United States Science Advisory Board EPA-SAB-EHC-99-003
Environmental Washington, DC November 1998
Protection Agency www.epa.gov/sab
&EPA AN SAB REPORT: REVIEW
OF THE HEALTH RISK
ASSESSMENT OF
1,3-BUTADIENE
REVIEW OF THE OFFICE OF RESEARC
AND DEVELOPMENT'S DRAFT HEALTH
RISK ASSESSMENT OF 1,3-BUTADIENE
PREPARED BY THE ENVIRONMENTAL
HEALTH COMMITTEE (EHC) OF THE
SCIENCE ADVISORY BOARD (SAB)
November 19, 1998
-------
EPA-SAB-EHC-99-003
Honorable Carol M. Browner
Administrator
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, DC 20460
Subject: Review of the Office of Research and Development's Draft Health
Risk Assessment of 1,3-Butadiene (EPA/600P-98/001A).
Dear Ms. Browner:
At the request of the Office of Research and Development (ORD), the Environmental
Health Committee (EHC) of the Environmental Protection Agency's Science Advisory Board
(SAB) reviewed the Agency's Draft Health Risk Assessment of 1,3-Butadiene. The Committee
met on April 30 and May 1, 1998 in Washington, DC.
The document reviewed by the Committee was developed by the EPA's Office of
Research and Development (ORD). The ORD published its first risk assessment of 1,3-Butadiene
in 1985. The first document covered cancer and mutagenicity and was prepared in response to a
request from the Office of Air Quality Planning and Standards to support the classification of
1,3-Butadiene as a Hazardous Air Pollutant. The recent draft 1,3-Butadiene document was
written in response to a request from the Agency's Office of Mobile Sources which, plans to use
the final document to support a future Air Toxics Rule. The current draft 1,3- review document
focuses on mutagenicity, carcinogenicity, and reproductive/developmental effects. The
1,3-Butadiene document reviewed by the Committee presents the Agency's first use of a
benchmark dose analysis for reproductive/developmental factors, and incorporates many new
studies published since the 1985 effort. The EPA concludes that this new information has
changed the weight of evidence for its findings re cancer. In addition, there are exposure data
available in an occupational study which are used to derive the cancer slope factor. The review
document is not a comprehensive health assessment however, and contains only an overview of
the ambient exposure and exposure of populations adjacent to emissions sources, without
including any actual exposure assessment as such.
In addition to a general review of the document by the Committee, the Office of Research
and Development specifically requested that the EHC provide comment on each of the following
Charge issues:
a) Review the health risk assessment for technical quality, comprehensiveness and
clarity.
b) Does the science support the classification of "known" human carcinogen?
c) Are the approaches taken to characterize plausible cancer risks reasonable given
-------
the science?
d) Are the conclusions and quantitative estimations for reproductive/developmental
effects adequately supported?
The EHC recognizes that preparing a health risk assessment (more correctly, a hazard
assessment - see below) of 1,3-Butadiene was a difficult and complicated task, given the large
amount of information on 1,3-Butadiene toxicity, epidemiology, and mechanisms available, as
well as the new and information being developed on a continual basis.
The Agency selected a cutoff date of January 31, 1997 for the inclusion of new
information in the revised draft document. However, a significant amount of new and important
information has been developed since then and is pertinent to a health risk assessment of
1,3-Butadiene. The Committee found that the report should reflect the most current research data
possible, including the recent evaluations by the International Agency for Research on Cancer
(IARC, 1998), and if finalized soon, the evaluation of Health Canada (Health Canada, 1997).
Therefore, with respect to the first charge question on the technical quality, comprehensiveness
and clarity of the document, the EHC found that the quality and comprehensiveness would be
greatly improved by including research data published in the peer-reviewed literature since the
cutoff date of January 31, 1997. The EHC notes that important research (such as the Delzell et
al, 1995) exposure re-estimation and pharmacokinetic modeling studies) are ongoing. Also, to
improve the clarity of the document, the Committee recommended several editorial changes, such
as the inclusion of summary tables in some of the chapters.
The majority of the Environmental Health Committee did not support classifying
1,3-Butadiene as a known human carcinogen, due to the lack of consistency between exposure
response rates for leukemia or lymphosarcoma when both the styrene-Butadiene rubber (SBR)
and monomer worker studies were considered in total. The majority opined that 1,3-Butadiene
should be classified as & probable human carcinogen. This opinion was based on several lines of
evidence: a) There was only one positive study on workers in the monomer process, and it
showed only a small excess of lymphosarcoma; in addition, this study has not been replicated.
(One would like to see at least a second independent confirmatory study before affirming that
there is "sufficient evidence of human carcinogen!city" regarding Butadiene and leukemia. Instead
a fairly large and reasonably sound second study shows no leukemia excess; while two smaller
ones found no evidence of leukemia risk); b) In the monomer study, there was no evidence of an
exposure response relationship for lymphosarcoma, nor was there evidence that those workers
with longer-term exposure had a higher risk of lymphosarcoma; and c) A large study of the
workers in the styrene-Butadiene rubber (SBR) industry showed an excess of leukemia.
However, since these workers were exposed to several different chemicals, the cancer excess
could not be attributed solely to 1,3-Butadiene. Furthermore, the findings concerning co-
exposure to styrene from the reevaluation of the exposure estimates in the Delzell et al. (1995)
study may also impact the risk assessment. The majority of the Committee felt that the finding of
"known human carcinogen" should solely be based on observational studies in humans, without
-------
regard to mechanistic or other information. Others on the Committee felt that Butadiene should
be identified as a "known human carcinogen" using the cumulative evidence from epidemiology,
animal cancer bioassays, and mechanistic studies as the basis for the judgment.
The Committee found the approaches taken to characterize plausible cancer risks to be
reasonable, but points out specific data that may have been misinterpreted by the Agency. In
particular, the discussion of metabolism and toxicokinetics failed to address critical differences in
metabolism of Butadiene in different species which could account for differences in tumor
susceptibility in different species. Inclusion of a discussion of state-of-the-art models for
Butadiene metabolic pathways and kinetics would significantly strengthen the scientific quality of
the document. Included in this discussion should be the strengths and weaknesses of the available
models for risk prediction. This fuller discussion should replace the simple statements made about
the inadequacy of the available models.
The Committee commends the Agency for looking at new approaches, such as the
benchmark dose procedure, to improve quantitative assessment of non cancer endpoints.
However, the Committee has submitted suggestions on how to further improve these approaches
and how to make these new approaches more clear, accurate and concise, including the following
recommendations:
a) Correct mathematical errors in the calculation of the benchmark concentration for
reproductive and developmental effects.
b) Address new dominant lethal studies.
c) The different toxicokinetic hypotheses and the hypothesized role of the various
metabolites should be discussed, at least qualitatively, with an indication of the
degree to which the assessment would be impacted if some of the hypotheses were
later proven true.
d) Inadequate justification is given for the application of the additional safety factor
for the benchmark dose.
e) The rationale for the selection of the toxic non-cancer endpoint that is utilized in
the derivation of the RfC is very important and should be more explicitly
explained.
f) As noted above, the review document is not a comprehensive health assessment
however, and contains only an overview of the ambient exposure and exposure of
populations adjacent to emissions sources, without including any actual exposure
assessment as such. Consequently, as the document now stands, it should be
retitled as a "Hazard Assessment" until such time as an exposure assessment
component can be incorporated. The document could be called "Health Risk
-------
Assessment of 1,3-Butadiene: 1. Hazard Assessment."
The Committee appreciates the opportunity to review the draft Health Risk Assessment of
1,3-Butadiene and looks forward to receiving a written response from the Assistant
Administrator, Office of Research and Development.
Sincerely,
/signed/
Dr. Joan M. Daisey, Chair
Science Advisory Board
Dr. Mark J. Utell, Chair,
Environmental Health Committee
Science Advisory Board
-------
NOTICE
This report has been written as part of the activities of the Science Advisory Board, a
public advisory group providing extramural scientific information and advice to the Administrator
and other officials of the Environmental Protection Agency. The Board is structured to provide
balanced, expert assessment of scientific matters related to problems facing the Agency. This
report has not been reviewed for approval by the Agency and, hence, the contents of this report
do not necessarily represent the views and policies of the Environmental Protection Agency, nor
of other agencies in the Executive Branch of the Federal government, nor does mention of trade
names or commercial products constitute a recommendation for use.
-------
ABSTRACT
The Environmental Health Committee (EHC) reviewed the EPA's updated draft health
risk assessment of 1,3-Butadiene, which had a cutoff date of January, 1997. A significant amount
of new and important information has been developed since then, and the Committee felt that the
report should reflect the most current research data.
The majority of the Environmental Health Committee did not support the proposed
classification of 1,3-Butadiene as a known human carcinogen due to the lack of consistency
between exposure response rates for leukemia or lymphosarcoma when both pertinent studies
were considered. The majority opined that 1,3-Butadiene should be classified as a probable
human carcinogen.
The Committee found the approaches taken to characterize plausible cancer risks to be
reasonable but points out specific data that may have been misinterpreted by the Agency. The
Committee supported the use of the benchmark dose procedure in developing Reference levels,
and suggested how to further improve the approaches for quantitative assessment of non-cancer
endpoints. Greater explanation is needed of the safety factors applied to the benchmark, and of
the newly proposed models, especially those modeling time to impact. Also, the EHC
recommends that the Agency explain, in more detail, the rationale for the selection of the toxic
non-cancer endpoint that is utilized in the derivation of the RfC.
Keywords: 1,3-Butadiene, EPA's proposed Cancer Risk Assessment Guidelines, known human
carcinogen, probable human carcinogen, lymphosarcoma, leukemia, reproductive/developmental
effects, pharmacokinetics risk assessment.
-------
U.S. ENVIRONMENTAL PROTECTION AGENCY
SCIENCE ADVISORY BOARD
ENVIRONMENTAL HEALTH COMMITTEE
1,3-Butadiene Panel
CHAIR
Dr. Mark J. Utell, University of Rochester Medical Center, Rochester, NY
MEMBERS
Dr. Cynthia Bearer, Case Western Reserve University, Cleveland, OH
Dr. Adolfo Correa, The Johns Hopkins University, Baltimore, MD (Did not attend meeting)
Dr. John Doull, University of Kansas Medical Center, Kansas City, KS
Dr. David G. Hoel, Medical University of South Carolina, Charleston, SC
Dr. Abby A. Li, Monsanto Company, St. Louis, MO
Dr. Michele Medinsky, Chemical Industry Institute of Toxicology, Research Triangle Park, NC
Dr. Frederica Perera, Columbia University, New York, NY (Did not attend meeting)
Dr. Lauren Zeise, California Environmental Protection Agency, Berkeley, CA
CONSULTANTS
Dr. Richard Albertini, University of Vermont, Burlington, VT
Dr. Elaine Faustman, University of Washington, Seattle, WA
Dr. Karl Kelsey, Harvard School of Public Health, Boston, MA
Dr. R. Jeff Lewis, Exxon Biomedical Sciences, Inc., East Millstone, NJ
Dr. Judith MacGregor, Toxicology Consulting Services, Rockville, MD
Dr. David Parkinson, L.I. Occupational and Environmental Health Center, Port Jefferson, NY
Dr. Roy Shore, New York University Medical School, New York, NY
Dr. James Swenberg, University of North Carolina, Chapel Hill, NC
in
-------
Science Advisory Board Staff
Ms. Roslyn A. Edson, Designated Federal Officer, U. S. Environmental Protection Agency,
Science Advisory Board (1400), 401 M Street, SW, Washington, DC 20460
Mr. Samuel Rondberg, Designated Federal Officer, U. S. Environmental Protection Agency,
Science Advisory Board (1400), 401 M Street, SW, Washington, DC 204601
Ms. Mary L. Winston, Management Assistant, Environmental Protection Agency, Science
Advisory Board (1400), 401 M Street, SW, Washington, DC 20460
JDid not attend the public meeting but provided editorial support for this report.
IV
-------
TABLE OF CONTENTS
1. EXECUTIVE SUMMARY 1
2. INTRODUCTION 4
2.1 Background 4
2.2 Charge 4
3. RESPONSE TO THE CHARGE 5
3.1 Technical Quality, Comprehensiveness and Clarity 5
3.1.1 Chapter 1 - Introduction 5
3.1.2 Chapter 2 - Overview of Exposure to 1,3-Butadiene 6
3.1.3 Chapter 3 - Metabolism and Pharmacokinetics 6
3.1.4 Chapter 4 - Mutagenicity 8
3.1.5 Chapter 5 - Reproductive and Developmental Effects 16
3.1.6 Chapter 6 - Toxicity in Animals 17
3.1.7 Chapter 7 - Epidemiologic Studies of Carcinogen!city 19
3.1.8 Chapter 8 - Pharmacokinetic Modeling 22
3.1.9 Chapter 9 - Quantitative Risk Assessment for 1,3-Butadiene 25
3.1.10 Chapter 10 - Weight of Evidence 32
3.1.11 Chapter 11 - Risk Characterization 32
3.2 Classification of 1,3-Butadiene as a Known Human Carcinogen 33
3.3 Approaches Taken to Characterize Plausible Cancer Risks 35
3.4 Conclusions and Quantitative Estimations for Reproductive/Developmental
Effects 38
4. SUMMARY OF RECOMMENDATIONS 40
APPENDIX A-TECHNICAL ISSUES A-l
GLOSSARY - ACRONYMS AND ABBREVIATIONS G-l
REFERENCES CITED R-l
-------
1. EXECUTIVE SUMMARY
The EPA Office of Research and Development prepared the draft Health Risk Assessment
of 1,3-Butadiene (USEPA, 1998a) in response to a request from the Office of Mobile Sources
(OMS). The document was requested by OMS in order to support a future Air Toxics Rule. The
review document is not intended to be a comprehensive health assessment and therefore, does not
contain any actual exposure assessment. The document focuses on mutagenicity, carcinogenicity,
reproductive, and developmental effects and presents the Agency's first benchmark dose analysis
for reproductive/developmental effects. In its document, the Agency states that the new studies
published since 1985 change the weight of evidence for cancer. Based on the weight of overall
evidence from human, animal, and mutagenicity studies, the Agency concludes that 1,3-Butadiene
is a known human carcinogen.
On April 30 and May 1, 1998, the Environmental Health Committee met at the EPA's
Waterside Mall complex in Washington, DC to review the Agency's draft Health Risk Assessment
of 1,3-Butadiene. In addition to a general review of the document by the Committee, the Office
of Research and Development specifically requested that the EHC provide comment on each of
the following aspects of the document:
a) Review the health risk assessment for technical quality, comprehensiveness and
clarity
b) Does the science support the classification of "known" human carcinogen?
c) Are the approaches taken to characterize plausible cancer risks reasonable given
the science?
d) Are the conclusions and quantitative estimations for reproductive/developmental
effects adequately supported?
The Committee acknowledges that this is an extremely difficult task given the large
amount of information on 1,3-Butadiene toxicity, epidemiology, and mechanism available with
new information being made available on a continual basis.
The Agency selected a cutoff date for the inclusion of new information in the revised, draft
document of January 31, 1997. However, a significant amount of new and important information
has been developed since then and is pertinent to a health risk assessment of 1,3-Butadiene. The
Committee felt that the report should reflect the most current research data, including the recent
evaluations by the International Agency for Research on Cancer (IARC, 1998) and on ongoing
evaluation in Canada (Health Canada, 1997). Therefore, with respect to the first charge question
on the technical quality, comprehensiveness and clarity of the document, the EHC found that the
quality and comprehensiveness would be greatly improved by including research data published in
-------
the peer-reviewed literature since the cutoff date of January 31, 1997. The EHC also notes that
important research such as the Delzell et al. (1995) exposure re-estimation and the
pharmacokinetic modeling studies is ongoing. To improve the clarity of the document, the
Committee recommended several editorial changes, such as the inclusion of summary tables in
some of the chapters.
The majority of the Environmental Health Committee did not support classifying
1,3-Butadiene as a known human carcinogen, due to the lack of consistency between exposure
response rates for leukemia or lymphosarcoma when both the styrene-Butadiene rubber (SBR)
and monomer worker studies were considered in total. The majority opined that 1,3-Butadiene
should be classified as a probable human carcinogen. This opinion was based on several lines of
evidence: a) There was only one positive study on workers in the monomer process, and it
showed only a small excess of lymphosarcoma; in addition, this study has not been replicated.
(One would like to see at least a second independent confirmatory study before affirming that
there is "sufficient evidence of human carcinogen!city" regarding Butadiene and leukemia. Instead
a fairly large and reasonably sound second study shows no leukemia excess; while two smaller
ones found no evidence of leukemia risk); b) In the monomer study, there was no evidence of an
exposure response relationship for lymphosarcoma, nor was there evidence that those workers
with longer-term exposure had a higher risk of lymphosarcoma; and c) A large study of the
workers in the styrene-Butadiene rubber (SBR) industry showed an excess of leukemia.
However, since these workers were exposed to several different chemicals, the cancer excess
could not be attributed solely to 1,3-Butadiene.
The Committee found the approaches taken to characterize plausible cancer risks to be
reasonable but points out specific data that may have been misinterpreted by the Agency. In
particular, the discussion of metabolism and toxicokinetics failed to address critical differences in
metabolites of Butadiene in different species accounting for differences in tumor susceptibility.
Inclusion of a discussion of state-of-the-art models for Butadiene kinetics would significantly
strengthen the scientific quality oft he document. Included in this discussion should be the
strengths and weaknesses of the available models for risk prediction. This fuller discussion should
replace the simple statements made about the inadequacy of the available models.
The Committee commends the Agency for looking at new approaches, such as the
benchmark dose procedure, to improve quantitative assessment of non cancer endpoints.
However, the Committee has submitted suggestions on how to further improve these approaches
and how to make these new approaches more clear, accurate and concise, including the following
recommendations:
a) There was an apparent mathematical error in the calculation of the benchmark
concentration for reproductive and developmental effects which must be
addressed. All calculations should be easy to follow, and of course, carefully
proofed.
-------
b) There are new dominant lethal studies, which are not included in the risk
assessment, that failed to replicate earlier findings.
c) A variety of viewpoints were expressed within the Committee over the extent to
which toxicokinetic analyses should be incorporated into the assessment. At a
minimum the different toxicokinetic hypotheses and the hypothesized role of the
various metabolites should be discussed, at least qualitatively, with an indication of
the degree to which the assessment would be impacted if some of the hypotheses
were later proven true.
d) Inadequate justification is given for the application of the additional safety factor
for the benchmark dose. Some of the Committee members could not understand
the rationale for its conclusion.
e) The rationale for the selection of the toxic non-cancer endpoint that is utilized in
the derivation of the RfC is very important and should be more explicitly
explained.
f) The review document is not a comprehensive health assessment and contains only
an overview of the ambient concentrations and exposure of those populations
adjacent to emissions sources, without including any actual exposure assessment as
such. Consequently, as the document now stands, it should be retitled as a
"Hazard Assessment" until such time as an exposure assessment component can be
incorporated. The document could be called "Health Risk Assessment of 1,3-
Butadiene: 1. Hazard Assessment."
-------
2. INTRODUCTION
2.1 Background
The draft Health Risk Assessment of 1,3-Butadiene (USEPA, 1998a) was developed by
the EPA Office of Research and Development at the request of the Office of Mobile Sources.
This document was requested to support a future Air Toxics Rule. The document was not
intended to be a comprehensive health assessment. Consequently, an actual exposure assessment
is not included in the document. The draft Health Risk Assessment of 1,3-Butadiene (USEPA,
1998a) focuses on carcinogen!city, mutagenicity, and reproductive/developmental effects. The
document presents the Agency's first benchmark dose analysis for reproductive/developmental
factors.
The draft document that was reviewed by the Committee updates a previously, published
document (USEPA, 1995). In the current draft document, the Agency has included many new
studies which have been published since 1985. Based on the weight of overall evidence from
human, animal, and mutagenicity studies, the Agency concludes, in the current draft document,
that 1,3-Butadiene is a known, human carcinogen.
The EHC subsequently met On April 30 and May 1, 1998, in Washington, DC to review
the Agency's draft Health Risk Assessment of 1,3-Butadiene document.
2.2 Charge
The Committee was charged to provide comments on each of the following aspects of the
document:
a) Review the health risk assessment for technical quality, comprehensiveness and
clarity (Address each chapter, but with specific reference to Charges b, c, and d).
b) Does the science support the classification of "known" human carcinogen?
c) Are the approaches taken to characterize plausible cancer risks reasonable given
the science?
d) Are the conclusions and quantitative estimations for reproductive and
developmental effects adequately supported?
-------
3. RESPONSE TO THE CHARGE
This report captures comments and recommendations reflecting a consensus by the
Committee, as well as specific technical comments and recommendations provided by individual
Committee Members (see Appendix A). It was not possible to review and achieve consensus for
each and every technical point provided in this Committee report. On major points, however, an
attempt has been made to reflect the consensus or range of views on the Committee.
3.1 Technical Quality, Comprehensiveness and Clarity
The Environmental Health Committee was asked to: review the health risk assessment for
technical quality, comprehensiveness, and clarity. The Committee provided comments and
recommendations for each chapter.
3.1.1 Chapter 1 - Introduction
The Introduction presents a clear and comprehensive summary of information on
1,3-Butadiene for the period from 1985 until January 31, 1997. It was concluded that apparent
differences in assessments by different groups might be explained by the availability of studies at
the time of evaluations, different cancer classification systems, and quantitative assessments done
for different purposes. Does this mean that they are all equally valid and scientifically defensible?
In other words can the statements made in Table 1-1 be restated in comparable terminology to
show that they are all compatible? If they are, as implied, then the document should demonstrate
this in subsequent chapters. If not, the Introduction should state so, and subsequent chapters
should explain why. The EHC also recommends that concentrations be expressed in one constant
unit throughout the document as well as in Table 1.
Every document must have a cut-off in time for completion. However, the International
Agency for Research on Cancer (IARC, 1998) has recently re-evaluated the evidence on the
carcinogenicity of 1,3-Butadiene, and Health Canada is completing a health evaluation on the
compound. It will be important to include the IARC evaluation in an updated version of the
health assessment, and should the Health Canada document be finalized soon, it should also be
included. For example, Table 1-1 should include the IARC evaluation (IARC, 1998), and if
available, the finalized Health Canada evaluation. Also, the Agency should clarify, in the Table,
whether the OSHA classification listed 1,3-Butadiene as a "known" or "potential" carcinogen.
It is stated that the profile for setting the American Conference of Governmental Industrial
Hygienists-Threshold Limit Value (ACGIH-TLV) for 1,3-Butadiene is not reviewed because it is
not a risk assessment. This may be true for the EPA definition of a risk assessment, but the
Agency should include some history of what has been used for the work environment, particularly
because of the high values in the past and the importance of epidemiological studies on worker
population.
-------
There should be a comment in the final document on the accessibility of unpublished data
that is referenced in Agency documents. This concern was based on the use of the Delzell (1995)
study in the Agency's health risk assessment of 1,3-Butadiene even though the review document
was not readily available to the public.
Finally, there is no mention in the Introduction of EPA's proposed Cancer Risk
Assessment Guidelines. While these guidelines have not been completely finalized, they are
sufficiently complete to be referenced and utilized as appropriate (EPA, 1996).
3.1.2 Chapter 2 - Overview of Exposure to 1,3-Butadiene
This chapter is intended to be an introductory review of possible sources of exposure to
1,3 Butadiene. It should be clearly indicated in the first paragraph that this chapter is not intended
to be a comprehensive review of exposure so that this chapter is not mistakenly used as a source
of data for enforcement purposes. An explanation should be included on how concentrations
have been measured over the years and how this might impact a comparison of exposure levels
measured in earlier years with more recent measurements. Ranges of concentrations are
important to include along with averages. The Committee recommends that all of the exposure
concentrations be expressed in one common unit throughout Chapter 2.
