EPA-450/5-82-003
RESPONSE TO PUBLIC COMMENTS
ON ERA'S LISTING OF BENZENE UNDER SECTION 112
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
May 1984
-.;.;'.. :.Y. i- , .Yfc. : OV. 1070
...jo, II, t>J«.04
-------
NOTE: Commenters are identified by the rulemaking docket numbers assigned
their submissions. The relevant dockets are as follows:
OAQPS-79-3 Benzene Listing and Maleic Anhydride Standard
A-79-27 Benzene Fugitive Emissions Standard
A-79-49 Ethylbenzene/Styrene Standard
A-80-14 Benzene Storage Vessel Standard
OAQPS-79-14 Airborne Carcinogen Policy
A number of commenters submitted multiple or duplicate comments to several of
the dockets. While EPA has made an effort to reference representative comment
sources, the citations should not be considered an exhaustive record of the
docket items addressing a particular issue.
-------
Table of Contents
1. INTRODUCTION 1
2.0 THE LISTING OF BENZENE UNDER SECTION 112 2
2.1 The Timing of the Benzene Listing Decision 3
2.1.1 The Pre-Listing Record 3
2.1.2 Relevance of Listing to the Proposed Airborne
Carcinogen Policy 6
2.1.3 The Requisite Intent to Propose Regulations 8
2.2 The Health Effects of Benzene 9
2.2.1 Reproductive and Teratogenic Effects 10
2.2.2 Chromosomal Effects 11
2.2.3 Carcinogeni city 14
2.3 The Health Basis for Listing 15
2.3.1 Epidemiological Studies 16
2.3.1.1 Infante et a]_. study 16
2.3.1.2 Aksoy et~aT_. studies 20
2.3.1.3 Ott etfah study 22
2.3.1.4 Other epidemiological studies 25
2.3.1.5 Epidemiological studies released following
the close of the comment period 27
2.3.2 Animal Studies 28
2.3.3 In Vitro Studies 31
2.4 Health Issues Relevant to the Benzene Listing Decision 31
2.4.1 The Non-Threshold Hypothesis 32
2.4.1.1 EPA's position on carcinogenic thresholds 32
2.4.1.2 Support for a carcinogenic threshold for
benzene 38
2.4.1.2.1 benzene metabolism studies 39
2.4.1.2.2 threshold-governed preconditions for
leukemia 40
2.4.1.2.3 epidemiology 45
2.4.1.3 EPA's conclusions 48
2.4.2 The Quantitative Estimation of Carcinogenic Risk 48
2.4.2.1 The assessment of human exposure 50
-------
2.4.2.2 The linear, non-threshold, dose response
model 54
2.4.2.3 Derivation of the unit risk factor for
benzene 56
2.4.3 Significance of the Estimated Carcinogenic Risk
from Benzene Exposure 64
2.5 Other Issues Relevant to the Listing of Benzene 70
2.5.1 The Adequacy of Other Standards Controlling Benzene..70
2.5.2 Selection of a Benchmark or De Minimi's Risk Target...72
2.6 EPA's Conclusions on the Listing of Benzene 72
3. THE SELECTION OF BENZENE SOURCE CATEGORIES FOR REGULATION 74
3.1 Selection of Five Source Categories for Initial
Regulation 75
3.2 Proposal of Standards: Significant Risk Judgment 75
3.3 Post-Proposal Review of Significant Risk Judgment 81
-------
1. INTRODUCTION
This document provides EPA's response to public comment on the listing
of benzene as a hazardous air pollutant under Section 112 of the Clean Air
Act. The document also includes EPA's response to comments on relevant
procedural issues raised in the course of the benzene rulemakings.
On June 8, 1977, EPA added benzene to the list of hazardous air
pollutants.1 On April 18, 1980, EPA proposed standards to limit benzene
emissions from maleic anhydride plants.2 With the proposal of the maleic
anhydride standard and in accordance with the provisions of Section 112,
the Agency invited public comment on the benzene listing decision. The
comment period for the maleic anhydride standard closed November 6, 1980.
EPA subsequently proposed emission standards for ethylbenzene/styrene
plants3 (December 18, 1980), benzene storage vessels4 (December 19, 1980),
and sources of benzene fugitive emissions5 (January 5, 1981). The last
of these comment periods closed on September 14, 1981. The benzene listing
1|J.S. Environmental Protection Agency, National Emission Standards for
Hazardous Air Pollutants. "Addition of Benzene to List of Hazardous Air
Pollutants." 42 FR 29332, June 8, 1977.
2U.S. Environmental Protection Agency, "Benzene Emissions from Maleic
Anhydride Plants; Proposed Rule" 45 FR 26660, April 18, 1980.
3U.S. Environmental Protection Agency, "Benzene Emissions from
Ethylbenzene/Styrene Plants; Proposed Rule and Public Hearing Announcement"
45 FR 83448, December 18, 1980.
^U.S. Environmental Protection Agency, "Benzene Emissions from Benzene
Storage Vessels; Proposed Rule and Notice of Public Hearing" 45 FR 83592,
December 19, 1980.
5U.S. Environmental Protection Agency, "Benzene Fugitive Emissions; Pro-
posed Rule and Notice of Public Hearing" 45 FR 1165, January 5, 1981.
-------
2
decision was addressed by commenters on the ethylbenzene/styrene, storage,
and fugitive regulations, as well as the maleic anhydride proposal.
Although certain comments are duplicated in two or more of the rulemaking
dockets, EPA has attempted to provide a complete listing of docket citations.
2.0 THE LISTING OF BENZENE UNDER SECTION 112
EPA listed benzene as a hazardous air pollutant based on "[scientific
reports [which] strongly suggest an increased incidence of leukemia in workers
exposed to benzene."! These reports included a review of benzene by the
National Academy of Sciences (NAS)2, updated criteria published by the
National Institute for Occupational Safety and Health (NIOSH)3 and a proposal
by the Occupational Safety and Health Administration (OSHA) for a revision
downwards of the existing workplace standard for benzene.4 While
acknowledging that ambient exposure to benzene normally occurs at levels
"substantially lower than those to which affected workers were exposed," EPA
maintained that "there is reason to believe that ambient exposures may
constitute a cancer risk and should be reduced."5
At the time of listing, EPA announced that it would review the scientific
data to determine the health risks from exposure to ambient levels of
1U.S. EPA, "Addition of Benzene to List of Hazardous Air Pollutants" 42 FR
29332, June 3, 1977.
^National Academy of Sciences, Health Effects of Benzene: A Review.
Washington, D.C., June 1976.
^National Institute for Occupational Safety and Health, "Update Criteria and
Recommendations for a Revised Benzene Standard," September 1976.
"^Occupational Safety and Health Administration, "Occupational Exposure to
Benzene; Emergency Temporary Standards" 42 FR 22516, May 3, 1977.
5EPA, (1977) p. 29332.
-------
3
benzene and invited public participation. The resulting EPA reports:
"Assessment of Health Effects of Benzene Germane to Low Level Exposures,"6
"Assessment of Human Exposures to Atmospheric Benzene,"? and "Carcinogen
Assessment Group's Final Report on Population Risk to Ambient Benzene
Exposures"^ form the basis for the majority of the public comments
directed at the listing decision.
Comments, largely from potentially affected industries and trade
associations, argued that the listing of benzene was ill-timed, unnecessary,
aid unjustified. The main thrusts of these arguments are that EPA failed to
develop an adequate record in advance of listing and that the record
subsequently prepared does not demonstrate that benzene at the levels
encountered in the ambient air warrants designation as a hazardous air
pollutant. Information presented in support of this position is summarized
in the sections that follow, along with EPA's response.
2.1 The Timing of the Benzene Listing Decision
2.1.1 The Pre-Listing Record
Several commenters felt that EPA did not make a convincing argument for
the listing of benzene as a hazardous air pollutant in the June 1977 notice
(A-79-27-IV-D-17-28,IV-F-l,IV-K-l). One reviewer contended that the "risk
assessments which [EPA] developed only after deciding to list should have
6U.S. EPA "Assessment of Health Effects of Benzene Germane to Low-Level
Exposures," Office of Health and Ecological Effects, Office of Research and
Development. September, 1978 (EPA-600/1-78-061).
7U.S. EPA, "Assessment of Human Exposures to Atmospheric Benzene," Office
of Air Quality Planning and Standards, June 1978, EPA 450/3-78-031.
8|J.S. EPA, "Carcinogen Assessment Group's Final Report on Population Risk
to Ambient Benzene Exposures," Roy Albert, Chairman, Carcinogen Assessment
Group, January 10, 1979, (EPA-450/5-80-004).
-------
4
been developed beforehand to inform [EPA's] decision" (OAQPS-79-3(Part II)-
IV-F-1 ,IV-F-9). Without such assessments performed in advance of listing,
"EPA had little or no basis for judging whether the health risk posed by
ambient benzene exposure was worthy of any regulation at all" (OAQPS-79-3(Part
O-IV-D-13; A-79-49-IV-D-9). Another commenter felt the "inadequate record"
constituted a "fatal flaw" in the timing of EPA's decision (A-79-27-IV-0-19).
«
One commenter noted that the lapse of over three years between the listing
of benzene and the proposal of the first emission standard "strongly suggested
. . . [t]hat the Administrator did not have the evidence to justify the June
1977 listing" (A-79-49-IV-D-11).
EPA Response:
The Clean Air Act requires EPA to list as hazardous air pollutants under
Section 112 those substances judged to cause or contribute to air pollution
"which may reasonably be anticipated to result in an increase in mortality or
an increase in serious, irreversible, or incapacitating reversible, illness."1
EPA based the decision to list benzene on a growing consensus in the scientific
and regulatory community, as evidenced by reports by the National Academy of
Sciences^, the National Institute for Occupational Safety and Health3, and
proposed regulations issued by the Occupational Safety and Health
Ifhe Clean Air Act as amended 1977, p. 37.
^National Academy of Sciences, Health Effects of Benzene: A Review.
Washington, D.C., June 1976.
^National Institute for Occupational Safety and Health, "Update Criteria and
Recommendations for a Revised Benzene Standard," September 1976.
-------
5
Administration^, that benzene was causally linked to the occurrrence of
leukemia in occupationally exposed populations. In EPA's judgment, leukemia
clearly fits the criteria described in Section 112 as a "serious, irreversible,
or incapacitating reversible, illness."
EPA's judgment that benzene present in the ambient air could "reasonbly
be anticipated" to pose a significant health hazard to the general population
relied on two arguments advanced in the listing notice: first, that benzene
was released to the air in hundred million pound quantities annually to which
"large numbers of people are routinely exposed" and, second, that EPA had
"adopted a regulatory policy which recognizes that some risk exists at any
level of exposure to carcinogenic chemicals."^ The latter referred to the
"Interim Procedures and Guidelines for Health Risk and Economic Impact
Assessments of Suspected Carcinogens" published by EPA in May, 1976.6
Based on the above, EPA believes that the decision to list benzene was
appropriate, fully-informed, and timely. The subsequent assessments of low-
level exposure and carcinogenic risk were intended, as indicated in the
listing notice, for use in "determining which sources of benzene emissions
must be controlled, and the extent of control needed."^ To the extent that
these documents addressed the criteria for listing benzene under Section 112,
they have affirmed EPA's decision.
^Occupational Safety and Health Administration "Occupational Exposure to
Benzene; Emergency Temporary Standards" 42 FR 22516, May 3, 1977.
5U.S. EPA, "Addition of Benzene to List of Hazardous Air Pollutants" 42
FR 29332, June 8, 1977.
6U.S. EPA, "Interim Procedures and Guidelines: Health Risk and Economic
Impact Assessments of Suspected Carcinogens," 41 FR 21402, May 25, 1976.
7EPA, 1977, p. 29333.
-------
6
EPA rejects the contention that the delay between listing and the proposal
of emission standards for benzene sources suggests that EPA lacked the
scientific evidence to justify the June 1977 listing. EPA's assessments of
the health effects of low-level exposure**, the extent of human exposure9
and the estimation of population risks^ were submitted for external review
before EPA's Science Advisory Board in December 1977, and publicly released in
•
September 1978, June 1978, and January 1979, respectively. The first emission
standard for benzene sources*! was not proposed until April 1980. Proposal
was not delayed by the lack of "evidence" for listing but rather by the complex
task of developing national emission standards.
2.1.2 Relevance of Listing to the Proposed Airborne
Carcinogen Policy
Three commenters maintained that the listing and rulemaking proceedings
for benzene was premature since they were based on a proposed EPA policy
regarding airborne carcinogens12 (A-79-27-IV-D-8,IV-D-25,IV-D-26; OAQPS-79-3
(Part D-IV-D-l.D-ll; A-80-14-IV-D-4,IY-D-11; A-79-49-IV-D-7). One commenter
felt that to proceed in advance of a final carcinogen policy would violate
8U.S. EPA, "Assessment of Health Effects of Benzene Germane to Low-Level
Exposures" Office of Health and Ecological Effects, Office of Research and
Development, September, 1978 ( EPA-600/1 -78-061 ).
9U.S. EPA, "Assessment of Human Exposures to Atmospheric Benzene," Office
of Air Quality Planning and Standards, June 1978, EPA 450/3-78-031.
1°U.S. EPA, "Carcinogen Assessment Group's Final Report on Population Risk to
Ambient Benzene Exposures," Roy Albert, Chairman, January 10, 1979, (EPA
450/5-80-004).
I.S. EPA, "Benzene Emissions from Maleic Anhydride Plants; Proposed Rule"
45 FR 26660, April 19, 1980.
12U.S. EPA, "Policy and Procedures for the Identification, Assessment, and
Regulation of Airborne Substances Posing a Risk of Cancer, Notice of
Proposed Rulemaking" 44 FR 58642, October 10, 1979.
-------
7
Section 307 (General Provisions Relating to Administrative Proceedings and
Judicial Review) of the Clean Air Act and Section 533 of the Administrative
Procedures Act (A-79-27-IV-D-8). One commenter felt that EPA exceeded its
legal authority and offended "good scientific practice" in utilizing the pro-
posed carcinogen policy to list benzene (A-79-27-IV-D-25; A-80-14-D-4,17-0-11).
EPA Response:
The listing of benzene (June 8, 1977) occurred prior to proposal of the
airborne carcinogen policy (October 10, 1979) and thus was not based on the
proposed policy. While the proposed benzene emission standards were developed
in general consistency with the proposed policy, standard development was not
based on the proposed policy, and the standard development methodology was
presented to be judged on its own merits.
EPA noted in the proposed policy that the procedures outlined would
"generally" be followed for actions taken in the interim between proposal and
promulgation of the policy. As a proposal, the policy does not bind EPA to
the procedures described, nor does EPA perceive a need to review previous
actions on the basis of the procedures outlined in this proposal. Further,
EPA recognizes that defense of a regulatory decision such as listing cannot be
based on a proposed policy and, thus, has not done so. However, EPA has made
decisions on benzene consistent with portions of the proposed policy. Still,
the listing of benzene and the proposal of emission standards are separate,
independent rulemakings.
Finally, EPA maintains that the appropriate standard against which to
evaluate the listing of benzene is Section 112. As described above, EPA is
persuaded that the decision to list benzene under Section 112 was neither
premature nor in excess of the Agency's legal authority.
-------
8
The commenter did not describe how EPA violated Section 307 of the Clean
Air Act or Section 533 of the Administrative Procedures Act, nor do the
statutory sections cited by the commenter appear to bear any relation to the
commenter's concern about the listing of benzene.
2.1.3 The Requisite Intent to Propose Regulations
One commenter asserted that the listing of benzene was invalid in that EPA
did not have "the requisite intent to propose regulations within 180 days"
(OAQPS-79-3(Part D-IV-D-10; A-79-49-IY-D-11). In support of this position,
the commenter referenced portions of the listing notice in which EPA solicited
information on the health effects of benzene exposure and identified
regulatory authorities other than the Clean Air Act that could be used in
addition to or in lieu of Section 112 to control benzene emissions. The
commenter considered these statements as evidence of EPA's "uncertainty" and
lack of intent.
EPA Response:
At the listing of benzene, EPA expressed the intent to propose "emission
control regulations as soon as possible."12 As the commenter points out, EPA
stated in the listing notice that it had "tentatively concluded that emission
reductions from some chemical manufacturing facilities, petroleum refineries,
and coke ovens may be necessary."I3 Neither the Clean Air Act nor
considerations of fairness support the proposition that the listing should be
invalidated if the Agency is unable to propose standards within the statutory
timetables.
12EPA, 1977, p. 29333.
-------
9
The notice of listing, while announcing EPA's intent to regulate the
identified source categories, also acknowledges the availability of other
regulatory authorities that could be used to control benzene emissions. The
Agency's willingness to consider alternative regulatory strategies at any
point in the regulatory process should not be considered a lack of intent but,
rather, a demonstration of flexibility. For the case in point, Section 112
specifically provides for a public hearing following proposal of emission
standards to entertain arguments that the substance is clearly not a hazardous
air pollutant.
EPA does not believe that the intent to regulate under Section 112 must be
free from uncertainty. In this regard, the Agency's request for information
to aid in the "responsible determination ... as to which sources of benzene
emissions must be controlled, and the extent of control needed"14 reflects not
the lack of intent but its execution.
