Document bao1d0D4nd5v69wpGrodDp2L1

-V vy , p y/V -,u >' ,/ /A .si' tl. . b' . ,_ ?V .v*' /j i<1 c/'-6~^ ?/u^vt7 *~ ""/' 3- -c BENZENE DRAFT f Yn /2 ^ t o / ! // ** y ^^ M /V*' V/ *5 V-1 p Thomas R. Bartman March 1981 Graduate School of Industrial Administration, Carnegie-Mellon University This work was done under a contract with the American Petroleum Institute. Summary and Conclusions Z*2* i I. Chronology 1 II. Health Effects of Benzene A. Summary B. Review C. Leukemogenicity Thorpe Study Infante Study ^ Aksoy Study Ott Study D. Non-Maiignant Blood Disorders Effects on Petal Development E. Chromosomal Effects III. Agency Quantitative Risk Assessment A. OSHA EPA B. Critique: OSHA EPA Appendix A: Review of CAG Risk Assessment Appendix B: Risk Assessment in Reply Brief for the Pederal Parties Appendix C: Outside Pressures on OSHA Appendix D: Background on EPA Risk Assessment 00 CONF TDFNT 1 Al YZOO15898 %. % Summary and Concluaions Benzene, one of the most extensively used petrochemicals, has adverse effects on the formed elements of the blood, and damages chromosomes. In 1977, OSHA issued an Emergency Temporary Standard and then a permanent standard lowering exposure in the workplace from 10 to 1 ppm, based on what it termed new evid nee of an excess incidence of leukemia among rubber workers. In 1977, EPA relied upon the same body of evidence to list benzene as a hazardous air pollutant. OSHA's new standard never took effect. The Emergency Temporary Standard was stayed and the permanent standard was vacated. The Court of Appeals for the Fifth Circuit held in 1978 that OSHA had failed to show a reasonable relationship betw en the benefits sought and the cost of lowering the standard. In 1980 a divided Supreme Court affirmed, holding that OSHA had not shown that benzene exposure under the existinq 10 ppm standard involved a significant risk; apparently, OSHA was required to show that significant benefits would result from lowering the standard. EPA subsequently held hearings on the first step of its regulation of atmospheric benzene exposure from various sources starting with maleic anhydride plants. Extensive evidence links benzene to leukemia and various nonmalignant blood disorders in humans and some evidence indicates a connection to fetotoxic effects in rodents. The failure to observe a leukemic response in animals until recently means n>0 06806? CONFTDFNTTA1 YZOO15899 that Bci ntists have lacked until now an important link estab lishing carcinogenicity and some have been cautious ii\ identify ing benzene as a carcinogen. Excess leukemia has been found in several epidemiological studies; but unfortunately the meaning of these studies and those finding no significant effect are obscured by the failure to measure actual exposure levels. Quantitative risk analysis is complicated by the lack of know ledge poncerning the physiological mechanisms by which benzene causes leukemia, by the lack of exposure data, by the possiblity of exposure to benzene from other sources, and other leukemog ns, and a latency period of up to twenty years. Studies of humans exposed to benzene for long periods show a reduction in the formed elements of the blood, a non-malignant blood disorder. This effect can develop into aplastic anemia and may promote the development of leukemia, but effects caused by low exposures appear to be reversible. Occupational studies appear to have found effects at 40 ppm and below, hut exposure can only be guessed. There is also evidence of chromosomal damage that can persist for long periods of time. Mon-malignant blood effects are confirmed by animal bioas says. There is some evidence of blood changes at an exposure of 44 ppm. Benzene has not produced leukemia in laboratory experi ments. Chromosomal changes in laboratory animals replicate human effects. There is limited evidence such damage occurs down to ii n 0680ft3 OONFrDFNTrAL YZ0015900 about 2 ppm. Th significanc of chromosomal changes is uncer tain. Benzene has been found to reduce litter size, retard embryonic development, and cause other fetotoxic effects in rodents. These effects appear to occur as low as 10 ppm. Although OSHA had recently rejected suggestions that it re vise the benzene.standard, in 1976 the newly designated Assistant Secretary of Labor for OSHA, Eula Bingham, began to organize an effort to revise the standard. Dr. Bingham saw benzene as a symbolic issue and placed a high priority on tightening th standard. She influenced Dr. Infante to compile and publish a preliminary assessment of his epidemiological study of rubber workers. Based on this evidence, OSHA published an Emergency Temporary Standard and rushed to establish a permanent standard. Allowing political, rather than scientific forces to raise the issue and then proceeding with unseemly speed to publish a standard led OSHA to ignore any calculations concerning th risk to workers and to reject an analysis submitted during the hear ing. Instead, OSHA argued that no level of exposure to a car cinogen -could be considered safe and that the standard should be set at the lowest level that was technically feasible and would not put the affected industry out of business. A quantitative examination of risk would have clarified the issues. Careful examination of the various epidemiological and bioassay studies would have avoided embarassing misinterpreta tions. Using risk assessment to set priorities would have focused OSHA's resources on more serious hazards. iii DO 06806.4 CONFTDFNTTAl YZ0015901 I By contrast, EPA based its proposed maleic anhydride plant standard on a combination of reviewing health effects, estimating human exposures to atmospheric benzene, and quantitative risk assessment. While agreeing with OSHA that there is no safe level of exposure to a carcinogen, EPA used quantitative risk assess ment to consider the risks and costs of exposure at various alternative, standards. EPA's proposed final standard.included an estimate of expected costs and benefits. While the analysis could have been improved in some aspects, EPA demonstrated the feasibility of a quantitative risk assessment for benzene and isolated many of the crucial issues in setting a standard. Whatever the ultimate fate of the proposed standard, the issues have1 been stated clearly, the scientific basis for the standard has been clarified, and both the 'scientific uncertainties and value conflicts can be addressed with greater clarity. ".> '* A . *<-* * iv 00 068065 conftofnttai YZ0015902 X REGULATORY CHRONOLOGY The present Occupational Safety and Health Administration (OSHA) standard for benzene was adopted from the concensus standard of the American National Standards Institute in 1971. Based on benzene's general toxic effects on the blood, the standard prescribes an 8-hour time weighted average (TWA) of 10 ppm and a ceiling concentration of 25 ppmr and permits excursions to a peak of 50 ppm for no more than 10 minutes per 8-hour period. In 1974, the National Institute for Occupational Safety and Health (NIOSH), the research complement of OSHA, compiled a criteria document on occupational exposure to benzene that recommended retention of the 10 ppm TWA and 25 ppm ceiling standard (NIOSH, 1976). In a letter dated April 23, 1976, the United Rubber, Cork, Linoleum, and Plastic Workers of Amer.ica urged the Secretary of Labor to issue an emergency standard regulating benzene (43 FR 5917). Secretary William J. Usery denied the request on Hay 18, 1976, noting the lack of sufficient scientific evidence to warrant'a change in the standard and indicating that other m ans of controlling benzene exposure were being taken (Usery, 1976). Later in 1976, a panel of scientists