Additional, more specific, comments are provided in Appendix A.
3.1.3 Chapter 3 - Metabolism and Pharmacokinetics
Chapter 3 reviews the metabolism and pharmacokinetics of 1,3-Butadiene. This chapter is
long, contains numerous incorrect or incomplete statements that need to be changed, and should
be updated. The title of the chapter should be Metabolism and Toxicokinetics, as Butadiene is
not used as a therapeutic agent. Also, the word, toxicokinetics, should replace pharmacokinetics
throughout the document.
The Health Canada assessment of Butadiene (Health Canada, 1997) and comments
submitted by the public during the open meeting (Himmelstein, 1998) will provide many of the
new references that need to be incorporated into the revision. Specific issues that should be
addressed in the revision of the risk assessment are provided below.
In evaluating the data on the metabolism of Butadiene and its epoxide metabolites across
species, including humans, the EPA concluded that there are no clear species differences in
Butadiene metabolism and that in many cases there is overlap in metabolic rates across species. In
forming their conclusion, the EPA misinterpreted the in vitro data and ignored the major
conclusion from the in vivo data. With respect to the in vitro data, the EPA compared maximal
rates of metabolism (i.e., Vmax) rather than basing their comparisons on a more appropriate
measure of metabolic rates, the ratio of Vmax/Km. This ratio is more appropriate because: a)
-------
species differences are observed in both metabolic parameters; and b) at non-metabolically
saturating concentrations (i.e., those concentrations relevant for humans) it is the ratio, not the
maximum rate, that dictates the rate of metabolism. With respect to the in vivo data, the EPA has
not adequately summarized the data from two independent laboratories that clearly and
unambiguously show large differences spanning orders of magnitude in the blood and tissue
concentrations of the Butadiene metabolite diepoxybutane. In particular, Table 3-8 is misleading.
This table includes data from the rat, mouse, and monkey for blood epoxybutene and
diepoxybutane concentrations from studies conducted by Bond et al. (1986) and Dahl et al.
(1991) which did not use gas chromatography-mass spectrometry (GC-MS) methods to
quantitate these metabolites (Bond et al., 1986; Dahl et al., 1991). Therefore, the values reported
by these investigators are unreliable. It would be preferable to, instead, compare the blood
concentrations reported by Himmelstein et al. (1994), Bechtold et al. (1995), and
Thorton-Manning et al. (1995a; 1995b; 1996) in one table. This comparison would serve two
purposes. First, it would highlight the interlaboratory reproducibility in the values reported for
these metabolites serving to increase the reliability of the data. Second, it would accentuate the
dramatic species differences in the circulating levels of diepoxybutane, lending support to the
hypothesis that this interspecies difference in metabolite formation underlies the differences in
susceptibility observed in the chronic studies.
Mutagenicity studies strongly suggest that diepoxybutane is a, if not, the critical
metabolite in Butadiene carcinogen! city. Thus, differences in levels of this metabolite formed in
vitro and in vivo are highly consistent with the observed species differences in carcinogen!city.
The Committee recommends that the Agency integrate and assess this information, develop
conclusions based on the weight of all the information presented, and present these conclusions in
the closing paragraphs of Chapter 3.
It is recognized that the EPA elected a cutoff date for inclusion of new information of
January 31, 1997. However, there is no question that a significant amount of new and important
information has been developed since this date that is particularly pertinent to Chapter 3 on
Metabolism and Pharmacokinetics. In short, the chapter on Metabolism and Pharmacokinetics
needs to be updated to include key references that have been published in the peer-reviewed
literature since January 31, 1997. Many of the data sets that have been published since this date
lend further support to the hypothesis that diepoxybutane is a critical metabolite involved in the
carcinogen!city of 1,3-Butadiene.
Although this chapter references important literature relative to the in vitro and in vivo
metabolism and Butadiene, it falls short in providing a comprehensive integration of the relatively
diverse data sets. A critical concept that appears lacking in this chapter is the fact that
diepoxybutane represents a critical metabolite of Butadiene and that there are significant species
differences both in vitro and in vivo in the formation of this important Butadiene metabolite. This
chapter neglects to note that metabolism and toxicokinetic studies of Butadiene conducted in
whole animals and in rodent and human tissues provide important insights into the likely critical
steps in the initiation of Butadiene carcinogen!city and importantly the identity of the most likely
-------
chemical species responsible for the development of tumors. For example, dosimetry data on both
epoxybutene and diepoxybutane following inhalation exposure to Butadiene clearly indicate that
blood concentration of epoxybutene were up to 8-fold higher in mice compared with rats and that
blood concentrations of diepoxybutane were nearly 40-fold higher in mice than in rats. Similarly,
tissue concentrations of epoxybutene ranged from 3-10 times higher in mice compared with rats
and tissue concentrations of diepoxybutane were up to 100 times higher in mice than rats.
Importantly, therefore the correlation between measured circulating blood and tissue levels of the
epoxides, particularly diepoxybutane and the observed development of tumors is clearly
suggestive of a role of diepoxybutane in the initiation of cancer. This important concept is not
presented in the chapter.
Moreover, in vitro data on the metabolism of Butadiene suggest that mice form
epoxybutene and diepoxybutane at a faster rate than rats or humans. Studies on the in vivo
metabolism and tissue concentrations of epoxybutene and diepoxybutane in mice and rats
following inhalation exposure to Butadiene are consistent with the in vitro studies on metabolism
of Butadiene. This is an important point and the chapter neglects to point out the close parallels
between the observations from in vitro studies and in vivo studies.
The Agency notes that Butadiene is an animal carcinogen and that the mouse is more
sensitive than the rat to Butadiene induced carcinogen!city. The Agency also notes that the
reasons for these interspecies differences are not understood at this time. However, the available
mechanistic data on the metabolism of Butadiene and its reactive epoxide metabolites supports the
hypothesis that interspecies differences in metabolic rates form the underpinning for the increased
sensitivity of the mouse compared with the rat. These differences in metabolic rates result in a
faster production and slower detoxification of the diepoxybutane in mice compared with rats with
resultant higher steady state levels of the diepoxybutane in blood and tissues in the mouse
compared with the rat following exposure to Butadiene. This observation coupled with the
demonstrated hundred-fold greater mutagenicity of the diepoxybutane compared with the
epoxybutene points to interspecies differences in the formation of diepoxybutane being critical to
interpretation of differences in response.
Additional, more specific, comments are provided in Appendix A.
3.1.4 Chapter 4 - Mutagenicity
This chapter summarizes the genotoxicity of 1,3 Butadiene (BD) and its metabolites for
mice, rats and humans, considering both cytogenetic endpoints and gene mutations in somatic and
germinal cells. The emphasis is on recent in vivo studies, although there is some mention of
results from in vitro assays.
The chapter would be improved by adding text tables summarizing key animal and human
findings derived from the entire body of information on BD. References to support key findings
should be included in these tables. Separate tables should be included for in vitro, animal and
-------
human findings, and similarities and species differences in response should be noted.
The chapter, as currently written, does not give sufficient emphasis and weight to the
positive heritable translocation studies in mice (given their potential relevance for human heritable
risks) ( Adler et al, 1998; 1995) and to several additional studies conducted in humans
(Pacchierotti, 1998). This finding is also discussed in Section 3.1.5 of our report, which
addresses the Reproductive and Developmental Effects chapter of the review document.
The conclusion section should be expanded to include what is known about the
mutagenicity of BD and its metabolites from the extensive literature. Careful editing of the final
document should be conducted to avoid missing dose units, units for mutant frequencies, and
similar omissions.
The binding of BD metabolites to DNA should be discussed in greater detail. Evidence
for reactivity with the DNA itself demonstrates that these BD intermediates reach their target
molecule for genotoxicity causing pre-mutagenic DNA lesions.
Early studies showed that DEB binds to N-7 guanine and that it forms inter-strand
crosslinks (Brooks and Lawley, 1961; Lawley and Brooks, 1967). Subsequently it was shown
that B6C3F1 mice and Wistar rats exposed to 14C labeled BD by inhalation have covalently bound
reactivity in liver DNA, as noted in Chapter 3 (Kreiling et al., 1986). The amounts bound in the
two species were comparable. The nature of the bound residues was not determined. The
complex kinds of adducts formed in DNA by BD metabolites are also under active investigation.
Specific enantio- and regioisomeric EB adducts formations have been shown (Koivisto et al.,
1997; Tretyakova et al., 1998). The N-7 position of guanine has been shown to be the most
reactive with EB, followed by the N-3 and N-l positions of adenine. EB adducts have also been
found at the N-6 position of adenine but this may represent a rearrangement of the N-l adenine
adduct. Adenine adducts may be important for the genotoxicity of BD as shown by mutational
spectral studies of this agent. (Leuratti et al., 1994; Cochrane et al., 1994).
The several nucleobase adducts in DNA formed by BD metabolites are being
characterized further by several groups (Leuratti etal, 1994; Neagu etal, 1994, 1995; Selzer
and Elfarra, 1996; Kumar et al., 1996). Furthermore, attempts are being made to use high
sensitivity methods for detecting DNA adducts in humans for biomonitoring purposes, including
the detection of urinary DNA adducts. However, despite the statement made in Chapter 11, page
11-9, para. 1, line 2, the Committee is not aware that DNA adducts have even been observed in
vivo in humans. The statement in this regard made on page 11-9 either needs a reference or
should be corrected.
The extensive mutagenicity results covered in the several reviews cited in Chapter 4
should be discussed in more detail. These results include positive Salmonella assays in the
presence of S9. It is important to note that human S9 also converts BD to mutagenic metabolites,
as determined in the Salmonella system (Arce et al., 1990). BD metabolites also have been
-------
positive for mutagenicity in a variety of microorganisms with or without metabolic activation.
Studies of mutation by BD at the tk locus in mouse lymphoma cells have been both positive
(Sernau et a/., 1986) and negative (McGregor et a/., 1991). BD, as opposed to its metabolites,
has not been found mutagenic in vivo in Drosophila melanogaster, as assessed by the sex-linked
recessive lethal mutation assay (Foureman et a/., 1994) or by the Wing Spot Test (Victorin et a/.,
1990).
The species comparisons should be considered in detail. There have been numerous
cytogenetic studies in vivo in rodents. These are covered in the cited reviews and, as correctly
stated in Chapter 4, support the dichotomy in carcinogenic response between these two species,
where mice are more responsive than rats. However, it should be noted that, with regard to
cytogenetic assays in vivo in these two species, there are no reports of positive results in rats
exposed to BD (as exposed to BD metabolites) but there are several reports of positive results in
mice.
It is important that the document reflect species differences in clastogenicity. The
difference in BD's apparent mutagenic potencies for mice and rats are worthy of comparison.
Studies can be considered as those measuring clastogenicity and those measuring specific gene
mutations. In regards to clastogenicity, for mice there are numerous reports of chromosome
aberrations, micronuclei and SCEs in somatic cells, i.e., in both lymphocytes and in bone marrow
cells, several showing effects at low doses. However, in rats, neither chromosome aberrations nor
micronuclei have been found in blood cells in vivo (i.e., in lymphocytes, bone marrow or
peripheral red blood cells) after BD inhalation (Arce et a/., 1990; Autio et a/., 1994), although
there is a report of a weak positive SCE response (Maki-Paakkanen et a/., 1993).
For germ cells, the contrast in clastogenicity results between these species is equally
striking. In mice, there are reports of dominant lethal effects with inconsistent results (references
given in the draft document), sperm head abnormalities (Morrissey et a/., 1990), micronuclei in
spermatids (Xiao and Tates, 1995; Tommasi et a/., 1998) and a positive Comet assay in haploid
and polyploid testicular cells (Brinkworth et a/., 1998). The dominant lethal studies discussed in
Chapter 5 should be cross-referenced in Chapter 4 as they are important to the discussion of
heritable chromosomal alterations induced by BD. Cytogenetic abnormalities in first-cleavage
embryos sired by male mice treated by BD inhalation have also been reported recently
(Pacchierotti et a/., 1998). By contrast, in the single study in rats, no dominant lethal mutations
were found (BIBRA, 1996).
Of special importance among the studies of clastogenicity are those of heritable
translocations in mice. The first study (Adler et a/., 1995) is discussed briefly in Chapter 4 . A
second study by the same group has recently been reported (Adler et a/., 1998). These studies are
particularly relevant to human health and risk assessments for human heritable damage and should
be emphasized. In fact, a human risk estimate has recently been reported using a parallelogram
approach that employed mouse somatic cell clastogenicity (micronuclei in bone marrow cells),
mouse germ cell clastogenicity (the heritable translocations) and human somatic cell clastogenicity
10
-------
(in lymphocytes from BD exposed workers) to estimate the heritable translocation risk for humans
(Pacchierotti et a/., 1998). The estimated doubling dose for human heritable translocations was
given as 1,100 ppmh (parts per million hours)
Species differences in mutagenicity should be detailed. Comparison of the specific locus
mutagenicity of BD between mice and rats is not nearly as striking as is the comparison for
clastogenicity. The studies summarized by Recio and Goldsworthy (1995) showing an increase in
BD-induced lacl mutations at A:T base pairs in bone marrow stem cells from young transgenic
B6C3F1 (BB) male mice are discussed in Chapter 4. However, it should be mentioned that an
earlier study in young male Balb/c X DBA/2 (CD2F1) transgenic mice (MM) found lacZ BD
induced mutations only in lung cells and not in liver or bone marrow stem cells (Recio et al.,
1992). There is also a report of a positive spot test in "T" stock mice that found an increase in in
vivo mutations in embryonic melanoblasts (Adler et al., 1994).
The several reports of positive hprt mutations in young B3C3F1 mice exposed to BD by
inhalation should be noted in the revised document. The Cochrane and Skopek (1994b) study
mentioned in Chapter 4 exposed pre-weanling mice and found dose-related mutational increases.
Again, there was a bias for A:T changes. The Meng study mentioned in Chapter 4 (Meng et al,
1996) also exposed young B6C3F1 male mice to BD and found increases in hprt mutations in
both thymic and splenic lymphocytes that persisted for several weeks after the exposures. This
work has now been accepted as a full publication and indicates that the hprt mutant frequency
(MF) in thymic lymphocytes rises from 2.2 x 10"6 to 11.3 x 10"6 at two weeks (maximal) and, in
splenic lymphocytes, from -1.8 x 10"6 to 19.7 x 10"6 at five weeks (maximal). The mutagenic
potency of BD in these mice, which is a measure of hprt mutant cell accumulation, was calculated
to be 69.62. The negative reports for in vivo hprt mutations in other strains of mice that were
older at the time of their BD exposures should also be noted. Tates et al. (1994) originally found
low order hprt mutagenicity in splenic lymphocytes of 10-12-week-old (102/E1 X C3FI/E1) Fl
mice , but later failed to find an increase in such hprt mutations in a second study of this strain or
in adult CD1 mice (Tates et al., 1998).
In vivo mutations in rats should be noted in the document. In contrast to the
clastogenicity studies, specific locus gene mutations have been found in vivo in rats exposed to
BD by inhalation. Meng et al. (1996) exposed young rats with results as noted in Chapter 4, i.e.,
increases in hprt mutations in thymic lymphocytes from 2 x 10"6 to 4.9 x 10"6 at three weeks after
exposure (maximal) and, in splenic T-cells, from -1.9 x 10"6 to 10.1 x 10"6 at four weeks after
exposure (maximal) (references given in chapter). The mutagenic potency of BD in rats was
calculated to be 15.85. Thus, the ratio of mutagenic potencies between mice and rats (mice/rats)
was calculated to be 4.4 or 5.0, depending on the weeks allowed for mutant cell accumulation.
Although BD is five times more mutagenic in mice than in rats, the magnitude of this difference is
not nearly as great as that reported for the carcinogenicity differential between these two species.
The many rodent studies of BD metabolites, some positive and some negative, show a
similar pattern and should be discussed in the document. In general, younger animals tend to give
11
-------
the positive results and, for mice, it is the B6C3F1 animals that are most often associated with
positive studies. The Cochrane and Skopek studies of EB and DEB were noted in Chapter 4.
Meng et al. (1996, 1997) have extended their studies in young mice (B6C3F1) and rats (Fischer)
to EB and DEB by inhalation and have shown increased hprt mutations in splenic TAcells in both
species. In rats, however, the response to EB was equivocal. In contrast to these positive results,
Tates et al. (1994) found no hprt mutation induction in older mice (102/E1 X C3H/E1) Fl and
CD1 or rats (Lewis) administered EB or DEB by injection or in the drinking water. However, as
noted in Chapter 4, rat germinal cells are at least as susceptible (or more so) to the clastogenicity
of BD metabolites as are mouse germinal cells.
In summary, there are striking species differences between mice and rats in the reported
studies as regards clastogenicity. There are numerous positive studies of this endpoint in mice,
but BD induced clastogenicity has not been demonstrated in vivo in rats. Of note, heritable
translocations have been demonstrated and confirmed following BD exposures to mice.
Clastogenicity has been observed in mice and rats exposed to EB or DEB. For gene mutations,
B6C3F1 mice appear to be more sensitive than other mouse strains to the mutagenic effects of
BD, EB or DEB. In general, mice are also more sensitive than rats to specific gene mutations,
although such mutations can be induced in younger rats. Mutations are most commonly seen in
both species in young animals, indicating that cell proliferation may be required to produce these
mutagenic effects.
Human Studies, because of their relevance to human health, are addressed in detail.
Chapter 4 includes most of the relevant mutagenicity (clastogenicity and specific gene mutations)
studies in human cells and/or in vivo in humans. In discussing the Cochrane and Skopek (1994)
study of hprt and tk mutations in vitro in human TK6 cells described in page 4-1, the EPA
document should note that mutation induction was assessed for EB, DEB and for Ebdiol (EBD).
The mutagenic potencies of these BD metabolites were found to be DEB > EB > EBD. It is
important to note that the metabolite EBdiol has been studied and has been found to be
mutagenic. The importance of EBdiol lies in its abundance. Therefore, even though Ebdiol may
be the least mutagenic of the metabolites, it may be the most abundant, and therefore may give the
most mutations.
Some of the in vivo cytogenetic studies in humans are covered in Chapter 4. The positive
challenge assay reported by Au et al. (1995) is noted, as is the negative report of chromosome
aberrations, micronuclei and SCEs in exposed workers by Sorsa et al. (1994). Not noted,
however, is the update by Sorsa et al. (1996) that reanalyzed these data according to GSTT1
status and found that the Tl null workers appeared to have increases in chromosome aberrations.
Also, the increases in chromosome aberrations but not in frequency of micronuclei reported by
Tates et al. (1996) and noted in Chapter 4 has been updated in a full manuscript that reports
increases in both chromosome aberrations and SCEs in BD exposed Czech workers (Tates et al.,
1998). The frequencies of micronuclei were not increased and the Comet assay was negative.
The in vitro findings of increased EB induced SCEs in lymphocytes from GST M null individuals
and increased DEB induced SCEs in lymphocytes from GST T null individuals are reported in
12
-------
Chapter 4 and are important, as is the observation that GST status does not appear to affect SCE
frequencies induced by EBD. A recent in vitro study observed chromosome specific aneuploidy
(for chromosomes 12 and X) in human lymphocytes in vitro following treatments with either EB
or DEB (Xi et al., 1997). The paper by Wiencke et al. (1995) demonstrating the affinity of the
glutathione transferase theta enzyme for the diepoxide metabolite of BD should be added.
The status of the in vivo hprt mutation studies in humans has been accurately reported in
Chapter 4. There are the three positive studies using the autoradiographic assay, as noted, and
the two negative studies that employed the cloning assay. Chapter 4 discusses the discordance in
these two groups of studies and concludes that, regardless of the reason for the difference, the
positive results with the autoradiographic assay probably reflect a mutagenic effect of BD in the
monitored workers. There was a difference in opinion among the Committee regarding the hprt
results. Some of the Committee found the Agency's conclusions to be reasonable while other
Members/Consultants were concerned that the autoradiographic results may be method-related.
The document's explanation as to possible reasons for the discordance between the
positive effects obtained by autoradiography and the negative results found by cloning for human
BD induced hprt mutations in vivo however should be rewritten. In considering the differences
between the Ward et al. (1994; 1996) positive autoradiographic studies and the Hayes et al.
(1996) and Tates et al. (1996) negative cloning assay studies, the draft document concludes that
"a simple explanation would be that the increase in the autoradiographic assay was due to clones
of mutants having arisen from earlier mutations." There was a diversity of opinion within the
EHC regarding the discordance between the positive effects obtained by the autoradiography and
the negative results found by cloning for human butadiene inducted hprt mutations in vivo. Some
Members of the Committee found that the Agency's explanation is probably not correct for two
reasons. First, the autoradiographic assay requires a technical step before the T-cells can be
assessed for hprt mutations, i.e., cryopreservation or some measure to arrest those few cell that
are cycling in human peripheral blood at the Gl-S interphase. This is simply a technical means to
insure that cycling non- mutant T-cells do not progress to their S-phase in vitro in the presence of
6-thioguanine and become labeled - even slightly labeled - and thereby become scored as mutants
in the assay. Such cells are not mutants but would appear to be so by virtue of their labeling, i.e.,
they are phenocopies. This phenomenon, which elevates scored variant frequencies, can be
eliminated by arresting the cycling cells at this phase of the cell cycle, from which they rapidly
proceed into S in vitro before the label is added. Thus, they miss the scoring window and are not
scored as variant cells.
The reason why this arresting or cryopreservation step is relevant to the issue of clonality
is that T-cells that are undergoing clonal expansions in vivo tend to be among the cycling cells. If
such large mutant clones were present, the cycling members of the clones would be eliminated
from the scoring window by the cryopreservation step (even though these would be mutants).
Therefore, if anything, clonal amplifications are less likely in the autoradiographic than in the
cloning assay. Said the other way around, if in vivo clonality were the reason for the higher
mutant frequencies in exposed vs. control workers, this phenomenon would have most likely
13
-------
affected the results in the cloning assay. This is the exact opposite to what was observed, i.e., the
increased mutant frequencies in exposed over controls were observed with the autoradiographic
assay. Another reason why in vivo clonality cannot account for the differences in results between
the assays is that molecular studies were not done. Such studies cannot be done for the
autoradiographic assay but can when using the cloning assay. Molecular analyses allows
detection of in vivo clonality using the cloning assay. Since no such studies were done, no
"corrections" were made and the point is moot.
There are other differences between the autoradiographic and cloning assays that could
account for the differences in results obtained with these two assays. The autoradiographic
method is a phenotypic assay, meaning that mutants (or variants) are consumed by the assay and,
therefore, cannot be analyzed at the molecular level. It is not, therefore, possible to demonstrate
that the observed mutants scored by this assay are actually genotypic mutants. This, however, is
not different from other phenotypic assays for mutations that are in wide use. To be scored in the
autoradiographic assay, variant cells have simply to synthesize DNA in their first S in vitro. No
cell division or growth is required. In addition, the cells must begin their DNA synthesis within
36-48 hours in culture. Therefore, a quite different T-cell subpopulation may be scored in the
autoradiographic than in the cloning assay. It is possible that, in the former, only that subset of
cells that are capable of division quickly in vitro is scored whereas, in the cloning assay, all T-cells
are scored. Alternatively, what is being scored in the autoradiographic assay may not be fixed
mutations but rather adduct-blocked RNA transcription. Another possibility is that the cell that
would go into a G2 block, and therefore not be measured by the cloning assay, are scored in the
autoradiographic assay because they are not required to go through G2 for scoring.