2.2 The Health Effects of Benzene
Public comments on the EPA report "Assessment of Health Effects of Benzene
Germane to Low-Level Exposure"1 focused on those areas of the benzene health
literature relevant to the evaluation of human health risks from ambient
exposure. These include: effects on reproduction and development (embryo-
toxicity and teratogenicity), effects on the cellular genetic material
(mutagenicity and chromosome breakage), and carcinogenicity. Although
other benzene-related effects, including conditions of the blood and blood-
forming system (hematotoxicity) cannot be ruled out as possible consequences
14Ibid.
ILLS. EPA, "Assessment of Health Effects of Benzene Germane to Low-Level
Exposure", Office of Health and Ecological Effects, Office of Research and
Development, September, 1978 (EPA-600/1-78-061).
-------
10
of chronic ambient exposure to benzene, EPA believes that the potential
seriousness of the genotoxic effects and the absence of identifiable
thresholds support a primary emphasis on these effects.
2.2.1 Reproductive and Teratogenic Effects
EPA concluded in the health assessment report that the health literature
"reports of effects of benzene on embryos are conflicting and inconclusive and
hence are not useful in evaluating the possibility that low ambient concen-
trations of benzene might have an effect on human embryogenesis."2 Based on
available animal studies of birth defects, EPA also concluded that "a role
for benzene in teratogenesis cannot be predicted with confidence at this
time."3
Some commenters took a stronger stance. One noted that "[n]o human
embryotoxicity or teratogenicity has ever been associated with any level of
benzene exposure. No animal study has shown that exposure to even high levels
of benzene caused a significant increase in the frequency of teratogenic
events" (OAQPS-79-3(Part I)-IV-D-13). While admitting that deficiencies
existed in the data base that needed to be filled, one commenter concluded
that "the available data do not identify an adverse effect of benzene on
reproductive capacity" (OAQPS-79-3(Part IJ-IV-D-9, (Part II)-IY-0-22,IV-F-1,
IV-F-8).
EPA Response:
EPA agrees with the commenters that the available data do not, at this
time, implicate benzene as a potential teratogen or embryotoxin in test
species. The risks of adverse fetal developmental or reproductive effects,
2Ibid, p. 20.
3Ibid.
-------
11
however, have not been adequately studied. No state-of-the-art multiple
generation reproduction studies involving benzene have been done and without
such studies it is not possible to determine the levels at which benzene
would have no observed effect.
From the available data concerning adverse reproductive effects of benzene
in humans, it is not possible to conclude that JTO_ adverse human reproductive
consequence results from ambient levels of benzene, since no well-designed and
executed epidemiological studies have been conducted. It is not known if
ambient levels of benzene have effects on many areas of human reproduction,
i.e., the processes of spermatogenesis, alteration in libido, changes in
menstrual cycle, change in the age of puberty or menarche, death of conceptus
manifested in early spontaneous abortion, and latent behavioral effects on the
neonate. Until such possibilities are explored, EPA believes that the
evidence for benzene-induced reproductive effects in humans must be regarded
as inconclusive.
2.2.2 Chromosomal Effects
The EPA health assessment concludes that "chromosome breakage and
rearrangement can result from exposure to benzene," that "[a] dose-dependent
relationship between exposure to benzene and amount of chromosome damage has
not been demonstrated," and that "[theoretical considerations and some
clinical observations suggest a relationship between chronic benzene ex-
posure, chromosome damage, and leukemia.'"*
While commenters did not disagree with EPA's conclusion that benzene can
cause chromosome breakage in humans, they were divided on the exposure levels
at which such damage occurred and on the implications of the observed
4U.S. EPA, (1978) p. 21.
-------
12
changes. One commenter maintained that such effects occurred only after
"excessive exposure" to benzene (OAQPS-79-3(Part IIJ-IV-F-1,F-8). Another
commenter asserted that these effects "result only from exposures in excess
of 10 parts per million" (OAQPS-79-3(Part I)-IV-D-13;A-79-27-IV-D-27; A-79-
49-IV-D-9).
One commenter challenged EPA's conclusion that a dose-dependent
relationship between benzene exposure and chromosome damage had not been
demonstrated, citing a study by Picciano^ in benzene-exposed workers. The
commenter maintained further that this study documented chromosomal effects at
exposure levels of benzene at and below 2.5 ppm'(OAQPS-79-3(Part I)-IV-D-8).
une commenter argued that "no reliable evidence" existed to link subclinical
benzene exposure to chromosome aberrations or to relate the observation of
chromosome breakage with human leukemia (OAQPS-79-3(Part I)-IV-D-13, (Part II)-
IV-F-l.F-8; A-79-49-IV-D-9).
EPA Response:
EPA does not agree that the data on human cytogenetic effects support a
conclusion that benzene-induced chromosome damage occurs only after "excessive
exposure." As described in the health assessment document, studies are
available which relate an increase in chromosome breakage to benzene exposure
well below the occupational standard of 10 ppm time-weighted average (TWA).6»7
, D. "Monitoring Industrial Populations by Cytogenetic Procedures"
in Proceedings of a Workshop on Methodology for Assessing Reproductive Hazards
in the Workplace, P.P. Infante and M.S. Legator eds. April 19-22, 1978.
^Kilian, D.J., and Daniel, R.C. "A cytogenetic study of workers exposed to
benzene in the Texas Division of Dow Chemical, U.S.A." February 27, 1978.
7Picciano, D. "Cytogenetic Study of Workers Exposed to Benzene" Env. Res.
19:33-38, 1979.
-------
13
With respect to a dose response relationship, EPA agrees that the Picciano
study indicates a dose-dependent relationship between exposure to benzene and
the amount of chromosome damage. As noted in the EPA health assessment
document, however, "[t]here is no correlation, . . . , between the degree or
length of exposure, the clinical symptoms, and persistence or extent of
chromosomal aberrations" [emphasis added].8 EPA believes that this study and
the study by Kilian and Daniel are appropriately considered evidence of an
association between benzene exposure and chromosome breakage and that the
lowest benzene levels at which significant increases in breakage were found
(1.0-2.5 ppm) are properly considered to reflect exposures below those
associated with clinical (here, hematotoxic) symptoms of toxicity.
EPA agrees that no direct evidence of a causal linkage between chromosomal
aberrations and leukemia exists. EPA remains concerned, however, by the
frequency of reports correlating chromosome abnormalities with cancer
incidence. In addition to benzene workers and leukemia, this association has
been pointed out in atomic bomb survivors with leukemia,9 in uranium miners
with lung cancer, in vinyl chloride workers with liver cancer, in luminous
dial painters with bone cancer, and in individuals developing visceral cancers
after methotrexate treatment for psoriasis.10
8U.S. EPA, (1978) p. 19.
^Bloom, A.D., Y. Nakagone, A. Awa, and S. Neriishi "Chromosome aberrations and
malignant disease among A-bomb survivors" Am. J. Public Health 60:641-644,
1970.
l°Mulvihi11, J.J. in Persons at High Risk of Cancer ed. Fraumeni, J.F. Jr.
Academic Press, N.Y., 1975, pp. 3-37.
-------
14
2.2.3 Cardnogenicity
Commenters did not challenge EPA's conclusion that "there is substantial
epidemiological evidence that benzene is a human leukemogen.''^ A number of
commenters, however, disagreed with EPA's conclusion that benzene posed an
increased risk of leukemia at the levels present in the ambient air. EPA
addresses these comments in Section 2.4 - Health Issues Relevant to the Benzene
Listing Decision.
One commenter took issue with EPA's conclusion that "there is no
convincing evidence that benzene causes neoplasias, including leukemia, in
animals,"12 citing two studies, one by Maltoni and Scarnato^ and one by
Snyder £t al_. I4 demonstrating benzene-induced tumors in rodents (OAQPS-79-
3(Part D-IV-D-8).
EPA Response:
The carcinogenicity studies on benzene in animals reported by Maltoni and
Scarnato (1979) and Snyder et al_. (1980) support the comment that a positive
tumorigenie effect of benzene is evident from these studies. The results of
these studies are addressed further under Section 2.3.2.
.S. EPA, "Carcinogen Assessment Group's Final Report on Population Risk
to Ambient Benzene Exposures", Roy Albert, Chairman, January, 1979
(EPA 450/5-80-004).
12u.S. EPA (1978) p. 1.
i, C. and C. Scarnato "First Experimental Demonstration of the
Carcinogenic Effects of Benzene" Estratto da "La Medicina del Lavoro"
70:5, 1979.
r, Carroll A. et al . "The Inhalation Toxicology of Benzene: Incidence
of Hematopoietic Neoplasms and Hematotoxicity in AKR/J and C57BL/6J Mice" Tox.
and Appl. Pharm. 54:323-331, 1980.
-------
15
2.3 The Health Basis for Listing
In June 1976, a committee of the National Academy of Sciences, in a report
commissioned by EPA, concluded that "based on the available literature, . . .
benzene must be considered as a suspect leukemogen."! The following year,
the Occupational Safety and Health Administration (OSHA) determined in its
issuance of an emergency temporary standard for occupational exposure to
benzene^ (May 3, 1977), that "accumulated studies strongly support the
conclusion that benzene causes leukemia in humans."^ These reports provided
scientific support for EPA's subsequent determination to list benzene as a
hazardous air pollutant.^
While leukemia is only one of several adverse health effects attributed
to benzene, the serious consequences of this disease and the uncertainties
regarding the existence of any risk free levels of exposure combine to make
it of central importance in any regulatory decision. EPA's health basis
for listing rested primarily on retrospective studies in occupationally
exposed human populations. Of these, three reports documenting an association
received greatest emphasis: Infante et_ aH_. ,5 Aksoy et_ aj_. ,6 and
^•National Academy of Sciences, Committee on Toxicology. "Health Effects
of Benzene: A Review" June, 1976.
^Occupational Safety and Health Administration" Occupational Exposure to
Benzene; Emergency Temporary Standards" 42 FR 22516, May 3, 1977a.
3Ibid, p. 22517.
4U.S. EPA, "National Emission Standards for Hazardous Air Pollutants; Addition
of Benzene to List of Hazardous Air Pollutants" 42 FR 29332, June 8, 1977.
5Infante, P.F., R. Rinsky, J. Wagoner, and R. Young, "Leukemia in Benzene
Workers" Lancet 2:76-78, 1977a.
6Aksoy, M., S. Erdem, and G. Dincol, "Types of Leukemia in Chronic Benzene
Poisoning, a Study in Thirty-four Patients. Acta Hematol. 55:65-72, 1976.
-------
16
Ott et_!l«7 In the interval since listing, animal data and additional
human data have become available which further support a causal relationship.
Commenters critical of EPA's decision to list benzene argued that the
three epidemiological studies suffered from design and methodological
flaws, the correction of which would tend to greatly reduce if not eliminate
the observed association. Several commenters also felt that EPA had
misinterpreted the study results and ignored other well-conducted studies
that reached significantly different conclusions.
The following sections summarize and respond to public comments on the
relevant epidemiological, animal, and in vitro studies, omitting those
comments addressing the issues of an effect threshold and the quantitative
estimation of risks which will be taken up in Section 2.4.
2.3.1 Epidemiological Studies
2.3.1.1 Infante et_ aj_. study
The work by Infante et_ aj_., a retrospective cohort mortality study
undertaken by the National Institute of Occupational Safety and Health
(NIOSH), was initially reported in 1977 with a completed follow-up published
in 1981.2 The study found a greater than five-fold excess risk of leukemia
among workers exposed to benzene during the period 1940-1949 in the "Pliofilm"
(rubber hydrochloride) production industry.
One commenter stated that the Infante work was "seriously flawed and
largely discredited," citing testimony from the public hearings on the OSHA
70tt, M.G., J.C. Townsend, W.A. Fishback, and R.A. Langner, "Mortality Among
Individuals Occupationally Exposed to Benzene" Exhibit 154, OSHA Benzene
Hearings, July 19-August 10, 1977.
8Rinsky, R.A., R. Young, and A. Smith, "Leukemia in Benzene Workers", American
Journal of Industrial Medicine 2:217-245, 1981.
-------
17
benzene standard^ and the Supreme Court's plurality decision on the OSHA
standard*0 (OAQPS-79-3(Part II)-IV-D-5;A-79-27-IV-D-8). More specifically,
commenters asserted that the study was flawed in two respects: the exposed
cohort was improperly defined; and the exposure levels assumed were erroneous
(OAQPS-79-3(Part I)-IV-D-13,(Part II)-IV-D-5,IV-F-1,IV-F-9;A-79-27-IV-D-8;A-
79-49-IV-D-9;A-80-14-IV-D-4,IV-D-l6).
One commenter maintained that the exposed cohort selected for the study
inappropriately excluded certain mechanical and "dry side" workers as well as
an unknown number of workers that left the plants' employment prior to 1944.
The commenter argued that the "dry side" group, in which no excess leukemia
had been observed, could have been exposed to benzene at levels that might
have reached 20 ppm (OAQPS-79-3(Part IIJ-IV-F-1,IV-F-9).
The same commenter expressed "major concern" with the study's "implication
that exposure levels were low." In support of the latter position, the commenter
supplied information from studies indicating that, in contrast to the authors'
conclusion that exposure levels during the study period were roughly equivalent
to prevailing occupational standards (100-10 ppm), the workers could have
been exposed to levels of 100-1000 ppm in the 1940's and as high as 355 ppm in
the 1970's with a mean of 30 ppm (OAQPS-79-3 (Part II)-IV-F-9;A-79-27-IV-D-24,
IV-F-1).
EPA Response:
Though EPA recognizes that the Infante et al. study has weaknesses, EPA
^Occupational Safety and Health Administration, Docket #H-059, Occupational
Exposure to Benzene, Proposed Standard, Transcript of Public Hearing, July 19-
August 10, 1977b.
10Supreme Court of the United States. Industrial Union Department v. American
Petroleum Institute et a]_., 448 U.S. 607 (1980).
-------
18
believes the characterization of the study as "seriously flawed and largely
discredited" is both unfair and untrue. Although the commenter does not
provide explanation of his criticism beyond references to the OSHA benzene
rulemaking, his remarks imply that the study is invalid due to erroneous
reporting of the exposure concentrations. EPA acknowledges, as did the
authors of the study, that the historical exposure levels cannnot be determined
with certainty. This fact, however, is irrelevant to the conclusion of the
study that exposed workers experienced a five-fold excess risk of leukemia
over the general population.
The issue of cohort definition in Infante et_ a]_. was discussed in
subsequent publications by the authors^»12 as well as the OSHA benzene
rulemaking.13 The authors argue that "dry side" workers "were never intended
for inclusion in the cohort following discussions with company personnel
indicating there was no benzene exposure on the dry side."I4 Subsequent
reports of benzene levels (three sample points) on the "dry side" by the
University of North Carolina^ were regarded as inadequately detailed "to
permit a valid interpretation."!^ The authors also contend that maintenance
^Infante, P.P., R. Rinsky, J. Wagoner, and R. Young, "Benzene and Leukemia"
the Lancet, October 22, 1977b.
12White, M.C., P. Infante, and 8. Walker, Jr. "Occupational Exposure to Benzene:
A Review of Carcinogenic and Related Health Effects Following the U.S. Supreme
Court Decision" Am. J. Indust. Med. l"233-243, 1980.
13OSHA, "Occupational Exposure to Benzene; Occupational Safety and Health
Standards" 43 FR 5918, February 10, 1978.
14Ibid, p. 5927.
^Environmental Survey. Occupational Health Study Group, University of
North Carolina, School of Public Health, 1974.
16Infante, 1977b, p. 868.
-------
19
personnel (pipefitters, mechanics) were appropriately excluded from the co-
hort "because company records did not show which men had responsibilities
in pliofilm production."^ Workers who left employment prior to 1944 "could
not be included because their personnel records were not in a retrievable
form."18
EPA considers the rationale for the selection of the Infante et_ a]_. cohort
as appropriate. EPA notes further that, as described in the completed follow-
up by Rinsky et_ _al_. as well as expert testimony offered by Dr. Marvin Sakol
at the OSHA benzene hearings,^ the strict cohort definition excludes several
additional cases of leukemia which "supports further the notion that there
existed a causal link between benzene exposure in those facilities and the
occurrence of leukemia."2*^
Rinsky et_ aj_. provides a thorough discussion of the available information
on the levels of benzene to which workers may have been exposed in the subject
facilities during the periods studied. The authors conclude that, "for the
most part, employees' 8-hour time-weighted averaged exposures were within the
recommended [occupational] standard in effect at the time. However, as is
characteristic of industrial processes, there were occasional excursions above
these limits."21 EPA concludes that, while intermittent levels may have
approached the values suggested by the commenters, the range of occupational
17Ibid.
18Ibid.
19OSHA, 1977b, Exhibit no. 61.
20Rinsky, p. 244.
21Ibid, p. 238.