at the National Academy of Sciences reviewed for EPA the information on health effects of benzene and concluded that benzene should be considered a suspect leukemogen (Committee on Toxicology, 1976). In August 1976, NIOSH submitted a revised criteria document that found benzene to be a leukemogen (43 FR 5917). NIOSH recommended that since no safe level for benzene exposure could be established, no worker should DO 068086 CONFTDFNTTAl YZ0015903 > 2 be exposed to more than 1 ppm. The Director of NIOSH recommended / a 1 ppm emergency standard to the Secretary of Labor by letter dated October 27, 1976 (43 PR 5917 at 5919). Based on the recommendations of NIOSH, OSHA issued voluntary \ "Guidelines for Control of Occupational Exposure to Benzene" on January 14, 1977 (43 PR at 5919). The guidelines, which were based on benzene's leukemogenicity, urged that the benzene concentration in the air not exceed an 8-hour TWA of 1 ppm per 8-hour shift. On April 15, 1977, NIOSH conveyed the preliminary findings of its epidemiological study of two Ohio rubber manufac turing plants to OSHA and again urged the setting of a 1 ppm emergency standard (43 PR at 5919). On Hay 3, 1977, OSHA issued a 1 ppm Emergency Temporary Standard (ETS) for Occupational Exposure to Benzene (42 PR 22516) to take effect on May 17.1/ The Pifth Circuit Court of Appeals, ruling on legal challenges,^ issued a temporary restraining order on May 20, 1977, and the ETS never went into effect. On May 27, 1977, OSHA published a proposed permanent stan dard incorporating the 1 ppm ceiling (43 FR 5917). The standard 37 Before issuing an ETS, OSHA must find "that employees are exposed to grave danger from exposure to substances det rmined to be toxic or physically harmful or from new hazards, and ... that such emergency standard is necessary to protect employees from such danger" 29 USC 5655(c). 2/ Petitioners, including the American Petroleum Institute and other industry groups, attacked the reduction fo the expo sure standard to 1 ppm, the prohibition of dermal contact, and engineering and work practice prescriptions. DO 0f,S0f.7 CONF TDFNT TAl YZ0015904 3 also prohibited any dermal contact with benzene. OSHA subse quently amended the standard to exempt operations where the only exposure was to liquids containing no more than 1% benzene and no more than .1% as of 1981. OSHA held hearings on the standard from July 19 through August 10 with witnesses from agencies# affected industries# unions# and other members of the public. OSHA also published a draft environmental impact statement for its standard on June 17 (42 PR 27455). During the same period in 1977# EPA was petitioned by the Environmental Defense Fund (EDF) to list benzene as a hazardous air pollutant under Section 112 of the Clean Air Act (April 4). Following consultations with OSHA on its regulatory action and with NIOSH and in response to the BDF petition# EPA listed benzene as a hazardous air pollutant on June 8# 1977 (42 FR 29332). During 1978, EPA issued an "Assessment of Health Effects of Benzene Germane to Low-Level Exposure" (EPA 1978a}# its Carcinogen Assessment Group's evaluation of benzene's leukemogenic risk and an "Assessment of Human Exposure to Atmospheric Benzene" (1978b). These documents formed the basis of EPA's plan to regulate benzene emissions from various sources# starting with a porposed benzene emission standard for maleic anhydride plants (1980). The Fifth Circuit ruled on the challenges to OSHA's 1 ppm permanent standard on October 5# 1978 (American Petroleum Institute V. OSHA# 581 F.2d 493 (5th Cir. 1978))# setting aside the DO 06B06R OONFTDFNTTAI YZ0015905 4 standard. The court cited the absence of substantial evidence showing a reasonable relationship between the measureable ben fits sought by reducing exposure and the significant cost of the regulation to the affected industries. The court also set aside OSHA's prohibition of any dermal contact, holding that the standard' was based on dated and inconclusive data. The Supreme Court heard oral argument October 10, 1979, in OSBA's appeal of the Fifth Circuit decision. The government argued that the benzene standard was based on the best available scientific`evidence, which supported the no-threshold Hypoth sis. It -further contended that both the lack of data on a safe lev 1 of exposure to benzene and the policy assumption that there is no safe level for a carcinogen required the standard to be set at A the lowest feasible level, namely 1 ppm. The government argued* further that the OSB Act did not require that a standard be bas d on a cost-benefit analysis. The American Petroleum Institute and associated groups argued that studies did not show there was an excess risk of leukemia at low concentrations of benzene and that OSHA had failed to show that the standard would produce ma terial or appreciable benefits. * The Supreme Court affirmed the Pifth Circuit's decision on July t, 1980 (Industrial Union Department, APL-CIO v. American Petroleum Institute, 100 S. Ct. 2844). Two other justices joined in Justice Stevens' plurality opinion holding that OSHA had made no showing that a significant health risk was associated with the V . . ** j no 068069 OONFTOFNTTAL YZOO15906 existing 10 ppm standard. Rather, the court held that OSHA had lowered- the benzene standard based on a series of assumptions about the risk at 10 ppm and about a reduction in risk that might result from a 1 ppm standard. The court found that OSHA lacks statutory authority to regulate to a completely risk-free 1 v 1. OSHA must show that appreciable benefits will result from reduc ing exposure. Three justices held that OSHA had made no att mpt to show that chronic exposure to 10 ppm of benzene would more likely than not result in a significant risk of material health impairment. Justice Powell stated in a separate opinion that while OSHA tried but failed to show by substantial evidence a hazard at 10 ppm, OSHA must also demonstrate that the economic effects of th standard bear a reasonable relationship to its expected ben fits. Justice Rehnquist voted to vacate the standard because Congress had unlawfully delegated authority with no guiding principles. Justice Marshall's dissent, joined by three justices, asserted that the plurality's requirement of a threshold showing of significant risk at 10 ppm had no statutory basis. As a first step in a program for reducing population expo sure to benzene emitted into the atmosphere from a range of sources, EPA proposed an Emission Standard for Hazardous Air Pollutants, Benzene Emissions from Maleic Anhydride Plants (45 PR 26660) on April 18, 1980. The proposed standard would require 97 percent control (100 percent for new plants) of benzene DO 068070 CONFTDFNT TAl YZ0015907 emissions from maleic anhydride plants.1/ Hearings on the proposed standard were held in August 1980. Comments by int res ted parties included the claim that using the same data and modified assumptions, EPA's risk assessment model shows that the risk of benzene exposure was not significant. EPA has subse quently not finalized its maleic anhydride standard but has proposed standards for other emission sources (Benzene Emissions from Benzene Storage Vessels, 45 FR 83952). II. HEALTH EFFECTS OF BENZENE A. Summary Epidemiological studies., animal bioassays, and other reports have implicated benzene, one of the most extensively used chemi cals, in increased incidences of leukemia, non-malignant blood disorders, and effects on chromosomes and embryonic development. Exposure to benzene at high concentrations has been accepted as leukemogenic in humans on the basis of numerous individual medical reports, discovered coincidences