The autoradiographic assay requires that peripheral blood T-cells be cryopreserved in
order to arrest cycling cells at the Gl-S interphase. If all cycling cells are not eliminated, the
autoradiographic assay might, because of artifacts, give an elevated reading because of the scoring
of phenocopies. This is explained above. As for all mutation assays, the autoradiographic assay
may have a scoring bias. However, the cloning assay too has an inherent observer bias and a
tendency of some technicians to score only larger mutant colonies.
In addition, with respect to the BD studies, the differences between the positive and
negative studies may be due, in part, to differences in study populations. The Ward et al. (1994;
1996) radiographic studies were conducted at a butadiene monomer production facility and an
SBR production facility in Texas. The Tates et al. (1996) study was conducted in the Czech
Republic (butadiene monomer production) and the Hayes et al. (1996) study in China
(polybutadiene rubber production). Smoking and other lifestyle and confounding parameters
would be different among these populations. In addition, the exposure assessment in these studies
were not precise and no attempt was made to determine mutation susceptibility profiles of the
monitored workers (It is of note, however, that one of the highest variant frequency values in the
Ward et al. (1996) study was in a GST T null individual.).
14
-------
Few conclusions are made at the end of Chapter 4. The quality of the conclusions would
be improved by expanding this section and adding statements summarizing what we know about
the mutagenicity of BD and its metabolites. The following list could serve as a guide:
a) Rodent studies indicate that some of the metabolites of BD are DNA binding
agents.
b) There are numerous in vitro assays in several model systems that demonstrate the
genotoxicity of BD and/or its metabolites.
c) in vivo studies in mice and rats show that the former is the more sensitive species
for the genotoxicity that follows exposure to the parent BD.
d) Young animals (including mice), and perhaps certain strains within a species, are
more susceptible to the genotoxicity of BD and/or its metabolites than are others
as indicated by hprt results.
e) Clastogenicity has not been demonstrated for either somatic or germinal cells in
rats exposed to the parent BD.
f) Heritable translocations are induced by exposure of male mice to BD by inhalation
(This is an important point for estimating heritable risks.). Equivalent data are not
available in the rat. Any estimation of human heritable risk must account for
species differences in metabolism.
g) BD metabolites give genotoxic effects in both rats and mice with mice being the
more sensitive to somatic effects but both species being equally susceptible (or rats
more so) to the germinal effects.
h) Human studies have shown that BD metabolites cause gene mutations in vitro in
human cells.
i) GST genotypes may affect susceptibility to the clastogenicity (and possibly
mutagenicity) of BD metabolites.
j) There is evidence of clastogenicity in vivo in human lymphocytes from exposed
workers.
k) Hprt mutations have been shown in vivo in human T-lymphocytes of exposed
workers as documented by the autoradiographic but not by the cloning assay for
these events.
15
-------
3.1.5 Chapter 5 - Reproductive and Developmental Effects
It was evident to the Committee that the EPA had a large amount of material to review
since that there were over 20 animal bioassays on this topic area. In addition, many new studies
have come in after the cut-off date for the assessment and, as noted below, this research will need
to be included in the report.
Because of the volume and the complexity of the studies and reports for both endpoints,
additional summary tables are necessary in Chapter 5. All pertinent studies that are discussed in
this chapter should be introduced in a summary table at the start of each section for
developmental and reproductive toxicity. A good example is Table 5-13 which was prepared for
summarizing the structure-activity information.
Chapter 5 needs to include both positive and negative studies in the assessment. For
example, text for Table 5-13 says that "since no non-neoplastic lesions were seen, then they were
not included in the table." The report should include positive and negative data so the reviewer
can develop a comprehensive understanding of the data base supporting the assessment. Also, the
health assessment needs to include more current research. Every effort should be taken to extend
the time for data inclusion to as close as possible to the release date of the assessment. For
example, because of the cut-off date, several new dominant lethal studies were not included.
These 2 new negative studies need to be included in the reassessment. The BIBRA Study III
(BIBRA, 1998), which was a repeat of BIBRA I (BIBRA, 1995), was negative and needs to be
considered in this health assessment especially since the BIBRA I study was used in one of the
quantitative assessments (BIBRA III) (BIBRA, 1996; 1998). The written statement from
Christian (1998) identifying and criticizing these studies should be considered in the revision of
the document.
It was difficult for reviewers of this document to integrate the toxicological findings.
A much greater emphasis on integrating findings across the chapters is needed. For example,
information on toxicokinetics and metabolism needs to be fully integrated into this chapter. The
information on toxicokinetics and metabolism in the review document appear to be almost an
after thought. Interesting data on specific ovotoxic metabolites are available, and should be
incorporated. The potential impact of including the PBPK model in the assessment of
reproductive and developmental toxicity needs to be discussed. Does this help to explain
dose response differences across species for ovotoxicity? The Agency should explain this in
the assessment. It is important not only to show data but to discuss the implication of the
data. Another example where integration is needed is with the general toxicity chapter with
Chapter 5. Was there any evidence for neurotoxicity from general toxicity testing to suggest
the need for developmental neurotoxicity assessment? What about effects in utero on ovarian
or testicular development? The heritable translocation data from Chapter 3 should be integrated
with Chapters.
16
-------
Because of the potential interrelatedness of the reproductive tumors and reproductive
organ atrophy data, the reproductive chapter should have some overlap with the chapter
discussing tumor results for these reproductive organs. Also, Chapter 5 should include a
discussion on the mechanisms responsible for atrophy and reproductive tumors, and provide
discussion on the biological significance of ovarian and testicular atrophy. This is especially true
as these endpoints are modeled in Chapter 9. How are exposed rodents different than control
rodents? The Agency should explain its rationale for looking at ovarian atrophy in aged rodents.
The following issues should also be addressed in revising Chapter 5:
a) The reader needs to understand the rationale for modeling the dominant lethal
data. The Agency should provide discussion that supports the use of these
endpoints in Chapter 9. Since new studies on heritable translocations are available,
these findings should be integrated with the chapter on mutagenicity. This will
provide stronger evidence for the potential of this compound or its metabolites to
produce heritable and multi-generational impact. The EPA should integrate the
data so that the total picture of research supports its identification of critical
endpoints and quantitative modeling.
b) For both the reproductive and developmental endpoints, it is necessary to list all
assumptions, identify where the assessor is uncertain and identify agency action
taken to respond to that uncertainty. The Agency should explain what the critical
data needs are that will address this uncertainty and should be as specific and
explicit as possible.
c) The Agency should avoid the use of speculative and editorial types of sentences.
A specific example is on Page 5-28, lines 32-33; the assessment speculates that the
compound does not cross the blood-testis barrier. Why does the Agency think that
this is true? Is this consistent with the chemistry of the compound (i.e., solubility)?
Is it consistent with the postulated effects of this compound on spermatozoa?
In summary, Chapter 5 does not prepare the reader for the discussion in Chapter 9 on the
quantitative risk assessment for 1,3-butadiene. For example, how does the biology information
affect your choice of models? In Chapter 5, the Agency briefly makes a statement about
thresholds, but does not explain implications for quantitative risk assessment. How did the
Agency decide to drop the highest dose levels? Was this because of excessive toxicity? If so, the
EPA should explain the rationale. What criteria will the EPA apply to let the reader know when
the Agency will drop the results on the basis of the doses?
3.1.6 Chapter 6 - Toxicity in Animals
This chapter provides detailed information about subchronic, chronic and carcinogenicity
studies published from 1985 to present. Only three studies are mentioned in the subchronic
17
-------
section. It is unclear why other repeat dose studies are not reviewed. Have they all been
incorporated in other chapters? Repeat dose in vivo mammalian studies of 1,3 butadiene would
be appropriate for inclusion in this section unless it is explicitly stated that they are covered
elsewhere in the document. It is also important to insure that the data from in vivo repeat dose
studies get thorough review because the Inhalation Reference Concentration (RfC) value should
be derived on the most sensitive adverse endpoint that is meaningful for human health. The most
sensitive non-cancer biological effect of 1,3 butadiene currently cannot be determined from the
health risk assessment document, as comparative assessments are not made in any of the chapters.
In the chronic and carcinogenicity sections, the use of frequent text tables helps to simplify
and to provide clarity to the presentation. Since the National Toxicology Program (NTP) chronic
bioassay is a part of EPA's quantitative risk assessment, the detailed coverage is warranted. The
NTP study (NTP, 1993) contains both chronic toxicity and carcinogenicity data and the study is
discussed in both areas of EPA's draft. In order to make the chapter more understandable, the
Committee recommends that the Agency present the study details presented in the first section
(chronic) and then refer to them in the carcinogenicity section rather than repeat the material in
slightly different words. As it reads now, the review document gives the impression that they are
different studies. The Agency should also use the same categories of tumors for presenting data
from the continuous treatment, 9 and 15-month interim sacrifices and stop-exposure study in
order to permit comparisons. Most of the time this was done. However, for target lymphatic
tumors the categories are different, preventing a direct comparison.
A table summarizing the positive oncogenic findings across all studies specifying the dose
tested, and the type of tumor that was significantly elevated, should be added because of the
numerous organs and tumor types involved. Data from the rat oncogenicity study should be
included as this information is important and is only indirectly mentioned in the risk assessment.
For clarity, it would be preferable to separate the carcinogenic evidence on the mammalian
metabolites of 1,3 butadiene from the data on related chemicals and place it in a subsection by
itself. The section on related chemicals should be updated to include the results from studies on
styrene, isoprene and any other relevant chemicals.
The quality of the chapter's Discussion and Conclusion section is good. The observation
that concentration, not time is a critical determinant of potency is interesting but it is not
supported by a comparison of the tumor data presented in Table 6-4 and Table 6-8 from the NTP
study (NTP, 1993) for continuous lifetime treatment and stop-exposure study. It is rare that
carcinogenic data are available from continuous, interim sacrifice and stop-exposure studies.
Further analysis of a comparison of the data from continuous and lifetime treatment would be
interesting to see if there is supporting evidence for a biological model of dose and time that could
be used in the risk assessment.
18
-------
3.1.7 Chapter 7 - Epidemiologic Studies of Carcinogenicity
It is unclear why this chapter presented details of all the historical reports for a given
butadiene (BD) exposed group, since more recent reports supersede the previous ones by virtue
of having longer follow-up and more numerous deaths. Chapter 7 should be restructured so that
it is a detailed report on the latest follow-up of each epidemiologic study, with perhaps any
additional analyses from previous reports that were not duplicated in the most recent report.
More specific recommendations for Chapter 7 follow.
a) The case for an association of butadiene exposure with lymphosarcoma/
reticulosarcoma (ICD 200, a no-longer-used subset of non-Hodgkins lymphoma;
hereafter called just "lymphosarcoma") is based on two studies of butadiene
monomer plants which the EPA draft report indicated had excesses of
lymphosarcoma. However, many of the findings did not support, or called into
question, the purported association, specifically:
1) Texaco Study: The largest study group was most recently reported by
Divine and Hartman (1996) and consisted of 2,795 workers with an
average of 32 years of follow-up. The overall observed/expected ratio
(O/E) for lymphosarcoma was 9/4.7 = 1.91 (95% CI = 0.87-3.6) which
was not statistically significant. The SMRs for those employed <5, 5-19
and 20+ years were 2.6, 1.8 and 0.8 respectively, which is largely counter
to expectations. The lymphosarcomas were concentrated among those first
employed during World War II (WWII)(O/E = 7/2.9 = 2.4, 95%
Confidence Interval (CI) = 1.0-5.0). One feature that suggests it might be
a real effect is that the WW II excess was limited to those in the group with
jobs that entailed higher exposures ("varied exposure" group). However,
this elevated risk among WW II workers showed an inverse association
with length of employment, in spite of the fact that the authors indicate that
high exposures still continued into the 1950's and 1960's— which does not
lend plausibility to the association. Although there was an overall excess of
lymphosarcoma in the group with jobs that entailed higher exposures (O/E
= 7/2.8 = 2.5, CI = 1.0-5.1) in the Divine-Hartman study, the subset of this
group employed for 10+ years showed no excess (O/E = 1/1.0). Divine
and Hartman created an index of cumulative butadiene exposure based on a
job exposure matrix that considered job class and calendar time. Using a
time-dependent Cox regression model for cumulative exposure vs.
lymphosarcoma, there was not even suggestive evidence for an association
between exposure and risk (Relative Risk (RR) = 1.00, 95% CI= 0.97,
1.04). On Page 7-32, Line 6 (and also Page 11-6, Line 22), in order to
present a balanced summary of findings on lymphosarcoma and monomer
production, it should be stated here that there was no indication of an
exposure-response association, based on the latest follow-up by Divine and
19
-------
Hartman (1996). Two other methods of analysis also reinforced the null
association. Regression analyses that modeled the time spent in each of
their six job classes as predictors were also conducted; again, there was no
indication of an association. In summary, the case for an association
between butadiene exposure and lymphosarcoma in this study is weak.
The EPA report of this study either fails to note or at least to consider the
implications of a number of the findings noted above that go against the
likelihood of a causal association between butadiene and lymphosarcoma.
In the chapter summary (Pages 7-31 - 7-32) the Agency cites positive
results from earlier follow-up studies of this cohort and one nominally
positive result from this report as their summary of the study, without
noting the important findings in the study that are not supportive of a
positive association.
2) Union Carbide Study: Ward et al. (1996) studied 364 employees who
worked at one of three units that had produced butadiene. For
lymphosarcoma the ratio of O/E = 4/0.69 = 5.77 (95% CI = 1.6-14.8).
These four cases had worked at a butadiene unit for 0.8, 2.9, 3.3 and 8.0
years, so the worktime of three of the four was relatively brief. Two of the
four, with butadiene exposures of 2.9 years and 3.3 years, had worked in
an acetaldehyde unit for 8 years and 29 years, and all four had exposure to
a variety of other chemicals at the facilities. The fact that two cases had
long-time exposure to acetaldehyde raises the possibility that those cases
may have been associated with acetaldehyde rather than butadiene, and this
possible confounder weakens the finding. The nature and extent of the
possible confounding by acetaldehyde are glossed over in the EPA 1,3-
butadiene report.
3) Shell Study: A third study of 614 employees at a butadiene monomer
production plant yielded negative results (Cowles et al., 1994). They were
followed for an average of 12 years after entry to the study (which, because
of the unusual criteria for study entry, was equivalent to 14-17 years of
follow-up after first exposure) and had an average of 7.6 years of butadiene
exposure. There were no deaths from lymphatic or hematopoietic cancer
(1.2 expected). This fact was not even mentioned in the EPA report (Page
7-8). This cohort provides no support for an association between
butadiene exposure and lymphosarcoma, although the fairly short
follow-up time and relatively small sample size means the negative results
should not receive a heavy weight. Nevertheless, the EPA statement that
"this study failed to provide any negative evidence towards the causal
association" (Page 7-32, Lines 15-16; also Page 7-9, Line 22-23 and Page
11-7, Lines 6-8) seems to be an overstatement; the study does provide
20
-------
some, albeit not compelling, negative evidence. In fact, it is at least
equivalent in size to the Ward et al. (1996) study that is highlighted as
providing positive evidence.
4) Delzell Study: A large study by Delzell et al. (1996) of 15,649
styrene-butadiene rubber (SBR) workers found no association between
butadiene exposure and lymphosarcoma. Overall the lymphosarcoma O/E
= 11/13.8 = 0.8 (95% CI = 0.4-1.4). In those with the greatest exposure
and latency (10+ years exposure and 20+ years since hire) there was no
elevation in lymphosarcoma risk (O/E = 4/3.9 = 1.0, CI = 0.3-2.6). The
other SBR studies are essentially earlier versions of this study, with minor
differences in the cohorts, so this study summarizes the lymphosarcoma
results for the SBR data.
It is also of note that the coding of the ICD 200 category "lymphosarcoma
and reticulosarcoma," as distinct from the ICD 202 category of other
non-Hodgkins lymphomas (NHL) for which no butadiene-related excesses
were found in the various studies, is very unreliable. In a review of medical
records for death certificates coded as lymphomas (ICD 200 or 202),
Matanoski et al., 1993 found that two were not lymphomas at all "and the
other 10 were so poorly classified into the 200 and 202 codes on the death
certificates as compared with the hospital records that we combined these
ICD categories" (Page 369) (Matanoski etal, 1993). This calls into
question the significance of the positive results reported above which were
based on death certificate diagnoses of lymphosarcoma. Had the other
categories of NHL (i.e., ICD 202) been put with this category, it is unclear
whether there would be any excesses of NHL.
b) Delzell et al. (1996) conducted a study of 15,649 men who had worked for at least
one year (during 1943-1991) at any of eight styrene-butadiene rubber (SBR)
plants. The study included all but one small plant of the earlier Johns Hopkins
study (a plant that did not begin SBR production until about 1970) and updated
the mortality experience at all plants studied, along with conducting a much more
detailed exposure reconstruction for these workers (Matanoski et al., 1990). The
cohorts differed slightly because of somewhat different definitions of eligibility,
but, for all practical purposes, the Johns Hopkins cohort is subsumed by the
Delzell cohort. Thus, there is essentially only one study of SBR workers, of which
the Delzell et al. (1996) and Macaluso et al. (1996) reports represent the most
recent follow-up — for up to 49 years with a mean of 25 years.
c) The Committee did not feel it was appropriate to "lump" lymphohematopoietic
tumors. Leukemia and lymphosarcoma are separate diseases.
21
-------
d) The Criteria for Causal Inference are biased in their presentation. They should be
revised to reflect the above points. It was generally agreed that the process used
for SBR met the criteria, but the monomer did not.
Additional, more specific, comments are provided in Appendix A.:
3.1.8 Chapter 8 - Pharmacokinetic Modeling
There was consensus that this chapter is out of date. The chapter should be updated and
should include the most recently published PBPK models. A major data gap is the lack of
modeling of 3,4-epoxy-l,2-butanediol (EBD). A model describing the kinetics of this metabolite
would be useful in relating external exposure concentrations to measures of internal dosimetry
such as hemoglobin adducts. Such a linkage would facilitate reconstruction of BD exposure
profile in exposed humans. A majority of the members were of the opinion that since DEB is very
likely involved in BD-induced carcinogenesis, current models which describe this metabolite
would be adequate for risk assessment. Other members felt that other metabolites, such as the
diol, may play an important role in carcinogenesis at some sites, and that these metabolites needed
to be addressed.
The technical quality and comprehensiveness of this chapter could be improved greatly by
the inclusion of the most recent relevant literature on butadiene toxicokinetic modeling. Specific
suggestions regarding a reorganization of Chapter 8 are offered below. Chapter 8 of the draft
document on toxicokinetic modeling does not reflect the current state of knowledge regarding
PBPK models for butadiene. Rather, this chapter describes and critiques the initial attempts of
several laboratories to independently develop "first-generation" PBPK models for butadiene.
These models were developed often without benefit of critical data such as solubility parameters
and metabolic rate parameters. Additionally, definitive data on concentrations of butadiene and
its metabolites in tissues of animals exposed to butadiene were largely lacking, precluding
rigorous model validation. As such, some of the conclusions that the investigators drew from
these early models are no longer relevant. It would be more appropriate to briefly acknowledge
these early and important contributions. Then, attention could be focused on a serious critique of
the more recent PBPK models (Sweeney et al., 1996; Csanady et al., 1996; Reitz et al., 1996; and
Sweeney et al., 1997). These later models represent a significant advance over the previous
generation PBPK models in that they describe the kinetics of diepoxybutane, a critical metabolite
of 1,3-butadiene.
Inclusion of a discussion of state-of-the-art models for butadiene kinetics would
significantly strengthen the scientific quality of the document. In particular, the numerous
statements throughout the document indicating that an adequate PBPK model for butadiene risk
assessment is not available should be revised to reflect the current state-of-the-art. At least two
models that describe the kinetics of butadiene, epoxybutene and diepoxybutane were available to
the EPA in 1996 (Csanady et al., 1996; Reitz et al., 1996) and one was published in 1997
(Sweeney et al., 1997). A majority of the Committee felt that any one of these models, combined
22
-------
with available in vitro metabolic rate constants could be used to obtain more refined human dose
estimates while other Committee members were concerned about the degree to which some
models were able to describe the available data, the extent to which parameters had to be changed
from measured values to obtain reasonable fits and the accuracy of the low dose extrapolation.
As noted by the Agency, these refined dose estimates could be used in the animal based risk
assessments for both cancer and reproductive toxicology endpoints. In accord with the revised
EPA Cancer Risk Assessment Guidelines and given what is understood regarding the mechanisms
of butadiene toxicity, a more scientifically based risk estimate for environmental exposure to
butadiene is the most appropriate course of action. In chapter 8, concern is expressed that serious
uncertainties exist pertaining to the model structure, parameter values and validation for the
various PBPK models. It is true that each of the models differ somewhat in the details and that
the parameter values chosen by various investigators are similar but not identical. However, these
models are similar enough that any of the models is capable of predicting blood concentrations of
butadiene, epoxybutene, and diepoxybutane following inhalation exposure to butadiene. This
suggests that whatever differences in the underlying model structure exist, these differences are
minor. Most importantly, each of these second generation PBPK models fully describes the
kinetics of diepoxybutane. A majority of the Committee felt that since it is likely that it is
diepoxybutane that is the critical metabolite for initiation of the carcinogenic effects following
exposure to butadiene, inclusion of this metabolite makes these models especially useful for risk
assessment. An alternative view on the Committee was that other metabolites, such as the diol,
may play an important role and should be addressed in developing models predictive of human
risk. Also, the models of the diepoxide require validation and further development to address
parameter and structural uncertainly as well as interindividual variability in human metabolism.
The conclusions drawn in Chapter 8 need to be revisited in light of information on the
kinetics and PBPK modeling of butadiene that is not presented in the draft report. In the
conclusions, the Agency notes five areas in which more research is needed including:
a) evaluation of diepoxybutane kinetics;
b) investigation of the validity of in vitro metabolic data for extrapolating in vivo
exposure;
c) clarification of values of various physiological parameters;
d) better characterization of the distribution values for the human metabolic rates; and
e) more measurement of tissue concentrations of metabolites for model validation.
Comments on each of these areas are noted below:
a) Revision of the draft risk assessment to include data published since January 31,
1997 will resolve this need. A paper by Valentine et al. (1997) reports on the
23
-------
pharmacokinetics of diepoxybutane in rats after IV injection, the most appropriate
route of administration for this non-volatile chemical. Sweeney et al. (1997) were
able to successfully simulate this data using a PBPK model developed for
butadiene, epoxybutene, and diepoxybutane. Given the high mutagenic potency of
diepoxybutane, it is unlikely that kinetic measurements of this chemical will ever be
made in humans in vivo. However, parallel studies using lung and liver tissue
samples have quantitated the rates of hydrolysis and glutathione conjugation of
diepoxybutane in rats, mice and humans (Boogaard et al, 1996; Boogaard and
Bond, 1996) and in vitro studies on the oxidation of epoxybutene have
characterized the rate of diepoxybutane formation (Seaton et al., 1995).
b) The validity of the in vitro -in vivo extrapolation for butadiene has been
investigated in a manuscript by Sweeney et al. (1997), in which the in vitro data
describing the rates of butadiene oxidation, epoxybutene oxidation, and hydrolysis
and glutathione conjugation of epoxybutene and diepoxybutane determined in vitro
were used directly in a PBPK model and the model predictions were validated
against in vivo data obtained from butadiene exposure of mice and rats. The
ability of the model to simulate both epoxybutene and diepoxybutane blood and
tissue concentrations following exposure to butadiene points to the usefulness of in
vitro parameters in \in vivo models. For carcinogenic or potentially carcinogenic
chemicals it is unlikely that direct toxicokinetic measurements will ever be made in
people. Therefore, the only viable option is to conduct in vitro experiments in
which metabolic rates are determined in human and animal tissue samples and then
use these metabolic rates in the context of the PBPK model to predict blood and
tissue concentrations of both animal species and humans. Acceptance of this in
vitro -in vivo extrapolation strategy would serve to encourage the collection of
appropriate data with the ultimate goal of incorporating more mechanistic
information into the risk assessment process.
c) Regarding the choice of model parameters, there is undue concern that all
investigators have not used the same values for physiological parameters. All
investigators developing butadiene PBPK models have selected specific parameter
values from a distribution of widely accepted parameters values for various
physiological parameters in these models. Selection of a single value for
ventilation, perfusion, blood flow or organ volume is in itself a simplification of the
biological system being modeled. Many of these parameters vary for a single
individual throughout the day. All of these parameters vary among individuals.