-------
20
standards for the periods studied (100-10 ppm) appears reasonable as an
estimate of the chronic exposure pattern. In this regard, EPA agrees with the
recent conclusion of the Benzene Work Group of the International Agency for
Research on Cancer (IARC) that "the excessive mortality from myelogenous and
monocytic leukemia had occurred among workers with occupational exposure to
benzene that was generally within accepted limits," recognizing that "the
possible contribution of the occasional excursions in exposure and of the
employment of some workers in other areas of the plant must be noted; and
may have made some contribution to the observed excess in mortality from
2.3.1.2 Aksoy et_ aj_. studies
Aksoy et_ aK studied the incidence of leukemia and other diseases among
workers occupationally exposed to benzene in the Turkish shoeworking
industry. 23, 24, 25 Based on case ascertainment by leukemia in the exposed
population compared to estimates for the general population of Western nations,
Aksoy et^ ail_. found a two-fold excess risk of leukemia among shoeworkers
with chronic exposure to benzene.
^International Agency for Research on Cancer, Monographs on the Evaluation of
the Carcinogenic Risk of Chemicals to Humans, 29:93-148, May 1982.
, M., K. Dincol, T. Akgun, S. Erdem, and G. Dincol "Haematological
effects of chronic benzene poisoning in 217 workers" Br. J. Ind. Med. 28:296-
302, 1971.
24Aksoy, M., K. Dincol, S. Erdem, and G. Dincol "Acute leukemia due to chronic
exposure to benzene" Am. J. Med. 52:160-166, 1972.
25Aksoy M., S. Erdem, and G. Dincol "Leukemia in shoeworkers exposed
chronically to benzene" Blood 44(6):837-841 , 1974.
-------
21
Many of the comments on the Aksoy et^ aj_. study relate to its use in the
assessment of population risks by EPA's Carcinogen Assessment Group. These
comments are addressed in Sections 3.4.1 and 3.4.2.
Although commenters generally agreed that the study was of value "in
reaffirming . . . that prolonged exposures to high concentrations of benzene
result in serious blood disorders including a small number of leukemias"
(OAQPS-79-3(Part IIJ-IV-F-1,IV-F-9), there were several specific criticisms
suggesting that the excess risk observed was exaggerated. Two commenters
argued that Aksoy et_ al_. relied on inappropriate figures (6 per 100,000) for
the background leukemia incidence, and that when a more reasonable estimate
derived from the experience of the European Standard Population (8-14 per
100,000) was used, the study no longer showed an excess incidence among the
exposed workers (OAQPS-79-3(Part D-IV-D-13, (Part II)-IV-F-1 ,IV-F-9; A-79-49-
IV-D-9). One commenter expressed concern that the age distribution of exposed
workers was not available and speculated that the margin for error in the
"official count" used as the denominator of the shoeworking population
(28,500) was "probably substantial" (OAQPS-79-3(Part IIJ-V-F-1.IV-F-9).
EPA Response:
EPA agrees that Aksoy's choice of the 6 per 100,000 background leukemia
incidence is subject to criticism since it is not easily attributed to the
Turkish rural population. It is also reasonable that the "official count" of
28,500 shoeworkers may be an underestimate and therefore overestimates the
excess leukemogenic risk in the exposed population. It is equally likely,
however, that Aksoy's methodology leads to an underestimate of the excess
risk. First, only those cases of leukemia of which the author was directly
aware as a medical practictioner were counted in the study. As Aksoy
testified before OSHA "undoubtedly there were other additional patients among
-------
22
shoeworkers who were not included in our study."26 Second, as EPA's health
assessment points out, "the distribution of cases reported by Aksoy et al.
strongly differs from that of leukemia in the general population. If the
relative incidence were computed solely for acute myeloblastic leukemia and
its variants [the forms of leukemia associated with benzene exposure], a
magnification of the risk in benzene-exposed shoeworkers would be observed."27
Finally, Aksoy has also indicated in testimony that the rural incidence
of leukemia in Turkey may be on the order of 3.0 per 100,000 or half of
what he had estimated originally.2^ This fact would also lead to an increase
in the calculated excess risk.
On the matter of the age distribution of the shoemaker population, EPA
agrees that such information would be helpful. As discussed in more detail in
Section 3.4, the limited age information available led EPA to incorporate an
age adjustment factor in the Agency's risk assessment. On the basis of new
information on the age structure of the male population of Turkey,2^ EPA now
believes this adjustment was unnecessary.
2.3.1.3 Ott et. al. study
Ott et_ al_.3° reported long-term mortality patterns and associated benzene
26OSHA, 1977b, Exhibit 61, p. 2.
27u.S. EPA "Assessment of Health Effects of Benzene Germane to Low-Level
Exposure," Office of Research and Development, EPA-600/1 -78-061 , September,
1978, p. 71.
28OSHA, 1977b, Exhibit 61, p. 2.
.S. Department of Labor, Bureau of the Census "County Demographic
Profiles, Turkey" ISP-DP-25, August, 1980.
300tt, G., J. Townsend, W. Fishbeck, and R. Lanqner "Mortality among
individuals occupationally exposed to benzene" Dow Chemical Company,
Midland, Michigan, 1977.
-------
23
exposure for a cohort of 594 chemical manufacturing workers. Three cases of
leukemia were observed where 0.8 were expected, an excess risk of 3.75. The
finding was statistically significant (p=0.047) in a one-tailed test of
significance.
One commenter criticized the statement in EPA's health assessment-^ that
the excess-leukemia incidence observed in the Ott et_ jjl_. study was only of
"borderline" statistical significance. The commenter noted that "[s]ince the
p value observed (0.047) is less than the p value (0.050) commonly used to
determine statistical significance, there is no basis for considering the
value borderline" (OAQPS-79-3(Part I)-IV-D-8). Other commenters argued
that the study should be appropriately regarded as "inconclusive"
(OAQPS-79-3(Part I)-IV-D-9,IV-D-13,(Part II)-IV-D22,IV-F-1.IV-F-9; A-79-49-
IV-D-9,IV-F-2). One commenter remarked that while the leukemia cases
were too few to draw "solid statistical conclusions," the Ott et a!.
study was the "best documented study of chronic exposures to benzene in
the literature to date" (OAQPS-79-3(Part II)-IV-F-l,F-9).
Commenters also contended that the exclusion of one decedent whose
leukemia was identified as a "significant other condition" rather than the
cause of death, eliminated the statistical significance of the study (OAQPS-
79-3(Part I)-IV-D-13). One commenter asserted that Ott _et_ J_]_. applied an
"inappropriate one-tailed [statistical] test" to determine significance and
that the use of an appropriate test (two-tailed) did not reveal a significant
association between the leukemia cases and exposure to benzene (OAQPS-79-
3(Part I)-IV-D-13).
31U.S. EPA, 1978, p. 83.
-------
24
The presence of confounding exposures to other potential carcinogens was
also noted by commenters as evidence that the study should not be viewed as
conslusive of a benzene-leukemia association. The same commenters noted that
the cases of leukemia occurred in workers exposed to lower levels of benzene
(2-9 ppm) than those encountered by many other individuals in the study
population (OAQPS-79-3(Part I)-IV-D-13,(Part II)-IV-F-l,IV-F-9).
EPA Response:
While EPA does not view the Ott et_ aj_. study, taken alone, as conclusive
evidence of an association between low-level (2-9 ppm) occupational exposure
to benzene and leukemia, the Agency believes that this work, in combination
with other findings in the benzene health literature, serves to reinforce the
public health concerns regarding benzene exposure.
EPA does not agree that the use of "borderline" in describing the
significance of the Ott et_ aj_. study is inappropriate since the value
calculated (.047) was very close to the predetermined limit (0.050) chosen
for the test. EPA does agree that the test, as constructed, supports a
finding of significance.
EPA disagrees that the use of a "two-tailed" test for significance would
be more appropriate than the "one-tailed" test employed by Ott et_ a\_. The
hypothesis to be tested is that benzene exposure increases the risk of
leukemia, not that the risk may increase or decrease. The benzene health
literature does not support a finding that benzene exerts a protective
influence on exposed individuals.
Omitting from the study the individual for which leukemia was not the
immediate cause of death, would not, in EPA's opinion, be an appropriate
change. In view of the well-established association between benzene and
-------
25
non-lymphatic leukemias, EPA believes that a case of myelogenous leukemia,
such as this, should not be ignored.
EPA does not view the extent of confounding exposures in Ott et_ aj_. as
severe. The authors did exclude from their analysis persons known to have
been exposed to levels of arsenicals, vinyl chloride, and asbestos, all of
which have been associated with human health effects. This eliminated 53
persons from consideration including one leukemia victim. The remaining
substances have not been shown to be associated with a risk of leukemia in
either man or animals. The inclusion of such exposed persons, therefore,
would not be likely to affect the target organ site for benzene in terms
of an increased risk.
According to the testimony of Ott et_ aU before OSHA, the "low levels of
potential benzene exposure relative to other employees in the cohort .
made a retrospective assessment of the possible relationship to benzene
exposure very judgmental."32 EPA, while recognizing this uncertainty, agrees
with the reservation expressed by OSHA in its benzene rulemaking that,
"because of the small population size as well as the possibility of
sensitivity of those individuals developing leukemia, it cannot be concluded
that these deaths are not caused by benzene exposure."33
2.3.1.4 Other epidemiological studies
Commenters cited other epidemiological studies in occupational populations,
notably the work of Thorpe,3^ in which no correlation between leukemia and
32OSHA, 1977 (b) Exhibit 154, p. 12.
33OSHA, 1978, 5928.
34Thorpe, J.J. "Epidemiologic survey of leukemia in persons potentially
exposed to benzene" J. Occup. Med. 16(6):375-382, 1974.
-------
26
benzene exposure was demonstrated (OAQPS-79-3 (Part I )-IV-D-9,IVD-13,(Part
II)-IV-F-1, IV-F-9; A-79-27-IV-D-24, IV-F-1; A-79-49-IV-D-9.IV-F2). The
Thorpe study found "no excess incidence of leukemia among petroleum
workers exposed to benzene levels estimated to range up to 20 ppm" (OAQPS-79-
3(Part I)-IV-D-13).
One commenter maintained that the epidemiological data do not support
EPA's position on benzene, citing a table prepared by the American Council on
Science and Health which indicated that the industrialized state of New Jersey
"has a lower leukemia mortality rate than other areas" (OAQPS-79-3(Part II)-
IV-D-24).
EPA Response:
EPA believes that the deficiencies in the Thorpe study preclude a judgment
that exposure to benzene below 20 ppm poses no risk of leukemia. The author
of this study dwells at considerable length on the shortcomings of the work,
the most important of which are 1) quantitative determinations of the extent
of exposure could not be done, 2) there was inadequate follow-up of members of
the cohort, and 3) problems existed with verification of the diagnosis of
leukemia.
Follow-up was left to each unit (plant) separately. Since the author did
none of it himself, the follow-up was poor. Many units had no mechanisms by
which the plant was notified of the death of an annuitant and, where
notification was made, often no cause of death was reported to the company.
Cases reported among annuitants were included, although it was recognized that
there may have been underreporting in this group. No mention was made
concerning follow-up efforts on former employees who did not qualify for an
annuity. Unfortunately, no table on completeness of ascertainment of vital
status was given.
-------
27
Other problems with this study involve the questionable practice of
reporting the pooled results of a study of eight separate and perhaps
considerably different plants. A significant risk that may have been present
in one or more of the plants could have been obscured by the inclusion of
populations of nonexposed individuals. Additionally, no consideration of
latent factors was presented; no effort was made by the author to require a
minimum time since onset of employment of individuals in the study or to
provide even cause-specific mortality by time since first employment.
Furthermore, the study has been criticized by Brown3^ on the basis of
possible underreporting of leukemia in the study population.
EPA agrees that no epidemiological data currently exist that link a
higher incidence of leukemia in a given community to industrial sources
of benzene. While the commenter did not provide the full reference cited
for his conclusion that New Jersey had "lower leukemia rates of white
males" than other areas, EPA does not regard this as good evidence of the
lack of an association. Certainly, if several different communities are
compared on the grounds of leukemia incidence alone, it would be likely,
by chance alone, that one or more would exhibit higher rates.
2.3.1.5 Epidemiological studies released following the close
of the comment period.
In December 1983, the Chemical Manufacturers Association submitted
to EPA and other regulatory agencies a mortality study of chemical workers
exposed to benzene.3** The study examined the causes of death for 7676
35Brown, S. "Letter to the editor concerning Thorpe article on leukemia and
potential benzene exposure" JOM 17(l):5-6, 1975.
36Chemical Manufacturers Association "An Industry-Wide Mortality Study of
Chemical Workers Exposed to Benzene" Environmental Health Associates, Inc.
December 8, 1983.
-------
28
chemical workers employed for at least 6 months between 1946 and 1975.
The 4602 employees from seven plants with a history of occupational
exposure to benzene were divided into two groups dependent on whether
their benzene exposures were "continuous" or "intermittent." The levels of
benzene exposure were estimated by company industrial hygienists. The
control group of 3074 workers with no known occupational exposure to
benzene was selected from the rolls of the same chemical plants.
The CMA study found significant increases in cancer deaths among the
exposed cohorts when compared to the control group. Deaths rates were
elevated for the general category of lymphopoietic cancers as well as for
sub-categories of leukemia and non-Hodgkin's lymphopoietic cancer.
A relative risk factor for leukemia could not be computed due to the absence
of cases in the control cohort although seven leukemia cases were observed
in the exposed cohort.
Comparison of cumulative exposure to benzene (part-per-million
months) with mortality from leukemia and all lymphopoietic cancers revealed
statistically significant dose response relationships.
2.3.2 Animal Studies
EPA originally concluded in the benzene health assessment that "there is
no convincing evidence that benzene causes neoplasias, including leukemia,
in animals."37 One commenter submitted that two animal studies had become
available demonstrating benzene-induced tumors in rodents (OAQPS-79-3(Part I)-
IV-D-8).
EPA Response:
EPA agrees that the reports by Maltoni and Scarnato and Snyder et al
37U.S. EPA (1978) p. 1.
-------
29
support the comment that a positive tumorigenic effect of benzene is evident
from these studies. The finding of a tumorigenie effect of benzene in other
mammalian species serves to strengthen the concern over benzene's effects on
human populations. The experimental results are briefly described below.
Maltoni and Scarnato (1979)38
Untreated control and benzene-treated groups consisted of 30 to 35
male and 30 to 35 female Sprague-Dawley rats per group. The animals were 13
weeks old when the study began. Benzene in olive oil was delivered by gavage
4 to 5 times per week for 52 weeks at doses of 50 and 250 mg/kg. Animals
were allowed to live until spontaneous death, and animals were subjected
to examination by necropsy and histopathology. An increased incidence of
Zymbal gland carcinomas, mammary gland carcinomas, and leukemia in treated
animals was observed as follows:
TUMOR INCIDENCE (No. with Tumors/No. Examined)
Dose Sex Zymbal Gland Mammary Gland Leukemias0
Carcinomas0 Carcinomas*0
250 mg/kg
50 mg/kg
Untreated
male
f ema 1 e
male
female
male
female
0/33
8/32 (p = 0.003)
0/28
2/30
0/28
0/30
0/33
7/32
0/28
4/30
0/28
3/30
4/33
1/32
0/28
2/30
0/28
1/30
*8esides mammary gland carcinomas, the incidence of mammary gland
fibroadenomas in female rats was as follows: 16/30 in controls, 21/30 in
low-dose animals, and 9/35 in high-dose animals.
°Note: Denominators indicate corrected number of animals alive after 20
weeks, when the first tumor (mammary fibroadenoma) was observed. Additional'
ly, skin carcinomas, angiosarcoma, and hepatoma were limited to two high-
dose females, one high-dose male and one high-dose male, respectively.
38Maltoni, C. and C. Scarnato "First Experimental Demonstration of the
Carcinogenic Effects of Benzene" Estratto da "La Medicina del Lavoro"
70:5, 1979.
-------
30
Snyder et al. (1980)39
Six-week-old male mice were used. Treated and untreated control groups
each consisted of 50 AKR mice and 40 C57BL mice. Test animals were exposed to
100 ppm benzene (AKR) or 300 ppm benzene (C57BL) vapor 6 hours/day, 5
days/week, for life. Median survival times for control and treated AKR mice
were 39 and 41 weeks and of control and treated C57BL mice were 75 and 41
weeks, respectively. The animals were given necropsy and histopathologic
examinations.
A carcinogenic effect of benzene was not observed for AKR mice.
Hematopoietic neoplasms were found in 8/40 treated C57B1 mice vs. 2/40 control
C57B1 mice. Hematopoietic neoplasms in treated mice were categorized as
lymphocytic lymphoma with thymic involvement in six mice, plasmacytoma
(myeloma) in one mouse, and leukemia with a hematocytoblast apparently as the
predominant cell type in one mouse. The hematopoietic neoplasms in the
control mice were described as lymphomas with no thymic involvement.
With respect to C57BL mice, bone marrow hyperplasia and splenic hyper-
plasia were also found in 13/32 treatment vs. 0/38 control and 16/32 treatment
vs. 2/38 control animals, respectively.
NTP Bioassay (1983)40
The National Toxicology Program, in October 1983, released preliminary
results of a two year bioassay of benzene in rats and mice. The study
found significant increases in neoplasms at multiple sites in both sexes
39Snyder, C.A. et al. "The Inhalation Toxicology of Benzene: Incidence of
Hematopoietic Neoplasms and Hematotoxicity in AKR/J and C57BL/6J Mice" Tox.
and Appl. Pharm. 54:232-331, 1980.