between the introduction of benzene and increased leukemia rates in occupational settings, and epidemiological studies. Although there is some evidenc of effects at lower exposures, leukemias have been generally shown following exposure to concentrations in the area of 100 ppm and ][7 While 80 percent of benzene in the ambient air results from mobile emissions, regulation of maleic anhydride plant emissions constituting 35 percent of emissions from chemical manufacturing was undertaken under 112 of the Clean Air Act, which covers only stationary sources. Do 068071 conf tdfnt T YZOO15908 7 over. Research, however, was until recently hampered by lack of an animal model and by uncertainties about the mechanism of benzene leukemogenicity. The significance of the few epidemio logical occupational studies is limited by the lack of dosage \ data. The three studies relied upon by EPA in its quantitative risk assessment involve additional uncertainties about the size of the population at risk and the role of unknown prior histories of workers. In sum, the available dose-response data are limited. The typical symptom of chronic benzene toxicity is reduction in the formed elements of the blood (pancytopenia). Blood effects can develop into fatal aplastic anemia and may partici pate in the development of leukemia, but effects caused by low exposures are apparently reversible. There is some evidence of blood changes in animals at an exposure of 44 ppm. Occupational studies lack accurate dosage data and have generally been based on small populations.' They have offered some evidence of effects at 40 ppm and lower. Specialized research into fetal malformation caused in rodents by benzene exposure has indicated chat benzene is probably not teratogenic. Exposure by inhalation and other means has, however, shown fetotoxic effects in rodents, such as reduced litters, retarded embryonic development, and fetal resorptions. Dose-response correlations for these effects are inconsistent, with some evidence of effects around 10 ppm. The limited results r>o ofro7f CONFTDFNTTAt 'YZ0015909 8 of related studies on humans suggest effects on reproductive organs in females. The meaning of the animal studies for re search into benzene effects on humans is complicated by the obvious differences between man and animal in terms of fetal development. Occupational and animal studies show that benzene, like other known carcinogens, causes chromosomal damage. Benzene exposure reduces DNA synthesis, an effect that evidently persists for considerable periods. Human studies have had to cope with small populations, uncertain dosages, and confounding factors such as age, the spontaneous occurrence of chromosomal changes, and damage resulting from laboratory techniques. There is some epidemiological evidence of chromosomal effects following occupa tional inhalation exposure of around 2 ppm. While the signifi cance of effects on chromosome number and structure is quite uncertain, such effects may participate in the transition from a non-malignant blood disorder to leukemia. B. Review of Health Effects Benzene is a clear, colorless liquid used extensively in the petrochemical and refining industries. An estimated 11 billion pounds of benzene were produced domestically in 1976. It is used as the raw material for other organic chemicals, is found in detergents and pesticides (and formerly solvents), and is a constituent of gasoline, generally in concentrations of less than two percent (in the United States). OSHA estimates that 190,000 DO 68073 donf Td FNTTAl YZ0015910 9 workers are occupationally exposed to benzene.4/ EPA estimates that 110 Billion people are exposed to atmospheric benzene at concentrations from a few parts per billion to .5 parts per mil lion due to manufacturing processes and automobile emissions.!/ Benzene,!/ which evaporates rapidly due to its low boiling point and high vapor pressure, enters the body primarily through inhalation. It diffuses rapidly through the lungs and is r adily absorbed. Its primary effect is on the bloodforming organs and the classic finding is a reduction in the number of platelets and red and white blood cells (pancytopenia). Benzene has also been causallv associated with damaqe to chromosomes and fetal develop ment. Recent OSHA and EPA attempts at reducing benzene exposure have been based on its suspected leukemogenicity. C. Leukemogenicity!/ The association of benzene and leukemia is based on numerous individual case reports. Girard (1970) surveyed leukemia patients and found a significant number with histories of benzene 4/ Feasibility study by A.D. Little & Co. 5/ in contrast to occupational exposures (8 hours per day for a limited number of days) in low parts per million, ambient air concentrations in the low parts per billion involve both 24 hour8 exposure per day all year and a population includ ing children and old people more sensitive than worker cohorts. / For a general review of benzene's health effects, see Laskin 1977, EPA 1978a and Snyder 1975. 7/ For a review of benzene leukemogenicity, see Goldstein and Leong in Laskin 1978, and EPA 1978a. pO 068074 CONFTDFNT1 At YZ0015911 A* 10 exposure. Host importantly, instances of increased leukemia rates have been found to follow chronologically outbreaks of industrial benzene poisoning. For example, Aksoy (1974a) dis cerned an increased incidence of leukemia among Turkish shoeworkers that coincided with the introduction of benzene in adhesives (resulting in exposure to a maximum exposure of 210-650 ppm), an increase which also subsided with the.substitution of toluene for benzene. Vigliani likewise noted an increased rate corresponding to specific outbreaks of benzene poisoning in Italian factories .(at concentrations of 200-500 ppm) (1976a). Vigliani estimated a 20-fold excess risk of leukemia for the exposed population of 5,000. Leukemia can be defined as a cancer of the white blood cells. It is characterized by the appearance of abnormal white cells in the circulating blood, replacement of the bone marrow with the leukemic cells, and widespread effects on the liver, spleen and other tissues. The occurrence of several different types of leukemia may relate to the duration of the disease (acute or chronic); the different types show differing inci dences, clinical development and mechanisms. While the prognosis for different types of leukemia varies (one factor being the ag at which it develops), adult acute leukemias, the type most often associated with benzene, are usually fatal. The mechanism of benzene leukemogenicity is unknown and this uncertainty hampers research into dose-response relationships and OO 068075 OONF TDFNTTAl YZ0015912 risks. Phenol and phenolic derivatives such as catechol and quinol are produced in the body as metabolites of benzene. It is hypothesized that a metabolite may be the actual leukemogen but this metabolite has not yet been Isolated. Further uncertainty surrounds benzene's association with the various types of leu kemia. European studies of substantial occupationally-exposed populations (Aksoyr Vigliani) have indicated a relationship between benzene and acute myelogenous and related (e.g., acute myelomonocytic) leukemias, and the association between benzen and acute forms seems well established. Other studies (Girard, 1979? Tareeff, 1963; McMichael, 1975) have found some correspon dence with chronic lymphocytic leukemia. The discrepancy in findings may be related to differing environmental factors or different exposure duration or concentrations (Laskin, 1975). Research has been seriously complicated by the unavail ability, until recently, of bioassays showing benzene's