The important point is that the selected values are physiologically realistic.
d) In the characterization of human metabolic rates, the statement "more research is
needed" appears to the reader to be gratuitous. As noted previously, Vmax and Km
values have been determined for a number of human samples. The fact is that the
data base for butadiene is particularly robust in terms of the number of samples
24
-------
characterized. If the EPA has concerns regarding the quantitative and qualitative
distribution of human metabolic rates, these concerns should be stated much more
explicitly and in sufficient detail that specific research could be conducted to meet
this need.
e) Finally, although more measurements of tissue concentrations of metabolites are
always desirable for model validation, more than 13 individual inhalation exposures
of mice and rats and monkeys to various concentrations of butadiene have been
conducted. Measurements at steady state and post exposure for butadiene and its
metabolites have been made in blood and tissues. These data have been
summarized by Himmelstein et al. (1997). The data are remarkably consistent
across laboratories and consistently point to dramatic species differences in
metabolism of butadiene. It is really not clear what additional data the Agency
might find useful for model validation.
Specific comments on Chapter 8 are displayed in Appendix A.
3.1.9 Chapter 9 - Quantitative Risk Assessment for 1,3-Butadiene
The Agency is to be commended for its efforts to develop a clear, well-reasoned
quantitative assessment of 1,3-butadiene cancer risk. The assessment is made particularly
difficult, by the continuing release of new and relevant scientific data, including toxicokinetic,
epidemiological and mechanistic data. Furthermore the analyses were undertaken after the release
and SAB review of the proposed EPA carcinogen guidelines but before they have been finalized.
The Agency is to be commended for looking at new approaches, such as the benchmark dose
procedures, to improve quantitative assessment on non-cancer endpoints. However, the
Committee has submitted suggestions on how to further improve these approaches and how to
make these new approaches more clear, accurate and consistent. Additional comments on
chapter 9 are given below.
a) Dose response analysis of the human carcinogenesis data
1) Quality of human data: The Committee agrees that the Agency's risk
assessment of 1,3-butadiene should include the latest follow-up of each of
the epidemiologic studies, including Delzell et al. (1995) and Matanoski et
al. (1997). Delzell fits a variety of dose response models, accommodating
a wide range of dose response curves. The models fit equally well, and as
noted by EPA as a whole were consistent with a considerable range of low
dose risk predictions. Thus the Delzell study does not permit one to
discriminate among models for the purpose of low dose prediction. In this
sense the data can be viewed as limited for the purpose of low dose risk
prediction. The reason why equivalent model fits are obtained requires
further explanation in order to convey a greater appreciation of the nature
25
-------
of the dose response data. For example, is this due to a relatively narrow
range of dose response data in the exposed cohort or to considerable
scatter?
The assessment notes that few subjects were exposed to benzene, but
benzene was not seen to confound the relationship between butadiene or
styrene and leukemia. Because benzene is widely recognized as a human
leukemiogen, further discussion of this point is needed (e.g., an indication
of whether the benzene exposures were insufficient to cause effects;
consistency with expectations based on a Chinese cohort).
In the section on uncertainties, the potential importance of exposure
misclassification in the Delzell study is discussed, but further discussion of
the uncertainties of dose reconstruction and the potential magnitude of
impact on the risk predictions is desirable.
2) Presentation of multiple model fits and low dose extrapolations: The
results of the multiple model fits derived by Delzell and colleagues are
tabulated, along with low dose risk predictions. While it is reasonable to
compare results to make the point that the data do not provide the basis for
discriminating among models of quite different shapes, once it is
established that this is the case the further presentation of low dose
analyses is not needed and unnecessarily complicates the presentation (e.g.,
in Table 9-3 and Table 9-4). Presentation of results for the linear model
would suffice, along with perhaps effect concentration estimates, within but
not outside the range of observation. The other point to be made, that the
exposure concentration at 1% risk level (EC01) results for the final versus
initial square root model differ by roughly 40 - 50% could also be made,
for example, in the text. It motivates the need for obtaining comparable
estimates for the linear model, by omitting styrene and race (as was done
for the "final" square root model).
The results for the square root model and the power model (which is
apparently best fit by a coefficient less than unity) are included.
Superlinear models such as these are not biologically plausible, and should
therefore not be used for low dose risk predictions. Synergy between
genotoxicity and cell proliferation frequently occurs as does saturation of
detoxification and metabolic activation. These factors result in sublinear
data sets. Metabolic data for butadiene clearly show saturation of
activation and the molecular dosimetry data are supralinear. An analogous
point was made by the Committee in commenting on proposed EPA
carcinogen guidelines regarding the use of the Weibull-in-dose model
without constraints for exponents less than unity for risk prediction. Lack
26
-------
of biological plausibility is another reason for not presenting low dose risk
predictions for the square root and power models.
The measure of deviance for the models seem to be within 0.4, while
normally one would say models differ if their difference in deviance is
about 4.0. Also, the best fit for the multiplicative model is sublinear; for
the power model it is superlinear and as noted by Delzell and colleagues,
the square root model fits best (although marginally so) - again, this
reinforces the notion that the data provide limited information on dose
response. Perhaps replacing the dose response figures with one figure
showing the data points with their confidence intervals, and superimposing
on the figure the various models fit would provide a better understanding
of why this is the case. The inclusion of data should clearly illustrate why
the data are equally explained by the wide range of models tried.
3) Dose rate, duration and timing: With benzene induced leukemia and
lymphoma the recency of exposure is an important feature of the dose
-time-response, with recent exposures most important for leukemia, and
distant, high exposure most important for lymphosarcoma. On Page 9-2,
the assessment notes the finding of Delzell and colleagues that excluding
exposures within 5-10 years of death slightly increased the exposure
response relationship but excluding exposures within the last 20 years
almost eliminated the relationship. It would be desirable for the Agency to
explore in detail the possible dose time relationships. On a related point,
Delzell and colleagues have recently reported on the possible importance of
peak exposures to 1,3-butadiene on leukemia risk (Delzell et a/., 1995). It
is important to explore the dose rate/time/response issues through careful
analyses of the Delzell data set.
4) Extrapolation of occupational results to general population: he available
data for risk assessment is for males, exposed as young and middle-aged
adults, and for a selected group in that they were workers in the jobs
studied. The extent to which this group can be used to represent the
general population should be addressed more carefully, with an attempt to
quantitatively address, where possible, differences between cancer potency
for the occupationally exposed and the general population. Such an
analysis should consider the following:
b) potential increased sensitivity for women and girls: No dose response information
appears to be available for women. The mammary gland was a sensitive site in
both rats and mice. Given the large public health concern for breast cancer in
women, and the findings in the bioassay, the issue should be explicitly addressed.
In terms of overall risk, female rats in the Hazelton Laboratories (Hazelton, 1981)
27
-------
study were clearly more sensitive than male rat, at least by an order of magnitude.
Although given the uncertainties in the quantitative analysis, this is less clear for
the mouse, cancer potency estimates were greater in the female compared to the
male by about a factor of 2. Although toxicokinetic differences are not apparent in
hemoglobin data for men and women with similar exposure, the potential for
increased susceptibility of mammary tissue cannot be excluded.
c) healthy worker effect: The extent to which a presumably small adjustment may be
appropriate should be addressed. The extent to which restrictions on smoking in
occupational settings with butadiene exposure may have contributed to the healthy
worker effect should be addressed by the Agency. Internal comparisons done by
Delzell et al. (1995) that were used to derive the human cancer potency estimate
are not influenced by the healthy worker effect since no external comparison
population is being used. However, the healthy worker effect is an important
consideration in extrapolating from the occupational setting to the general
population, and this could not be assessed through an internal comparison.
d) exposure during a working life versus other life stages: Issues falling under this
category to address include the potential for inherent increased susceptibilities at
different ages (e.g., in utero, and during infancy, childhood and old age), and those
related to the stages and mechanisms of carcinogenesis (e.g., time of exposure
versus observation; after a point, diminution of risk with time since exposure).
Analyses of the Delzell et al. (1995) and perhaps Matanowski et al. (1997) data,
and observations from the large Chinese benzene/leukemia cohort may shed light
on the second issue. The available data on butadiene obviously does not provide
direct information on the first issue, but data on leukemogenesis from other agents
may fill this gap.
e) other potentially susceptible subpopulations: The potential range of susceptibility
within the general population should be explored quantitatively, by for example,
considering the impact of polymorphisms on low dose risk, and addressing the
extent to which certain potentially susceptible groups are contained within the
Delzell cohort.
The document would be improved by presenting greater detail on the derivation of
lifetime risks from continuous exposure from the fits to the occupational data.
Perhaps this could be done in an appendix.
1) Choice of maximum likelihood estimate over Upper Confidence Limit
(UCL) on potency/LEG (95% lower confidence interval): Maximum
likelihood estimates (MLEs) rather than lower confidence bound on the
effective concentration (EC) (or upper confidence bound on potency) are
used in the final characterization of the dose response derived from the
28
-------
human data. It has been the practice of the Agency to use maximum
likelihood estimates in potency derivations from human data and upper
confidence bounds on derivations from animal data. The rationale
provided in the 1,3-butadiene document is that the simple linear model fit
to the human data does not have the instabilities that can be associated with
polynomial fits, and that there is far less uncertainty in the potency estimate
derived from human data. In this particular case there are a number of
unaddressed issues suggesting that, at least, the EC01 may be
underestimated for the general population. These include exposure
misclassification in the Delzell cohort, the lack of availability of data on all
but males exposed occupationally during adulthood, and the lack of
availability of a fit of a "final" linear model (comparable to the model for
the square root model [omitting styrene and race]). In addition, as
previously stated in this report, there is a possibility that there was an
overestimation due to exposures being underestimated in the Delzell et al.
(1995) study. It would be preferable to explicitly take these factors into
account, to the extent possible; it is unlikely that application of a
confidence bound will provide adequate correction.
2) Availability of Delzell study: Although the Delzell study has been published
(Delzell et al., 1996; Macaluso et al., 1996), some of the key analyses that
aid in interpreting the data are available only in an unpublished technical
report and thus is of limited availability to the public. At the meeting the
Committee heard that there was peer review of the unpublished technical
report. A report so critical to the dose response analysis and so heavily
cited in the document should be made widely available to the public, along
with description of the peer review conducted by the Agency.
f) Dose response analysis of animal cancer bioassay data:
1) Time dependent analyses: The default, time dependent analyses of the
cancer bioassay data were carefully done and are well presented.
Because the cancer incidence rates are large for some tumor sites, errors
are introduced if it is assumed the polynomial represents a true model of
multistage carcinogenesis, as Moolgavkar (1994) has pointed out. This
should be acknowledged in discussions which infer number of stages on the
basis of modeling results (e.g., Page 9-20). The errors may be relatively
small given the availability of data at relatively low doses; it would be
desirable to attempt to gain an understanding of the magnitude of the error
for these fits.
29
-------
The method used to develop the upper 95% confidence bound for the sum
of the incremental lifetime unit cancer risk for humans (q^ parameter
across sites is reasonable. An alternative approach for future assessments
would be to derive the distribution for the ql parameter for each tumor site
and use Monte Carlo simulation or numerical techniques to obtain the
distribution of the sum. The result should differ marginally from the one
presented, and thus it is not been suggested that the analysis be redone for
this case.
2) Lack of site concordance across species: While leukemia is the observed
endpoint in humans, leukemias are not observed in the animal bioassay in
either rats or mice, in either sex. The rationale for the default inferences
regarding site concordance, and the inclusion of sites such as the Harderian
gland and forestomach should be clearly stated. Ovarian tumors are
observed, along with ovarian atrophy in the animal bioassay. The degree to
which the toxicity may have played a role in certain sites observed in the
bioassay and the extent of relevance to humans exposed at lower doses
should be discussed.
3) Toxicokinetics: A variety of viewpoints were expressed within the
Committee over the extent to which toxicokinetic analyses should be
incorporated into the assessment for use in interspecies and high to low
dose extrapolation. The Agency was criticized by some for not
incorporating the results of the recent models. Others noted significant
deficiencies in the proposed models and the very recent developments
suggesting the field was undergoing rapid development, and found the
models speculative. Nonetheless, the different toxicokinetic hypotheses
and the hypothesized role of the various metabolites should be discussed,
and at least qualitatively, with an indication of the degree to which the
assessment would be impacted if some of the hypotheses were later proven
true. The discussion should acknowledge the variety of competingview-
points and hypotheses of researchers in the field. Where information on
toxicokinetics and mechanism provide adequate understanding of findings
at certain sites within one rodent and not the other, this should be given.
g) The assessment of the developmental and reproductive toxicity endpoints:
Improvement in non cancer risk assessment, such as use of effect levels as
suggested in the benchmark dose procedures, was supported by the Committee.
However, since these are relatively new procedures, the Agency must meet a high
standard of clarity and transparency in their initial applications as presented in this
this assessment. The Committee encourages the EPA to review the accuracy of
their calculations. The Committee also requests that additional discussion be
added to the document the to explain the use of EPA's newly proposed models,
30
-------
especially those modeling time to impact, and to provide additional explanation of
the safety factors applied to the benchmark calculations resulting from the
modeling exercises:
1) A clear rationale for the selection of one particular point of departure over
another is needed - for each case for which the default is applied and for
cases deviating from the default.
2) Where it is not obvious the biological significance of particular endpoints
for the assessment should be conveyed.
3) When applying novel approaches, such as the time-to-response modeling of
atrophy, the advantages and limitations in utilizing the approach in the
context of reference concentration estimation should be clearly understood
and presented.
4) The incorporation of information on toxicokinetics (e.g., ovotoxicity of
diepoxide) should be used where possible on at least a qualitative if not
quantitative basis to inform the choice of endpoints, modeling and
uncertainty factors.
5) The exclusion of points at high doses contributing to a poor fit is a
reasonable approach, and typically can be explained, for example, in terms
of toxicities other than the one being modeled, or toxicokinetics. When the
approach is applied where possible the specific rationale for the case at
hand should be presented. The use of the procedure and the rationale
provided for should be consistent across endpoints and across assessments
for the same scenarios.
6) The EHC did not come to a consensus about the propriety of using a risk
reduction factor, when reference concentrations are based on benchmarks
associated with risk of effect. There was agreement that if this factor is
used, the rationale for such use needs to be clearly stated.
7) Analyses of this type applied to these endpoints may be unfamiliar to others
working in the general area. Confusion over the presentation of some
results (e.g., logarithm versus linear scales) was obvious in some of the
comments received. Also, calculations should be easily followed. Some of
the Committee noted, as did several consultants presenting comments to
the Committee, that mathematical errors were evident. EPA should
carefully review this section of the document. Tables and figures should be
free standing, with statistics referenced.
31
-------
Additional comments and recommendations for Chapter 9 appear in Appendix A.
3.1.10 Chapter 10 - Weight of Evidence
The Committee did not reach a unanimous opinion about whether one can conclude that
1,3- butadiene is a known human carcinogen. There was a consensus that there is sufficient
evidence to say that working in synthetic rubber production is causally associated with leukemia.
The majority felt that there were conflicting results among SBR and monomer workers for
leukemia and a lack of compelling evidence for a relationship between lymphosarcoma and
butadiene exposure. There was concern that an increased risk of leukemia was not seen in the
epidemiology studies of workers in the butadiene monomer industry. To most Committee
members, this lack of association significantly reduced confidence in the assumption that
1,3-butadiene was the causative agent for leukemia in the SBR industry study. In addition,
confounding or coexposure to other chemicals could not be ruled out with confidence.
There were a variety of views amongst the EHC as to how the call should be made on
whether a substance can be considered a "known human carcinogen." One view was that the
human data - from observational studies of cancer in humans - must stand on its own to make the
finding of "known human carcinogen," without regard to mechanistic or other information. A
second view was that the cumulative evidence, from human and animal studies, as well as
mechanistic data, particularly as it relates to human findings, should be used as the basis for the
judgment. The majority of the Committee felt that the judgment should be made on the basis of
human cancer observations alone and the evidence was not sufficient for 1,3-butadiene. A few of
the Committee members considered the human evidence in and of itself sufficient or that the
cumulative evidence was sufficient to make a finding of "known human carcinogen." Other
Committee members considered the body of mechanistic data to be indicative of the fact that
significant interspecies differences in response to 1,3-butadiene exist between rodents and
humans. For these members, the mechanistic data were consistent with 1,3-butadiene not being
classified as a known human carcinogen.
With respect to the narrative discussion in the evaluation, the Committee felt that it should
more reflect the range of opinion on the matter when discussing the human findings. Clearly
1,3-butadiene, when mutagenic or clastogenic, is so through its metabolism. This finding should
be provided in the evaluation, as well as a statement regarding the mechanistic studies most
relevant to humans.
The weight of evidence is confined to addressing cancer endpoints. The reproductive
endpoint is the basis for RfC calculations and, therefore, also should be addressed.
3.1.11 Chapter 11 - Risk Characterization
The majority of the Environmental Health Committee did not consider 1,3-butadiene to be
a known human carcinogen due to the lack of consistency between exposure and leukemia or
32
-------
lymphosarcoma data when the styrene-butadiene rubber (SBR) and monomer worker studies were
considered in total. The majority opined that 1,3-butadiene should be classified as a probable
human carcinogen whereas a minority felt that the science supported the classification of
1,3-butadiene as a known human carcinogen. Most of the Committee felt that occupational
exposure to the SBR process was a known cause of cancer.
The majority of the EHC was reluctant to classify 1,3-butadiene as a known human
carcinogen because it was felt that there was no consistent relationship between exposure and
leukemia or lymphosarcoma when the SBR and monomer studies were considered in total. Only
one study population (SBR industry) had credible leukemia excess related to exposure. Leukemia
was not elevated or related to estimated exposure in butadiene monomer industry.
Lymphosarcoma was only elevated in short term, not long term workers in the monomer industry
and also not elevated in the SBR industry. Hence, there was no dose response for the
lymphosarcoma. The lymphopoietic cancers should be considered separately when assessing
consistency across studies. The majority of the Committee felt that the finding of "known human
carcinogen" for the SBR process could be based solely on observational studies in humans,
without regard to mechanistic or other information. A minority of the Committee held that
1,3-butadiene should be identified as a "known human carcinogen" because the cumulative
evidence, from epidemiology, animal cancer bioassays, and mechanistic studies should be used as
the basis for the judgment.
Finally, the Committee was unable to follow the logic in the use of the "risk reduction
factor" that was applied to the benchmark dose based-RfC. This section of the EPA document
should be rewritten to make the adjustment process clear.
Futher details and recommendations will be found in Appendix A.
3.2 Classification of 1,3-Butadiene as a Known Human Carcinogen
There was a majority opinion within the Committee that the extant science supports the
SBR process exposures as a known human carcinogen, but does not support such a finding for
exposures to butadiene monomer. The lack of positive findings regarding lymphosarcoma in the
large Delzell cohort and the lack of positive findings regarding leukemia in the three butadiene
monomer studies weaken the case for a straightforward causal association between these
endpoints and 1,3-butadiene exposure. The EPA report suggests these discrepancies may be due
to: a) butadiene "dose rate,"; b) different confounding factors; or c) different "co/modifying"
factors (Page 11-7). The first reason seems very implausible. There is no known example of a
chemical or substance that causes different cancers depending on dose rate, especially at the
relatively low dose rates in these studies (unlike the rodent studies with hundreds of ppm). The
second reason raises the possibility that one or both of the putative associations may not be real
but may be caused by confounding by some unknown factor(s). This reason weakens the case for
causality. The third reason, that there may be some cofactor or modifying factor, suggests that
even though butadiene might induce cancer in the presence of some cofactor or modifying factor,
33
-------
it would not do so in the absence of that factor. If this is the case, then it would be inappropriate
to generalize to other exposure scenarios for the general public since they would likely not have
concomitant exposure to the cofactor. There is some evidence to support this later point in the
studies of Leavens et al. (1997) in which mice were exposed to mixtures of 1,3-butadiene and
styrene as well as 1,3-butadiene or styrene alone. Genotoxicity was noted in mice exposed to
1,3-butadiene; cytotoxicity was noted in mice exposed to styrene; both cytotoxicity and
genotoxicity were noted in mice exposed to the mixture. This study provides some mechanistic
support for the observation of leukemia in SBR workers but not in butadiene monomer workers
as it is consistent with biologically-based models that demonstrate the requirement for both
cytotoxic and genotoxic events for the development of cancer. In short, this inconsistency in
findings, whatever its basis, weakens the case for causal associations and/or that such associations
can be generalized to exposure of the public to butadiene.
The case for an association between lymphosarcoma and butadiene exposure is weakened
by the fact that in the main study (Divine et al., 1996) that purportedly showed the association,
there was no indication of an exposure-response relationship nor was there evidence that those
with longer-term exposure had a higher risk of lymphosarcoma.
The report claims (on page 11-3, lines 11-13) that '"sufficient evidence' of human
carcinogen!city is based on more than 10 epidemiological studies examining five different groups
of workers" and summarizes them in Table 11-1. But there are effectively only 4 groups of
workers, not 5, since the Matanoski and Delzell cohorts have a high degree of overlap (i.e., about
95% of the Matanoski cohort is included in the Delzell cohort). In addition, the "more than 10"
studies are just earlier reports of the same cohorts and do not add anything to an inference of
causality beyond that seen in the most recent follow-ups. Of the four independent studies, one
small one (Cowles et al., 1994) is completely negative for lymphosarcoma and leukemia; another
small one was positive for lymphosarcoma, but based on only 4 cases, and showed no excess of
leukemia (Ward et al., 1996); a relatively large study was mostly negative, but somewhat
suggestive of a lymphosarcoma excess, and provided no support for a leukemia excess (Divine et
al., 1996); and one large study was positive for leukemia but not for lymphosarcoma (Delzell et
al., 1996).
In summary, the weight of epidemiological evidence does not support an association
between butadiene exposure and lymphosarcoma/reticulosarcoma. While the Delzell et al. (1996)
and Macaluso et al. (1996) are large and methodologically sound studies, one would like to see at
least a second independent confirmatory study before affirming there is "sufficient evidence of
human carcinogen!city" regarding butadiene and leukemia. Instead, one sees a fairly large and
reasonably sound study that shows no leukemia excess (Divine and Hartman, 1996) plus two
smaller ones with no evidence of leukemia risk, and these weaken the case.