40NTP Carcinogenesis Technical Report No. 289: "Bioassay of Benzene for
Possible Carcinogenesis," U.S. Department of Health and Human Services,
Board Draft, October 19, 1983. Peer review scheduled for June, 1984.
-------
31
of both species. Sites affected included the hematopoietic systems and
Zymbal glands in rats and mice; adrenal capsules, lung, liver, ovary, and
mammary glands in mice; and skin and oral cavity in rats.
2.3.3 In Vitro Studies
One commenter stated that studies performed in the early 1970's in the
USSR showed biological effects of benzene exposures down to 0.625 ppm. The
commenter asserted that these effects, rather than leukemia, should form
the basis for listing benzene (OAQPS-79-3(Part IIJ-IV-F-10).
EPA Response:
The study cited, though not clearly identified, involved decreased
phagocytic activity (the ability of cells to engulf foreign material).
Decreased phagocytic activity in the presence Df decreases in blood
concentrations of platelets, leukocytes, and red blood cells may be an
important indication of leukemogenic risk. Since decreased phagocytic
activity may be of nonspecific origin and may also be reversible, however,
this observation alone does not lend itself as the basis for a regulatory
decision. EPA believes this study requires confirmation before relying
on the reported results.
2.4 Health Issues Relevant to the Benzene Listing Decision
EPA listed benzene as a hazardous air pollutant based on evidence linking
occupational benzene exposure with leukemia and on the knowledge that large
numbers of people are exposed to and, therefore, may be at risk from, benzene
emitted into the ambient air by a variety of stationary sources. This
rationale assumes that 1) it is reasonable to conclude that a causal
relationship continues to exist at the significantly lower exposure levels
characteristic of the ambient air, and 2) that the magnitude of the
relationship warrants efforts to reduce human exposure.
-------
32
A number of commenters took issue with EPA's judgment, both in regard to
the listing of benzene and the proposal of the four emission standards,
arguing that an exposure threshold for benzene-induced leukemia exists
below which there is no health risk and that, even granting an association
with leukemia at ambient levels, the magnitude of the health risks to
exposed populations is negligible. Due to the number of commenters and
similarity of the comments, individual docket numbers are not listed.
The comments focus on EPA's presumption that effect thresholds do not
exist for carcinogens (the non-threshold hypothesis) and the methodology
used by EPA's Carcinogen Assessment Group (CAG) in deriving quantitative
estimates of benzene leukemogenic risks. A summary of public comments
addressing the issue of a carcinogenic threshold for benzene follows the
statement of EPA's position on carcinogenic thresholds.
2.4.1 The Non-Threshold Hypothesis
2.4.1.1 EPA's position on carcinogenic
thresholds
In evaluating the public health hazards associated with exposure to
carcinogens, EPA has maintained that in the absence of sound scientific
evidence to the contrary, such substances must be considered to pose
some finite risk of cancer at any exposure level above
!u.S. EPA, "Health Risk and Economic Impact Assessments of Suspected
Carcinogens; Interim Procedures and Guidelines" 41 FT 21401, May 25, 1976.
2u.S. EPA, "Policy and Procedures for the Identification, Assessment, and
Regulation of Airborne Substances Posing A Risk of Cancer; Notice of Pro-
posed Rulemaking" 44 FR 58642, October 10, 1979.
3Interagency Regulatory Liaison Group (IRLG): Consumer Product Safety
Commission (CPSC), Environmental Protection Agency (EPA), Food and Drug
Administration (FDA), and Food Safety and Quality Service (FSQS) "Scien-
tific Bases for Identification of Potential Carcinogens and Estimation
of Risks" 44 FR 39858, July 6, 1979.
-------
33
This conviction has been shared by, among others, the Occupational
Safety and Health Administration (OSHA),4 the Consumer Product Safety
Commission (CPSC), the Food and Drug Administration (FDA), the Food Safety
and Quality Service, the President's Regulatory Council,^ and the National
Academy of Sciences.^
Support for the non-threshold hypothesis for carcinogens derives from
both scientific and practical considerations. As summarized by the Interagency
Regulatory Liaison Group (IRL6):
"[t]he self-replicating nature of cancer, the multiplicity
of causative factors to which individuals can be exposed,
the additive and possibly synergistic combination of
effects, and the wide range of individual susceptibilities
work together in making it currently unreliable to predict a
threshold below which human population exposure to a
carcinogen has no effect on cancer risk."'
The mechanism by which a carcinogen acts is of particular importance in
postulating whether or not an effect threshold exists. The National Academy
of Sciences (NAS) has observed:
"Whether or not a particular effect follows a dose-
response relationship that has a threshold depends entirely
on the mechanism of the effect. Many effects have
thresholds. For example, the gastrointestinal-radiation
syndrome, acute drug toxicity, and radiation or drug control
of some tumors, all have dose-response curves that show
thresholds. The curves are sigmoid, and below a particular
4U.S. Occupational Safety and Health Administration "Identification,
Classification, and Regulation of Potential Occupational Carcinogens"
45 FR 5002, January 22, 1980.
5Regulatory Council "Statement on Regulation of Chemical Carcinogens;
Policy and Request for Public Comments" 44 FR 60038, October 17, 1979.
6Safe Drinking Water Committee, National Research Council "Drinking Water
and Health" National Academy of Sciences, Washington, D.C., 1977.
7IRLG, p. 39876.
-------
34
dose there is a zero probability of producing the effect,
response relationship that has a threshold depends entirely
on the mechanism of the effect. Many effects have
thresholds. For example, the gastrointestinal-radiation
syndrome, acute drug toxicity, and radiation or drug control
of some tumors, all have dose-response curves that show
thresholds. The curves are sigmoid, and below a particular
dose there is a zero probability of producing the effect,
because the effect requires many independent events and will
not occur until the number of such events exceeds some
critical value. The gastrointestinal-radiation (or drug)
syndrome is a case in point. An animal will not die until
the number of intestinal crypt cells that have been killed
exceeds a value that is critical to the integrity of the
organ. Any radiation or drug dose that kills fewer cells
than this critical number can be considered to be safe (at
least for this one syndrome).
We are used to thinking in terms of thresholds and
sigmoid dose-response curves. For example, if it costs
$4,000 to buy an automobile, we do not imagine that we will
have a 50% chance of buying the same vehicle for $2,000. If
100 aspirin tablets constitute a lethal dose, we do not
calculate that we will have a 1% chance of dying if we
swallow a single tablet. Because we know the mechanisms
underlying these events, we expect thresholds to the dose-
response curves, and indeed they are evident.
However, other effects may well not have threshold dose-
effect relationships. If an effect can be caused by a
single hit, a single molecule, or a single unit of exposure,
then the effect in question cannot have a threshold in the
dose-response relationship, no matter how unlikely it is
that the single hit or event will produce the effect.
Mutations in prokaryotic and eukaryotic cells can be
caused by a single cluster of ion pairs which were
produced by a beam of ionizing radiation. We would
expect that mutations can be caused by a single molecule
or perhaps group of molecules in proximity to the DNA.
The necessary conclusion from this result is that the
dose-response relationship for radiation and chemical muta-
genesis cannot have a threshold and must be linear, at least
at low doses.
It is one step further to correlate mutagenesis with
carcinogenesis. Nevertheless, the evidence is strong that
there is a close relationship between the two [references].
We therefore conclude that, if there is evidence that a
particular carcinogen acts by directly causing a mutation in
the DNA, it is likely that the dose-response curve for
carcinogenesis will not show a threshold and will be
linear with dose at low doses."8
8Safe Drinking Water Committee, p. 11-19.
-------
35
Additional evidence for a linear-carcinogenic response at low dose comes
from studies which suggest that cancers may arise from the "transformation" of
a single cell.^»^ As Crump summarized this work in testimony before OSHA:
"If individual cancers arise from an original, single,
transformed cell, then the statistical nature of the
carcinogenic dose response will be governed by the extreme
tail of the transformation response distribution. The
effect of this is to make virtually any process of discrete
events approximately linear at low dose. Two primary
observations indicate the single-cell origin of cancers. In
women who are heterozygous for electrophoretic variants of
X-linked glucose-6-phosphate dehydrogenase, cancers are
uniformly of one phenotype or the other, whereas a
comparable amount of normal tissue is composed of a mixture
of cells of the 2 phenotypic classes. Further evidence for
the single-cell origin of cancers comes from experimental
efforts in which transformed cells are transplanted into
whole animals. Although there is much controversy associated
with various aspects of this line of research, it seems
that the ability of a single cell to give rise to a cancer
is well demonstrated. Thus, 2 lines of evidence indicate
that cancer can be most reasonably assumed to arise from
events associated with or occurring inside single cells."*!
EPA's presumption that any exposure to a carcinogen poses a health risk is
not intended to foreclose discussion or ignore evidence of real or practical
effect thresholds for such substances. In this regard, there are a number of
theories that postulate the existence of thresholds. These include
consideration of the body's defense and repair capabilities (immuno-
surveillance, detoxification, and DNA repair) and reports of the regression
of pre-neoplastic lesions with the cessation of exposure. Observations of
an inverse relationship between dose and the latency period for tumor
^Fiaklow, P.J. "The origin and development of human tumors: studies with
cell markers" New Eng J. Med. 291:26-35, 1974.
l°Gartler, S.M. "Utilization of mosaic systems in the study of the origin
and progression of tumors" in Chromosomes and Cancer J. German, ed. Wiley
Interscience, New York p. 313-334, 1974.
UOSHA, p. 5126.
-------
36
expression have been proposed as evidence of practical thresholds where the
dose corresponds to a latency that exceeds the lifespan of the individual.
Proponents have also suggested, as indirect evidence of thresholds, the
carcinogenicity at high doses of certain substances for which a biological
requirement exists. Threshold levels have, in addition, been inferred from
"negative" epidemiological and animal studies.
While EPA agrees that the evidence for real or practical carcinogenic
thresholds should play a role in hazard evaluation, the Agency is persuaded
that the utility of such information in establishing "no effect" levels is
seriously limited. Although protective mechanisms such as DNA repair are
reasonably effective, it is generally recognized that few, if any, biological
processes are 100% efficient.^2 Similarly, while a decrease in dose could
possibly result in an increase in the median time-to-tumor to greater than a
lifespan, the typical distribution of tumors across age groups would still
result in "early" cancers arising.
The evidence for practical thresholds is also questionable. There is no
reason to believe that biologically required substances which have been found
to be carcinogenic at high levels may not pose some cancer risk at the levels
in which they are normally found in the body. In the same way, the failure to
detect a positive association in an animal bioassay or epidemiological study
does not constitute evidence of a no-effect level. As the National Academy of
Sciences (NAS) has noted:
"... the observation of no positive responses does not
guarantee that the probability of response is actually
zero. From a statistical viewpoint, zero responders out of
a population of size N is consistent at the 5% significance
level with an actual response probability between zero and
12OSHA, p 5126, 5129.
-------
37
approximately 3/N (e.g., when N = TOO and zero responders
are observed, the true probability of response may be as
high as 3%)."13
Finally, EPA concurs with the NAS that theoretical arguments for the
existence of carcinogenic thresholds must be tempered by the knowledge that
the exposed human population is a "large, diverse, and genetically
heterogeneous group exposed to a variety of toxic agents. Genetic variability
to carcinogenesis is well documented (Strong, 1976),L"143 and it is also known
that individuals who are deficient in immunological competence (for genetic or
environmental reasons) are particularly susceptible to some forms of cancer
(Cottier et al_-. 1974)[15].«16
OSHA noted in its summary of public hearings on an occupational
carcinogen policy:
"A number of witnesses testified that, even if thresholds
could be established for the circumstances in which animals
are exposed only to single carcinogens, this would have
little or no relevance to risk assessment for humans, who
are exposed to many carcinogens, either simultaneously or
sequentially. Specifically, several witnesses pointed out
that there is already a relatively high incidence of cancer
in the human population. Hence many individuals are already
at or close to the threshold for certain processes involved
in cancer development, so that incremental exposure to even
Drinking Water Committee, p. 11-26.
14Strong, L.C. "Susceptible subgroups" presented at NIEHS Conference on
the Problems of Extrapolating the Results of Laboratory Animal Data to Man
and of Extrapolating the Results from High-Dose Level Experiments to Low
Dose Level Exposures" Pinehurst, N.C., March 10-12, 1976.
15Cottier, H., M.W. Hess, H.U. Keller, P. Luscieti, and B. Sordat
"Immunological deficiency states and malignancy" in: Interaction of Radiation
and Host Immune Defense Mechanisms in Malignancy. Proceedings of a conference
at Greenbrier, W. Va., March, 1974, p. 30-44.
16Safe Drinking Water Committee, p. 11-21.
-------
38
small quantities of an agent that accelerates these
processes would be expected to lead to an increase in the
frequency of cancer."^
The NAS has further elaborated:
"In considering the possibility of thresholds for
carcinogenesis, it is important to understand that there is
no agent, chemical or physical, that induces a form of
cancer in man that does not occur in the absence of that
agent. In other words, when there is exposure to a
material, we are not starting at an origin of zero
cancers. Nor are we starting at an origin of zero
carcinogenic agents in our environment. Thus, it is likely
that any carcinogenic agent added to the environment will
act by a particular mechanism on a particular cell
population that is already being acted on by the same
mechanism to induce cancers. This reasoning implies that
only if it acted by a mechanism entirely different from that
already operating on the tissue could a newly added
carcinogen show a threshold in its dose-response curve."18
In summary, EPA is persuaded that the non-threshold hypothesis is, for
carcinogens, a reasonable and appropriate presumption that must be overcome by
sound scientific evidence before any exposure to such substances can be
concluded to be without health risk. At the same time, however, EPA regards
relevant evidence of the ability of biological systems to mitigate
carcinogenic insults as important considerations in the evaluation of
the health hazard.
2.4.1.2 Support for a carcinogenic
threshold for benzene
Comments challenging EPA's non-threshold presumption for benzene argued
that the Agency had failed to consider convincing evidence that a leukemogenic
threshold for benzene does exist and that this threshold was well above any
ambient levels that might be encountered by the general population. In
17OSHA,p.5135.
18Safe Drinking Water Committee, p. 11-20.
-------
39
support of this position, commenters cited studies of benzene metabolism,
alternative mechanisms for cancer induction, and evidence derived from
epidemiological studies.
2.4.1.2.1 Benzene metabolism
studies
One commenter cited the work of Rickert and Irons^ as evidence that
exposure to levels of benzene below 10 ppm does not produce any adverse health
consequences in human cells (OAQPS-79-3(Part I)-IV-D-13).
EPA Response:
Rickert studied benzene metabolism in rodents annd human cells in vitro to
determine the concentrations of toxic benzene metabolites that might occur in
the bone marrow of humans exposed to benzene.^ He concluded that the
metabolite concentrations in rats and human tissue are of the same order of
magnitude at similar benzene doses. Irons used this information to compare
the metabolite concentrations expected at various benzene exposures with those
at which the first signs of hematotoxicity (lymphocytopenia) occurred.3 He
found "that a significant difference exists between the projected concen-
tration of benzene metabolites in bone marrow, as calculated for a 6-hour
exposure to 10 ppm benzene in vitro, and the concentration of the same
metabolites which produce a demonstrable effect on a sensitive population
of human cells in vitro."^
^•Rickert, D. and R. Irons (oral statements) from U.S. EPA, "Public Hearing:
National Emission Standards for Hazardous Air Pollutants: Benzene Emissions
from Maleic Anhydride Plants" August 21, 1980 (Transcript pp. 5-24).
2Ibid, p. 6.
3Ibid, p. 15.
4Ibid, p. 22.
-------
40
Although EPA regards this work, published after the release of the
health assessment document, as generally supportive of the concept of
a threshold for lymphocytopenia and other hematotoxic effects which may
result from benzene exposure, EPA does not agree with the inference drawn
from this study that exposures below 10 ppm pose no health risk. The in vitro
system used may not represent the most sensitive human population at risk of
hematotoxic effects. Further, it is not clear that effects such as
lymphocytopenia must precede the induction of leukemia, nor has it been
established that the benzene metabolites studied are related to the onset of
leukemia.
2.4.1.2.2 Threshold-governed preconditions
for leukemia
Several commenters submitted that EPA's presumption of low level benzene
risk ignored alternative mechanisms for carcinogenesis, applicable to benzene,
for which effect thresholds appear likely. One commenter asserted that, while
a substance's ability to directly alter genetic material could be viewed as
support for a non-threshold mechanism, there is "no evidence that [benzene]
reactds] with ONA" (OAQPS-79-3(Part D-IV-D-9, (Part II)-IV-0-22). According
to the commenter, "benzene induces neoplasia through cell injury" to the bone
marrow. The injury is "followed by regeneration of the bone marrow and
myelogenous leukemia in a small number of cases." "During exposures of man to
benzene levels in the air of 10 ppm or less, the metabolic detoxification
reactions maintain the levels [of benzene] and its metabolites to be
sufficiently low in the blood to be below the threshold for any effect on
the bone marrow or metabolic effects on lymphocytes" (OAQPS-79-3(Part IJ-IV-D-9,
(Part II)-IV-D-22).