leukemogenicity. Benzene was one of the two human carcinogens for which no analogous effect in animals had been established. The lack of animal data did not negate evidence of benzene's human leukemogenicity, since relatively few studies of chronic administration with long-term follow-up have been done and it was thought that there may be no analog in rodents of benzene leukemogenicity (Goldstein, 1977). However, this lack of confirmation from animal studies caused some scientists to hesitate in acknowledg ing benzene leukemogenicity. Haltoni (1979) has found a signi ficant incidence of Zymbal gland tumors in benzene-exposed rats DO 06fi07(S CONFIOFNT TAl YZ0015913 - 12 - and mice. The preliminary results of animal testing at New York University. which may show a leukemogenic effect in animals. have been recently published. There are few large-scale epidemiological studies on benzene leukemogenicity. Studies of leukemogenicity have usually been in the form of individual medical reports, which provide a poor basis for calculating the incidence of leukemia. Similarly, the ubiquity of benzene complicates estimating the population at risk. It has been difficult to isolate benzene as the only contaminant, particularly in occupational settinqs (Aksoy. 1972; Vigliani. 1976). The potentially long latency period preceding leukemic manifestations -- periods of 15 and 27 years have been reported!/ -- complicates research, requiring retrospective studies covering long periods. Retrospective studies offer no means of confirming past diagnoses or of validating historic, technically inadequate measures of dose. Because benzene expo sure could have resulted from any number of sources and might have occurred long before diagnosis of the effect, the link between benzene and mortality may be overlooked. Past epidemiological studies have lacked precise measures of benzene exposure. They have confirmed the qualitative 8/ See for example. E.C. Vigliani. 1976b. By contrast, the latency period for leukemia from ionizing radiation is Only 2-4 years (BEIR III). If the latency period for benzene leukemia is this short, the dose-response curve would be different from that estimated by EPA and other investigators. DO 068077 CONFTDFNTTAl YZ0015914 13 association of benzene and leukemia. Since the ambient dose of benzene is so low, resulting in a normal incidence of leukemia of about 8 deaths per 100,000 persons per year, it would be hard to evaluate an isolated, possibly fortuitous case. Large sample sizes would be required for significant results. The unavaila bility of large samples for which there are controls of envir nmental factors, accurate dosage data, and a sufficient time period to take latency effects into account, contributes to the uncertainty of research into benzene leukemogenicity. Thorpe Study Among the limited number of epidemiological studies ar several sponsored by the petroleum industry finding no increased incidence of leukemia and thus complicating the evaluation of benzene's effects (Stallones, 1977; Tabershaw, 1976). For example, Thorpe's study (1974) for Exxon on the refining industry and covering the 10-year period 1962-1971, examined employment records of eight Exxon affiliates in Europe with a combined population and duration of exposure of 380,000 man-years. He found 18 cases (23.23 expected) of diagnosed leukemia occurring at four of the plants, an incidence judged equivalent to that in the general population of comparable age. Thorpe acknowledged several limitations of his study, including uncertainty in dosage data and the inadequacy of follow-up on retirees which might result in the under reporting of cases. Exposure levels were known only for certain job DO 0FR07R CONFTDFNTTAl YZ0015915 14 categories and known only roughly as "much less than the recommended level of 2S ppm." Likewise, the duration of exposure and the extent of exposures to different levels as a result of job mobility are not known. In addition to the uncertainties not d by Thorpe, it seems likely that the eight different plants were subject to different environmental factors and population (ethnic-genetic) characteristics, and the reliability of the diagnosis of leukemia would vary. Finally, job mobility and lack of data made unclear how much of the population was chronically exposed to benzene at all. The Thorpe study indicates at least that there was no out standing leukemic incidence in the refining industry. While it does not eliminate a risk of leukemia due to benzene, the Thorpe study can be used to place an upper bound on the quantitative effect of benzene. Infante Study The epidemiological study by Peter F. Infante (1977a) of NIOSH involved a cohort combining workers from two Ohio plants producing a natural rubber cast film known as "Pliofilm" by "nearly identical" processes. Benzene was established to be the only significant airborne contaminant.^ \/ The Infante study was heavily relied on in OSHA's promulga tion of an Emergency Temporary Standard and was a focus of industry attacks on the permanent standard in the 1978 OSHA hearings DO 068079 00NFTDFNTTA1 YZ0015916 f - 15 - The study cohort comprised 748 workers employed and exposed to benzene at Akron and St. Mary's between January 1, 1940 and December 31, 1949. Follow-up covered the period from first employment to June 30, 1975. The period covered would allow development of possible long term latent effects. Causes of death were derived from death certificates, but deaths before 1950 were excluded due to a lack of information on the caus of death. This exclusion, and the assumption that the 25 percent of the population not covered by the initial followup were alive, were judged to underestimate the risk of leukemia. As controls. Infante chose 1,447 white males from a fiberglass manufacturing factory situated in a geographically similar area and the United < States white male population during the 1940s, adjusted for age. Levels of exposure to benzene before 1946 are unknown (OSHA, 1978). In 1946, installation of an exhaust ventilation system significantly reduced exposure. After a survey of the Pliofilm process in 1946, the Industrial Commission of Ohio reported that "tests were made with benzol detectors and the results indicate that concentrations have been reduced to a safe level and in most instances range from zero to 10 or 15 parts per million."12/ A total of 112 surveys were conducted in one plant in the period between -1963 and 1974. Infante notes that these and earlier To/ Industrial Commission of Ohio, Report, 1946, cited in Infante, 1977. DO 068080 CONFTOFNTT At YZ0015917 surveys indicate that employee exposure was generally within the recommended limits of the respective year. Infante found 140 deaths from all causes compared to 187.6 expected from his U.S. white male control qroup. He attribut d the lower mortalitv rate of the rubber workers to incomplet follow-up* and it may also be due to the "healthy worker effect," the select character of workers in terms of health compared to the heterogenicity of the general population. Infante found, however, a significant excess of hematopoietic malignancies, chiefly of leukemia deaths. Seven workers died from myelogenous or monocytic leukemias, as compared to an expected 1.38 deaths (p<.002) among all white males, a 5-fold excess risk, and an expected 1.48 (p<.002) deaths among the fiberglass workers. In a supplemental letter to The Lancet Infante (1977b) referred to reports of additional leukemias among Pliofilm workers. An Akron hematologist found two leukemias in workers who, he claimed, probably belonged in the Akron side of Infante's cohort. Infante did not including these additional deaths