A majority of the EHC members felt that the body of mechanistic data on butadiene does
not support the classification of known human carcinogen. The Agency notes (page 9-52) that
there are large unexplained differences in the response of rats and mice to butadiene and states
34
-------
that the specific mechanisms of 1,3-butadiene induced carcinogenesis are unknown. However,
there is strong evidence from both metabolism and genetic toxicology studies that diepoxybutane
is a critical metabolite in the carcinogenic process. The extensive capability of mice to form the
diepoxide metabolite and the extensively higher levels of the diepoxide metabolite in blood and
tissues of mice compared with rats, most likely forms the basis for this dramatic species difference
in carcinogenic response. This is an important point because it suggests that the diepoxide
metabolite may be the best dosimeter for assessing risks for humans exposed to butadiene.
Toxicokinetic modeling cannot prove or disprove the relationship between chemical
exposure and toxic or carcinogenic effect. However, these models can be developed to test
quantitative hypotheses regarding proposed mechanisms of toxicity or carcinogenicity of
chemicals. The Agency notes that "pharmacokinetic modeling of 1,3-butadiene has not elucidated
the reasons for interspecies differences in carcinogenic response between rats and mice" and "mice
and rats also exhibit substantial quantitative differences in their metabolism of 1,3-butadiene to
potentially reactive metabolites. Unfortunately, existing pharmacokinetic models have been
unable to explain the species differences in carcinogenic response." These statements are not true
if one considers the current PBPK models that describe the disposition of butadiene, epoxybutene
and diepoxybutane. These PBPK models clearly simulate the dramatic species differences in
tissue and blood concentrations of diepoxybutane between rats and mice observed in vivo. The
models suggest that species differences in response to butadiene are most likely related to
differences in rates of formation and removal of diepoxybutane. Experimental data and PBPK
model simulations indicate that mice produce far greater concentrations of this reactive metabolite
compared with rats.
These PBPK models can be, and have been, extended to humans. In vitro data on rates of
butadiene, epoxybutene, and diepoxybutane metabolism obtained in human tissue samples can
also be used to predict blood and tissue concentrations of both epoxide metabolites in humans.
When these simulations are conducted using average values for human metabolic rates results
indicate that human diepoxide concentrations would be orders of magnitude less than those of
mice and lower but much more similar to concentrations predicted for rats. Nonetheless,
1,3-butadiene produces cancer in rats, albeit at higher dose levels than for mice, so similarity of
human and rat metabolism would still contribute to the weight of evidence. Some Committee
Members pointed out limitations in the models and the underlying assumptions regarding choice
of human parameters and activity of metabolites. These Members also point to the considerable
other mechanistic data (which are described in comments on Chapter 4).
3.3 Approaches Taken to Characterize Plausible Cancer Risks
The database for butadiene is very challenging and robust, but all of the important issues
are not settled. The EPA must state the facts in proper perspective and should not be afraid to
say that more information is needed. The Agency must clearly address known issues, search for
(currently) unknown issues, and separate the important data from the unimportant.
35
-------
Risk estimates use external butadiene exposures as the dose estimate given the
sophisticated PBPK models currently developed and the availability of in vitro metabolic
parameters in rodents and humans an approach that used an estimate of internal dose . The
Agency notes (Page 11-12, Lines 1-3) that a review of the available pharmacokinetic data and
models reveal that the state of this science is currently inadequate for either explaining
interspecies differences or improving on default dosimetry assumptions. As noted previously in
this report, this statement should be revised to reflect the available scientific data and
pharmacokinetic models. The fact that human variability in metabolic response is noted should
not be viewed as a limitation for these models, but instead as an opportunity to reduce uncertainty
and characterize variability. If metabolic activation of butadiene is necessary to produce genetic
damage, and ultimately carcinogenicity, which is not an unreasonable hypothesis, then individuals
with the highest metabolic capacity for activation and the lowest metabolic capacity for
deactivation would be at most risk for exposure to butadiene. The PBPK model could be used to
determine the extent to which metabolic differences among humans is the dominant process
controlling the magnitude of the effective dose at the target site.
The Agency states (page 9-16, lines 1-7) that risk assessments based on rat carcinogenicity
data "are not considered the most appropriate estimates of human risks .... EPA believes that the
mouse is likely to represent a better rodent model for human cancer risk assessment from
1,3-butadiene." By contrast, the EHC believes that available mechanistic data on the formation of
epoxybutene and diepoxybutane obtained in rat, mouse and human tissue samples suggest that the
rat is a more appropriate model for assessing risks for humans than is the mouse. Choosing a rat
versus a mouse as the most appropriate animal for assessing risks in humans may, in fact, be an
oversimplification. As noted above, the development of PBPK models that are capable of
predicting concentrations of the reactive metabolites in the target tissues and availability of
distributions of human metabolic rate constants would allow one to not rely on either mouse or
rat per se. Instead one can use all of the available mechanistic data within the context of a PBPK
model to predict the butadiene doses in humans necessary to yield epoxide concentrations in
tissues similar to those predicted for rats or mice exposed to carcinogenic concentrations of
butadiene.
The Agency notes (page 9-17, lines 6-11) that no attempt was made to adjust for internal
doses of reactive 1,3-butadiene metabolites because the PBPK data were inadequate to develop a
reliable PBPK model. As noted previously PBPK models not reviewed by the Agency in this draft
document have been developed that are capable of predicting the butadiene epoxybutene and
diepoxybutane blood and tissue concentrations in rats, mice and humans following exposure to
butadiene. In this case, a human PBPK model could be readily used to obtain either point
estimates using average human values or a range of estimates by conducting Monte Carlo
simulations to sample from a distribution of available parameter values for humans for both
metabolic rate parameters and physiological parameters. A plausible assumption is that for a
given diepoxybutane tissue concentration the human response would be equivalent to the rodent
response. Then, concentrations of butadiene necessary to elicit tissue doses of diepoxybutane in
humans equivalent to diepoxybutane concentrations in mice or rats at doses which yielded tumors
36
-------
could be calculated. PBPK models of these and the diepoxybutane and other active metabolites
can then be applied to human risks predictions. The advantage of using a PBPK model is that
mechanisms underlying absorption, distribution, metabolism and elimination of butadiene,
epoxybutene, epoxybutanediol and diepoxybutane which are either similar or different across
species and across doses would be accounted for. Given the availability of these
PBPK models, this approach would be far superior to using a simple arithmetic adjustment for
continuous daily exposure versus exposure under bioassay conditions.
On page 9-25, lines 1-7 the EPA notes that "mice and rats also exhibit substantial
quantitative differences in their metabolism of 1,3-butadiene to potentially reactive metabolites.
Unfortunately, existing pharmacokinetic models have been unable to explain the species
differences in carcinogenic response." A majority of the Committee assert that there are several
published PBPK models that have been able to successfully explain species differences in the
carcinogenic response exposure to 1,3-butadiene based on differences in the metabolism of
1,3-butadiene to reactive metabolites while other Committee members were concerned about the
degree to which some models were able to describe the available data, the extent to which
parameters had to be changed from measured values to obtain reasonable fits and the accuracy of
the low dose extrapolation. In particular, the highly mutagenic diepoxybutane is formed to a
much greater extent in mouse tissues compared with rats.
Again on page 9-25, lines 15-19, the EPA notes that "ideally a PBPK model for the
internal dose of the reactive metabolites would decrease some of the quantitative uncertainty in
interspecies extrapolation. However, current PBPK models are inadequate for this purpose." The
available pharmacokinetic models and emerging data on human pharmacokinetics should be
further explored for use in risk assessment. As noted in the previous comments, some of the
Committee members assert that PBPK models that are capable of interspecies and low dose
extrapolation for this purpose do exist. Other Committee members assert that although the PBPK
models require validation and development, they may prove useful in the near term. In either
case, models should be further developed to address parameter and structural uncertainty as well
as interindividual variability in human metabolism.
On page 9-51 the Agency asserts that NOAEL, LOAEL, EC 10, or LEC10 should be
converted to appropriate human equivalent exposures before using these exposure levels as points
of departure and that theoretically this is best accomplished using a PBPK model. The Agency
also notes that the current PBPK models are inadequate for use in risk assessment. However, as
noted previously, some of the Members felt that there are several available PBPK models that are
capable of predicting diepoxide concentrations in target tissues for rats, mice and humans and that
any one of these models could be used to obtain a more appropriate human equivalent exposure.
Other Committee members felt that, at a minimum, the different toxicokinetic hypotheses and the
hypothesized role of the various metabolites should be discussed, at least qualitatively, with an
indication of the degree to which the assessment would be impacted if some of the hypotheses
were later proven true. In the 1,3-butadiene health risk assessment, the Agency should address
the variety of viewpoints and hypotheses of research in the field of PBPK modelling.
37
-------
The EPA report pays relatively little attention to the issue of how much peak exposures to
butadiene may have influenced leukemia risk in the Delzell study. It indicates (page. 7-22, lines
30-33) that there was an association of peak exposures (defined as >100 ppm) with leukemia but
dismisses it as an "irregular" association. The Acquavella (1998) commentary on the EPA draft
that was sent to the EHC presents a table showing how much peak exposures affect the
association of butadiene ppm-years with leukemia risk (Page 10). The formal linear regression
estimates with and without adjustment for peak exposures are not presented, but it is apparent
that the regression estimate would be appreciably less when peak exposure was adjusted for.
Specifically, for the three highest cumulative exposure groups, the excess relative risks with and
without control for peak exposures were 0.0 and 1.0 respectively for 20-99 ppm-years, 0.3 and
1.4 respectively for 100-199 ppm-years, and 1.5 and 3.6 respectively for 200+ ppm-years. In
each dose group, adjustment for peak exposures reduced the leukemia risk substantially. Since
butadiene exposures to the public will almost never approach the peak exposure range, a more
appropriate model for risk would factor out the peak-exposure component.
Regarding the Delzell analysis of butadiene exposure vs. leukemia, (on Page 9-2, Lines
24-25) it is noted that "excluding exposures within 20 years of death weakened and almost
eliminated the relationship...." This indicates that in modeling lifetime risk, a model that assumes
a limited effect time (i.e., that leukemia risk during a given year of age is affected largely by the
butadiene exposures received during the previous, say, 20 years, and only slightly or not at all by
more distant ones) should be considered. This "windows of exposure" model has precedents,
e.g., lung cancer risk from radon has been modeled in this way in a National Academy of Sciences
report (NAS, 1988) because lung cancer risk was little affected by radon exposures in the distant
past; leukemia risk from radiation is highly elevated at 5-10 years after irradiation but there is little
elevation by 20-30 years after irradiation (NAS, 1990). If this model were considered for
projecting lifetime risk, it would show appreciably less risk from chronic exposures than does the
present one, which assumes that excess relative risk at, say, age 70 is an additive function of all
the exposure accumulated in the previous 69 years.
3.4 Conclusions and Quantitative Estimations for Reproductive/Developmental Effects
The Committee supports the Agency's use of benchmark dose procedures and the
modeling of reproductive toxicity endpoints. The EHC also supports the continuing attempts to
develop new strategies, such as those presented in the report to address ovarian, uterine and
testicular atrophy, to quantitatively address reproductive endpoints in risk assessment documents.
This is one of the first such assessments and some general suggestions are made regarding
conduct of the analyses and the presentation of results for the analyses:
a) A clear rationale for the selection of one particular point of departure over another
is needed - for each case for which the default is applied and for cases deviating
from the default.
38
-------
b) Where it is not obvious the biological significance of particular endpoints for the
assessment should be conveyed.
c) When applying novel approaches, such as the time-to-response modeling of
atrophy, the advantages and limitations in utilizing the approach in the context of
reference concentration estimation should be clearly understood and presented.
d) The incorporation of information on toxicokinetics (e.g., ovotoxicity of diepoxide)
should be used where possible on at least a qualitative if not quantitative basis to
inform the choice of endpoints and modeling.
e) The exclusion of points at high doses contributing to a poor fit is a reasonable
approach, and typically can be explained, for example, in terms of toxicities other
than the one being modeled, or toxicokinetics. When the approach is applied
where possible the specific rationale for the case at hand should be presented. The
use of the procedure and the rationale provided for should be consistent across
endpoints and across assessments for the same scenarios.
f) The Committee did not agree about the propriety of using a risk reduction factor,
when reference concentrations are based on benchmarks associated with risk of
effect. The rationale for use of such a factor needs to be clearly laid out.
g) Analyses of this type applied to these endpoints may be unfamiliar to others
working in the general area. Confusion over the presentation of some results (e.g.,
logarithmic versus linear scales) was obvious in some of the comments received.
Also, calculations should be easily followed, and of course need to be carefully
proofed by someone other than the one making them. Tables and figures should
be free standing, with statistics referenced.
Some of the Committee members are of the opinion that the quantitative estimations for
reproductive/developmental effects could benefit greatly from the application of PBPK modeling
to estimate the effective dose at the target site. The Agency notes (Page 9-46, lines 6-11) that
"ovarian atrophy has been shown to be related to the amount of the diepoxide metabolite in the
tissue. Modeling of the ovarian atrophy and uterine atrophy data was considered based on
internal dose of the diepoxide metabolite, however an adequate model was not available to
estimate levels of the diepoxybutane." There are several PBPK models that are currently published
that allow calculation of the dose of the epoxide metabolite in target tissue. Any one of these
models could be used to determine the internal dose of the diepoxide metabolite and provide a
more refined estimate of risks for this reproductive endpoint.
39
-------
4. SUMMARY OF RECOMMENDATIONS
The Committee's principal findings and recomendations are:
a) The EHC recommends that the Agency's updated, draft health risk assessment of
1,3-butadiene reflect the new information that has been published in the
peer-reviewed literature since the Agency's cut-off date of January 31, 1997. The
inclusion of this new information would greatly improve the technical quality and
comprehensiveness of the report.
b) Since critical research like the Delzell et al. (1995) exposure reestimation and the
pharmacokinetic modeling has not yet been completed, the Committee
recommends that the Agency's risk assessment of 1,3-butadiene be labeled as an
interim hazard assessment (The document could then be called "Health Risk
Asssessmment of 1,3-Butadiene: 1. Hazard Assessment").
c) The majority of the Committee recommends that the Agency classify 1,3-butadiene
as a "probable human carcinogen" (rather than a "known human carcinogen")
because there was not consistency between the exposure-response rates for
leukemia or lymphosarcoma when both the SBR and monomer worker studies
were considered in total. Also, the role of confounders and cofactors was unclear.
d) The Committee found the approaches taken to characterize plausible cancer risks
to be reasonable but points out specific improvements that can be made.
e) The Committee supports the Agency's use of benchmark dose procedures and the
development of mathematical models for reproductive endpoints. To improve the
analysis and the clarity of the results, the EHC offered specific recommendations
focused on:
(1) providing a clear rationale for selecting one particular point of departure
over another;
(2) conveying the biological significance of particular endpoints of the
assessment;
(3) presenting the advantages and limitations of novel approaches such as the
time-to-response modeling of atrophy;
(4) incorporating information on toxicokinetics on at least a qualitative, if not
quantitative basis to inform the choice of endpoints;
40
-------
(5) excluding points at high doses that contribute to a poor fit;
(6) explaining the rationale for using a risk reduction factor; and
(7) proofing calculations and correcting them where needed.
f) In Chapter 1, Introduction, the Agency should state whether the different cancer
classification systems and quantitative assessments are equally valid and
scientifically defensible, and explain the rationale for these judgements. In
addition, the Agency should comment on its Cancer Risk Assessment Guidelines.
g) In Chapter 2, Overview of Exposure to 1,3-Butadiene, the Agency should clearly
indicate that the chapter is not intended to be a comprehensive review of exposure.
The chapter should include an explanation on how concentrations have been
measured over the years and how this might affect a comparison of exposure levels
measured in earlier years with those from more recent years. Specific
recommendations for improvements for chapter 2 are included in Section 3.1.2.
h) There are several recommendations regarding Chapter 3, Metabolism and
Pharmacokinetics. The main recommendations include the following with
additional comments included in Section 3.1.3.
(1) The title of Chapter 3, Metabolism and Pharmacokinetics, should be
changed to Metabolism and Toxicokinetics because 1,3-butadiene is not
used as a therapeutic agent.
(2) The word, toxicokinetics, should replace the word, pharmacokinetics,
throughout the document.
(3) Many new studies are not incorporated into the chapter. The Health
Canada assessment and the comments that were submitted by Dr.
Himmelstein provide many of those new references (Himmelstein, 1998).
(4) The Agency should revisit its statement regarding species differences in
butadiene metabolism, taking into account the most recent information on
species differences in the production of diepoxybutane and other reactive
metabolites.
(5) The chapter should point out the close parallels between the observations
in \in vitro studies on metabolism and tissue concentrations of epoxybutene
and diepoxybutane in mice and rats and in vivo studies on the metabolism
of butadiene.
41
-------
i) The Committee's recommendations for Chapter 4, Mutagenicity, are numerous.
Some of the recommendations are given below and additional recommendations
are included in Section 3.1.4:
(1) The statement on Page 11-1 regarding the lack of sufficient data to
determine if children or other subpopulations are afffected differently by
exposure to 1,3-butadiene should be revisited in light of the studies by
Nelson et al. (1995), Wiencke et al. (1995) and Kelsey et al. (1995).
Nelson et al. (1995) found the glutathione transerase theta (GSTT1) to be
highly polymorphic, due to wide variation in its ethnic distribution.
(2) Most of the extensive work on mutagenicity prior to 1994 should be
included.
(3) The chapter should include text tables that summarize key animal and
human findings derived from the entire body of information on butadiene.
(4) The tables should include references to support key findings.
(5) There should be separate tables for in vitro, animal and human findings.
(6) The similarities and species differences in response should be noted.
(7) The chapter should include more emphasis on the positive heritable
translocation studies in mice because of their potential relevance for human
heritable risks, and to several additional studies conducted in humans.
(8) The conclusion section should be expanded to include what is known about
the mutagenicity of 1,3-butadiene and its metabolites.
(9) The missing dose units, missing units for mutant frequencies and other
similar omissions should be added.
(10) The explanation as to possible reasons for the discordance between the
positive effects obtained by autoradiography and the negative results found
by cloning for human 1,3-butadiene induced hprt mutations in vivo may be
incorrect and should be reconsidered.
(11) The conclusion section should be expanded by adding statements
summarizing what is known about mutagenicity of 1,3-butadiene and its
metabolites.
42
-------
j) Some of the recommendations for Chapter 5, Reproductive and Developmental Effects,
are included in recommendations cited in (e). Additional recommendations are provided
in detail in Section 3.1.5 and are summarized below.
(1) All of the pertinent studies should be introduced in a summary table at the
beginning of each section.
(2) Both positive and negative studies should be included so the reader can
develop a comprehensive understanding of the data base supporting the
assessment.
(3) More current research should be included, especially the new dominant
lethal studies.
(4) Chapter 5 should be integrated with the other chapters, especially the
chapters on pharmacokinetics and metabolism, animal toxicity, and
quantitative risk assessment.
(5) The Agency should identify where it is uncertain with regard to conclusions
about the reproductive and developmental endpoints, identify the action it
will take to respond to the uncertainly, and should include all of the
assumptions regarding uncertainty in the respective chapter.
k) The Committee's recommendations on Chapter 6, Toxicity in Animals, are listed in
Section 3.1.6 and include the following:
(1) The rationale for the selection of the toxic non-cancer endpoint that is
utilized in the derivation of the RfC is very important and should be more
explicitly explained.
(2) It is unclear whether all of the repeat dose studies have been reviewed by
the Agency. The EPA should incorporate the repeat dose in vivo
mammalian studies of 1,3-butadiene in Chapter 6 unless these are covered
elsewhere in the document and the Agency explicitedly so states.
(3) The Agency should present the NTP study details in the first section
(chronic) of Chapter 6 and then refer to them in the carcinogenicity section.
(4) The EPA should use the same categories of tumors for presenting data
from the continuous treatment, 9 and 15-month interim sacrifices and stop-
exposure study in order to permit comparisons.
43
-------
(5) A table summarizing the positive oncogenic findings across all studies
specifying the dose tested, and the type of tumor that was significantly
elevated, should be added since there were numerous organs and tumors
involved.
(6) The carcinogenic evidence on the mammalian metabolites of 1,3- butadiene
should be separated from the data on related chemicals and placed in a
subsection by itself.
(7) The rat carcinogenicity data should be presented in similar detail to the
NTP data so that the reader does not have to find the old risk assessment
to see the data.
(8) The observation that concentration, not time is a critical determinant of
potency is not supported by a comparison of the tumor data presented in
Table 6-4 and Table 6-8 from the NTP (1993) study for continuous lifetime
treatment and stop-exposure study.
1) The Committee recommendations for Chapter 7 include:
(1) A statement that, for the Delzell et al. (1996) study, there was no excess
among those hired before 1950 (Observed/Expected = 17/16.4 = 1.04)
when one would expect the highest exposures, but there was an excess
among those hired during 1950-59 (Observed/Expected = 20/10=2.0,
Confidence Interval = 1.2-3.1)
(2) A statement regarding the inappropriateness of "lumping" the
lymphohematopoietic tumors should be added.
(3) A statement regarding the possible role of confounding should be included
in the document.
m) The recommendations for Chapter 8 are provided in Section 3.1.8 and make the
following revisions:
(1) The most recent relevant literature on butadiene toxicokinetic modeling
should be included.
(2) The Agency should revisit its conclusion in Chapter 8 once it includes the
recent PBPK models as explained in Section 3.1.8.
44
-------
n) The recommendations for Chapter 9 are provided in Section 3.1.9 and include the
following:
(1) The data of Matanowski et al. (1997) should be considered,
2) The Agency should explain or provide further explanation on the following:
(i) why equivalent model fits for the Delzell data are obtained in order
to convey a greater appreciation of the nature of the dose response
data.
(ii) the role of benzene as a confounder.
(iii) the potential importance of the uncertainties of dose reconstruction
and the potential magnitude of the impact on the risk predictions,
(iv) where possible, the rationale for the exclusion of points at high
doses contributing to a poor fit.
(v) information on toxicokinetics (e.g., ovotoxicity of diepoxide).
o) Chapter 10, Weight of Evidence, should be rewritten to reflect the range of
opinion regarding the human findings. In addition, the finding that 1,3-butadiene,
when mutagenic or clastogenic, is so through its metabolism should be provided in
both the evaluation and in the statement regarding the mechanistic studies that are
most relevant to humans.
p) Specific recommendations for Chapter 11, Risk Characterization, are provided in
Section 3.1.11 and include the following:
(1) The statement that the conclusion of "sufficient evidence" of human
carcinogen!city is based on more than 10 epidemiologic studies examining
five different groups of workers should be rewritten since it is misleading.
(2) In discussing the excess leukemia risk in the nested case-control study in
the Matanoski et al. (1990) study, it is important to indicate that there was
no excess of leukemia observed in the cohort study.
(3) The chapter should clearly state that the leukemia excess observed in the
Delzell et al. (1996) study has not been replicated in a completely different
study population. In addition, a weight of evidence approach should
incorporate all three studies rather than to emphasize the one positive
study.
45
-------
(4) The strength of the association evaluation for the lymphosarcomas and the
leukemias should focus on these different cancers separately.
(5) The EPA should incorporate the PBPK modeling into Chapter 11 if it has
time to do so or should at a minimum, discuss future possible directions
and consider alternative ways (e.g., reduced safety factor) to account for
the species differences.
(6) Some members felt that the additional safety factor of 3 to move from an
effect dose to a no effect dose should be removed because it is
inappropriate.
(7) The document should point out important data gaps in our knowledge and
research needs.