-------
41
*
Similarly, commenters argued that the documented association between
hematotoxic effects (usually decreases in the levels of various formed
elements in blood: cytopenia, pancytopenia, and lymphocytopenia) and leukemia
supports the finding that such effects may be a necessary precondition for
leukemia (OAQPS-79-3-IV-D-13;A-79-27-IV-D-24;A-79-49-IV-D-9). In this regard,
one commenter quotes Goldstein's observation that "there [do] not appear to
•
be any proven cases in which leukemia began in the absence of previous
cytopenia."^ Commenters contend that because "pre-leukemic" changes such
as cytopenia "do not occur below about 35 ppm," this exposure level or, more
conservatively, a level of 20 or 10 ppm constitutes an effective threshold
below which benzene "presents no health risk whatsoever" (OAQPS-79-3(Part I)-
IV-D-13,(Part IIJ-IV-F-1 .IV-F-9; A-79-27-IV-0-24,IV-0-27,IV-D-29, IV-F-1;
A-79-49-IV-D9,IV-D-ll,IV-D-12,IV-F-l,IV-F-2;A-80-14,IV-D-l,IV-D-3,IV-F-l).
EPA Response:
While EPA agrees that the non-genetic or "epigenetic" mechanism con-
stitutes a possible explanation for the way in which cancers could arise
in the absence of direct interaction with genetic material, the Agency is not
persuaded, based on the largely theoretical evidence provided, that such a
mechanism is applicable in the case of benzene. For similar reasons, the
Agency continues to regard as inconclusive the contention that hematotoxic
effects must necessarily precede the development of leukemia in benzene-
exposed individuals.
Covalent bonding (reaction) with DNA is generally regarded as evidence that
an agent may have the ability to "transform" a normal cell into an abnormal,
^Goldstein, B. "Hematotoxicity in Man," A Critical Evaluation of Benzene
Toxicity, S. Laskin and B. Goldstein, ed., 1977, p. 165.
-------
42
and possibly cancerous, cell via a somatic mutation. The absence of such
bonding or its non-detection, however, does not demonstrate that substances
such as benzene may not interact directly with genetic material to produce
aberrant cells. In fact, there is evidence that benzene, at levels as low as
1-2.5 ppm, produces significant increases in chromosome abnormalities in bone
marrow cells including chromosome breaks and marker chromosomes (rings,
dicentrics, translocations, and exchange figures).6»? Whether such changes
are appropriately considered mutations or simply toxic events is dependent on
the fate of the affected cell. As the Occupational Safety and Health
Administration (OSHA) has pointed out in its benzene rulemaking: "If the
alteration in the chromosomal material results in an inhibition of further
cellular division, then in terms of its reproductive potential, the cell is
dead and the damage inflicted may be classified as a toxic event. However,
if the alteration is replicated, this may constitute a persistent gross
mutation. The finding of gross chromosomal damage in bone marrow cells
clearly demonstrates that despite competing detoxification reactions . . .
benzene, or a reactive metabolite, is able to overwhelm protective defense
mechanisms and enter the nucleus of hematopoietic cells."8
The quote attributed to Goldstein noting that "there [do] not appear to
be any proven cases in which leukemia began in the absence of previous cyto-
penia" is correct but incomplete. Later in the page Goldstein cautions
^Kilian, D.J., and Daniel, R.C. "A cytogenetic study of workers exposed to
benzene in the Texas Division of Dow Chemical, U.S.A." February 27, 1978.
^Picciano, D. "Cytogenetic Study of Workers Exposed to Benzene" Env. Res.
19:33-38, 1979.
8U.S. Occupational Safety and Health Administration "Occupational Exposure
to Benzene; Occupational Safety and Health Standards" 43 FR 5918,
February 10, 1978.
-------
43
that this interpretation is "open to speculation, especially in view of the
paucity of routine laboratory data preceding the onset of leukemia."^
The lack of information, as well as the retrospective nature of most of the
analysis, make it difficult to substantiate a precedent relationship between
hematotoxic effects and leukemia. In this regard, OSHA has observed:
"... since the mechanism by which benzene induces leukemia
has not been elucidated, it is possible that leukemia
develops, not in response to the pancytopenic effects of
benzene, but rather to the direct carcinogenic effect on the
marrow hematopoietic stem cells not necessarily accompanied
by any other evidence of marrow effect ... In such
events, protection against non-neoplastic blood disorders
would not rule out subsequent development of leukemia."10
Similarly, Browning, in 1965, noted: "benzene leukemia is frequently
superimposed upon a condition of aplastic anemia, but it can develop without
a preceding peripheral blood picture characteristic of bone marrow
aplasia."^
Finally, EPA is not persuaded that the "thresholds" identified by
commenters for benzene-induced "injury" are sound. First, it is not clear
that techniques such as peripheral blood counts and aspiration of bone marrow
are capable of detecting injury to the hematopoietic system, particularly
when the normal ranges of such counts are broad. As EPA stated in the
benzene health assessment:
"there is a relatively large reserve of bone marrow, which
is capable of perhaps a six fold greater output of mature
cells in a normal situation. Accordingly, counts of the
9Goldstein, p. 165.
10OSHA, p. 5929.
^Browning, (1965), in U.S. Occupational Safety and Health Administration,
Docket H-059, Ex. 2-31, 1977.
-------
44
circulating cells do not give an adequate index of very
early benzene hematotoxicity."12
and also:
"a decrease in bone marrow cellularity is not always noted
in individuals with hematotoxicity associated with benzene
exposure nor in all animals treated with benzene. This may
be attributable to individual variation, to ineffective
response of poietic bone marrow to benzene, or to the
sampling error inherent in the presumption that a local
aspirate of bone marrow represents all hematopoietic
tissue."13
Second, injury may be occurring at levels below those at which cyto-
penia is observed. The National Academy of Sciences in its review of
benzene, commented on a report of benzene-induced chromosome abnormali-
ties: "Vigliani and FornH1*] reported chromosomal aberrations of both
the stable and unstable type. In general, the chromosome aberrations
were higher in peripheral blood lymphocytes of workers exposed to benzene
than in those of controls. This was true even in the absence of overt
signs of bone marrow damage"^ [emphasis added].
As noted above, Picciano and Kilian and Daniel have also reported
significant increases in chromosomal aberrations, an effect whose toxic
potential cannot be ignored, in workers exposed to benzene at levels
12U.S. EPA, "Assessment of Health Effects of Benzene Germane to Low-Level
Exposure," Office of Health and Ecological Effects, September, 1978
(EPA-600/1 -78-061), p. 64.
13Ibid, p. 49.
i, E.G. and A. Form' "Benzene, chromosome changes, and leukemia"
J. Occup. Med. 11:148-149, 1969.
15National Academy of Sciences-National Research Council "Health Effects of
Benzene: A Review" for U.S. EPA (EPA 560/5-76-003) June 1976, p. 7.
-------
45
substantially below the 10 ppm submitted as the lowest level for a
"threshold" for benzene-induced effects.
2.4.1.2.3 epidemiology
Commenters found support for a benzene carcinogenic threshold in
epidemiological studies which did not find a significant association between
benzene exposure and leukemia, in control or non-exposed populations for
*
which a case for benzene exposure could be made, and among exposed
populations following exposure reduction efforts (OAQPS-79-3(Part I)-IV-D-
9,IV-D-ll,IV-D-l3,(Part II )-IV-D-22,IV-F-l,IV-F-9;A-79-27-IV-D-24,IV-D-27,
IV-D-28,IV-F-l;A-79-49-IV-D-10,IV-D-11,IV-D-12,IV-F-l,IV-F-2;A-80-l4-IV-F-l),
Commenters cited "negative" epidemiological studies, in particular, work by
Thorpe,I6 Tabershaw,17 and Stallones,^ in support of the conclusion that
chronic exposure to benzene levels below 20 ppm (the average exposure level
assumed by Thorpe) does not pose a risk of leukemia.
Referring to the Infante et_ aj_. study of workers in the Pliofilm
industry, several commenters maintained that the population of "dry side"
workers excluded from Infante's exposed cohort on the grounds that they
were not exposed to significant benzene levels, in reality "were in
contact with benzene levels up to 20 ppm" and could be considered an
exposed population. The fact that Infante e_t_ a_l_. found no leukemia
deaths among the members of this population as a result of a cursory
16Thorpe,"Epidemiologic Survey of Leukemia in Persons Potentially Exposed
to Benzene," 16 JOM 375 (1974).
17 Tabershaw Cooper Associates, A Mortality Study of Petroleum Refinery
Workers Project OH-1 (1974) (OSHA Benzene Record, Ex. 2-59).
^Stallones, R.A. and 0. Syblik, Report on Mortality from Leukemia
(1977) (OSHA Benzene Record, Ex. 115, C.2).
-------
46
examination of death certificates, was submitted to EPA as additional
evidence that benzene exposure below 20 ppm does not constitute a
health hazard.
Both the Infante et_ aj_. and the Askoy et_ ^1_. studies were cited by
commenters as examples of the disappearance of an epidemiological associa-
tion subsequent to reductions in exposure. In the case of Aksoy et al.,
commenters contended that the excess leukemias "disappeared when the
shoeworkers phased out pure benzene as a solvent and replaced it with
gasoline," despite the fact that the benzene levels in gasoline could
continue to provide exposures of "20 ppm or more" (OAQPS-79-3(Part I)-IV-D-
13). The disappearance of an effect in the Infante et_ aK population
was attributed to compliance with the lowered occupational standards.
EPA Response:
As indicated in Section 3.3.1.4 above, EPA does not believe that the
"negative" epidemiological studies interpreted by commenters as evidence of
no-effect levels permit a firm conclusion regarding a carcinogenic threshold
for benzene. In the larger context of the utility of negative epidemiological
studies, EPA, as a member of the Interagency Regulatory Liaison Group
(IRLG), has concluded:
"Absence of a positive statistical correlation does not by
itself demonstrate absence of a hazard. Substances
distributed widely in commerce or in the environment are
particularly difficult to study by epidemiologic methods
unless high risk ratios are observed, because it is often
impossible to identify unexposed groups as controls or to
separate groups with high and low exposure. The problem of
adequate controls is further compounded by the long latency
of cancer, during which multiple opportunities exist for
-------
47
exposure to other potentially carcinogenic substances and
modifying factors."*9
This general view is shared by OSHA as well as the National Cancer Ad-
visory Board:
"In principle, it is impossible to prove a negative with any
study of a finite size. An epidemiologic study, like any
other screening test, is an instrument of limited
sensitivity. Thus, however well a study may be carried out,
the most that can be concluded reliably.is that it failed to
show an effect within the limit of sensitivity imposed by
the study design."^0
"Negative epidemiological data may not establish the safety
of suspect materials. Negative data on a given agent
obtained from extensive epidemiological studies of
sufficient duration are useful for indicating upper limits
for the rate at which a specific type of exposure to that
agent could affect the incidence and/or mortality of
specific human cancer."21
While EPA agrees that follow-up studies such as those undertaken on the
Infante e^t ^1_. and Aksoy et^ ^1_. populations may be useful in demonstrating
risk reductions, they are not appropriate support for a position that risks
have been eliminated, particularly in view of the possibility with benzene-
associated leukemia that the initiating exposure may precede the onset of
disease by several years.22
19Interagency Regulatory Liaison Group (CPSC, EPA, FDA, FSQS) "Scientific
Bases for Identification of Potential Carcinogens and Estimation of
Risks." 44 FR 39858, July 6, 1979.
20OSHA, "Identification, Classification and Regulation of Potential
Occupational Carcinogens; Final Rule" 45 FR 5001, January 2, 1980.
^National Cancer Advisory Board "General Criteria for Assessing the Evidence
for Carcinogenicity of Chemical Substances: Report of the Subcommittee on
Environmental Carcinogens" J.N.C.I. 58:461-465, 1977.
22OSHA, 1978, p.5930, referencing docket exhibits 2B-528, 217-152, 2-50,
2A-272.
-------
48
2.4.1.3 EPA's Conclusions
Having reviewed the public comments, EPA concludes that the evidence
submitted in support of a real or practical threshold for benzene-induced
leukemia is not sufficient to overcome EPA's presumption that benzene may pose
a finite risk of leukemia at any level of exposure greater than zero.
Although commenters have sought to demonstrate that benzene may cause
leukemia via a non-genetic mechanism which requires threshold-governed tissue
injury prior to leukemia induction, and that levels of benzene below this
threshold are non-injurious or otherwise detoxified, EPA regards this
evidence as largely theoretical in nature and inconclusive.
EPA believes that the support for a "hematotoxic" threshold as protective
against leukemia induction is speculative for two reasons: first, because
neither the mechanism for benzene-induced leukemia nor that for blood
disorders have been elucidated, and, second, because information is available
that other effects of potential adverse health consequence have been shown
to occur at levels lower than those postulated as hematotoxic thresholds.
Finally, EPA does not agree that the non-positive epidemiological studies offer
a means of establishing credible no-effect levels.
2.4.2 The Quantitative Estimation of
Carcinogenic Risk
EPA initially published interim guidelines for the conduct of quantita-
tive risk assessments for carcinogens in May, 1976.1 in -\gjg, these were
succeeded by the report of the Work Group on Risk Assessment of the
ill.S. EPA "Health Risk and Economic Impact Assessments of Suspected
Carcinogens; Interim Procedures and Guidelines" 41 FR 21402, May 25, 1976.
-------
49
Interagency Regulatory Liaison Group (IRLG)^ of which EPA was a member. Con-
cerning the quantitative estimation of cancer risks, the IRLG observed:
"In some instances a regulatory agency may be required, or
may find it useful, to estimate quantitatively the cancer
risk of such a substance in exposed humans if the compound
is assumed to be a human carcinogen.
Quantitative assessment of human cancer risk may be based
on epidemiologic or animal data. In either instance,
methodologic problems arise because of the need to
extrapolate from effects observed under one condition and
level of exposure and in one population group or biologic
system to arrive at an estimate of the effects expected in
the human group or individual. Because extrapolations are
involved, uncertainties are necessarily attached to the
cancer risk estimates that can be made with current
methodologies. Furthermore, uncertainties arise from other
sources, particularly from attempts to identify accurately
conditions and levels of exposure of the human group or
individual.
Despite the uncertainties, risk estimates can be and are
being made, not only by some regulatory agencies but by
other scientific bodies. Because of the uncertainties,
however, and because of the serious public health
consequences if the estimated risk were understated, it has
become common practice to make cautious and prudent
assumptions wherever they are needed to conduct a risk
assessment. This approach has a precedent in other areas of
public health protection where similar problems arise
because of gaps in knowledge. Thus current methodologies,
which permit only crude estimates of human risk, are
designed to avoid understatement of the risk. It must be
recognized, however, that in some circumstances this cannot
be guaranteed because of other factors that may enhance
human response, such as synergistic effects. Thus risk
assessments should be used with caution in the regulatory
process."3
EPA prepared, in conjunction with the listing of benzene under Section
112 and the development of emission regulations, an assessment of the
2Interagency Regulatory Liaison Group (IRLG), (CPSC, EPA, FDA, FSQS)
"Scientific Basis for Identification of Potential Carcinogens and
Estimation of Risks" 44 FR 39858, July 6, 1979.
3Ibid, p. 39871.
-------
50
population risk to ambient benzene exposures.4 The assessment was based
on an extrapolation of the human leukemogenic risk drawn from available
epidemiological evidence in combination with an assessment of human
exposure to benzene emitted into the air by industrial sources.5
Although a few commenters objected to the performance of a risk assess-
ment, arguing that the underlying uncertainties were too great to permit a
meaningful result, most respondents were in favor of attempting to estimate
population risks. In an extensive critique of EPA's assessment, however,
commenters disagreed with EPA on a number of scientific and technical grounds,
ranging from the appropriateness of the dispersion model used in estimating
ambient benzene levels to errors in the assumptions made in deriving an
estimate of benzene's leukemogenic potency. Commenters argued that the
correction of such errors would result in an overall leukemogenic risk from
benzene sources substantially below that predicted by EPA, and, in fact, small
enough to be regarded as a "statistical artifact" for which regulatory
attention was unwarranted.
2.4.2.1 The assessment of human exposure
The original assessment of human exposure to benzene was performed by
the Stanford Research Institute (SRI) under contract to EPA. A number of
commenters on the benzene listing and proposed standards criticized the SRI
assessment as relying on outdated emissions, employing an upwardly biased
exposure model, failing to account for population mobility, omitting
plant-specific information, and erroneously including plants no longer
^U.S. EPA "Carccinogen Assessment Group's Final Report on Population Risk
to Ambient Benzene Exposures" (EPA-450/5-80-004) January, 1979.
5|J.S. EPA "Assessment of Human Exposures to Atmospheric Benzene" (EPA-
450/3-78-031) June, 1978.