in his findings since they were not located by his own study. He indicated, however, that a more complete follow-up on his cohort and a recalculation of the years-at-risk reduced the number of expected deaths from 1.38 to 1.25 for white males. Infante's report of a significant excess incidence of fatal leukemias in an industrial setting in which, benzene was claimed to be the only comtaminant, supports the qualitative association DO 068081 CONFTOFNTTAl YZ0015918 17 of benzene and leukemia. The study's limited usefulness to th setting of standards and to estimation of a dose-response curv arises from the paucity of its dosage data. Infante gives no information on the duration of worker exposure. It is unci ar why Infante did hot determine the length of a worker's tenure in a Pliofilm production job from the same employment records by which he placed the worker in that process. V/ infante suppli 8 no information on the varying durations of exposure due to possi bly considerable job mobility (McMichael, 1975). He evidently lacked records of prior employment that might have been the source of worker's exposure to benzene or other contaminants. Exposure data are limited and uncertain. Exposure levels before 1946 were not recorded. The Ohio Industrial Commission reported an exposure level of 10-15 ppm in most places and "safe" levels everywhere. These "safe" levels were those recommended for the various years* a level of 100 ppm from 1941-1946, 50 ppm TWA in 1947 and 35 ppm in 1948. The fact that recommended xposure levels decreased over the period of study is no guarante that the actual exposure declined. The total of 112 air surveys in one plant between 1963-1974 amount to less than one samplinq per month. Further, though uniform exposure could be determined for most production areas, certain sites in the production process were subject to significantly higher exposures. Us of TlV Indeed the OSHA hearings indicated that workers were included in the study on the basis of as little as one day of employment exposure. DO 06808? CONF TDFNTTAL YZ0015919 18 respirators was recommended for such sites but there is no assurance that respirators were used at these times. There was testimony of unrecorded temporary excursions significantly abov the recommended safe level, up to 350 ppm and beyond, resulting, . for example, from accidental drenching. Finally, there is some doubt about whether sampling and measurement techniques in the 19408 were capable of accurately measuring levels as low as 10 ppm. Infante's cohort of white men directly exposed to benzene was poorly defined. The use of populations from two separat plants with different amounts of available exposure data leaves some doubt on the comparability of environmental and occupational exposure data.!2./ infante plausibly justified the combining of two separate plants as a means of siqnificantlv enlaroina the studv population with workers all involved in essentially iden tical processes. Infante's definition of his cohort also excluded the group of "dry side" employees whose jobs involved some regular, though not constant, benzene exposure. Infante explained that this job category was outside the intended scope of the study and that there had been no figures showing any possible benzene exposures in these non-production line jobs. However, mention of the 1 ukemia incidence in this population exposed at a lower level would T27 Tabershaw (1977), point out that one plant also manufactured products other than Pliofilm, and that turnover was qreater at one Plant than at the other. DO 068083 CONFTDFNTTAL YZ0015920  19 have expanded the significance of his findings. Infante does not describe the job categories used or his criteria for eliminating other jobs.JLl/ Infante18 estimate of the exposure levels of workers in his \ study was low. The Infante study documents the qualitative asso ciation of benzene and leukemia, but provides little evidence on doses that may be associated with leukemia. Aksoy Study Dr. Muzaffer Aksoy (1971, 1972a, 1972b, 1974a, 1974b, 1976, 1977, 1978a, 1976b) has published a series of studies of health effects among Turkish shoeworkers using benzene-based adhesives.2i/ Observing leukemic patients in Istanbul hospitals, his primary conclusion was that an increase in cases coincid d with the introduction of benzene as a solvent for adhesives. He has reported (1971) that of 217 apparently healthy shoework rs selected at random and compared to a matched group of 100 medical personnel and students, there were 51 cases of haematological abnormalities associated with acute benzene poisoning. The T37 Critics further noted a failure to account for the fact that the worker population must have been reduced in the number of healthy males due to the draft that took place during the study period. This reduction in the number of healthy workers was likely small. There was also some question about the nosological accuracy in the study's recording of death certificate figures, which again, is likely a small element of uncertainty. 14/ Aksoy's findings were relied on by both OSHA and BPA in set ting standards and Aksoy was presented as an OSHA witn ss at its 1978 administrative hearings. D0 068084 00NFTDFNTTA! YZ0015921 20 - exposure was between 30 and 210 ppm for three months to 17 years.i/ In the study specifically used in CPA's risk assessment for its large sample size and quantitative importance, Aksoy (1974b) reported 26 cases of leukemia amonq shoeworkers admitted to Istanbul hosoitals durina the period 1967-1 973.1He considered this figure conservative since he had not seen all the leukemia cases in Istanbul and not all cases would be re ported in an area with a low level of health care. The concen tration of benzene resulting from the use of adhesives in poorly ventilated workshops was estimated to be as much as 210-650 ppm. The duration of exposure ranged from four months to 15 years, with a mean exposure period of 9.7 years. The range in ages at the time of diagnosis was 16-58 (average 34.2). Official records indicated that 28,500 workers were involved in shoe manufactur ing, leadinq Aksoy to calculate an incidence of leukemia of 13 per 100,000, a significant excess over six per 100,000, which he accepted as the incidence for Western nations, or 2.5-3, the estimated incidence of leukemia for Turkey. Aksoy's larqe cohort and siqnifleant, perhaps even conserva tive number of cases, support the qualitative association between J_5/ Hatching workers with medical personnel does not, however, control well for differences in income and lifestyle. .16/ Three cases were acute lymphoblastic leukemia, which are as yet inconclusively associated with benzene (Goldstein, 00 068085 OONFTOFNTTAI YZOO15922  - 21 - benzene exposure and leukemia. However, his rough dosage figures are difficult to quantify in establishing a dose response rela tionship. Also, comparison of these results to an American occu pational setting is problematic. Aksoy's exposure data are likely based on his examining the workplace and history of a small fraction of the shoeworker popu lation. There is thus no information on the duration of exposure which must vary in this heavily populated, unregulated industry. The lack of data on length of exposure and employment makes it difficult to judge the role of latent.effects that may occur after the study's completion. The possibility of undiscovered effects may contribute to ah underestimation of incidence. Indeed, Aksoy's 1979 follow-up reported a number of additional leukemia cases and thus a higher incidence. Aksoy gives only a rough ranqe, a minimum of 210-650 ppm, for the workshop concentration of benzene. Excursions