46
-------
APPENDIX A-TECHNICAL ISSUES
Detailed Comments on Specific Aspects of the Draft Document
Chapter 2
a) Section 2.2.1. Butadiene monomer production facilities need to be discussed
especially since the epidemiology section discusses exposure in these facilities
b) Section 2.2.1. The impact of compliance with the 1994 Hazardous National
Emissions Standard for Hazardous Air Pollutants and the polymers and resins
Maximum Achievable Control Technology (MACT) on air emissions should be
discussed.
c) Section 2.3.3. This section discusses open burning of tires. Data on controlled
burning of tires should also be included and distinguished from open burning.
Specifically, results from a pilot study conducted by the U.S. EPA (USEPA, 1994)
should be included. In this study, 1,3 butadiene could not be detected from
controlled combustion of tire-derived materials.
d) Section 2.4. For a better perspective, where possible, tables should indicate the
number of samples that were below the detection limit as well as the ranges of
exposure concentrations. More recent data on emissions, such as the 1995 and
1996 Toxics Release Inventory, (USEPA, 1997; 1998b), and the EPA's 1996
nationwide emissions inventory for butadiene (when released) should be included.
If available, exposure concentrations inside automobiles should be added.
e) Section 2.5. In this discussion of exposures, similarities and differences of general
population and occupational cohort exposures should be discussed so that the
limitations of extrapolation from occupational setting to exposure of the general
population are understood. For example, short peak exposures in industrial setting
vs. chronic low levels of exposure to general population may impact the validity of
an extrapolation of occupational data on exposure-response relationships.
Chapter 3
a) Page 3-1, Line 15: Much of the literature used the abbreviation BD for 1,3-
butadiene. It should be less confusing if BDiol is used for the 3-butene-l,2-diol.
This section does not even mention the 3,4-epoxy-l,2-butanediol (EBD). The
latest research on molecular dosimetry strongly suggests that EBD is the major
electrophile that binds to DNA and hemoglobin. EBD should be readdressed in
the final version of this risk assessment.
A-l
-------
b) Page 3-2, Line 14: Recent data shows that the trihydroxybutane (THE) adducts
are clearly the predominant adducts in DNA and hemoglobin. These can arise
from either 3,4-epoxy-l,2-butanediol or l,2:3,4-diepoxybutane (DEB). Based on
tissue measurements of DEB and EB, the ratios of epoxybutene (EB) to
DEB/EBD(THB) adducts strongly suggest that EBD is the primary source of the
THE adducts. The EBD metabolite has not been quantitated following in vivo
exposure. This represents a major gap in our knowledge that needs to be
acknowledged
c) Page 3-2, Line 14-18: The toxicokinetic data clearly does not support the
statement that the possible crotonaldehyde metabolites are "Of greater
significance." This editorial comment should be removed from the document, since
no causal role of these metabolites in butadiene mutagenicity or carcinogen!city has
been shown.
d) Page 3-39, Line 18-25: The toxicokinetics of butadiene are much more
complicated than are discussed. The molecular dosimetry of DNA adducts
following 4 weeks of exposure clearly shows that the first oxidation to EB and its
subsequent binding to DNA is linear over a range of 20-625 ppm in rats and mice
(Swenberg, et a/., 1998). It is rapidly converted to BDiol and then to EBD. To a
lesser extent, EB is oxidized to DEB. Both EBD and DEB form THE adducts,
although most of these come from EBD. The formation of THE adducts is
saturated at 62.5 ppm in the rat, so that exposure to higher amounts such as
1000-8000 ppm results in little more THE adducts than does exposure to 62.5
ppm. In contrast, the mouse shows a biphasic response for the formation of THE
adducts, with a steep slope between 0 and 62.5 ppm and a lesser slope from
62.5-625 ppm. The mouse does not show total saturation of the formation of
THE adducts, suggesting that it has a second enzymatic pathway that is still active
at high exposures. These data were presented at the SAB meeting, were presented
at the Society of Toxicology (SOT) and Health Effects Institutes (HEI) annual
meetings in 1998, and will be submitted for publication this summer (Tretyakova et
a/., 1998; Swenberg et a/., 1998).
e) Page 3-40, Line 13-14: This sentence is unclear. What is metabolic capacity of
EB? Is this formation or further metabolism? Available DNA data suggests that
formation continues to occur in an exposure related manner.
f) Page 3-45: The section on Discussion and Conclusions should be revised. Lines
8-10 are oversimplified. Line 12 is wrong — 1,3-butadiene epoxide should be
identified as butene diol. Lines 19-23 clearly do not reflect molecular dosimetry
data and actually appear to be reversed. Line 27 is wrong. Butene diol is not
toxic. EBD needs to be added to this section, as it is likely to be a major
metabolite.
A-2
-------
g) Page 3-3: This chart has several errors that need correction. 3-Butene-l,2-diol is
not reactive and should not have a box around it. Furthermore, the arrow for
Reaction 11 is going in the wrong direction. Crotonaldehyde and acrolein are
reactive and should have boxes around them. The urinary metabolites M-I and
M-II should be shown.
h) Page 3-27: (9) This reaction seems to be mislabeled. It appears that this refers to
BDiol GSH-BDiol.
i) Page 3-38, Line 21-22: It is stated that rat-excreted 1,3-dihydroxypropanone may
be derived from hydroysis of diepoxybutane. However, since most
1,3-dihydroxypropanone probably comes from EBD this statement gives a wrong
impression.
k) Page 3-41, Line 8: Butene diol is not a hydrolysis product of DEB. This should be
EB.
1) Page 3, Line 41-42: The sections on DNA and hemoglobin adducts are seriously
out of date. Data from several laboratories have shown that the THE adducts are
predominant over EB adducts. At high concentrations of butadiene (625-1000
ppm), the ratio of THB/EB is 1.5-4 for rats and 3-9 for mice. This drastically
changes at nonsaturating exposures of 20 ppm and 62.5 ppm, where rats have 27.5
and mice have 43-47 times more THE adducts. The same finding has been
demonstrated for human hemoglobin adducts. THB-Valine adducts are formed
about 40 times more frequently than EB-valine adducts. The EPA document only
mentions EB adducts. A molecular epidemiology study of hemoglobin adducts
was recently completed at the National Cancer Institute (Swenberg et a/., 1998) in
a Chinese butadiene worker study. Nearly all blood samples examined, whether
from exposed or unexposed individuals, had measurable THB-Valine adducts. The
number of adducts in unexposed individuals averaged -40 pmol/g globin. There
are similar data for U.S. research workers. With exposure to butadiene estimated
to be 1-3.5 ppm, the number of adducts increased 2-3-fold (R2 = 0.33). Of interest
is the finding that glutathione S-transferase theta (GSTT1) genotype had no effect
on THB-Valine adducts. Since the GSTT1 null genotype has clearly been
associated with increased susceptibility to DEB-induced sister chromatid
exchanges (SCEs) and is primarily located in erythrocytes, this observation
supports EBD as the primary human electrophile forming THB-Valine adducts.
m) Page 3-45: The Discussion/Conclusions section of the risk assessment should also
address what is known about genetic polymorphisms that are likely to affect
individual susceptibility to butadiene and its metabolites. Several genes appear to
be important. Inherent susceptibilities have been shown for both DEB and EB
A-3
-------
Chapter 5
(Weincke and Kelsey, 1993), which may be due to glutathione S-transferase theta
(GSTT1) status. Also, glutathione S-transferase ji(GSTMl) appears to be an
important detoxifying factor for EB, so that GSTM1 null individuals would be
expected to have greater effects following formation of EB. Unfortunately, no
data have been published on the effects of GST polymorphisms on EBD. This is a
gap in our knowledge. Genetic polymorphisms have also been identified for
epoxide hydrolase (EH) and CYP 2E1 that would be expected to affect
susceptibility to butadiene and its metabolites. The role of these proteins in the
toxicokinetics of numerous chemicals is reasonably well known. Three studies
(Csanday etal., 1992; Seatone^a/., 1995; and Duescher and Elfarra, 1994) have
shown in vitro using rodent and human tissue samples that CYP2E1 plays a role in
the oxidation of both BD and EB. It is possible to expect that polymorphisms that
reduce EH activity will increase susceptibility to butadiene. Likewise, rapid CYP
2E1 metabolizers would be expected to be at greater risk.
a) Pages 5-1, lines 29-31: Reword this sentence as shown below (changes in italics)
to make it neutral: "Because the results were not analyzed statistically and other
details regarding the duration of the mating periods were not present, it is not
possible to conclude that 1,3-butadiene either had or did not have an effect on
fertility in rats."
b) Pages 5-9, lines 27-29: Under what condition is 108± ppm 4 vinyl-1-cyclohexene
identified? What is the stock solution? The missing information must be added.
c) Pages 5-25, tables 5-13: Why are only some of the studies of structurally related
compounds listed on this table? Why are negative observations not included? The
explanation given on pp.5-24, lines 17-19 was weak. Since neoplastic lesions in
reproductive organs were seen, these findings should be added here, albeit
identified as neoplastic or non neoplastic. The title of this table does not limit
effects on neoplastic status. However, the table should not just give positive
studies but also include negative observations, like those in rats. The observations
of decreased Graafian follicules should be included if significant (note that
significance status was not given in lines 25-27).
d) Pages 5-28, lines 15-17: The summary introduces new structure-activity
information from Maronpot, 1987. This data should be introduced and integrated
with Chapter 5 prior to the summary. See also summary lines 7-33.
e) Pages 5-28: The summary speculates about reduced steroidogenesis, has any
study measured steroid levels?
A-4
-------
f) Pages 5-28, lines 32-33: The Committee did not see any data presented to
support this speculation, in fact the observed effects of 1,3-butadiene on
spermatozoa and spermatids (discussed in the next paragraph) argues against this
"protective" effect of this barrier. The Committee recommends the removal of this
sentence.
g) Table 5-1: Table 5-1 has numbers in parenthesis that are not defined.
h) Section 5.1.6: Sections such as this would be clearer if a data table would
accompany the text. For example, see Tables 5-4 through 5-12.
i) Page 5-28, lines 11-17: The paragraph on ovarian lesions fails to present the well
established mechanism for ovarian toxicity and carcinogenicity. DEB is highly
toxic to the ovary, which makes it non-responsive to FSH. Continual elevation of
FSH results in the carcinogenicity.
Chapter 7
a) Page 7-3, Line. 33: The latency period of 10-19 years was left out.
b) Page 7-5, Line. 14-15: Singling out an intermediate subgroup that gave a
suggestive elevation in risk, when subgroups with more exposure did not, is a
questionable scientific procedure (i.e., picking and choosing the ad hoc results that
support a particular point of view).
c) Page. 7-20, Line. 1-5 & Page 7-33, Lines 9-12: It seems curious to report
subgroup analyses based on just 3 of the 8 plants, especially when the reason given
for choosing them ("three plants who had geometric means of exposure" out of the
7 plants with measurements) seems irrelevant.
d) Page 7-21, Line 30-32: The document states at this point that "When this
subcohort was further restricted..." Please define the subcohort that is discussed,
and also state the endpoint that is being discussed.
e) Page 7-23, Line 25-26: One important feature was that the job- exposure matrices
(JEM) estimates were specific for calendar time. This should be noted.
f) Page 7-31, Line 32-33: It is inappropriate to cite positive findings from some
earlier follow-up of the cohort when these were not confirmed by the latest
follow-up.
g) Page 7-33, Line 1-8 and Page 11-5, Line 12-25: In order to present a balanced
review of the Matanoski et al. (1990) case-control study, the re-analysis by Cole et
A-5
-------
al. (1993) that yielded a null RR should also be reported, with an indication of
how highly sensitive the results were to the particular cutpoints chosen
The Cole-Acquavella (Acquavella et al., 1994) results are mentioned only
obliquely, and the discrepancy is not articulated and its implication are not
considered.
h) Page 7-34, Line 33,35: To be more balanced, the report should indicate that the
Meinhardt (1982) results were not statistically significant.
i) Page 7-36, Line 23: This sentence should be deleted because it suggests that the
stop-exposure studies conducted by Melnick et al. (1990) confirm the findings of
excess lymphosarcoma among short-term monomer workers in the Divine et al.
(1996) study. This statement is incorrect.
j) Page 7-36, Line 23: There is no way that the stop exposure studied in mice
"confirm" the short term worker effect. Long term workers employed at the same
time and later did not develop increased lymphosarcomas.
j) In the total cohort the leukemia rate was somewhat elevated (O/E = 48/36.6 =
1.31, 95% CI = 1.0-1.7) with some 11 excess leukemias (Delzell etal, 1996).
Surprisingly, there was no excess among those hired before 1950 (O/E = 17/16.4=
1.04) when one would expect the highest exposures, but there was an excess
among those hired during 1950-59 (O/E= 20/10.0= 2.0, CI= 1.2-3.1). Research is
ongoing as to whether the risk elevation beginning in 1950 is a function of
concomitant exposure to DMDTC (Dimethyldithiocarbamate) in the SBR process
beginning in about 1950, but it is premature to judge that hypothesis at this time.
k) There should be a table to present the actual exposure-response data for the
Macaluso et al. (1996) study of butadiene exposure and leukemia risk, rather than
just burying the values in the text (e.g., Page 7-22, Line 22 and Page 7-25, Line 7).
These are among the most important numbers in the report, so they should be
prominently displayed.
1) According to Macaluso et al. (1996\ the 0-dose group contained a substantial
number of salaried workers but the other groups apparently did not. This could
potentially bias the exposure-response risk estimate, and it should be mentioned.
m) There seems to be a substantial discrepancy between two sets of risk estimates
given for this study in the report on Page 7-22 (Line 22) and Page 7-25 (Line 7),
as shown in the table below. One wonders if the second set (the lower half of the
table), which appear to be the one used by the EPA, may be incorrect. Notice that
in the upper half of the table, the RR for >0-19 ppm-years is 1.1, whereas in the
A-6
-------
Chapter 8
a)
lower half, the two cells that cover the range of >0-19 ppm-years have RRs of 2.0
and 2.1 — rather implausible values, especially the RR of 2.0 for <1 ppm-year. The
middle range is likewise higher in the lower half of the table (RR = 2.4 for 20-79,
vs. 1.8 for 20-99 ppm-years in the upper half). The RRs in the high exposure
range also differed notably: RR = 4.5 for 80+ ppm-years in the lower half of the
table, but in the upper half the RRs for 100-199 and 200+ ppm-years are 2.1 and
3.6. The reason for the discrepancies is not clear; the only difference noted in the
derivation of the two sets of estimates is that the upper set adjusted for years since
hire and calendar period, whereas the lower one did not. If one of these two sets
is not in error, then the discrepant results suggest that the results must be very
sensitive to the particular cutpoints, confounders used, etc., which would argue for
observing caution so as not to extrapolate from the most extreme results. It is of
note that most of the analyses reported by Delzell et al. (1996) in their technical
report, e.g., Tables 60-62, are similar to the upper half of the table below, rather
than the lower half.
Table 1, below, displays the two sets of estimates of leukemia relative risks (RR)
(in relation to butadiene exposure) reported from the Macaluso et al. (1996) study.
Exposure range
(ppm-years) 0 >0-19 20 - 99 100 - 199 200+
RR
Exposure range
(ppm/years)
RR
1.0
0
1.0
1.1
<1
2.0
1.8
1-19
2.1
2.1
20-79
2.4
3.6
80+
4.5
Table 1 (from Delzell etal, 1996)
It is unclear from the EPA report as to how high the correlation between butadiene
and styrene exposure levels was in this study. However, one would expect the
correlation to be fairly large. Since there is also some indication that styrene may
be associated with leukemia in this study, partialling out the effect that is
attributable to butadiene is problematic, particularly if there is more reliability or
accuracy in assigning butadiene exposures than in assigning styrene exposures, or
vice versa, since variations in reliability/accuracy could drive the proportion of
variance attributed to one chemical vs. the other in regression analyses.
Page 8-1, lines 6-7: The PBPK models cited by the EPA in this report should be
updated. More mechanistic and sophisticated models are now available including
A-7
-------
Reitz etal. (1996), Csanady etal. (1996), Sweeney et al. (1996; 1997), and Kohn
(1997).
b) Page 8-2, lines 1-32: The model of Hattis and Wasson (1987) was developed
before there were any experimental data on the rates of 1,3-butadiene activation
and detoxication in experimental animals and before there were any experimentally
determined data for partition coefficients, estimates of solubility that are used in
PBPK models. Given the acknowledged limitation of this model, it would be more
appropriate to simply mention that this model is one of the first PBPK models to
be developed without going into details regarding the model predictions. This is
particularly important given the conclusion of Hattis and Wasson that "differences
in pharmacokinetics failed to account for differences in carcinogenesis between
mice and rats and that with respect to risk assessment, uncertainties in PBPK
modeling are trivial compared with the differences in apparent sensitivities between
these species" (Page 8-2, lines 18-32). The overwhelming body of experimental
data on the toxicokinetics of 1,3-butadiene and its metabolites collected in rats and
mice since the publication of the Hattis and Wasson model have shown that
differences in pharmacokinetics between these two species can account for species
differences in carcinogenesis. It is also significant that the Hattis and Wasson
model was an unpublished report and never benefited from peer-review. It would
be more appropriate for the EPA to devote a paragraph to the presentation of the
Hattis and Wasson model similar to what was done for the Hallenbeck (1992)
model (page 8-4, lines 3-9).
c) Page 8-4, lines 10-33: It might be appropriate for the EPA to cite the Kohn
(1996) model instead of the 1993 model primarily because the model of Kohn and
Melnick (1993) relied on theoretically derived partition coefficients rather than
experimentally determined partition coefficients. The use of these calculated
partition coefficients resulted in overpredictions of the concentrations of
1,3-butadiene in tissues, especially fat.
d) Page 8-7, lines 27-34: As noted above, the use of empirically derived calculated
partition coefficients rather than experimentally determined values led Kohn and
Melnick in their 1993 model to conclude that storage in fat is a significant fraction
of the retained 1,3-butadiene, especially in rats and humans. It is generally
recognized that this conclusion is based on the use of calculated rather than
experimentally measured partition coefficients. Inclusion of this paragraph in the
chapter may lead readers not knowledgeable in 1,3-butadiene toxicokinetics to
believe that this model prediction is accurate. While the EPA does discuss this
limitation (page 8-8, lines 1-13), it is still misleading for the EPA to devote
significant discussion to models in which theoretical values are used when later
experiments determined the values to be inaccurate. It would be much more
prudent for the EPA to report the most recent PBPK models noting when
A-8
-------
necessary that these laboratories had also participated in the development of earlier
models.
e) Page 8-12, lines 21-23: The EPA notes that "Johanson and Filser are reportedly
working on a corresponding PBPK model for humans but it has not yet been
published." At this point it would be most appropriate for the EPA to cite the
Csanady et al. (1996) PBPK model for 1,3-butadiene that includes model
predictions for man.
f) Page 8-15, lines 9-11: The EPA notes that a limitation of the model of Evelo et
al. (1993) is that metabolism of butadiene is limited to the lung and the liver. This
should not be viewed as a limitation. The objective of PBPK models is to account
for the most significant mechanistic steps in the disposition of chemicals with the
goal of predicting the concentration time profile of the toxic agent in either the
target tissue or a suitable surrogate for the target tissue such as blood. It is not
practical nor is it necessarily advantageous to develop a model that incorporates all
pathways in the disposition of a chemical however minor. The simplest models
that are the most useful for risk assessment are most likely to have the greatest
value since these models will have the fewest number of parameters that require
independent experimental determination in animals and humans. Thus, it is not
clear why the Agency feels it is necessary in the case of butadiene to account for
metabolism of this chemical in all tissues of the body. Extensive modeling efforts
have determined that inclusion of metabolic activation in the liver, the major organ
for metabolism of butadiene, and in the lung, a target organ for mice, are the most
appropriate from a mechanistic standpoint.
g) Page 8-15, line 15: The EPA notes that the most recent PBPK model published
for 1,3-butadiene is the model of Medinsky et al. (1994). This statement is correct
only in the context of the January 31, 1997 cutoff date for consideration of reports
in this document. However, the statement is misleading as there are a number of
other PBPK models that have been published since 1994, including models
published in 1996 and 1997. The EPA is urged to revise this chapter to include
these more recent models. These models not only consider the metabolism of
1,3-butadiene, but also include the disposition of its two epoxide metabolites,
epoxybutene and diepoxybutane, thereby making the models more appropriate for
use in risk assessment.
h) Page 8-16, lines 6-14: The EPA notes that "in the model of Medinsky et a/., the
microsomal concentrations reported by Csanady et al. (1996) were not used to
scale metabolic rates. Instead, literature values for microsomal concentrations
were used." While the Agency correctly summarizes the approach taken, the
rationale for taking this approach is not appropriately presented. The objective of
the study reported by Csanady et al. (1992) was to determine the rates of
A-9
-------
butadiene oxidation in microsomes from rodents and humans. To achieve this
objective it is essential that purified microsomes be obtained from liver, but it is not
essential that all of the microsomal protein in the liver be accounted for. Thus,
Csanady et a/., 1992 report microsomal yield which is less than the total
microsomal protein content of liver or lung. Other investigators, in contrast, have
sought to determine the total amount of microsomal protein in liver or lung. These
investigators have used other techniques for this assessment. Extrapolating in
vitro results expressed per milligram of/microsomal protein to the entire organ
requires knowledge of the total amount of microsomal protein in an organ not
simply the yield of microsomal protein obtained in the biochemical experiment.
Thus, the appropriate approach for scaling in vitro rates to the whole animal is to
use total microsomal total protein content rather than yield. The literature values
used by Medinsky et al. (1994) for total microsomal protein content were similar
to those used by Johanson and Filser (1993) and Kohn (1997). In contrast, Kohn
and Melnick (1993) used the values of microsomal yield reported by Csanady et al.
(1992).
i) Page 8-19, Section 8.3, Summary: The EPA notes that "pharmacokinetic
modeling of 1,3-butadiene has not elucidated the reasons for interspecies
differences in carcinogenic response between rats and mice." This statement is not
true if one evaluates the current PBPK models that describe the disposition of
1,3-butadiene, epoxybutene and diepoxybutane. These PBPK models clearly
demonstrate that the dramatic species differences between rats and mice in
response to 1,3-butadiene are most likely related to species differences in rates of
formation and removal of the diepoxybutane metabolite. Experimental data and
PBPK model simulations indicate that mice produce far greater concentrations of
this reactive metabolite compared with rats. Sweeney et al. (1997) have used in
vitro metabolism data collected in tissues from rats, mice, and humans directly into
a PBPK model to make predictions regarding the epoxide concentrations in blood
and tissues following exposure to 1,3-butadiene. Using in vitro derived
parameters they were able to adequately simulate the pharmacokinetics of
1,3-butadiene, epoxybutene, and diepoxybutane. This ability to use in vitro data to
make in vivo predictions suggests that in vitro data on rates of 1,3-butadiene,
epoxybutene, and diepoxybutane metabolism obtained in human tissue samples can
also be used to predict blood and tissue concentrations of both epoxide metabolites
in humans. When these simulations are conducted, using average values for human
metabolic rates, the results indicate that diepoxybutane concentrations in humans
would be orders of magnitude less than those of mice and lower than, but much
more similar to, concentrations predicted for rats. Thus, the EPA should revise
this chapter to include a discussion of the toxicokinetic models for 1,3-butadiene
that are now capable of simulating not only the disposition of 1,3-butadiene, but
also its two most important epoxide metabolites.
