4
261
290.9
290.9
290.9
290.9
391
58.7
p) 74.1
F) 102.1
158.2
Potency
index
1x10+6
8x10+3
2x10+1
1x10+3
5x10+2
2x10+1
1x10+3
3x10+3
5x10+2
4x10+2
2x10+6
7x10+1
2x10+3
4x10+3
9x10+2
Order of
magnitude
(Iog10
i ndex)
+6
+4
+1
+3
+3
+ 1
+3
+ 3
+ 3
+3
+6
+2
+3
+4
+3
as = Sufficient evidence; L = Limited evidence; I = Inadequate evidence.
(continued on the following page)
-------
TABLE 14. (continued)
Compounds (
N-ni trosopyrrol i di ne
N-nitroso-N-ethylurea
N-nitroso-N-methyl urea
N-nitroso-diphenylamine
PCBs
Phenols
2,4,6-Tri chl orophenol
Tetrachl orodibenzo-
co p-dioxin (TCDD)
Tetrachl oroethylene
Toxaphene
Tri chl oroethyl ene
Vinyl chloride
Level
of evidence3
:AS Number
930-55-2
759-73-9
684-93-5
86-30-6
1336-36-3
88-06-2
1746-01-6
127-18-4
8001-35-2
79-01-6
75-01-4
Humans
I
I
I
I
I
I
I
I
I
I
S
Animals
S
S
S
S
S
S
S
L
S
L/S
S
Grouping
based on
I ARC
criteria
2B
2B
2B
2B
2B
2B
2B
3
2B
3/2B
1
Slope
(mg/kg/day)-1
2.13
32.9
302.6
4.92x10-3
4.34
1.99x10-2
1.56x10+5
6.0x10-2
1.13
1.2x10-2
1.75x10-2(1)
Molecular
wei ght
100.2
117.1
103.1
198
324
197.4
322
165.8
414
131.4
62.5
Potency
i ndex
2x10+2
4x10+3
3x10+4
1x100
1x10+3
4xlQO
5x10+7
IxlOl
5x10+2
2x100
1x100
Order of
magnitude
(Iog10
index)
+2
+4
+4
0
+3
+1
+8
+1
+3
0
0
aS = Sufficient evidence; L = Limited evidence; I = Inadequate evidence.
Remarks:
1. Animal slopes are 95% upper-limit slopes based on the linearized multistage model. They are calculated based on
animal oral studies, except for those indicated by I (animal inhalation), W (human occupational exposure), and H
(human drinking water exposure). Human slopes are point estimates based on the linear nonthreshold model.
2. The potency index is a rounded-off slope in (mmol/kg/day)-l and is calculated by multiplying the slopes in
(mg/kg/day)-l by the molecular weight of the compound.
3. Not all of the carcinogenic potencies presented in this table represent the same degree of certainty. All are
subject to change as new evidence becomes available.
-------
DRAFT—DO NOT QUOTE OR CITE
provide sufficient evidence of carcinogenicity and have sufficient exposure
i nformation.
The potency index for 1,3-butadiene based on the NTP mouse inhalation study
(NTP, 1984) is 5.4 (mmol /kg/day)'1. This is derived as follows: the upper-limit
incremental unit risk estimate from the inhalation study is 2.9 x 10~5 (yg/m3)-1.
Transforming this to mg/kg/day, the conversion factor is
1 yg/m3 = 1 yg/m3 x 20 m3/day x 10'3 mg/yg x 1/70 kg = 2.86 x 10'4 mg/kg/day
Then
= 2.9 x 1Q-5 (yg/m3)'1 x 2.86 x 10'4 mg/kg/day = 1.0 x 10'1 (mg/kg/day)'1
Multiplying by the molecular weight of 54.1 gives a potency index of 5.4 x 10°.
Rounding off to the nearest order of magnitude gives a value of 101, which is
the scale presented on the horizontal axis of Figure 2. The index of 5.4 lies
at the top of the fourth quartile of the 54 chemicals that the CAG has evaluated
as suspect carcinogens. Thus, in terms of potency alone, 1,3-butadiene would
place among the weakest of these carcinogens. However, the fact that 1,3-buta-
diene causes so many fatal tumors in animals and sharply decreases the latency
period increases concern beyond that based simply on relative potency.