at high levels (possibly 5,000 ppm in a hot climate) in the form of drenchings are likely in workshops described as "not hygienic and poorly ventilated" (Aksoy, 1974) but there are no data. Finally, there are no figures on the change in benzene concentration as the use of the solvent began, peaked, and faded during the period of observation. Thus, Aksoy's estimate of exposure levels is rough and possibly underestimated, but this possibility is hard to judge. DO 068086 CONFTDFNTTAl YZOO15923 - 22 Aksoy's results are difficult to apply to the American occupational and general populations due to the differing expo sure experiences that presumably differ between the populations. For example, Aksoy has estimated the benzene level in the shoeworkers' shops to be 15-30 ppm outside of working hours. It is likely that a great many shoeworkers labor in their living quarters or in the immediate vicinity. The 15-30 ppm concentra tion would approximate a background exposure to which Turkish shoeworkers are exposed for much of the period outside of working hours. Background exposure for the general population might well differ importantly from that of the U.S. in a context wh re benzene was not banned or regulated until 1969. By contrast, ambient benzene levels in the U.S. generally peak at under 10 ppb. It would seem even more difficult to isolate benzene as th only source of carcinogenic exposure. Indeed, Aksoy described the workshops as "not hygienic," though he himself could find no other contaminant in the workplace. His actual survey of shops was evidently limited. Aksoy's calculation of the incidence of leukemia is uncer tain. His figure of 13 per 100,000 is based on a population at risk that is difficult to quantify. The figure of 28,500 docu mented shoeworkers is likely a rough, conservative estimate: if the actual number were twice as high, the incidence rate for his worker population would be halved. His figure of 6 to 8 expected no 0RR0R7 OONFTDFNTTAl YZ0015924 23 leukemia deaths per 100,000 is that for the general population; since the incidence of leukemias increases with age, this figure is high for a younger, worker aged population, and 13 per 100,000 is conservative. The lack of figures on exposure level and duration is crucial. Aksoy gives no details about how his estimate of a maximum range of 210-650 ppm was reached!!/ and, in view of likely excursions, his range may be conservative. The probable under-reporting of leukemia cases (suggesting a higher incidence) and the great uncertainty regarding dose magnify the uncertainty of Aksoy's calculation of leukemia incidence. Ott Study Dr. M. Gerald Ott's (1978) epidemiological study!!/ of workers exposed to benzene in the petrochemical industi^ examin d 594 workers at a Dow Chemical plant. The study covered employ ment between 1940 and 1973. Ott notes that since 355 workers entered employment before 1950, sufficient allowance was made for the development of latent effects. On the basis of employment records, Ott located workers in particular departments and 17/ Indeed, Aksoy uses different exposure levels in other studies of shoeworkers (e.g., 150-210 ppm in 1972b). H does not specify whether these figures are based on closer examination of a fewer number of workplaces. 18/ This industry-sponsored study was published after the Infante results and the OSHA BTS promulgation. It was cited by indus try in the OSHA hearings as a negative study and was used by EPA to broaden the data base in its calculation of the b nzene risk assessment. DO 088088 CONFTDFNTTAl YZ0015925 - 24 associated them with particular durations of exposure and ranges of benzene concentration (from 2 to 25 ppm) .JJ/ Ott found no statistically significant excess rate of death from any caus other than leukemia. He identified two deaths due to leukemia, and a third in which leukemia was noted as a significant condi tion, the primary cause of death being broncho-pneumonia. This result, compared to an expected incidence of .8, based on the U.S. white male rate^O/ yielding a result of borderline statistical significance (p<.047).--/ i Ott's access to work histories and industrial hygiene records gives a better estimate of benzene dosage than the other studies and ott has tabulated his findings in detail. Ott points out the uncertainty due to the likelihood that long-term employees were exposed to a wide range of chemicals and exposure levels* two of the leukemia cases had had previous exposure to potentially carcinogenic chemicals. 1$/ Fiffcy-three workers were excluded from dose-response calcula tions due to known exposure to arsenicals, asbestos, and high vinyl chloride levels. ,20/ He presumably adjusted the figures for the white male popula tion for the age of the workers and for the period of the study. Ott used no other occupationally or geographically comparable control. .21/ Evaluation of two of the leukemia cases is, in fact, complica ted by documented exposure of one to vinyl chloride and an xposure level up to 200 ppm of benzene, while the other had worked in veneer manufacturing which is associated with in creased risk of myelocytic leukemia. Also, the cause of death in the third case was given on the certificate as bronchopneumonia with leukemia as a significant concurrent condition. DO 068089 CONFTDFNTTAl YZOO15926 V - 25 - Ott does not discuss air sampling methods* His historic re cords o hygienic surveys are presumably subject to doubt based on the limited technical capability of measuring low concentra tion in the 1940s, the difficulty of accounting for temporary excursions, and the possibly inadequate number of air samples. The major limit on the study's utility arises from its small sample size. While the resultant leukemia deaths are suggestive, their statistical inconclusiveness makes use of the study's figures difficult. This study has both the lowest concentration of benzene and the smallest worker population of the three studies. The low incidence of leukemia reported by the study is not necessarily surprising or significant. The Ott study is important because it is the only one with documented exposures below 25 ppm. D. Non-Malignant Blood Disorders The effects of benzene on the hematopoietic tissue, which have been recognized for 100 years and which served as the basis for past benzene standards, can be readily produced in laboratory animals (Leong, 1977). An acute dose affects the central nervous system, quickly inducing a light anesthesia followed by depres sion and respiratory failure. The narcotic threshold is 1,000 ppm and acute toxic effects follow exposure to concentrations below 10,000 ppm. Studies have shown that a subacute, chronic dose produces changes first in the bone marrow, slowing the maturation process of red blood cells, and next inhibits th production of white blood cells (Leong, 1977). AO 068090 OONFTDFNTTAI ^ YZ0015927 -26- Atteropte to correlate concentration and effect in chronic, inhalation studies have had inconsistent results.22/ Wolf et al. exposed rats, rabbits and guinea pigs for seven to eight hours per day, five days per week, to benzene concentrations of 80 and 88 ppm. The study found a significant incidence of leukopenia and testicular and splenic degeneration. Exposing rats to mean levels of 47 ppm (range 33-55) and 44 ppm (range 40-50) for s ven hours a day, Deichmann et al. (Deichmann, 1963) found a signifi cant reduction in the white blood cell count, with females appar ently more sensitive. The study, however, lacked parallel controls and the animals were subject to significant variability in weight and age. By contrast, Jenkins t al. (Jenkins, 1970), exposing various species to benzene levels