A-10
-------
j) Page 8-19, lines 33-34: The EPA statement that uncertainties in the existing
PBPK models and data make them unreliable for use in risk assessment must be
revisited. Given that several investigators have been able to use these models and
the underlying metabolic data to predict butadiene, epoxybutene, and
diepoxybutane concentrations in rodents exposed to butadiene, it is not clear what
additional data and what uncertainties need to be resolved prior to the use of these
models in risk assessment. For example, Sweeney et al. (1997) have used in vitro
metabolism data obtained from rat, mouse, and human tissues directly in a PBPK
model to make predictions regarding the epoxide concentrations in blood and
tissues following exposure to 1,3-butadiene. Using in vitro derived parameters
they were able to adequately simulate the toxicokinetics of butadiene,
epoxybutene, and diepoxybutane.
k) Page 8-20, line 5-21: In this paragraph the EPA presents a number of criticisms
regarding the parameters used in the PBPK models that are not necessarily
accurate. For example, the EPA notes that "with respect to parameter values,
there are disagreements about the ventilation rate and about metabolic
parameters." As noted above, while it is true that each investigator uses a different
value for ventilation rate or metabolic parameter, the values used by all
investigators are within the normal range associated with these parameters. It is
also not unreasonable to expect that if the EPA were to use a PBPK model in risk
assessment that other point estimates also within the range of reported values for
these parameters would be chosen by the EPA.
The EPA also notes that there is a paucity of human in vitro data for extension of
the PPBK model to humans and that the few measurements that have been made
on a few metabolic parameters show a high amount of variability. Relative to
other chemicals there is an extensive amount of experimental data on the rates of
metabolism of 1,3-butadiene by human tissue samples. Each of the metabolic
pathways important in the disposition of 1,3-butadiene and its metabolites have
been quantitated with Vmax and Km values obtained from multiple human samples.
These pathways include oxidation of 1,3-butadiene, oxidation of epoxybutane,
hydrolysis of epoxybutane, glutathione conjugation of epoxybutane, glutathione
conjugation of diepoxybutane, and hydrolysis of diepoxybutane. Means and
standard deviations for these parameters have been calculated because multiple
human samples have been used. Thus, sample distributions can be generated from
which population values can be obtained. A PBPK model applied to human risk
assessment could employ either average values to obtain deterministic predictions
or Monte Carlo simulation techniques to get probabalistic estimates of the range of
responses of hundreds of simulated humans. The latter approach using Monte
Carlo simulation would provide some estimate of the potential variability in human
response to inhaled 1,3-butadiene in addition to an estimate of the response of the
most sensitive humans. Regarding the large amount of variability associated with
A-ll
-------
these metabolic parameters when measured in humans, this reflects the inherent
variability in the expression of various xenobiotic metabolizing enzymes in the
human population, a fact that has been well documented in the literature. Thus, it
is not unexpected that assessment of metabolic capacity from tissues from multiple
humans should yield a range of outcomes.
1) Page 8-20, lines 22-34: The EPA notes that the existing models have been
subjected to a very limited validation mostly by comparison of simulation results
with chamber uptake data. This statement is true regarding the PBPK models that
were published in 1994 and earlier (e.g., the models reviewed in this current
document). However, since the publication of these first generation models,
multiple inhalation toxicokenetic studies have been conducted in rats and mice
where blood and tissue concentrations of 1,3-butadiene, epoxybutene, and
diepoxybutane have been quantitated following inhalation exposure to
1,3-butadiene. The second generation models that include the prediction of not
only epoxybutene but also diepoxybutane have utilized this recent in vivo
toxicokenetic data for model validation.
The EPA also notes that "for PBPK models to be more reliable, they should also
be validated against tissue concentration data for various metabolites and various
tissues. More recently these data have become available although they must be
interpreted with caution because it appears that metabolites in some of the tissues
are subject to further metabolism during the lag time between termination of
exposure and measurement of tissue concentrations." This statement is true and
ironically the implications of post exposure metabolism were first recognized when
one of these second generation PBPK models (Sweeney et a/., 1996) failed to
adequately simulate tissue concentrations of epoxide metabolites. When the
authors modified the initial conditions of the model to reflect the time lag between
termination of exposure and measurement of tissue concentrations and the capacity
of the tissues to metabolize 1,3-butadiene post exposure, they were successful in
predicting the actual measured epoxide tissue concentrations.
m) Page 8-21, lines 4-23: The EPA concludes that the existing PBPK models and
data cannot explain the interspecies differences in 1,3-butadiene carcinogenicity.
As noted previously, the first PBPK models discussed in this report did not include
the toxicokinetics of diepoxybutane. More recent second generation PBPK
models that include the formation and elimination of this metabolite are successful
in simulating in vivo data in both rats and mice for 1,3-butadiene for epoxybutene
and diepoxybutane concentrations in blood and tissues. Both model predictions
and experimental data indicate that the dramatic interspecies differences in
carcinogenic response to 1,3-butadiene can, in fact, be explained by the dramatic
interspecies differences in circulating concentrations of the diepoxybutane.
A-12
-------
Chapter 9
a) Page 9-4: Since previous EPA lifetime risk estimates have used a 70-year
time-frame, it would seem appropriate to follow this precedent for the sake of
comparability, rather than use an 85-year estimate.
b) Figures 9-9 to 9-14: These figures should be prepared so that the axis are
understandable without reading the full text.
c) Page 9-27, line 33: Why does this sentence say 600 ppm and above? Why
doesn't it say 625 ppm, a dose that was tested?
d) Page 9-5 & 9-7: There is a discrepancy in the 95% lower confidence interval with
a 1% level of risk, LECl5 value between the text and the figure (0.12 ppm vs.
0.066 ppm). Two of the figures appear to be mislabeled.
e) Page 9-13, lines 18-19: It was difficult to follow the logic in the last sentence.
This sentence should be modified to provide necessary rationale to follow this
choice.
f) Page 9-36, lines 13,14: What is the rationale for this statement? Either the
explanation should be expanded or the statement should be deleted.
g) Tables 9-13 thru 9-15: Statistics should be included for these summary tables.
h) Figures 9-9 thru 9-14: For clarity, the axis should be labeled so that reader can
easily convert dose to ppm without going back to the text. Also figures, like
tables, should indicate when exposures were adjusted to 24 hour daily exposures.
i) Section 9.3: Although the 95% lower confidence interval with a 10% level of risk
(LEC10) values are taken for the "point of departure" for the RfC's calculated for
the benchmark based approach, minimal discussion was given as to why the LEC10
was chosen over the 95% lower confidence interval with a 5% level of risk (LEC5)
nor why the 95% lower confidence interval with a x% level of risk (LECX) versus
EDX values are chosen. Justification and rationale for this issue for all modeled
endpoints should be provided.
j) Pages 9-49, lines 33-34: Is this statement true for dominant lethal effects as well
as fetal weight reduction? This Committee urges caution in such general
statements or authors should provide specific justification of these statements.
k) Pages 9-42, lines 5-7: Text cites Allen et al. (1994b) as source of information
supporting the use of LEC10 as being "at or below the range of detectable
A-13
-------
responses." Since this paper dealt only with developmental toxicity data, this
statement as well as those later in this section are extrapolations from that
research. The following text should be reworded, "Other studies are supportive of
this statement. For example, the statistical power of detection of this study design
supports this statement."
1) Pages 9-44, Table 9-19: A footnote should be added to describe what the Z
statistic is and how Q O-2 are obtained. This section was very unclear.
m) Pages 9-40 thru 9-44, Section 9.3.4, pp. 5-3 thru 5.5 and Section 5.1.4: The
Agency needs to address more fully the statistical and biological significance of the
testicular atrophy. The footnote on page 5-4, Table 5-1 states that statistics were
not conducted on the testicular lesions yet in Section 9.3.4 this endpoint becomes a
study for modeling. Is the background rate for this lesion in the control B6C3 Fl
mice low compared to historical mouse population statistics?
n) Pages 9-46, line 1-2: In this example, the text provides a reason for discarding the
top doses in the modeling however, this decision is inconsistently applied in the
modeling Section 9.3 as evidenced in Table 9-16. The text should provide some
common guidance on what will be done regarding dropping higher dose levels
from modeling calculations.
o) Pages 9-46, line 10-11: Statement in lines 10-11 appear to differ from more
recent research published on PBPK models. Can this health risk assessment go
further in using the data about known ovotoxicity of the diepoxide? See examples
provided in testimony from the Chemical Manufacturers Associations's documents.
(CMA, 1998a; 1998b)
p) Pages 9-51, lines 2-4: The Committee was unconvinced about the superior nature
of the time-to-response modeling that was conducted for the LEC10 determination
for atrophy. Additional discussion is needed to support this statement especially
given the limitations of the biological time-to-response data for this endpoint.
q) Pages 9-51, lines 16-26: The EHC agrees with these limitations but would then
use these points to justify using PBPK modeling to improve the target organ and
time concentration curves relative to these specific reproductive versus
developmental endpoints. The text has provided the justification but the
assessment falls short of acting on these suggestions.
r) Pages 9-51, lines 27-31: The Committee agrees on these issues. Please see the
earlier comments on how to constructively address these points.
A-14
-------
s) Mathematical errors need to be corrected:. It is absolutely necessary that all
calculations in Section 9 be carefully reviewed. The Committee noted numerous
problems in this section and refer USEPA to the written statement of R. Seilken
(1998) that is especially relevant to this issue. There may be confusion with some
numbers presented as natural logs versus non-log numbers. Please proof very
carefully. All calculations must be easy to follow by a general scientific audience.
Chapter 9 did not meet this goal.
Chapter 11
a) Page 11-1: The statement regarding the lack of sufficient data to determine if
children or other subpopulations are affected differently by exposure to
1,3-butadiene should be revisited in light of the studies by Nelson et al. (1995),
and Wiencke et al. (1995). Nelson et al. (1995) found the glutathione transferase
theta (GSTT1) to be highly polymorphic with wide variation in its ethnic
distribution (Nelson et a/., 1995). Wiencke et al. (1995) have shown the
association between the genotype for the production of glutathione transferase
theta and genetic anomalies in sister-chromatidic exchange induced by metabolites
of 1,3- butadiene: epoxybutene and diepoxybutane. This implies that the protein
produced by the gene is important in conjugating both of these metabolites.
GSTT1 only affects DEB and GSTM1 only affects EB. These are two proteins
from two genes. Given the profound racial distribution of the polymorphism, it is
important to note that this may account again for the significant portion of the
alterations in the metabolism of 1,3-butadiene. Consequently, the enzymes
responsibilities for the metabolic conversion of 1,3-butadiene to its mono- and di-
epoxide forms as well as its diol form are highly polymorphic. This implies that
there may be differential susceptibility to the genotoxic effects of exposure to
butadiene. Investigations of this are rapidly moving ahead. There is some
indication from field studies that these polymorphisms may contribute to directly-
measurable genetic effects (Sorsa et al., 1994). Hence, risk assessment and future
studies of this compound should take note of this and adjust as is appropriate.
b) Page 11-3, line 11-13 and Table 11-1: EPA states that the conclusion of
"sufficient evidence" of human carcinogen!city is based on more than 10
epidemiologic studies examining five different groups of workers. This statement
is misleading because it implies that there is a consistency of results across several
studies of equal caliber examining completely different populations. The
predominant emphasis should be on the methods and findings of the two latest
studies of butadiene monomer and SBR workers. The emphasis on the early
epidemiologic studies is misleading to the extent that it gives the impression of
consistency across several study populations in the SBR and buta diene monomer
industries. There is really only one study population in each industry that provides
appreciable information. Each of these studies consolidated and/or more
A-15
-------
accurately refined the populations used in the earlier studies. This needs to be
clarified because it is central to the understanding of how consistent the
relationship is between 1,3 butadiene and cancer. For example, EPA should
explain that the Delzell et al. (1995; 1996) studies included all of the eligible
population from the Matanoski et al. (1990) study and the Meinhardt et al. (1982)
study. The leukemia finding in the Matanoski et al. (1990) study should not be
presented as if it were a separate finding in a completely different population from
the Delzell study. In addition, the Delzell study supersedes this and other previous
studies and rectifies many of the limitations and errors of the earlier studies.
c) Page 11-5, line 12-34: The Matanoski study is presented as if it were a study on a
completely different population from the Delzell studies. In discussing the excess
leukemia risk in the nested case-control study in the Matonoski study, it is
important to indicate in line 14 that there was no excess of leukemia observed in
the cohort study (standardized mortality ratio was 1.0 representing 22 observed,
22.9 expected). This is an important major point to be made up-front in this
section because it helps explain the scientific debate later referred to in lines 26-
28. The evidence linking butadiene exposure and cancer is still strongest for
leukemia based on one large, high quality cohort study of SBR workers (Delzell
studies) which supercedes the Matanoski study.
d) Pages 11-6 to 11-7: It is never clearly stated that the leukemia excess seen in the
Delzell study has not been replicated in a completely different study population.
The studies of butadiene monomer workers and of other butadiene exposed
workers report null results for leukemia.
e) Page 11-7, line 17: The lymphosarcomas and leukemias are discussed as if they
could be considered as the same type of cancer. It is implied that two different
populations have an excess in lymphohematopoietic cancers thereby demonstrating
consistency across studies. The strength of association evaluation should focus on
these different cancers separately.
The EPA review states on page 11-7, lines 20-26 that the monomer workers
exposed to shorter periods of time probably had higher exposures than workers
exposed for longer periods of time. This statement needs to be removed as there is
no evidence for this. The available evidence is insufficient for a causal relationship
between butadiene and lymphosarcoma. Lymphosarcoma was elevated for short
term exposed workers but not among long term exposed workers indicating a lack
of dose-response. Of equal importance, lymphosarcoma/non-Hodgkins lymphoma
(NHL) was not elevated in the SBR cohort which had excess leukemia. These
points must be discussed in the EPA document. The possibility that it is the SBR
process and not butadiene alone that may explain excess leukemia should be
discussed regardless of final decision on cancer classification. Delzell's finding that
A-16
-------
the leukemia excess concentrated among workers who began employment in the
1950's and not those that worked exclusively in 1940's led to hypothesis by Irons
and Pyatt (1998) that DMDTC might be a contributing factor in the leukemia
excess. This hypothesis needs to be discussed as it is part of the scientific
literature.
f) Page 11-8, Table 11-2: Some of the entries in this table seem unbalanced.
(1) Matanoski (1993) is cited as showing "7 to 9 times higher relative odds for
leukemia" without mentioning that others analyzed the same data using a
slightly different cutpoint and found an odds ratio of <1.0.
(2) Most of the dose-response tabulations by Delzell et al. (1996) do not show
mortality ratios in the highest dose groups as high as the ones cited.
(3) It states that a dose-response relation for monomer plant workers "Cannot
be demonstrated due to lack of quantitative exposure data," when in fact
several dose-response analyses were performed (albeit with an imperfect
dose metric) and showed not even a hint of an association.
g) Page 11-9, Line 24: The Committee classified the SBR process as a known
human carcinogen.
h) Section 11.3.3, Page 11-9: This section needs to more accurately reflect the
scientific literature that demonstrates a clear difference in metabolic activation
between humans and rodents and should discuss how PBPK modeling can refine
the risk assessment process. If EPA cannot incorporate PBPK modeling into risk
assessment within the mandated time-contraints, EPA should discuss future
possible directions and consider alternative ways (reduced safety factor) to account
for the species differences.
i) Page 11-10, line 36 to Page 11-11, lines 1-2: The EPA states that the tumor type
in rodents most analogous to the lymphohematopoietic cancers is the lymphocyte
lymphomas. To properly discuss the strengths and weaknesses of this statement,
EPA should discuss the data generated by Irons et al. (1996) which shows that T-
cell lymphoma in mice is due to a specific population of stem cells in the mouse
bone marrow that is not present for humans or rat bone marrow cells. EPA should
also point out that the link between 1,3-butadiene exposure and lymphoma is weak
as there was no consistent dose-response relationship. Additionally, the Agency
should show consistency in combining different tumor types across documents.
For example, in the revised cancer risk assessment guidelines a case study for an
aromatic hydrocarbon (presumably benzene) is presented where mice are shown to
develop lymphomas following exposure whereas humans develop acute
A-17
-------
myelogenous leukemia. In this case, the Agency did make the distinction that the
response in the animal models and the response in the humans was different. The
Agency should acknowledge that there is considerable disagreement as to whether
these two tumor types, lymphomas and leukemias, derive from the same origin.
The potential impact of task specific peak exposures must be addressed. The
primary metric for SBR workers is based on time weighted average exposure. But
SBR workers frequently get the majority of their exposures during a small fraction
of the work day during the conduct of specific tasks. The leukemogenic effect was
associated with those jobs that involved high peak exposures in latex sampling
(laboratory workers and in vessel cleaning for maintenance laborers). The general
population is typically exposed to lower ambient levels. An additional uncertainty
that should be discussed is the uncertainly of extrapolating from an occupational
setting where peak exposures occurred to the generally low ambient levels of
exposure.
j) Page 11, Lines 11-13: The statement that the evidence regarding human
carcinogenicity is based on ten studies is deceptive, because the reports are not
independent, but most reports are updates of previous ones. In addition, there are
basically only four independent cohorts, not five, because the Delzell study
included about 95% of the workers in the Matanoski studies.
k) Section 11.5, Pages 11-13-11-14: This section needs to be re-evaluated in light of
the two more recent negative dominant lethal studies. A weight of evidence
approach should be taken looking at all 3 studies rather than to emphasize the one
positive study.
1) Section 11.5 Pages 11-14, lines 1-3: This sentence should be rewritten as fact not
as speculation.
m) Pages 11-14, lines 10-16: These two sentences should be rewritten or removed.
Text discusses concept of "meaningful increases in risk." What is the definition of
this phrase? Is this in the USEPA guidelines? This phrase appears to be an
editorial. For this reviewer to understand that "...effects are not expected in
humans exposed to low environmental exposures.", the assessment must provide a
detailed rationale that follows from Chapters 5 and 9. This summary statement
was not supported in the document.
n) Section 11.6, Pages 11-15: This summary section needs to be integrated with the
biology discussed in Chapter 5 and quantitative assessments in Chapter 9. No new
"surprise," non-justified statements should appear in this section.
A-18
-------
o) Section 11.6: The importance of the positive heritable translocation studies in
mice should be more heavily emphasized.
p) Section 11.6, Page 11-15, line 10-17: The additional safety factor of 3 to
extrapolate from a LOAEL to NOAEL is inappropriate and should be removed.
Inadequate justification is given for the application of the additional safety factor
for the benchmark dose. The EHC could not understand the rationale for its
inclusion. This section must discuss the rationale for using the "hybrid model" for
continuous data analysis and exactly what the hybrid model is so that the risk
assessment can be completely transparent. On the surface it appears that the
approach yielding the lowest LEG was selected. Some of the EHC
Members/Consultants recommend that the Agency use the EC10 (central estimate)
as the point of departure rather than the LEC10. Other EHC Members/Consultants
did not agree with the recommendation.
q) Section 11.8 on Page 11.16: The Agency should expand on future research needs
to fill gaps in knowledge. The list is inadequate.
A-19
-------
GLOSSARY - ACRONYMS AND ABBREVIATIONS
ACGffl
TLV
Bdiol
C
DEB
DMDTC
EB
EBD
Ecx
EHC
ED
EBdiol
GC-MS
GSTT1
GSTM1
HEI
ICD
JEM
*V
LECp
LOAEL
MACT
mg
MLE
NCI
NHL
NOAEL
NTP
O/E
QMS
ORD
PBPK
ppm
ppmh
Qi
RfC
RR
SBR
SCE
American Conference of Governmental Industrial Hygienists
Threshold Limit Value
3-butene-l,2-diol
confidence interval
1,2,3,4-diepoxybutane
Dimethyldithiocarbamate
1,2 epoxy-3-butene
3,4-epoxy-l,2-butanediol
effective concentration at x% risk, ranging from 0.1% to 10%
Environmental Health Committee
effective dose
1,2 dihydroxy-3-4 epoxybutane
gas chromatography - mass spectrometry
glutathione S-transferase theta
glutathione S-transferase |i
Health Effects Institute
International Classification of Diseases
job-exposure matrices
substrate concentration at one-half maximum velocity
95% lower confidence intervals associated with a risk (p), ranging from 1%
to 10%
lowest-observed-adverse effect
maximum achievable control technology
milligram
maximum likelihood estimate
National Cancer Institute
non-Hodgkins lymphoma
no-observed-adverse effect
National Toxicology Program
ob served/expected
Office of Mobile Sources
Office of Research and Development
physiologically-based pharmacokinetic
parts per million
parts per million per hours
the incremental unit cancer risk for humans
Inhalation Reference Concentration
relative risk
styrene butadiene rubber
sister chromatid exchange
G-l
-------
SOT Society of Toxicology
THE trihydroxybutane
TRI Toxics Release Inventory
UAB University of Alabama at Birmingham
Vmzx maximum velocity for an enzyme-mediated reaction
WWII World War II
G-2
-------
REFERENCES CITED
AcquavellaJ. F. Delzell E., and P. Cole. 1994. Butadiene and lymphatic and hematopoietic
cancer. Epidemiology 5:262-263.
Acquavella, J.F. 1998. Comments on EPA Health Risk Assessment for 1,3-butadiene: external
review draft (EPA/600/P-98/001A, January 1998), April 17, 1998 submitted to the EPA
on behalf of the International Institute of Synthetic Rubber Institute.2
Adler, ID., Cao, J., and P. Gasner. 1994. Mutagenicity of 1,3-butadiene inhalation in somatic
and germinal cells of mice. Mutation Research 309(2):307-314.
Adler, ID., Filser, J.G., and P. Gassner. 1995. Heritable translocations induced by inhalation
exposure of male mice to 1,3-butadiene. Mutation Research 347(3,4): 121-127.
Adler, ID., Filser, J., Gonda, H., and G. Schriever-Schwemmer. 1998. Dose response study for
1,3-butadiene-induced dominant lethal mutations and heritable translocations in germ cells
of male mice. Mutation Research, Special Issue 397:85-92.
Allen, B.C., Kavlock, R.J., and C.A. Kimmel. 1994. Dose-response assessment for
developmental toxicity, II. Comparison of generic benchmark dose estimates with
NOAELs. Fundamentals of Applied Toxicology 23:487-495.
Arce, G.T., Vincent, D.R., and MJ. Cunningham. 1990. In vitro and in vivo genotoxicity of
1,3-butadiene and metabolites. Environmental Health Perspectives 86:75-78.
Au, W.W., Bechtold, W.E., and E.B. Wharton, Jr. 1995. Chromosome aberrations and response
to gamma-ray challenge in lymphocytes of workers exposed to 1,3-butadiene. Mutation
Research 334:125-130.
Autio, K., Renzi, L., and J. Catalan. 1994. Induction of micronuclei in peripheral blood and bone
marrow erthyrocytes of rats and mice exposed to 1,3-butadiene by inhalation. Mutation
Research 309(2):315-320.
Bechtold, W.E., Strunk, and M.R. Thornton-Manning, Jr. 1995. Analysis of butadiene,
butadiene monoxide and butadiene diepoxide in blood by gas chromatography/gas
chromatography/mass spectroscopy. Chemical Research Toxicology, 8:182-187.
Not published in the peer reviewed literature
R-l
-------
BIBRA. 1998. Brinkworth, M.H., Anderson, D., and J.A. Hughes. Genetic effects of
1,3-butadiene on the mouse testis. [BIBRA Mouse Study No. 3] Mutation Research -
Speciallssue 397(l):67-75.
BIBRA. 1995. The detection of dominant lethal mutations and foetal malformations, tumours
and chromosome damage in the offspring of male mice treated sub-chronically with
butadiene. BIBRA Report No. 1060/5/95 (Draft, April, 1995). [BIBRA Mouse Study
1] BIBRA International, Surrey, Great Britain.