-------
51
using benzene (OAQPS-79-3(Part II)-IV-F-l,IV-F-9;A-79-27-IV-D-27;A-79-
49-IV-D-9). Several commenters were supportive of an alternative methodology
submitted by Systems Applications, Inc. (SAI) (OAQPS-79-3(Part D-IV-D-9,
IV-D-13 (Part II)-IV-0-22,IV-F-1, I V-F-8.IV-F-9; A-79-49-IV-D-9). Following
EPA's subsequent use of the SAI model in the development of proposed rules,
one commenter questioned the use of a 20 kilometer radius in developing the
exposure estimates (OAQPS-79-3(Part IJ-IV-D-8).
EPA Response:
EPA agrees that the SAI exposure methodology offers some improvements
over the exposure methodology developed by SRI for the benzene assessment.
SAI developed their methodology under contract to EPA in response to a need
for a rapid, computer efficient method for conducting exposure assessments.
This methodology with the additional data submitted in the course of the
comment periods on the benzene proposals, has been used to revise the
exposure estimates and risk assessments for the promulgated standards.
While the SAI methodology possesses certain features that enable EPA to
incorporate more plant-specific information in the estimation of exposure, EPA
does not agree that the assessments should routinely require plant-specific
data on population density and operating parameters. For source categories
with large numbers of sources or which lend themselves to the development of
model plant configurations, a source-by-source analysis would be costly and
usually unnecessary.
The SAI methodology does provide for the inputting of actual plant
locations with specific emission and some source characteristics and
utilizes local or regional multi-year meteorological data available from
the National Climatic Center. Where plant specific information is obtainable
-------
52
within reasonable resource limits, EPA will employ it in estimating
exposure.
Although the SAI methodology has supplanted the methodology initially
used by EPA to estimate benzene exposures, EPA does not agree that the
SRI model, for the purposes intended, is grossly inaccurate or upwardly
biased.
The SRI report was intended to be a rough-cut estimate of national-
level exposures to ambient air concentrations of benzene caused by air
emissions from various types of sources. The purpose of the report was
to help EPA decide which benzene sources to study in more depth to determine
the extent of regulation needed under the Clean Air Act, and to help EPA
determine the order in which the studies would be conducted. Those
studies, which accompany the development of regulations under Section 112
of the Clean Air Act, address far more explicitly the sources of benzene
selected for regulation and the public exposures to benzene associated
with those sources. The nature of many of the comments received suggests
that the commenters did not understand EPA's intended use of the report
and of the intentionally rough-cut approach considered appropriate for
that use.
Because the collection and use of source-specific data would have re-
quired a great deal more time and money than were warranted by the purpose
for the report, many assumptions were used. The assumptions applied to most
sources did not seriously affect the comparability of the results, nor the
usefulness of the report. For example, plants were assumed to operate
continuously at rated capacities. To have collected data on the actual
operating hours and capacities, and the year-to-year variations, for all
of the plants encompassed by the SRI report and to have developed and used
-------
53
an analytic methodology with the complexity needed to use these data
would have required an effort greatly disproportionate to the purpose
for the report.
Because the assumptions used applied to all plants, it is doubtful
that a significant relative error resulted and that the usefulness of the
report was lessened. The emission rates used were derived by two con-
tractors: SRI International and PEDCo; the docket numbers for their
reports are OAQPS-79-3(Part I)-II-A-028 and II-A-021, respectively. Both
contractors estimated emissions for an "average" plant. Information on
emissions and emission factors was gathered from government publications,
technical journals, industry publications and documents, data from state
agencies, and conversations with representatives from industry and govern-
ment. Because of the many sources of information, it seems unlikely that
a consistent over-estimation of emission factors occurred. The annual
average used in the report represents an omni-directional average around
the source and is neither the highest nor the lowest annual average esti-
mated. The conversion of one-hour averages to annual averages is based
on a one-hour-to-eight-hour scaling factor of 0.5, and eight-hour-to-
annual scaling factor of 0.1, and a factor of 0.4 as the ratio of the omni-
directional concentration to the highest annual concentration, for a
total scaling factor of 0.02; the commenter who suggested using 0.01
did not support the suggestion with a convincing technical argument.
The population assumed to be exposed to the estimated concentrations
was represented by a city-average density of people in the area of
the concentrations; the purpose of the report did not justify the effort
needed to define exact plant locations and the distributions and mobility
of the surrounding populations.
-------
54
EPA agrees that much of the SRI report is difficult to evaluate. This
report was one of EPA's first attempts at estimating exposure, and the
methodologies'were not as fully described as one might wish. As explained,
the report was not meant to be a definitive statement on exposure to ben-
zene, but to be a guide to follow-on studies. EPA believes the report
was adequate for its intended use.
In the SAI model, the selection of a 20 kilometer limit on exposure
estimation in the vicinity of stationary sources is based on modeling
considerations. Twenty kilometers was chosen as a practical modeling
stop-point. The results of dispersion models are felt to be reasonably
accurate within that distance. The dispersion coefficients used in
modeling are based on empirical measurements made within 10 kilometers of
sources. These coefficients become less applicable at long distances from
the source, and the modeling results become more uncertain.
2.4.2.2 The linear, non-threshold, dose
response model
Comments were generally critical of the use by EPA's Carcinogen Assessment
Group (CAG) of a linear, non-threshold model to derive a benzene unit risk
factor. One commenter stated:
"[t]he CAG methodology includes the assumption of no
threshold and the validity of the linear model extrapolated
toward zero. Neither of these assumptions are we willing to
accept. Rejection of the first asssumption would lead to a
risk estimate of zero additional risk. Rejection of the
second assumption would produce a risk assessment result
lower than that derived with CAG's pessimistic linear no-
threshold model" (OAQPS-79-3(Part II )-IV-F-l,IV-F-9).
Other commenters viewed the model as unjustified, "inherently conservative"
and likely to yield an upper limit of the health risks (OAQPS-79-3(Part
I)-IV-0-13;A-79-27-IV-D-27,29;A-80-14-IV-D-10a,13).
-------
55
EPA Response:
While EPA agrees that the linear, non-threshold model is conservative
in nature and would tend to provide a reasonable upper bound to the
statistical risk range, the Agency does not believe that the assumptions
upon which it is based are unreasonable or that the results of its use
are exaggerated. As the IRLG has observed:
"[t]he mathematical procedures per se are intended to
provide upper limit estimates of rilTT from statistical
standpoint. However, the risk estimates as applied to
humans should not be regarded as upper limit estimate
because of large biologic uncertainties."^
The dose response model with linearity at low dose was adopted
for low dose extrapolation by EPA because at the time of its introduction,
it had the best, albeit limited, scientific basis of any current mathematical
extrapolation model.7 EPA described this basis most recently in a
Federal Register notice announcing the availability of Water Quality
Criteria Documents:
"There is really no scientific basis for any mathematical
extrapolation model which relates carcinogen exposure to
cancer risks at the extremely low levels of concentration
that must be dealt with in evaluating the environmental
hazards. For practical reasons, such low levels of risk
cannot be measured directly either using animal experiments
or epidemiologic studies. We must, therefore, depend on our
current understanding of the mechanisms of carcinogenesis
for guidance as to which risk model to use. At the present
time, the dominant view of the carcinogenic process involves
the concept that most agents which cause cancer also cause
irreversible damage to DNA. This position is reflected by
the fact that a very large proportion of agents which cause
cancer are also mutagenic. There is reason to expect that
6IRLG, p. 39873.
'Crump, K., D. Hoel, C. Langly, and R. Peto "Fundamental carcinogenic
processes and their implications for low dose risk assessment"
Cancer Res. 36:9 p. 2973-2979, 1976.
-------
56
the quanta! type of biological response that is
characteristic of mutagenesis is associated with a linear
non-threshold dose-response relationship. Indeed, there is
substantial evidence from mutagenesis studies with both
ionizing radiation and with a wide variety of chemicals that
this type of dose-response model is the appropriate one to
use. This is particularly true at the lower end of the
dose-response curve; at higher doses, there can be upward
curvature, probably reflecting the effects of multistage
processes on the mutagenic response. The linear non-
threshold dose-response relationship is also consistent with
the relatively few epidemiological studies of cancer
responses to specific agents that contain enough information
to make the evaluation possible (e.g., radiation-induced
leukemia, breast and thyroid cancer, skin cancer induced by
aflatoxin in the diet). There is also some evidence from
animal experiments that is consistent with the linear non-
threshold hypothesis (e.g., liver tumors induced in mice by
2-acetylaminofluorene in the large scale EDgi study at the
National Center for Toxicological Research, and initiation
stage of the two-stage carcinogenesis model in the rat liver
and mouse skin)."°
2.4.2.3 Derivation of the unit risk factor
for benzene
Commenters argued that, in addition to the conservative nature of the
model used, the assumptions made by EPA (CAG) in the derivation of a unit
leukemia risk factor for benzene represented "serious misinterpretation"
of the underlying epidemiological evidence (OAQPS-79-3(Part I)-IV-D-13,
(Part II)-IV-F-l,IV-F-9;A-79-27-IV-D-24,27;A-80-14-IV-D-10a,21). Among
the specific criticisms were that the CAG: 1) inappropriately included in
its evaluation of the Infante et_ aj_. study two cases of leukemia from
outside the cohort, inappropriately excluded a population of workers that
had been exposed to benzene, and improperly assumed that exposure levels
were comparable with prevailing occupational standards; 2) accepted, in the
Aksoy et_ a]_. studies, an unreasonable undercount of the background leukemia
8U.S. EPA "Water Quality Criteria Documents; Availability" 45 FR 79319,
November 28, 1980, p. 79359.
-------
57
incidence in rural Turkey, made a false adjustment for age, and underestimated
the exposure duration; and 3) included the Ott e_t al_. study in the analysis
despite a lack of statistical significance. Several commenters also
referenced an alternative analysis developed by Lamm for the American
Petroleum Institute^ as a more appropriate approach to the derivation of
the unit risk factor (OAQPS-79-3-IV-F-9;A-79-27-IV-D-24;A-80-14-10a,21).
EPA Response:
In regard to the Infante et_ al_. study, EPA generally concurs, as discussed
previously, with the definition of the exposed and non-exposed cohorts
provided by the authors.^JO From the testimony of Dr. Marvin Sakol at the
OSHA benzene hearings, however, it appeared that Infante et_ al_. may have
overlooked several additional cases of DiGuglielmo's Disease that should have
been included in the Infante et^ al_. study among benzene-exposed workers.^
Infante himself did not disagree in the ensuing questioning. The Agency
reviewed all such cases at the time and decided that only two additional cases
should have been included. In preparing the Agency risk assessment, these
two were added for the purpose of obtaining a better estimate of risk. Sub-
sequently, it was found that one of the two was employed in dry-side pliofilm
8aLamm, S., Professional Consultants in Occupational Health, Inc. 1980.
Oral presentation to the U.S. EPA on behalf of the American Petroleum
Institute, August 21, 1980. Transcript pp.41a-49 and docket reference
OAQPS 79-3(Part ID-IV-F-9.
^Infante, P.P., R. Rinsky, J. Wagoner, and R. Young "Leukemia in Benzene
Workers" Lancet 2:76-78, 1977a.
lORinsky, R.A., R. Young, and A. Smith "Leukemia in Benzene Workers"
American Journal of Industrial Medicine 2:217-245, 1981.
^Occupational Safety and Health Administration, Docket #H-059, Occupational
Exposure to Benzene, Proposed Standard, Transcript of Public Hearing, July 19-
August 10, 1977b, Exhibit No. 61.
-------
58
operations and thus did not qualify for inclusion.
Dry-side workers were excluded from the study cohort by Infante et al.
from the very beginning as a result of discussions with company personnel
and on-the-spot surveys leading to the judgment that employees in non-production
areas were not exposed to benzene.12 EPA is persuaded that these workers were
appropriately excluded. As discussed in detail by Rinsky et.a1_.13 in the
completed follow-up of the Infante ejt al_. cohort, short-term excursions above
the standards may have occurred, but from the data it is reasonable to
conclude that benzene exposure was generally within accepted limits.
Concerning the Aksoy study, EPA initially felt that an age adjustment
factor of 1/2 was appropriate in view of the apparent age differences between
the Turkish shoeworkers and the Turkish general population. Based on an
average age at diagnosis of 34.2 years in several leukemia cases, and
additional information from an earlier paper by Aksoy e_t al_.l4 that the
mean age of 217 apparently healthy male shoeworkers was 24.7 years with an
age range of 12 to 58 years, the Agency assumed that the vast majority of
Turkish shoeworkers were under age 40. EPA further assumed that the Turkish
rate of 2.5 to 3.0 per 100,000 given by Aksoy e_t al_. pertained to all ages
combined and that the age structure of the Turkish population was like that
of the United States. It seemed appropriate to compare the incidence of
leukemia in Turkish shoeworkers with the incidence of leukemia in a similarly
aged segment of the Turkish population. Given that the U.S. incidence of
leukemia in persons under age 40 was one-half that of the United States for
12lnfante et al_. p. 77.
13Rinsky, e_t al_. p. 225.
14Aksoy, M. et al. "Haematological effects of chronic benzene poisoning in
217 workers"~B~r~J. Ind. Med. 28:296-302, 1971.
-------
59
all ages combined,^ £p/\ applied a factor of 1/2 to the Turkish rate. The
Agency believed that this procedure, although crude, was preferable to no
age adjustment under the circumstances.
The Agency now recognizes that the addition of an age adjustment factor
was unnecessary. Since the 1979 risk assessment, the Agency has acquired new
information regarding the age structure of the male population of Turkey in
the year 1970 (Bureau of the Census 1980). More than 93% of male Turkish
citizens were under age 60 in October 1970 compared with 70% of U.S. males
under age 60 during the same period. This younger age distribution produced
an indirect age-adjusted leukemia rate for Turkish males that was 34% less
than that of the U.S. males,^ assuming that U.S. age-specific leukemia rates
were similar to those of Turkey. Unfortunately age-specific leukemia rates
are not available for Turkey, but it appears that the age structure of the
Turkish population is heavily weighted toward younger ages which are subject
to lower, more uniform rates of leukemia. In the U.S., the white male age-
specific leukemia incidence is fairly uniform up to age 49 with a gradual
increase from 4.9 per 100,000 to 11.9 per 100,000 by age 59. Since Turkish
shoeworkers and the vast majority of the Turkish population are subject to the
same uniform rates of leukemia in the categories under age 60, an age
15Cooke, J. "The occurrence of leukemia" Blood 9:340-347, 1954.
1685+ (U.S. poph=1971 male white
I (U.S. pop)-,- x (U.S. Rate)j=8.56 U.S. pop. July 1, 5-yr age groups.
i=0
(U.S. Age Specific Rate)j= 1971
male white U.S. incidence rates
ICD 204-207 by 5-yr age groups.
85+
I (Turkey pop)ix(U.S. Rate)j=5.46 (Turkish pop)j = 1970 male Turkish
i=0 pop. October, 5-yr age groups.
-------
60
difference between these two groups would produce little differential in
the overall age-adjusted leukemia incidence in the two groups. Therefore,
the 1/2 age adjustment factor is not necessary.
The commenters misinterpreted the exposure estimates made by the Agency
for the Turkish shoeworkers. EPA assumed that the low level of 15 to 30
ppm measured outside working hours was an estimate of the minimum concentration
during working hours when solvents were not actually being used, and this was
averaged with the maximum concentration during working hours (210 ppm) when
solvents were being used.1? Dr. Aksoy indicated in his testimony at the OSHA
benzene hearings that attempts to measure benzene levels were made only when
solvents were actually in use.18 The Agency judged that peak measurements are
not accurate estimates of a workday average.
The average exposure duration for the shoeworkers was estimated by the
American Petroleum Institute (API)19 to be 23.7 years, based on the assumption
that they all worked continuously from an average age of 26.3 years until age
50. The Agency judges this to be an overestimate because of the likely high
turnover rate of workers in the many small shoe businesses in Turkey.
Again citing the 1971 Aksoy et_ ^1_. study, the author characterizes a
selected group of 217 apparently healthy male shoeworkers as having "rather
heavy exposure to benzene" over a period ranging from 3 months to 17 years.
The assumption is implicit that none of these 217 shoeworkers who were
"healthy males" were exposed over 17 years. The midpoint of this range is
17Aksoy et al_.
18OSHA, p. 157.
19Lamm, p. 23.
-------
61
3.6 years. This is probably a more unbiased estimate of the average exposure
duration than that of the API or the earlier EPA estimate.
Based on consideration of the comments, the Agency has recalculated the
benzene quantitative risk estimate. In making the new estimate the following
changes have been made, referring to the original Carcinogen Assessment Group
report^ as the basis for comparison.
As stated above, the Agency has removed one case of leukemia from those
attributed to the Infante et_ al_. study, changing the number of cases from nine
to eight. The exposure estimate in the Infante et_ _§]_• cohort has been re-
evaluated, as stated below, under a different set of plausible assumptions
about exposure. This causes a change in the lifetime average exposure from
2.81 ppm to 2.73 ppm.
In the Aksoy et_ aj_. studies the factor of 1/2 has been removed from the
relative risk calculations, as explained above. Adjustments also have been
made in the average exposure durations for the workers. As a result, the
estimate of lifetime average exposure for the shoeworker has been reduced
from 4.2 to 2.2 ppm.
The Ott et_ aJN study was not eliminated from the risk assessment in spite
of its borderline statistical significance, because it provided the best
documented exposure data available from any of the three studies. No change
was necessary in the calculations based on this study.