of 256, 30 and 17.6 for eight hours a day, 5 days a week, 30 exposures total, found no statis tically significant changes at any level. The evaluation of these results is complicated by the problem of extrapolating from animal to human effects.22/ The blood counts normal to different laboratory animals will vary, and hormonal changes and even the time of day may distort blood counts, but such factors have generally not been recorded in past H7 he studies briefly cited are representative of research into dose-response relationships. More complete surveys are fpund in Goldstein and Leong, 1977, and in EPA 1978a. 23/ Ppr a general discussion of this problem see Risk Assessment document. DO 06R091 OONFTDFNTIAl YZ0015928 - 27 studies (Leong 1977). Animal studies have given insight into the mechanism of benzene effects on the blood/ but in view of their inconsistent results and their uncertainties/ past animal studi s have presented inconclusive dose response data. The results of human epidemiological studies are complicated by a worker's likely exposure to multiole aaents in the work place. [As in the case of research into benzene leukemias, however, the correlation between benzene and blood effects is based on a strong temporal correspondence between the incid nee of pancytopenia with benzene use.l Benzene has also been iso lated as the common denominator in various occupational settinqs where blood effects were found. Benzene has induced in animals blood effects not reproducible with other chemical aaents (Goldstein 1977). Chronic benzene exposure is associated with the occurr nc of pancytopenia/ the reduction of all formed elements in the blood and interference with cell production in the marrow. Some evidence suaaests that exposure shortens the survival of circulatina cells (EPA. 1978a). Benzene exposure also reduces the counts of particular cell types or combinations of cell types. Symptoms of mild pancytopenia are lassitude/ tiredness, dizzi ness. headaches/ and shortness of breath. Recovery is likelv in a mild case of oanevtooenia if benzene exposure is halted, although there have been reports of leukemia developing after apparent full recovery (EPA 1978a). DO 06809? 00NFIDFNTTA1 YZOO15929 28 Severe pancytopenias may be accompanied by hemorrhagic ffects due to thrombocvtooenia (reduction in platelets necessary to clotting), with the possibility of fatal bleedinq. Aplastic anemia, which has a 50 percent mortality rate, mav result, aenerallv at hiah exposure levels. The influence of benzene on blood forming tissues mav affect the human immunological system, potentially resulting in immuniological defenses being over whelmed bv other disease. Finallv. it is possible that benzeneinduced blood disorders are related to the development of leu kemia. Viqllani (1976a) has speculated that leukemia is the terminal stage of pancytopenia. It is possible that leukemia is in fact always; preceded by pancytopenia. Benzene's effects on the blood mav allow the development of acute leukemia bv hamoerinq the blood's immune surveillance function that normally weeds out abnormal cells. However, Goldstein (1977)notes evidence that a relatively' small percentage of workers with pancytopenia will develop serious effects, particularly if occupational conditions inmrove. Dose resDonse data are limited as in the case of benzene leukemia studies. Goldstein (1977) has suggested that the tru incidence of benzene-induced oanevtooenia may be underreported because mild blood disorders will not be noticed unless they are looked for. Similarly, benzene's role in a particular hematqlogical case may be overlooked, due to the ubiquitousness of benzene. 00 068093 conftdfnttai YZ0015930 29 Goldwater 11941) studied 332 male rotogravure workers who were exposed to benzene levels o 11-160 ppm for at least six months and generally about three years. He found six cases of benzene poisoning severe enough to require hospitalization. Goldwater found hematological abnormalities in various tests of the blood of benzene workers compared to a control group of 81 unexposed workers. Aksoy's (1971) study of 217 apparently healthy shoeworkers, exposed to a maximum of 210-650 ppm for uncertain periods, found 51 with benzene-associated abnormali ties, including 41 cases of leukopenia. In one of the only studies of long-term effects, Pagnotto (1977) did a follow-up of an earlier survey of 38 workers in the rubberworking industry exposed from one to 24 years to average concentrations between five and 50 ppm (90 ppm in the case of one worker). Pagnotto found a small number of cases with abnormal blood changes and one case of mild benzene poisoning at an exposure level averaging 40 ppm. There were, thus, no significant effects 13 years after the use of. solvents containing benzene had been discontinued. Finally, a Korean study (Chang, 1972) of 119 workers at low benzene levels found hematological effects at 20 ppm and attemp ted to extrapolate down to effects at 10 ppm. The study, how ever, is very difficult to evaluate due to insufficient informa tion on exposures and on the work force, aside from difficult! s of comparing its Korean population to an American occupational cohort, in sum, epidemiological studies have generally lacked data on dosage and have been based on small populations. 00 068094 CONFTDFNTTAl YZ0015931 - 30 While the association of benzene and changes in the human hematopoietic system seems well established, with some evidence of effects at relatively low doses, the data are insufficient for a dose-response curve. The relationship of benzene-induced blood changes to leukemic developments is unclear and there is little study of long term effects of slight blood changes. The mild disorders that result from low doses appear reversible. Effects on Fetal Development Teratological studies have examined the effects of benzene on fetal development.The usual protocol involves exposur of rodentB during the middle-trimester of gestation, the period of organ formation. The fetus is examined on the 20th day of development for external, skeletal, and visceral malformation. The significance of teratological studies in animals for re search into the effects of benzene on humans is complicated by the difference in metabolism and embryonic development in ani mals, specifically rodents, which are used for reasons of cost and practicality. Rats and mice have respiratory rates five to 10 times the rate of humans, which may mean that an ambient concentration gives the animals a higher dose. In rats, the maternal blood does not mix with that of the fetus to the ext nt it does in humans; the different placenta further complicates TT7 FoF a review of teratological studies, see Consumer Products Safety Commission, 1980. This is a new area of research; little data was available to EPA's assessment of health effects and OSHA did not cover it. DO 068095 CONFTDFNTTAl YZOO15932 31 comparing findings in rats to human effects.Finally, rats and mice show spontaneous abnormalities similar to effects caused by benzene. A sizeable control population is required to deter mine the role of these abnormalities, which may result from environmental factors like fasting. Watanabe (1970) subcutaneously injected pregnant mice with a benzene dose of three ml/kg of body weight during organogenesis and found numerous abnormalities, including cleft jaw and missing lower jaw. These effects may be due to the acute dose of benzene and a method of exposure that does not correspond to the way hu mans are exposed. The study also lacked a control group. Nawr t (1979), having introduced doses of benzene from .3 to 1 ml/kg into the stomachs of mice (gavage) during the period of organo genesis, reported significant lethal maternal effects and fetal resorptions at .5 and 1 ml. Fetal weight gain was significantly reduced at all doses. He found no statistically significant change in the incidence of malformations. Though the method of xposure,-gavage, would give a less acute dose than an injection, it differs from inhalation and complicates evaluation of the results. Inhalation studies have produced inconsistent findings. The Hazelton Laboratories study of rats (1977) reported a signifi cantly lower fetal and maternal weight gain at benzene concentra tions of 50 to 500 ppm, a significant incidence of reduced \ 25/ And there is evidently no human equivalent to the finding of fetal resorptions in rats. Do 068096 cONFTDFNTTA! YZOO15933 32 crown-rump distance at 500 ppm, and retarded fetal ossification. At 500 ppm it also found a statistically insignificant incidence of skeletal malformations, suggesting a teratogenic potential.