4.4.5. Summary of Quantitative Estimation
Based on the linearized multistage model, a 95% upper-limit incremental
unit risk was calculated for 1,3-butadiene using the geometric mean of the
95% upper-limit incremental risk estimates from the pooled male and pooled
female significant tumor responses of the NTP mouse study. The mean value of
q^ = 6.5 x 10"2 (ppm)"1 was then used to predict human responses in several
90
-------
DRAFT—DO NOT QUOTE OR CITE
epidemiologic studies, and the predicted and actual responses were then com-
pared. The comparisons were hampered by a scarcity of information on the epi-
demiology concerning actual exposures, age distributions, and work histories.
In addition, because there was no consistent cancer response across all of
the studies, the most predominant response, cancer of the lymphatic and hemato-
poietic tissues, was chosen as being the target for 1,3-butadiene. Based on
the comparisons between the predicted and observed human response, the extrapo-
lated value from the animal data appeared high by a factor of about 3, but in
view of the uncertainty and apparent inconsistency of the epidemiologic data,
no better estimate can be made at this time.
In addition to a 95% upper-limit incremental unit risk, a measure of car-
cinogenic potency was determined for 1,3-butadiene. Among the 54 chemicals
that the CAG has evaluated as suspected carcinogens, 1,3-butadiene ranks fairly
low in potency, placing at the top of the fourth quartile. Based on the wide
spectrum of cancers and sharply decreased latency associated with these cancers,
however, 1,3-butadiene should evoke much more concern than the potency numbers
alone indicate.
91
-------
DRAFT—DO NOT QUOTE OR CITE
REFERENCES
Andjelkovich, D., J. Taulbee, and M. Symons. 1976. Mortality experience of a
cohort of rubber workers, 1964-1973. J. Occup. Med. 18:387-394.
Andjelkovich, D., J. Taulbee, M. Symons, and T. Williams. 1977. Mortality of
rubber workers with reference to work experience. J. Occup. Med. 18:387-
394.
Beaumont, J.I. and N.E. Breslow. 1981. Power consideration in epidemiologic
studies of vinyl chloride workers. American Journal of Epidemiology
114:725-734.
Bolt, H.M., G. Schmiedel, J.G. Filser, H.P. Rolzhauser, K. Lieser, D. Wistuba,
and V. Schurig. 1983. Biological activation of 1,3-butadiene to vinyl
oxirane by rat liver microsomes and expiration of the reactive metabolites
by exposed rats. J. Cancer Res. Clin. Oncol. 106:112-116.
Carpenter, C., C. Shaffer, C. Weil, and H. Smith. 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.
Checkoway, H., and T.M. Williams. 1982. A hematology survey of workers at a
styrene-butadiene synthetic rubber manufacturing plant. Am. Ind. Hyg.
Assoc. J. 43:164-169.
Checkoway, H., T. Wilcosky, P. Wolf, and H. Tyroler. 1984. An evaluation of
the association of leukemia and rubber industry solvent exposures. Am.
J. Ind. Med. 5:239-249.
Conner, M., J. Luo, and 0. Gutierrez de Gotera. 1983. Induction and rapid
repair of sister-chromatid exchanges in multiple murine tissues in vitro
by diepoxybutane. Mutat. Res. 108:251-263.
Cox, C.R. 1972. Regression model and life tables. J. Roy. Stat. Soc. B
34:187-220.
Crouch, C., D. Pullinger, and I. Gaunt. 1979. Inhalation toxicity studies
with 1,3-butadiene, Part 2: Three-month toxicity study in rats.
Am. Ind. Hyg. Assoc. J. 40:796-802.
Crump, K.S. 1979. Dose-response problems in carcinogenesis. Biometrics
35:157-167.
Crump, K.S., and Watson, W.W. 1979. GLOBAL79. A FORTRAN program to extra-
polate dichotomous animal carcinogenicity data to low dose. Natl. Inst.
of Environ. Health Science Contract No. l-ES-2123.
Crump, K.S., Guess, H.A., and Deal, L.L. 1977. Confidence intervals and test
of hypotheses concerning dose-response relations inferred from animal
carcinogenicity data. Biometrics 33:437-451.
92
-------
DRAFT—DO NOT QUOTE OR CITE
Crump, K.S., and R.B. Howe. 1983a. The multistage model with a time-dependent
dose pattern: applications to carcinogenic risk assessment. Unpublished
report. EPA Contract No. 68-01-3111.
Crump, K.S., and R.B. Howe. 1983b. ADOLL1-83 and ADOLL2-83, FORTRAN programs
for implementing a multistage model with a time-dependent dose pattern.
Unpublished instructions for use of computer programs.