The detailed calculations in the derivation of the new risk estimate
follow:
20U.S. EPA "Carcinogen Assessment Group's Final Report on Population
Risk to Ambient Benzene Exposures" Roy Albert, Chairman, January, 1979
(EPA 450/5-80-004).
-------
62
1. Infante et_ al_. (1977)
The new relative risk is
R = (7 + D/1.25 = 6.4
Selecting the midpoint of the interval 1940 to 1949 as the most likely
point of entry into the cohort (an equal number was assumed to enter before
1945 as after 1944), assume further that the average exposure during each time
interval was as stated in the initial CAG report except that, prior to 1947,
the average exposure was on the high side, 100 ppm. Then the time-weighted
average exposure over the period 1945 to 1975 would be 29.02 ppm. The
equivalent continuous lifetime exposures corresponding to these workplace
exposure estimates are:
29.02 x 240/365 x 1/3 x 30.5/71 = 2.73 ppm (8.9 mg/m3)
Thus, the revised leukemia rate per lifetime average ppm in the
atmosphere becomes:
3 = 0.006732 x (6.4 - l)/2.73 = 0.0133
2. Aksoy etli. (1974, 1976); OSHA (1977)
Retaining the yearly incidence rate in Turkish shoeworkers of 13.15 per
100,000 as estimated by Aksoy et_ a_1_., and assuming that the incidence rate in
Turkey in the age range 10 through 50 [the age range of Turkish shoeworkers
and also the range into which most Turks are likely to fall based on
projections of the Bureau of the Census (1980)] is 2.5 to 3.0 per 100,000.21
21OSHA, p. 7.
-------
63
I = (2.5 + 3.0)/2 x 24/50 = 1.32 per 100,000
where 24/50 is the proportion of nonlymphatic leukemias in a group of 50
nonexposed leukemia cases.22 Then the estimate of the relative risk for
benzene exposed shoeworkers is:
R = 13.15/1.32 = 9.96
Furthermore, if we assume the same average working hour exposure (X£ =
63.6 ppm), a more realistic 8-hour working day, a 240-day working year, an
average age at the end of the observation period of 54 years (the mean
survival time of the average Turkish resident), and an estimated 8.6
years of exposure (0.25 + 17)/2 = 8.6 from Aksoy et_ al_. (1971), then
X2 = 63.6 x 8/24 x 240/365 x 8.6/54 = 2.22 ppm (8.2 mg/m3)
and finally
B = 0.004517 x (9.96 - l)/2.22 = 0.0182
3. Ott et_ al_. (1977)
The risk assessment has not changed; therefore B = 0.0464.
Summary:
The geometric mean of the three risk estimates is:
B = 3 / (0.0133)(0.0182)(0.0464) = 0.0223
The revised estimate constitutes a decrease of 7% from the Agency's earlier
estimate of 0.024. Risk assessments prepared independently by other
22lbid.
-------
64
interested organizations using the same three studies exhibit a range of
values for the potency factor as follows:
Carcinogen Assessment Group (CAG) revised 0.022/ppm
U.S. EPA (11/13/81)
Center for Policy Alternatives (CPA) 2/8/80 0.026/ppm
Massachusetts Institute of Technology
Consumer Product Safety Commission 1980 0.027/ppm
(CPSC)
Office of Toxic Substances (OTS) 12/12/79 0.024/ppm
U.S. EPA
Office of Policy Analysis (OPA) 11/20/79 0.0057/ppm
U.S. EPA
Steven Lamm 8/21/80 0.0022/ppm
American Petroleum Institute (API)
Carcinogen Assessment Group (CAG) 1/10/79 0-.024/ppm
U.S. EPA
Of the six independent assessments prepared previously, two reflect
considerably lower risk estimates. EPA believes that this spread of values
can be ascribed to differences in assumptions and viewpoints of the authors.
Both the OPA and API estimates, which are significantly lower than the
others, are based on criticisms of the original CAG estimate that the risks
were overestimated. In contrast, the criticisms of CPA, CPSC, and OTS were
mixed in their effect, in that some resulted in raising the risk estimate and
others resulted in lowering the estimate.
2.4.3 Significance of the Estimated Carcinogenic
Risks from B'enzene Exposure
Based on EPA's estimates of carcinogenic risk or on the alternative
calculations submitted to the Agency for consideration, a number of commenters
asserted that the risk of developing leukemia from exposure to benzene in the
ambient air was too small to warrant regulatory consideration under Section
-------
65
112. Specifically, commenters argued that the regulation of benzene under
section 112 would have "no meaningful impact on the occurrence of leukemia in
the general population." (OAQPS-79-3 (Part I)-IV-D-9,(Part II)-IV-F-l,IV-F-
9). In support of this position, commenters cited EPA's estimate that roughly
80% of ambient benzene emissions were attributable to mobile sources that
would not be regulated under section 112 and noted that the number of leukemia
cases predicted by the EPA assessment to occur as the result of benzene
emissions from stationary source categories represented "less than one-tenth
of one percent (of) the normal leukemia mortality risk in the U.S. population,
... a result so small as to be indistinguishable from a risk of zero"
(OAQPS-79-3(Part I )-IV-D-13,(Part IIJ-IV-F-1JV-F-9; A-79-49-IV-D-9); A-79-
27-IV-D-10,18,27,IV-F-l; A-80-14-IV-D-10a,IV-F-l_.
Several commenters referenced, as evidence of the insignificance of the
ambient benzene risk, the comparable or higher risks associated with
activities such as skiing, hunting, and sky diving (OAQPS-79-3(Part I)-IV-D-
19) and with involuntary hazards such as drowning and electrocution (OAQPS-793
(Part I)-IV-D-13,(Part II)-IV-F-lfIV-F-9).
Commenters also maintained that the estimated risks posed by benzene
emissions were at or below levels recognized by EPA and other Federal agencies
as acceptable goals or targets for regulation (OAQPS-79-3(Part I)-IV-D-13).
One commenter noted that the Supreme Court had recently ruled that
absent a "clear mandate" from Congress to eliminate all risk, the statutory
term "safe" (regarding exposure levels), rather than meaning "absolutely
risk-free," implied a level that protects against a "significant risk of
harm." The commenter noted further that the benzene risks estimated by
EPA are not "significant" as that term has been used by the Court (A-79-27-
IV-K-1).
-------
66
EPA Response:
EPA does not agree with the commenter's assertions that the health
risks posed by benzene emissions from all stationary sources are insignificant
or that the regulation of benzene under section 112 is, therefore, unwarranted.
This view is based on a number of considerations including quantitative
estimation of the health risks.
EPA remains persuaded that the well-documented evidence of benzene's
leukemogenicity, the volume of stationary source emissions, the size and
distribution of exposed populations, and the numerical estimates of health
risks support the determination that exposure to benzene emitted from
stationary sources "may reasonably be anticipated to result in an increase in
mortality or an increase in serious irreversible, or incapacitating re-
versible, illness."!
With an estimated 9.9 billion pounds (4.5 million Mg) produced in 1981,
benzene ranks 16th among all chemical products in the U.S.2 Benzene is the
highest ranked chemical that has been causally linked to cancer in humans.
The physical properties of benzene: a low boiling point (80.1°C), high
vapor pressure (100mm Hg at 26.1°C), and low atmospheric reactivity support
the assumption that benzene disperses in the air in much the same manner as
an inert gas; that is, it neither has any appreciable weight nor adds any
buoyancy to the dispersing medium.3 in evaluating the dispersion pattern,
1The Clean Air Act, Section 112(a)(l).
"^Chemical and Engineering News, May 3, 1982, p. 11.
3u.S. EPA "Atmospheric Benzene Emissions" October, 1977. (EPA-450/3-77-
029) p. 3-2.
-------
67
EPA has noted that "the primary consideration should be the type of source
from which the benzene is released and how it affects plume rise."4
EPA estimates that in excess of 120 million pounds (55,000 Mg) of
benzene are emitted annually to the ambient air from stationary industrial
sources. The sources are primarily plants involved in benzene production,
other chemical manufacturing, and the storage and distribution of benzene
and gasoline. At these sources, benzene is emitted from the process vents
and storage tanks, and liquid transfer operations as well as well as from
leaks in process components such as pumps and valves.
According to EPA estimates, 30 to 50 million people live within 20
kilometers of stationary sources that emit benzene. Levels of benzene
have been monitored in the vicinity of benzene-emitting facilities at
levels as high as 350 ppb (1117 ug/m3) with median values of 3.0 ppb
(9.6 ug/m3).6
EPA regards the emissions of benzene from some stationary source
categories and potential human exposure to these emissions as significant.
The fact that mobile sources emit more benzene than stationary sources
has no bearing on the significance of the benzene emissions from stationary
sources, since these sources also emit large quantities of benzene. The
4Ibid, p. 3-5.
^U.S. EPA, "Volatile Organic Chemicals in the Atmosphere: An Assessment
of Available Data" Office of Research and Development, 1983
(EPA-600/3-83-027(A)}.
-------
68
fact that specific standards have not been proposed for mobile sources
does not imply that the Agency has reached a conclusion on the significance
of the health risks associated with these sources. As commenters pointed
out, mobile sources are not regulated under section 112, but under Title II
of the Clean Air Act. One control technology applicable to benzene emissions
from mobile sources, as for other hydrocarbon compounds, is installation of
a catalytic converter. Benzene emissions from automobiles are substantially
reduced (along with other hydrocarbon compounds) by the installation of
catalytic converters required under Title II. EPA projects that by 1985,
mobile source benzene emissions will have been reduced by 69 percent
compared with those in the baseline year (1970) of the enactment of the
Clean Air Act and by 1990 by 83 percent.6
EPA does not agree that benzene does not warrant regulation because
such regulation will not have a meaningful impact on the occurrence of
leukemia in the general population. With the exception of established causal
relationships with benzene and certain hereditary factors, the causes of
leukemia are not known. The fact that only a small proportion of leukemias
may, at present, be preventable does not argue that reasonable control
measures should not be taken to reduce that proportion.
Further, EPA does not agree that the presence of other unregulated
or tolerated health risks, equal or greater in magnitude than those
estimated for exposure to benzene, obviates the need for regulation.
Activities such as hunting and skiing are essentially voluntary in nature with
the associated risks well advertised. The risk of being struck by lightning,
^U.S. EPA, Memorandum from Charles L. Gray to Donald R. Goodwin. Subject:
"Mobile Source Benzene", (Fall, 1982).
-------
69
while largely involuntary, would be difficult to reduce effectively. For
benzene, however, a large component of the health risk is involuntary.
At the same time, reasonable actions are available that can reduce the
risks from benzene exposure. EPA questions the appropriateness of weighing
risks that are accepted voluntarily or that have little opportunity for
mitigation against risks largely beyond the individual's control but for
which societal remedies are readily available.
Commenters have also chosen to make comparisons based on the "average"
lifetime risks or the expected number of leukemia deaths attributable to
benzene emissions, arguing that an "average" lifetime risk of leukemia
from ambient levels of benzene of 1 per 100,000 (10~5) does not constitute
a significant hazard and has, in fact, been accepted by EPA as well as
other Federal agencies as an appropriate goal for regulation. Aside from
the technical and philosophical difficulties inherent in the selection
and verification of such goals described in Section 3.5.2 below, EPA has
not selected a specific "goal" for carcinogenic risks from hazardous
pollutants and, further, disagrees with the choice of the "average"
lifetime risk as an appropriate measure of individual risk. EPA believes
that the determination that a substance poses a significant health risk
via the ambient air must include consideration of the magnitude of the
hazard to those individuals and sub-populations most exposed to emissions
of the substance. In the case of benzene, the estimated lifetime risks
for these populations are substantially higher than the "average" risks
cited by commenters. Current EPA estimates for most exposed individuals
living in the vicinity of source categories for which standards are being
developed range from a leukemia risk of 150 per 100,000 for benzene
fugitive sources to 830 per 100,000 for coke by-product plants.
-------
70
Finally, EPA does not agree with the commenter's conclusion that the
risks estimated by EPA are not "significant". EPA assumes that the
commenter refers to the decision in Industrial Union Department, AFL-CIO
v. American Petroleum Institute, 448 U.S. 607 (1980). In EPA's judgment,
the Agency's regulation of benzene is consistent with that decision.
In conclusion, EPA continues to believe that benzene emissions from
•
some stationary source categories represent a significant risk of leukemia to
exposed populations, particularly to those individuals and sub-populations
residing near major point sources. This belief rests on the documented
evidence that benzene is a human leukemogen, on the magnitude of benzene
emissions to the ambient air from stationary sources, on estimates of the
health risks to exposed populations, including consideration of the
uncertainties associated with quantitative risks estimates.
2.5 Other Issues Relevant to the Listing of Benzene
2.5.1 The Adequacy of Other Standards Controlling
Benzene
Several commenters asserted that the listing of benzene was un-
necessary in view of the "network of regulatory programs already put into
effect to control ambient benzene exposures" (OAQPS-79-3(Part I)-IV-D-10,
IV-D-13(Part II)-IV-F-1 ,IV-F-9; A-79-49-IV-0-10,IV-F-1 ,IV-F-2). Commenters
observed that these programs not only make it unnecessary for EPA to adopt
a new layer of control but also take benzene out of the statutory defini-
tion of "hazardous air pollutant" under Section 112. Commenters also noted
that a minimal proportion of the nation's population (0.2 percent) is exposed
to benzene emitted from sources which will not be controlled by non-attain-
ment provisions of state implementation plans.
-------
71
EPA Response:
The regulatory programs to which the commenters refer were not put into
effect to control ambient benzene exposures. The programs were put into
effect to attain and maintain the national ambient air quality standard for
ozone. The health effects from exposure to ozone are very different from the
health effects from exposure to benzene; ozone-caused health effects are
serious, but there is no evidence that exposure to ozone causes cancer. There
is, therefore, no scientific or technical basis for believing that attaining
and maintaining the national ambient air quality standard for ozone will
ensure that the public is adequately protected from exposure to benzene.
It is true that controlling volatile organic compounds (VOC) emissions
to attain and maintain the ozone standard often results in a degree of
control over benzene emissions, because benzene is often emitted with the
VOC's being controlled. EPA did not, as one commenter suggests, "ignore"
this fact. The effectiveness of existing State standards and control
devices in place for any other reason have been estimated. In fact, the
amount of control currently in place for three benzene source categories for
which standards were previously proposed, maleic anhydride and EB/S process
vents and benzene storage vessels, is relevant to the Agency's proposed
conclusion that benzene emissions from these source categories no longer
warrent Federal regulatory action. One cannot reasonably assume, however,
that the extent and stringency of the control of VOC emissions equates to
adequate control of all benzene emissions nationwide. None of the regulatory
programs to which the commenter refers was designed to provide the public
with protection from exposure to benzene. For example, the State regulations
which control VOC emissions are Federally required only for those areas
of the State where they are needed to attain and maintain the ozone
-------
72
standard, and, in those areas of the State where such regulations are
required, the regulations need be applied only to enough VOC sources with
enough regulatory stringency to attain and maintain the ozone standard.
Such regulations do not necessarily provide adequate control of all
stationary benzene sources. The Agency disagrees with the commenters'
assertions that existing regulatory programs for ozone/VOC make it unnecessary
to regulate benzene.
2.5.2 Selection of a Benchmark or De Minimis Risk
Target
Commenters suggested that EPA should adopt an acceptable carcinogenic risk
target for benzene and other airborne carcinogens, citing precedents in other
EPA and Federal rulemakings (OAQPS-79-3(Part I)-IV-D-13(Part II)-IV-F-1,IV-F-
9; A-79-49-IV-F-1,IV-F-2). In the absence of agreement on such de mim'mis
levels, commenters contended that "important distinctions among public health
risks may go unrecognized and agency resources may thereby be squandered on
insigificant hazards" (OAQPS-79-3(Part I)-IV-0-13).
EPA Response:
Although EPA finds the concept of an "acceptable" or "de minimi's" risk
level appealing, the Agency perceives substantial difficulty in determining an
explicit level. EPA agrees that a lower range of risk health estimates
(incidence and maximum risk) may be identifiable where it is judged that
the estimated risks do not pose such a public health problem as to warrant
Federal regulation. This, in conjunction with other factors such as
achievable emissions and health risk reductions, may persuade the Administrator
that a source category is not appropriate to regulate under Section 112.
2.6 EPA's Conclusions on the Listing of Benzene
Based on available information and the consideration of public comments,
-------
73
EPA has reached the following conclusions with regard to the listing of
benzene under Section 112.
1. EPA continues to believe that the decision to list benzene under
Section 112 was appropriate, fully-informed and timely. EPA does not
consider the arguments that the Agency's scientific assessments were incomplete
at listing as evidence of an "inadequate record." Similarly, EPA is not
persuaded that the listing decision was premature, as commenters contended,
in view of the unfinished status of the airborne carcinogen policy or the
subsequent delay between listing and the proposal of emission standards.
2. EPA concludes that the weight of scientific evidence supports the
finding that benzene is a human leukemogen. While benzene exposure may lead
to other adverse health effects in humans, EPA believes it is appropriate,
in the absence of an established exposure threshold below which no carcino-
genic risk exists, to regard the potential leukemogenicity of benzene as the
critical health effect for the purposes of regulatory decision making.