BIBRA. 1996. The detection of dominant lethal mutations and foetal malformations and
chromosome damage in the offspring of male mice treated sub-chronically with butadiene
by inhalation - second study. BIBRA Report No. 1542/1/96. [BIBRA Mouse Study 2]
BIBRA International, Surrey, Great Britain.
Bond, J.A., Dahl, A.R., and R.F. Henderson. 1986. Species differences in the disposition of
inhaled butadiene. Toxicology and Applied Pharmacology 84:617-627.
Boogaard, P.J., and J.A. Bond. 1996. The role of hydrolysis in the detoxification of
l,2:3,4-diepoxybutane by human, rat, and mouse liver and lung in vitro. Toxicology and
Applied Pharmacology 141: 617-627.
Boogaard, P.J., Sumner, S.C.J., and J.A. Bond. 1996. Glutathione conjugation of
l,2:3,4-diepoxybutane in human liver and rat and mouse liver and lung. In vitro.
Toxicology and Applied Pharmacology 136:307-316.
Brooks, P. and P.D. Lawley. 1961. The reaction of mustard gas with nucleic acids in vitro and
in vivo. BiochemicalJournal, 77: 478-484.
CMA. 1998a. Comments of the Olefms Panel of the Chemical Manufacturers Association on
EPA's Draft Health Risk Assessment for 1,3-Butadiene, April 17, 1998 submitted to the
National Center for Environmental Assessment, USEPA, Washington, DC.
CMA. 1998b. Comments of the Olefms Panel of the Chemical Manufacturers Association on
EPA's Draft Health Risk Assessment for 1,3-Butadiene, April 29, 1998 submitted to the
National Center for Environmental Assessment, USEPA, Washington, DC.
Christian, M. 1998. Excerpts of M. Christian's Testimony to OSHA, Appendix Hln _Comments
of the Olefms Panel of the Chemical Manufacturers Association on EPA's Draft Health
Risk Assessment for 1,3-Butadiene, Chemical Manufacturers Association, April 17, 1998,
submitted to the National Center for Environmental Assessment, USEPA, Washington,
DC.
R-2
-------
Cochrane, I.E., and T.R. Skopeck. 1994. Mutagenicity of butadiene and its epoxide metabolites:
II. Mutational spectra of butadiene, 1,2-epoxybutene, and diepoxybutane at the hprt locus
in splenic T cells from exposed B6C3FJ mice. Carcinogenesis 15:719-723.
Cole, P., Delzell, E., and J. Acquavella. 1993. Exposure to butadiene and lymphatic and
hematopoietic cancer. Epidemiology 4:96-103.
Cowles S.R., Tsai, S.P., Snyder P, and C. Ross. 1994. From a cohort of long term workers
involved in 1,3-butadiene monomer production. Occupational and Environmental
Medicine, 51:323-329.
Csanady, G.A., Guengerich, P.P., and J.A. Bond. 1992. Comparison of the biotransformation
of 1,3-butadiene and its metabolite, butadiene monoepoxide, by hepatic and pulmonary
tissue from humans, rats, and mice. Carcinogenesis 13:1143-1153.
Csanady G.A., et al. 1996. A physiological toxicokinetic model for 1,3-butadiene in rodents and
man: blood concentration of 1,3-butadiene, its metabolically formed epoxides and
hemoglobin adducts - relevance of glutathione depletion. Toxicology 113:300-305.
Dahl, A.R., Sun, D.J., and L.S. Birnbaum. 1991. Toxicokinetics of inhaled 1,3-butadiene in
monkeys: comparison to toxicokinetics in rats and mice. Toxicological and Applied
Pharmacology, 110:9-19.
Delzell, E., Sathiakumar, N., Macaluso, M., Hovinga, M., Larson, R., Barbone, F., Beall, C., and
P. Cole. 1995. A follow-up study of synthetic rubber workers. Final report prepared
under contract to International Institute of Synthetic Rubber Producers, October 2, 1995.
Delzell E, Sathiakumar N., and M. Hovinga. 1996. A follow-up study of synthetic rubber
workers. Toxicology, 113:182-189.
Divine B.J., and C.M. Hartman. 1996. Mortality update of butadiene production workers.
Toxicology 113:169-181.
Evelo, C.T.A., Oostendorp, J.G.M., and W.F. ten Berge. 1993. Physiologically based
toxicokinetic modeling of 1,3-butadiene lung metabolism in mice becomes more important
at low doses. Environmental Health Perspectives 101:496-502.
Foureman, P., Mason, J.M., and R. Valencia. 1994. Chemical mutagenesis testing in Drosophila.
Environmental and Molecular Mutagenesis 23:51-63.
Hallenbeck, W. 1992. Cancer risk assessment for the inhalation of 1,3-butadiene using
physiologically based pharmacokinetic modelling. Bulletin of Environmental
Contamination and Toxicology 4:66-70.
R-3
-------
Hattis, D. and J. Wasson. 1987. Pharmacokinetic/mechanism-based analysis of the carcinogentic
risk of butadiene, PB88-20817, Rockville, MD: National Institute for Occupational Safety
and Health, 1-122.
Hayes, R.B., Xi, L, and W. Bechte. 1996. Hprt mutation frequency among workers exposed to
1,3-butadiene in China. Toxicology 113:100-105.
Hazelton. 1981. 1,3-butadiene: inhalation tetratogenicity study in the rat (final report),
Harrowgate, England: Hazelton Laboratory, addendum no. 2788-522/3
Health Canada. 1997. Canadian Environmental Protection Act Priority Substances List,
Supporting Documentation Health-Related Sections, 1,3-Butadiene, Draft, December
1997.
Himmelstein, M.W., Turner, M.J., and B. Asgharian. 1994. Comparison of blood concentrations
of 1,3-butadiene and butadiene epoxides in mice and rats exposed to 1,3- butadiene by
inhalation. Carcinogenesis, 15:1479-1486.
Himmelstein, M.W., Acquavella, J.F., Recio, L., Medinsky, M.A., and J.A. Bond. 1997.
Toxicology and Epidemiology of 1,3-Butadiene, CRC Reviews in Toxicology. 27:1-108.
Himmelstein, M. 1998. Comments on Chapter 3 — Metabolism and Pharmacokinetics, Health
Risk Assessment for 1,3-Butadiene (External Review Document), submitted by the
Chemical Manufacturers Association to Dr. Aparna M. Koppikar, Project Manager for
1,3-Butadiene, Technical Information Staff (8623-D), National Center for Environmental
Assesment - Washington Office, Washington, DC, April 29, 1998
IARC. 1998. IARC Monographs on the evaluation of carcinogenic risks to humans,
re-evaluation of some industrial chemicals. 71 [106-99-0] in preparation. The IARC
Working Group that evaluated 1,3-butadiene met on February 17-24, 1998 in Lyons,
France. The findings were presented at the EHC's April 30-May 1, 1998 meeting by Dr.
James Bond of the Chemical Industry Institute of Toxicology.
Irons, R., Colagiovanni, D., and W. Stillman. 1996. Murine thymic lymphoma is associated with
a species-specific hematopoietic progenitor cell subpopulation. Toxicology 113:59-67
Irons, R. and Pyatt, D. 1998. Dithiocarbamates as potential confounders in butadiene
epidemiology. Carcinogenesis 19, 4:101-104.
Johanson, G. and J.G. Filse. 1993. A physiologically based pharmacokinetic model for butadiene
and its metabolite butadiene monoxide in rat and mouse and its significance for risk
extrapolation. Archives of Toxicology, 67:151-163.
R-4
-------
Koivisto, P., Sorsa, M., and F. Pacchierotti. 1997. 32P-postlabelling/HPLC assay reveals an
enantioselective adduct formation in N7 guanine residues in vivo after 1,3-butadiene
inhalation exposure. Carcinogenesis 18( 2):439-443.
Kohn M.C. and R.L. Melnick. 1996. Effects of the structure of a toxicokinetic model of
butadiene inhalation exposure on computed production of carcinogenic intermediates.
Toxicology 113:31-39.
Kohn, M.C. and R.L. Melnick. 1993. Species differences in the production and clearance of 1,3-
butadiene metabolites: a mechanistic model indicates predominantly physiological, not
biochemical, control. Carcinogenesis, 14:619-628.
Kreiling, R., Laib, RJ., and H.M. Bolt. 1986. Alkylation of nuclear proteins and DNA after
exposure of rats and mice to (l,4-super(14C)l,3-butadiene). Toxicology Letters
30(2):131-136.
Kumar, R., Vodicka, P., and P. Koivisto. 1996. 32-P-postlabelling of diastereomeric 7-alkyl-
guanine adducts of butadiene monoepoxide. Carcinogenesis 17:1297-1303.
Lawley, P.D.,and P.B. Brooks. 1967. Interstrand cross-linking of DNA by difunctional
alkylating agents. Journal of Molecular Biology 25:143-160.
Leavens, T.L., Farris, G.M., and R.A. James. 1997. Genotoxicity and cytotoxicity in male
B6C3F1 mice following exposure to mixtures of 1,3-butadiene and styrene.
Environmental Molecular Mutagenesis 29:335-345.
Leuratti, C., Jones, N.J., and E. Marafante. 1994. DNA damage induced by the environmental
carcinogen butadiene: Identification of a diepoxybutane-adenine adduct and its detection
by super (32) P-postlabelling. Carcinogenesis 15(9): 1903-1910.
McGregor, D., Brown, A., and P. Cattanach. 1991. Responses of the L5178Y mouse lymphoma
forward mutation assay: v. gases and vapors. Environmental and Molecular
Mutagenesis 17:122-129.
Macaluso M., Larson R., and E. Delzell. 1996. Leukemia and cumulative exposure to butadiene,
styrene and benzene among workers in the synthetic rubber industry. Toxicology
113:190-202.
Maki-Paakkanen et al. 1993. Cited in: Norppa, H. and M. Sorsa. 1993. Genetic toxicity of
1,3-butadiene and styrene. IARC Sci. Publication No. 127:185-193.
R-5
-------
Matanoski G.M., Santos-Burgoa, C., and L. Schwartz. 1990. Mortality of a cohort of workers
in the styrene-butadiene polymer manufacturing industry (1943-1982). Environmental
Health Perspective 86:107-117.
Matanoski G., Francis M., Correa-Villasenor A., Elliot E., Santos-Burgoa, C., and L. Schwartz.
1993. Cancer epidemiology among styrene-butadiene rubber workers. In: SorsaM,
Peltonen K, Vainio H, Hemminki K, eds. Butadiene and Styrene: Assessment of Health
Hazards. Baltimore: IARC Scientific Publications No. 127:363-374.
Matanoski, G., Elliot, E., and X. Tao. 1997. Lymphohematopoietic cancers and butadiene and
styrene exposure in synthetic rubber manufacture. Ann NY Acad Sci pp. 157-168.
Matanoski, G., Elliot, E., Tao, X., Francis, M., Correa-Villasenor, A., and C. Santos-Burgoa.
1998. Lymphohematopoietic Cancers and Butadiene and Styrene Exposure in Synthetic
Rubber Manufacture, unpublished. Johns Hopkins University, Department of
Epidemiology, Baltimore, Maryland.
Medinsky, M.A., Leavens, T.L., and G.A. Csanady. 1994. in vivo metabolism of butadiene by
mice and rats: a comparison of physiological model predictions and experimental data.
Carcinogenesis 15:1329-1340.
Meinhardt, T.J., Lemen, R.A., and M.S. Crandall. 1982. Environmental epidemiologic
investigations of the styrene-butadiene rubber industry: mortality patterns with discussion
of the hematopoietic and lymphatic malignancies. The Scandinavian Journal of Work,
Environment, and Health 8: 250-259.
Melnick, R.I., Huff, I.E., and BJ. Chou. 1990. Carcinogenicity of 1,3-butadiene in
C57BL/6xC3HFl mice at low exposure concentrations. Cancer Research 50:6592-6599.
Meng, Q., Recio, L., and A. Reilly. 1996. The in vivo mutagenic potential of 1,3-butadiene,
evaluated at the hprt locus of T-cells, is greater in mice than in rats exposed concurrently
by inhalation. Abstract presented at the Twelfth Health Effects Institute Annual
Conference.
Meng, Q., Henderson, R., Dahl, A., and V.E. Walker. 1997. Species comparisons of the
mutagenic potency of epoxybutane and diepoxybutane in hprt of splenic T-cells following
exposures of female F344 rats and B6C3F1 mice. 13th Health Effects Institute
Conference, May 4-6, 1997, Annapolis, MD (Abstract p. 37).
Moolgavkar, S.H. 1994. Biological models of carcinogenesis and quantitative cancer risk
assessment. Risk Analysis 14(6):879-882.
R-6
-------
Morrissey, R.E., Schwetz, B.Z., and P. Hackett. 1990. Overview of reproductive and
developmental toxicity studies of 1,3-butadiene in rodents. Environmental Health
Perspectives 86:79-84.
NAS. 1988. Health Risks of Radon and Other Internally Deposited Alpha-Emitters (BEIR IV).
National Academy of Sciences/National Research Council. Washington, DC. National
Academy Press.
NAS. 1990. Health Effects of Exposure to Low Levels of Ionizing Radiation (BEIR V).
National Academy of Sciences/National Research Council. Washington, DC. National
Academy Press.
Neagu, I, Koivisto, P., aand C. Neagu. 1994. Alkylation of purine bases with 3,4-expoxy-l-
butene, a reactive metabolite of 1,3-butadiene. IARC Sci. Publication No. 125:419-422.
Neagu, I, Koivisto, P., and C. Neagu. 1995. Butadiene monoxide and deoxyguanosine
alkylation products at the N7-position. Carcinogenesis 16:1809-1813.
Nelson, H., Wiencke, J., and D. Christian!. 1995. Ethnic differences in the prevalence of the
homozygous deleted genotype of glutathione S-transferase theta. Carcinogenesis
16(5):1243-1245.
NTP. 1993. National Toxicology Program (NTP). U.S. Public Health Service, U.S.
Department of Health and Human Services. Toxicology and Carcinogenesis studies of
1,3-butadiene (CAS No. 106-99-0) in E6C3Fl mice (inhalation studies). NTP TR 434,
ResearchTriangle Park, North Carolina.
Pacchierotti, F. 1998. Genetic effects of 1,3-butadiene and associated risk for heritable damage.
Mutation Research. - Special Issue 397(1):93-115.
Pacchierotti, F., Tiveron, C., and R. Ranald. 1998. Reproductive toxicity of 1,3-butadiene in the
mouse:cytogenetic analysis of chromosome aberrations in first-cleavage embryos and flow
cytometric evaluation of spermatogonial cell killing. Mutation Research - Special Issue,
397(l):55-66.
Recio, L., Osterman-Golkar, S., and G. Csnady. 1992. Determination of mutagenicity in tissues
of transgenic mice following exposure to 1,3-butadiene and N-ethyl-N- nitrosourea.
Toxicology and Applied Pharmacology 117:58-64.
Recio, L. and T. Goldsworthy. 1995. The use of transgenic mice for studying mutagenicity
induced by 1,3-butadiene. Toxicology Letters 82:607-12
R-7
-------
Reitz, R.H., Gargas, M.L., and S.M. Hays. 1996. Expansion of the CUT PBPK model for
butadiene (Description of diepoxide kinetics). Research Report submitted to EPA May
29, 1996.
Seaton, M.J., Follansbee, M.H., and J.A. Bond. 1995. Oxidation of l,2-epoxy-3-butene to
l,2:3,4-diepoxybutane by cDNA-expressed human cytochrome P450 2E1 and 3A4 and
human, mouse and rate liver microsomes. Carcinogenesis 16:2287-2292.
Seilken, R. 1998. Sielken Report-II: Corrected Calculations of the Benchmark Concentrations
(BMCs) for Male-Mediated Developmental Toxicity, Appendix E in Comments of the
Olefms Panel of the Chemical Manufacturers Association on EPA's Draft Health Risk
Assessment for 1,3-Butadiene. Chemical Manufacturers Association, April 17, 1998.
Selzer, R. and A. Elfarra. 1996. Characterization of N super(l)- and N super(6)-adenosine
adducts and N super(l)-inosine adducts formed by the reaction of butadiene monoxide
with adenosine: Evidence for the N super(l)-adenosine adducts as major initial products.
Toxicology, 9(5):875-881.
Sernau, J., Cavagnaro, J., and P. Kehn. 1986. 1,3-butadiene as an S9 activation-dependent
gaseous positive control substance in L5178Y cell mutation assays. Environmental
Mutagenesis, Suppl. 8:75-76 (Abstract 203).
Sorsa, M., Osterman-Golkar, S., and K. Peltonen. 1996. Assessment of exposure to butadiene in
the process industry. Toxicology 113:77-83.
Sorsa, M., Autio, K., and N.A. Demopoulous. 1994. Human cytogenic biomonitoring of
occupational exposure to 1,3-butadiene. Mutat. Res. 309:321-326.
Sweeney L.M. 1996. Physiologically-based pharmacokinetic modeling of blood: tissue epoxide
measurements for butadiene. Toxicology 113:318-321.
Sweeney L.M. 1997. Physiologically-based pharmacokinetic modeling of 1,3-butadiene,
1,2-epoxybutene and 1,2,3,4-diepoxybutane toxicokinetics in mice and rats.
Carcinogenesis 18:611-625.
Swenberg, J.A., Christova-Gueorguieva, N.I., Scheller, N., and A. Ranasinghe. 1998.
Quantitation of expoxybutene and trihydroxybutane hemoglobin adducts within exposed
and control mice, rats and humans. Presented at the Health Effects Institute Annual
Conference, Boston, MA, April 5-7, 1998.
Tates, A.D., van Dam, F.J., and A.T. Natarajan. 1994. Frequencies of hprt mutants and
micronuclei in lymphocytes of cancer patients under chemotherapy: A prospective study.
Mutation Research, 309(1):299-306.
R-8
-------
Tates, A.D., van Dam, F.J., and F.A. de Zwart. 1996. Biological effect monitoring in industrial
workers from the Czech Republic exposed to low levels of butadiene. Toxicology
113:91-99.
Tates, A.D., and FJ. van Dam. 1998. Comparison of induction of hprt mutations by
1,3-butadiene and/or its metabolites 1,2-epoxybutene and 1,2,3,4- diepoxybutane in
lymphocytes from spleen of adult male mice and rats in vivo. Mutation Research - Special
Issue 397(\):2\-36.
Tretyakova, N., Lin, Y., and R. Sangaiah. 1997. Identification and quantitation of DNA adducts
from calf thymus DNA exposed to 3,4-epoxy-l-butene. Carcinogenesis 18(1): 137
Tretyakova, N.Y., Ealker, V.E., and J.A. Swenberg. 1998. Formation and persistence of DNA
adducts in liver of rats and mice exposed to 1,3-butadiene. Toxicological Science
42:180.
Thornton-Manning, J.R., Dahl, A.R., and W.E. Bechtold. 1995a. Disposition of butadiene
monoepoxide and butadiene diepoxide in various tissues of rats and mice following a
low-level inhalation exposure to 1,3-butadiene. Carcinogenesis 16:1723-1731.
Thornton-Manning, J.R., Dahl, A.R., and W.E. Bechtold. 1995b. Gender differences in the
metabolism of 1,3-butadiene in Sprague-Dawley rats following low-level inhalation
exposure. Carcinogenesis 16:2875-2878.
Thornton-Manning, J.R., Bechtold, W.E., and R.F., and R.F. Griffith. 1996. Tissue
concentrations of butadiene epoxides in rodents following multiple and single exposures to
1,3-butadiene (BD) by inhalation (Abstract). Toxicologist 30:215.
Tommasi, A.M., deConti, S., and M.M. Dobrzynska. 1998. Evaluation and characterization of
micronuclei in early spermatids of mice exposed to 1,3-butadiene. Mutation Research -
Special Issue 387(l):45-53.
USEPA. 1985. Mutageni city and carcinogeni city assessment of 1,3-butadiene. Office of Health
and Environmental Assessment, Office of Research and Development, Washington, DC.
EPA/600/8-85/004F.
USEPA. 1996. EPA's proposed guidelines for carcinogen risk assessment. USEPA, Office of
Research and Development, Washington, DC. Federal Register 61(79):17960-18011.
R-9
-------
USEPA. 1994. Pilot-Scale Evaluation of the Potential for Emissions of Hazardous Air Pollutants
from Combustion of Tire-Derived Fuel, U.S. Environmental Protection Agency, Control
Technology Center, EPA-600/R-94-070 (NTIS PB-94-169463), April 1994.
USEPA. 1997. 1995 Toxics Release Inventory (TRI) - Public Data Release Book. USEPA,
Washington, DC, April 1997. EPA745-R-97-005. [The TRI is available on the TOXNET
computer system. Registered National Library of Medicine online services users can
access the TOXNET system via the modem using the FTS Toll-free number
(1-800-525-0216) or through the Internet (Telnet directly @ toxnet.nlm.nih.gov).
USEPA. 1998a. Health Risk Assessment of 1,3-Butadiene, External Review Draft, USEPA
Office of Research and Development, EPA/600/P-98/001A, January 1998.
USEPA. 1998b. 1996 Toxic Release Inventory - Public Data Release - 10 Years of Right to
Know. USEPA, Washington, DC. EPA745-R-98-005, May 1998.
Valentine, J.L., Boogaard, P.J., Sweeney, L.M., Turner, MJ,. Bond, J.A., and M.A. Medinsky.
1997. Disposition of butadiene epoxides in Sprague Dawley rats. Chemico. -Biological
Interactions 103-115.
Victorin, K., Busk, L., and H. Cederberg. 1990. Genotoxic activity of 1,3-butadiene and
nitrogen dioxide and their photochemical reaction products in Drosophila and in the
mouse bone marrow micronucleus assay. Mutation Research 228:203-209.
Ward, J.B., Jr., Au, W.W., and M.M. Ammenheuser. 1994. Hprt Mutant lymphocyte frequencies
in workers at a 1,3-butadiene production plant. Environmental Health Perspectives
102:79-85.
Ward E.M., Fajen J.M., Ruder A.M., Rinsky R., Halperin W., and C. Fessler-Flesch . 1996.
Mortality study of workers employed in 1,3-butadiene production units identified from a
large chemical workers cohort. Toxicology 113:157-168.
Wiencke, J.K., Pemble, S., Ketterer, B., and K.T. Kelsey. 1995. Gene deletion of glutathione S-
transferase theta: correlation with induced genetic damage and potential role in
endogenous mutagenesis. Cancer Epidem Biomarkers andPrev 4:253-259.
Xi, L., Zhang, L., and Y. Wange. 1997. Induction of chromosome-specific aneuploidy and
micronuclei in human lymphocytes by metabolites of 1,3-butadiene. Carcinogenesis
18(9):1687-1693.
Xiao, Y., and A.D. Tates. 1995. Clastogenic effects of 1,3-butadiene and its metabolites 1,2-
epoxybutene and 1,2,3,4-diepoxybutane in splenocytes and germ cells of rats and mice in
vivo. Environmental and Molecular Mutagenesis, 26(2): 97-108.
R-10
-------
DISTRIBUTION LIST
Administrator
Deputy Administrator
Assistant Administrators
EPA Regional Administrators
Director, National Center for Environmental Assessment
Director, Office of Science Policy, ORD
EPA Laboratory Directors
EPA Headquarters Library
EPA Regional Libraries
EPA Laboratory Libraries
Library of Congress
National Technical Information Service
Congressional Research Service
------- |