Dean, B.J., and G. Hodson-Walker. 1979. An in vitro chromosome assay using
cultured rat-liver cells. Mutat. Res. 63T329-337.
de Meester, C., F. Poncelet, F. Roberfroid, and M. Mercier. 1978. Mutage-
nicity of butadiene and butadiene monoxide. Biochem. Biophys. Res.
Commun. 80:298-305.
de Meester, C., F. Poncelet, F. Roberfroid, and M. Mercier. 1980. The
mutagenicity of butadiene towards Salmonella typhimurium. Toxicol.
Lett. 6:125-130.
Doll, R. 1971. Weibull distribution of cancer. Implications for models of
carcinogenesis. J. Roy. Statistical Soc. A 13:133-136.
Dripps, R.D., Eckenhoff, J.E., and Vandam, L.D. 1977. Introduction to
anesthesia, the principles of safe practice. 5th Ed. W.B. Saunders
Company, Philadelphia, pp. 121-123.
Hazleton Laboratories Europe, Ltd. 1981a. 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).
Hazleton Laboratories Europe, Ltd. 1981b. 1,3-Butadiene: Inhalation terato-
genicity study No. 2788 522/3. Prepared for the International Institute
of Synthetic Rubber Producers, New York, NY. (Unpublished).
Hemminki, K., K. Falck, and H. Vainio. 1980. Comparison of alkylation rates
and mutagenicity of directly acting industrial and laboratory chemicals.
Arch. Toxicol . 46:277-285.
Huff, J.E., R.L. Melnick, H.A. Solleveld, J.K. Haseman, M. Powers, and R.A.
Miller. 1985. Multiple organ carcinogenicity of 1,3-butadiene in B6C3F}
mice after 60 weeks of inhalation exposure. Science 227:548-549.
International Agency for Research on Cancer (IARC). 1982. IARC monographs
on the evaluation of the carcinogenic risk of chemicals to humans.
Supplement 4. Lyon, France.
Jersey, G., M. Balmer, J. Quast, C. Park, D. Schuetz, J. Beyer, K. Olsen, S.
McCollister, and L. Rampy. 1978. Two-year chronic inhalation toxicity
and carcinogenicity study of monomeric styrene in rats. Final report.
Dow Chemical Co., Midland, MI.
93
-------
DRAFT--DO NOT QUOTE OR CITE
Lawley, P.O., and P. Brookes. 1967. Interstrand cross-linking of DNA by
difunctional al'/lating agents. J. Mol. Biol. 25:143-160.
Lilis, R., and W.J. Nicholson. 1976. Cancer experience among workers in a
chemical plant producing styrene monomers. Proceedings of Styrene-
Butadiene Briefing, Covington, KY, April 30, 1976. DHEW (NIOSH) Pub.
No. 77-129. Cincinnati, OH, pp. 22-27.
Linet, Martha S. 1985. The leukemias: epidemiologic aspects. Oxford
University Press, Oxford, England.
Luker, M.A., and 8.J. Kilbey. 1982. A simplified method for the simulta-
neous detection of intragenic and intergenic mutations (deletions) in
Neurospora crassa. Mutat. Res. 92:63-68
Malvoisin, E., G. Lhoest, F. Poncelet, M. Roberfroid, and M. Mercier. 1979.
Identification and quantitation of l,2-epoxybutene-3 as the primary
metabolite of 1,3-butadiene. J. Chrom. 178:419-429.
Malvoisin, E., and M. Roberfroid. 1982. Hepatic microsomal metabolism
of 1,3-butadiene. Xenobiotica 12:137-144.
Mantel, N., and Schneiderman, M.A. 1975. Estimating "safe" levels, a
hazardous undertaking. Cancer Res. 35:1379-1386.
Matanoski, G.M., L. Schwartz, J. Sperrazza, and J. Tonascia. 1982. Mortality
of workers in the styrene-butadiene rubber polymer manufacturing industry.
Johns Hopkins University School of Hygiene and Public Health, Baltimore,
MD. (Unpublished).
McMichael, A.J., R. Spirtas, J.F. Gamble, and P.M. Tousey. 1976. Mortality
among rubber workers: Relationship to specific jobs. J. Occup. Med.
18:178-185.
McMichael, A.J., R. Spirtas, and L.L. Kupper. 1974. An epidemiologic study
of mortality within a cohort of rubber workers, 1964-72. J. Occup. Med.
16:458-464.
McMichael, A.J., R. Spirtas, L.L. Kupper, and J.F. Gamble. 1975. Solvent
exposure and leukemia among rubber workers: an epidemiologic study.