3. EPA does not consider the criticism of the Infante, Aksoy, and Ott
epidemiological studies sufficient to change the Agency's conclusions
regarding the leukemogenicity of benzene or to exclude any of these studies
from consideration in the development of quantitative estimates of risk.
4. EPA regards the evidence submitted in support of a carcinogenic
threshold for benzene insufficient to overcome the Agency's presumption that
carcinogens such as benzene may pose health risks at any exposure level
above zero.
5. EPA concludes that benzene is emitted to the air from stationary
sources in quantities exceeding 120 million pounds annually and that large
numbers of people are routinely exposed to these emissions.
6. EPA believes that quantitative estimates of the carcinogenic risks
associated with benzene exposure should play a role in the regulatory de-
cision process.
7. EPA does not consider the methodology employed in the estimation
of the carcinogenic strength (potency) of benzene as unreasonable or
exaggerated.
8. EPA concludes that the emissions of benzene from certain stationary
industrial sources pose significant risks to human health. EPA rejects the
argument that the level of these risks is de mim'mis or that regulatory
consideration is unwarranted in view of other, higher risks presumed to be
socially tolerable.
-------
74
3. THE SELECTION OF BENZENE SOURCE CATEGORIES FOR REGULATION
EPA has proposed standards for four source categories of benzene
emissions: maleic anhydride process vents, ethylbenzene/styrene (EB/S)
process vents, fugitive emission sources, and benzene storage vessels. A
standard is being proposed for a fifth source category, coke by-product
plants. Comments submitted on all four proposed standards have contended
that each of the source categories regulated does not pose a significant
risk to public health, and therefore, does not warrant regulation. (OAQPS-
79-3(Part II)-IV-0-9,IV-0-22,IV-F-l,IV-F-9; A-79-27-IV-D-24,IV-0-27,
IV-D-28,IV-F-1,IV-L-1; A-79-49-IV-D-7,IV-D-10,IV-D-12; A-80-14-IV-D-10a,13,
16,IV-F-1). Similar preproposal comments have been received on the coke
by-product source category. Arguments advanced in support of this position
include: the relative insignificance of stationary source emissions of
benzene versus mobile source emissions; the low level of the estimated
risks from benzene compared to other public health risks; and the negligible
impact of benzene control on the total U.S. leukemia incidence. EPA's
response to these comments is provided in Section 2.4.3 above.
Other commenters maintained that, even if the source categories being
regulated were judged to be significant at the time of proposal, the
emissions from these source categories are now actually, much lower than
projected at proposal and, thus, no longer pose a signficant risk.
One commenter expressed concern that EPA's risk estimates did not
not include consideration of non-carcinogenic effects (A-79-27-IV-D-31).
EPA Response:
EPA believes that two of these source categories still warrant
regulation. The rationale for this determination is presented in the
-------
75
sections that follow.
3.1 Selection of Five Source Categories for Initial Regulation
Following the listing of benzene as a hazardous air pollutant, EPA
divided the stationary sources of benzene emissions into 12 source categories.
After evaluating these categories, EPA selected five for initial regulation:
process vents at maleic anhydride and EB/S plants, benzene fugitive
emissions sources, benzene storage vessles, and coke by-product plants.
EPA is collecting additional data on the remaining seven source
categories to use in deciding whether or not standards development is
warranted for them.
3.2 Proposal of Standards: Significant Risk Judgment
The information used in selecting the five source categories for
initial regulation was preliminary information, based on screening studies
of the identified source categories. During standards development prior to
proposal, EPA gathered more detailed and refined information. The new
information necessitated revisions in emissions estimates for the five
source categories with some estimates increasing and others decreasing.
Examples of the information used to upgrade emissions estimates include
emissions test data, updated status on the number of operating plants, and
more precise information on the control devices already installed on these
plants.
In addition to upgrading the emissions estimates, EPA used the more
precise emissions data to revise the quantitative risk estimates. Table I
presents information for each source category, based on the emissions
status of that source category at the time the standards were proposed.
-------
76
TABLE I. BASELINE IMPACTS OF BENZENE SOURCE CATEGORIES AT PROPOSAL AND NOW
Benzene
emissions
(Mg/year)
Benzene
and other
VOC
emissions
{Mg/year)
Number of Leukemia
affected incidence/yr 1 3
plants Maximum lifetime risk 2 3 (Cases per year)
Benzene Fugitive
At proposal
Current
Maleic Anhydride
At proposal
Current
Ethylbenzene/Styrene
At proposal
Current
Benzene Storage
At proposal
Current
8,300
7,900
5,800
960
2,400
210
2,200
620
13,200
12,600
7,400
1,250
6,240
330
2,200
620
130
128
7 (5)4
7 CD4
1.7 x 10-4 to 1.2 x 10-3
1.5 x 10-3
2.3 x 10-4
7.6 x lO-5
13 (12)4 6.2 x 10-4 to 4.4 x 10-3
13 (3)4 1.4 x 10-4
126
126
1.5 x 10-4 to 1.0 x 10-3
3.6 x lO'5
0.15 to 1.1
0.45
0.46
0.029
0.027 to 0.20
0.0057
0.12 to 0.82
0.043
n 20 kilometers of plant.
ZMaximum lifetime risk is the estimated probability that the people exposed continuously for 70 years to the highe
maximum annual average ambient concentration of benzene will contract leukemia as a result of that exposure.
3The ranges of maximum lifetime risk and annual leukemia incidence at proposal presented in this table represent
quantification of two sources of uncertainty in the risk estimates: the variation in the dose/response relation-
ships among the three occupational studies upon which the unit risk factor is based, and a part of the uncertainty
associated with the assessment of human exposure. Described in the text are additional sources of uncertainty tha
are not quantifiable. Also described are certain analytical assumptions made necessary by a general lack of data
the complexity of source/individual interactions. It is EPA's position that the approach taken in the estimation
health risks is both balanced and rational and that the risk estimates derived are reasonable surrogates for the
magnitude of the health hazard from exposure to benzene. For this reason, and to avoid the implication that a ran
of estimates bounds the total uncertainty, the final benzene risk numbers are presented as point estimates of the
leukemia risk. For purposes of comparison, the proposal ranges may be converted into rough point estimates by
multiplying the lower end of the range by a factor of 2.6.
^Includes all plants; number in parenthesis denotes number of plants with uncontrolled emissions which would be
controlled by the standard.
-------
77
EPA's estimates of the health risks associated with exposure to benzene
emitted from stationary sources have played a major role in the final benzene
decisions^ The following paragraphs describe the basis for these estimates
and the scientific and technical uncertainties which accompany them.
The estimated carcinogenic risks posed by benzene emissions are
characterized in two ways: as the predicted annual incidence of leukemia
(expressed as cases per year), and as the lifetime risk of leukemia for
individuals exposed to the highest predicted annual average ambient benzene
concentrations (expressed as a probability). "Annual incidence" represents
the aggregate risk for the population residing within a specified distance
of emitting sources. "Maximum lifetime risk" represents a plausible upper
bound of the probability of contracting leukemia for those individuals
assumed to be exposed for a lifetime to the highest average benzene
concentrations predicted to occur in the ambient air in the vicinity of
emitting sources.
While EPA has sought, through quantitative estimates of risk, to provide
information on the magnitude of the health hazard, the numbers do not repre-
sent actual measurements of the health risks. Rather, they are surrogates
for such risks and rely on a number of assumptions, that, depending on the
perspective of the reviewer, could be considered to bias the estimates
towards underprediction or overprediction of the actual risks. EPA's
analytical assumptions and the major sources of uncertainty in the risk
analysis are described below.
The health risks estimated for benzene source categories are comprised
of three components: the unit risk factor, based on a dose-response function
derived from epidemic!ogical data; the exposed population estimated from
-------
78
census data; and the benzene ambient concentrations, derived from dispersion
modeling of emissions. The uncertainties in the health risk estimates
arise from the assumptions necessarily made to derive these components.
At proposal of the benzene standards, the risk estimates presented in
the Federal Register appeared as ranges. The ranges represented consideration
of two sources of uncertainty in the risk estimates for benzene: variations
in the dose/response relationships among the three epidemiological studies
(Infante '77, Aksoy '76, and Ott '77) upon which the unit risk factor is
based, and a part of the uncertainty associated with the evaluation of human
exposure.
Other uncertainties inherent in the risk estimates could not be
quantified and did not appear in the proposal ranges. These included the
appropriateness of extrapolating the leukemia risks identified for
occupationally exposed populations (generally healthy, white males) to the
general population (including women, children, non-whites, the aged, and
the unhealthy) for whom susceptibility to a carcinogenic insult could
differ markedly. The presence of more susceptible subgroups within the
general population would argue that an occupationally-derived risk factor
may underpredict actual risks.
On the other hand, general population exposures to benzene are much
lower than those experienced by the exposed workers in the occupational
studies, often by several orders of magnitude. In relating the occupational
experience to the general population, EPA has applied a linear, non-threshold
model that assumes that the leukemia response is linearly related to benzene
dose, even at very low levels of exposure. There are biological data
supporting this approach, particularly for carcinogens. There are also
-------
79
data, however, which suggest that, for many toxic chemicals, dose/response
curves are non-linear with response decreasing faster than dose at low
levels of exposure. Compared to many of the non-linear alternative models,
EPA's approach is generally considered to be conservative, that is, it
could potentially overestimate actual risks.
There are sources of uncertainty, as well, in the estimation of human
exposure to benzene emitted from stationary sources. As with the derivation
of the unit risk factor, the analytical assumptions required by unavailable
or incomplete information could affect exposure estimates in either direction.
EPA estimates ambient benzene concentrations in the vicinity of emitting
sources through the use of atmospheric dispersion models. Modeled ambient
benzene concentrations depend upon: (1) plant configuration, which is
impractical to determine for a large number of plants; (2) emission point
characteristics, which can be different from plant to plant and are imprac-
tical to obtain for a large number of plants; (3) long-term emission rates,
which may vary over time and from plant to plant; and (4) long-term meteoro-
logy, which is seldom available for a specific plant. The particular
dispersion model used can also have a significant impact on the results. A
popular model, the Industrial Source Complex (ISC) model, is highly regarded
as a predictive tool but requires detailed, source-specific inputs and is
often too resource-intensive for modeling a large number of sources. Less
complex models introduce uncertainty by requiring a greater number of
generalizing assumptions.
For benzene, the dispersion models assume that the terrain in the
vicinity of the sources is flat. For sources located in complex terrain,
this assumption could result in an underestimation of the maximum annual
-------
80
concentration. On the other hand, EPA's benzene exposure models assume
that the exposed population is immobile and outdoors, continuously exposed
to the predicted concentrations. To the extent that benzene levels indoors
are lower or that people leave the area, EPA's estimates may overpredict
exposure. At the same time, since the model does not take into account
migration into an exposure area, the population exposed over time could be
underestimated.
Finally, EPA's estimates of the health risks from exposure to benzene
consider only leukemia as a health endpoint. Other health effects, such as
aplastic anemia and chromosomal aberrations have been observed at occupational
exposure levels. The risk estimates also do not include the possibility of
synergistic effects with other pollutants. While such effects or other
non-detected health impacts could occur as a result of ambient exposure to
benzene, EPA does not have data which would support modification of the
analysis. In any event, EPA regards it as unlikely that the dose/response
relationships for such effects, if identifiable, would be more conservative
than the non-threshold, linear assumption adopted for leukemia.
The uncertainties inherent in the estimation of benzene health risks
have led some commenters on EPA's proposed rules to suggest that the risk
estimates are inappropriate for use in regulatory decision making. Although
EPA acknowledges the potential for error in such estimates, the Agency has
concluded that these estimates, subject to the reasonableness of the
methodology employed in their calculation and the data available, represent
a reasonable surrogate for the actual magnitude of the health hazard,
and as such, should play an important role in the regulation of hazardous
pollutants.
-------
81
As indicated above, EPA, at proposal, presented the risk estimates as
ranges based on the quantification of a part of the underlying uncertainty.
Upon consideration, EPA has concluded that the presentation of the risk
estimates as partial ranges does not offer significant advantages over the
presentation as the associated point estimates of the risk. Further, the
proposal ranges for benzene make risk comparisons among source categories
more difficult and tend to create a false impression that the bounds of the
risks are known with certainty. For these reasons, the benzene risks in
Table I are presented as point estimates of the leukemia risk. As noted
above, EPA believes that these estimates represent reasonable surrogates of
the magnitude of the actual human cancer risk posed by benzene emitted from
the source categories evaluated. For comparison, the proposal ranges may
be converted into rough point estimates by multiplying the lower end of the
range by a factor of 2.6.
3.3 Post-Proposal Review of Significant Risk Judgment
Some commenters on the proposed standards indicated that benzene
emissions were actually much lower than estimated at proposal, citing
factors such as increased controls, plant closures, reduced production
capacity, and lower emission factors. In support of their contentions, they
submitted detailed plant-specific information and results of emission test
programs.
Based on this updated information, EPA has revised benzene emissions
for the various source categories (see Table I). The maleic anhydride
emissions estimates now include consideration of all new controls, plant
closures, and changes in feedstock. The EB/S emissions estimates are those
provided by the industry, based on plant-specific information. (In addition,
-------
82
EPA-assumed flare efficiency has been revised to 98 percent from 60 percent.)
New benzene emission factors have been developed for benzene storage tanks
and refined for benzene fugitive sources.
Based on these revised emissions estimates, EPA reconsidered whether
benzene emissions from maleic anhydride process vents, EB/S process vents,
benzene fugitive emission sources, and benzene storage vessels still warrant
Federal regulation under Section 112. The factors considered by EPA are
described in the following paragraphs. (The selection of coke by-product
recovery plants for regulation is discussed in the preamble to the proposed
standard for that source category and is not discussed further here).
Benzene fugitive emissions, which are not substantially different than
they were when judged to be significant at proposal, contribute 7,900
Mg/yr; this figure reflects current controls. (EPA adjusted the control
level for petroleum refineries in nonattainnment areas to reflect controls
required by States in accordance with EPA's Control Techniques Guidelines
(CTG) document. This adjustment reduced emissions, but the reduction was
offset to some extent by refinements in emissions factors.) Approximately
20 to 30 million people live within 20 kilometers of the 128 plants with
these fugitive emissions. These people are exposed to higher levels of
benzene than is the general population. Due to the lack of a demonstrated
threshold for benzene's carcinogenic effects, these people not only incur
a higher benzene exposure but also run greater risk of contracting leukemia
due to that exposure.
EPA revised the quantitative risk assessments for this source category
based on the updated emissions estimates, the revised risk factor, and the
more detailed SAI human exposure model. The lifetime risk of contracting
-------
83
leukemia for the most exposed individuals is estimated to be about 1.5 x 10~3
for benzene fugitive emission sources, and the increased leukemia incidence
as a result of exposure to the current figitive emissions is estimated to
be about 0.45 cases per year. As explained earlier in this section, there
is considerable uncertainty associated with the calculation of leukemia
incidence and maximum lifetime risk numbers.
The number of process units emitting benzene fugitive emissions is
anticipated to grow from about 240 to 310 units. These new sources probably
would increase the number of people exposed to benzene emitted from this
source category and increase the estimated leukemia incidence accordingly.
Based on the human carcinogenicity of benzene, the magnitude of benzene
fugitive emissions, the estimated ambient benzene concentrations in the
vicinity of the plants with fugitive emissions, the proximity of people to
these plants, the resulting estimated maximum individual risks and estimated
incidence of leukemia cases in the exposed population, the projected
increase in benzene emissions as a result of new sources, the estimated
reductions in emissions and health risks that can be achieved, and
consideration of the uncertainties associated with the quantitative risk
estimates (including effects of concurrent exposures to other substances
and to other benzene emissions), EPA finds that benzene emissions from
oenzene fugitive emission sources pose a significant cancer risk and that
the establishment of a national emission standard under Section 112 is
warranted. (Part II of this notice responds to comments on the proposed
standards for this source category and promulgates final regulations.)
Several other factors were also considered which support this finding.
First, if no standards were promulgated, several existing plants would
-------
84
remain uncontrolled or poorly controlled. Some benzene fugitive emissions
sources are located in nonattainment areas and are controlled to some extent
in accordance with the CTG; others are in attainment areas where no control
is required. Control techniques are readily available to reduce uncontrolled
emissions from benzene fugitive emission sources at reasonable costs. Second,
nationwide standards would ensure that existing sources are controlled on a
continuing basis. Third, if no standard were promulgated, new sources
could remain uncontrolled or poorly controlled, thereby increasing cancer
risks.
The revised estimated baseline emission and health impacts for maleic
anhydride and EB/S process vents and benzene storage vessels have decreased
significantly since proposal of the standards for these source categories.
These impacts are presented in Table I. Because of this decrease and the
small additional reduction in health risks that could be achieved, the Agency
has concluded that these source categories no longer warrant federal
regulation under Section 112. The basis for this decision is discussed in
the FEDERAL REGISTER notice proposing withdrawal of the proposed benzene
standards for these three source categories.
------- |