J. Occup. Med. 17:234-239.
Meinhardt, T.J., R.A. Lemen, M.S. Crandall, and R.J. Young. 1982. Environmental
epidemiologic investigation of the styrene-butadiene rubber industry. Scand,
J. Work Environ. Health 8:250-259.
Monson, R.R. 1979. U.S. death rates programs, versions USDR 27 and USDR 38.
Harvard University School of Public Health, Boston, MA. (Unpublished).
National Toxicology Program (NTP). 1984. Toxicology and carcinogenesis
studies of 1,3-butadiene (CAS 106-99-0) in B6C3F1 mice (inhalation
studies).
94
-------
DRAFT--DO NOT QUOTE OR CITE
Olszewska, E., and B.J. Kilbey. 1975. The mutagenic activity of diepoxybu-
tane in yeast. Mutat. Res. 33:383-390.
Ott, M.G., R.C. Kolesar, H.C. Scharnweber, E.J. Schneider, and J.R. Venable.
1980. A mortality survey of employees engaged in the development or
manufacture of styrene-based products. J. Occup. Med. 22(7):445-460.
Perry, P., and H.J. Evans. 1975. Cytological detection of mutagen-carcinogen
exposure by sister chromatid exchange. Nature 258:121-125.
Poncelet, F., C. de Meester, M. Duverger-van Bogaert, M. Lambotte-Vandepaer,
M. Roberfroid, and M. Mercier. 1980. Influence of experimental factors
on the mutagenicity of vinylic monomers. Arch. Toxicol. Suppl . 4:63-66.
Sankaranarayanan, K. 1983. The effects of butylated hydroxytoluene on radi-
ation and chemically-induced genetic damage in Drosophila melanogaster.
Mutat. Res. 108:203-223.
Sankaranarayanan, K., W. Ferro, and J.A. Ziljlstra. 1983. Studies on mutagen-
sensitive strains of Drosophila melanogaster. III. A comparison of the
mutagenic sensitivities of the ebony (UV and X-ray sensitive) and Canton-S
(wild-type) strains to MMS, ENU, DEB, DEN, and 2,3,6-Cl3-PDMT. Mutat. ~
Res. 110:59-70.
Shugaev, B. 1969. Concentrations of hydrocarbons in tissues as a measure of
toxicity. Arch. Environ. Health 18:878-882.
Spirtas, R. 1976. Mortality among rubber workers: relationship to jobs with
styrene-butadiene exposures. Proceedings of NIOSH Styrene-Butadiene
Briefing, Covington, KY, April 30, 1976. DREW (NIOSH) Pub. No. 77-129.
Cincinnati, OH, pp. 9-21.
Taulbee, J., 0. Adjelkovich, T. Williams, J.F. Gamble, and P. Wolf. 1976.
A study of possible associations between exposure to SBR processes and
mortality from leukemia and related diseases based on toxicologic,
industrial hygiene and epidemiologic considerations (for workers in
the 1951 and 1964 cohorts and deaths 1964-1973). Proceedings of NIOSH
Styrene-Butadiene Briefing, Covington, KY, April 30, 1976. DHEW (NIOSH)
Pub. No. 77-129. Cincinnati, OH, pp. 113-129.
Vogel, E. 1976. Mutagenicity of carcinogens in Drosophila as a function of
genotype-controlled metabolism. In: F.J. de Serres, J.R. Fouts, J.R.
Bend, and R.M. Philpot (eds.). In vitro metabolic activation in muta-
genesis testing, pp. 63-79. Amsterdam Elsevier/North-Holland Biomedical
Press.
Voogd, C.E. J.J. van de Stel, and J.J.J.A.A. Jacobs. 1981. The mutagenic
action of aliphatic epoxides. Mutat. Res. 89:269-282.
Wade, M.J., J.W. Moyer, and C.H. Hine. 1979. Mutagenic action of a series of
epoxides. Mutat. Res. 66:367-371.
95
-------
DRAFT--DO NOT QUOTE OR CITE
Zaborowska, D., Z. Swietlinska, and J. Zuk. 1983. Induction of mitotic
recombination by UV and diepoxybutane and its enhancement by hydroxyurea
in Saccharomyces cerevisiae. Mutat. Res. 120:21-26.
Zimmering, S. 1983. The mei-9a test for chromosome loss in Drosophila: A
review of assays of 21 chemicals for chromosome breakage. Environ.
Mutagen. 5:907-921.
U.S. GOVERNMENT PRINTING OFFICE-5bl
96
------- |