^/ Two studies undertaken by Litton Bionetics exposed rats to 10 and 40 ppm (1977), and 8.7 and 43.4 ppm (1978) for six hours per day. The investigators found a significant increase in fetal resorp tions (a sensitive measure of fetotoxicity) at the 40, 10 and 8.7 ppm levels, but no other significant fetal or maternal effects. By contrast, research by Green (1978) exposed pregnant rats to benzene levels of 100, 300 and 2,200 ppm for six hours per day and found a significant reduction of crown-rump length only at the highest level. Green noted an increased incidence of skele tal abnormalities at the middle and high dose levels and particu larly in female fetuses. The study found no increase in resorp tions at any level. Similarly, Murray (1979) found a significant increase in minor skeletal abnormalities in mice and rabbits at a benzene concentration of 500 ppm, a level which caused some maternal toxicity. Murray notes that malformations occur in mice spontaneously at a low incidence rate. Results of these studies must be considered in light of the technical parameters of teratological study. Exposure is focused on the period of organogenesis for the purpose of studying abnormal growth. This ten-day exposure during the period of 26/ Critics have suggested that the control groups used were im properly small. HO 068097 OONFTDFNTTA! YZ0015934 33 gestation is unlikely to have any close correspondence to human occupational exposure, which would occur before and after as well as during pregnancy. The duration of exposure is not long enough to produce the hematological effects of benzene exposure,.?!/ such as dyscrasias, that might well be dangerous to ^n utero development in cases of aplastic anemia. (Deichmann (1963), for example, required five weeks exposure at 40 ppm to produce blood changes in mice.) Human exposure to benzene would likely occur before as well as during pregnancy.22/ In an animal study involving exposure prior to pregnancy, Gofmekler (1968) exposed female rats inhaled benzene levels of 6.3 to 209.7 ppm for 10-15 days before impregnation. Gofmekler found that litter size dropped proportionally with increased exposure at 19.5 ppm and above and found no pregnancies at the highest dose. He found no significant effect on fetal weight or survival, or on the inci dence of external fetal abnormalities. He noted that weights of fetal organs varied with the dose and found one case in ten of fetal resorption at 19.8 ppm. Zn sum, studies of benzene's effect on embryonic development indicate that benzene is apparently not a teratogen in rats at exposure levels under 500 ppm. Benzene's fetotoxic effects seem well established and occur at lower levels, possibly as low as 10 27/ . Deichmann (1963) 28/ Pregnant rats are not exposed before organogenesis because the fetus may be killed, vitiating the teratological study. DO 068098 OONFTOFNTTAl YZOO15935 34 ppm. The fact that this line of research must rely on animals complicates its significance for the study of benzene effects on humans. EPA (1980) has cited two studies related to reproductive ef fects. Hett and Maak (1938) found degeneration of ovarian folli cles and effects on ova in mice at exposure levels at which blood dyscrasias were produced. These were apparently toxic levels. Vara and Kinnunen (1946) studied gonadal effects in 30 occupa tionally exposed females. Twelve workers reported menstrual peculiarities that they themselves attributed to their employ ment. The researchers also gave chronic injections of benzene to rabbits until blood changes occurred and found similar effects on reproductive organs, including reduced ovary size in some animals. Neither the occupational reports nor the animal tests offer dosage data. While these reports suggest areas for furth r study, and roughly correlate with general findings of fetotoxicity in animals, they are insufficiently described and have no dose-response significance. E. Chromosomal Effects The study of the effects of benzene exposure has turned more recently to the examination of its potential cellular and chromo somal effects.21/ The theory of somatic mutation has been 29/ For a more complete survey, see Wolman, 1977. This area of research is relatively new. OSHA's review of benzene health effects did not explore it in depth. DO 068099 CONFTDFNTTAl YZOO15936 35 associated with the initiation or maintenance of malignant change. There aref in fact, indications that many agents causing chromosomal damage (like radiation, viruses, and chemicals) are carcinogenic.' There is indirect evidence in that the three human syndromes of chromosomal instability are both highly sensitive to cla8togenlc (causing chromosomal breakage) agents and are charac terized by "unstable" chromosomal changes such as ring structures and fragments, the type of changes most often associated with the toxicity of chemicals like benzene (Vogel, 1979). The signifi cant latency period that can precede benzene's effect would b consistent with original damage done to the genetic system by benzene. The effect of benzene as a mitotic poison has been well documented (Wolman, 1977). Experiments with labeled thymidine have shown a reduced uptake and thus a decreased rate of DNA synthesis ^n vivo in rat and rabbit bone marrow.30/ This effect corresponds to a depressed formation of marrow cells resulting from benzene toxicity and has been demonstrated in vitro in human marrow. The most common finding of benzene cytogenetic studies has involved chromosomal alterations. Kissling (1972) found a significantly increased rate of chromosomal aberrations in rabbit lymphocytes after 18 weeks of chronic subcutaneous injections of 30/ The results in rabbit studies were variable but still significant. 00 0^8100 OONF TDFNT JAI YZOO15937 - 36 * .2 ml/ kg of body weight per day. They noted an increase in % the number of aberrations from 5.9 percent at the outset to 58 percent at the end of the injections. The study found a 36 percent-incidence of such effects persisting two months after the end of the treatment. They compared their cytogenetic findings to those associated with dyscrasias induced by ionizing radiation. The significance of an increased incidence of chromosomal breaks is not certain: breaks may be repaired, but a larger incidence may increase the possibility of mutation. The persis tence of changes after exposure at least suggests that daroag has occurred (Forni; 1971a). Chromosome changes have also been associated with specific cancers. For instance, 50 percent of patients with acute myelogenous leukemia and a history of solvent exposures have been found to have cytogenetic alterations (Mitelman, 1978). But the incidence of aberrations accompanying specific cancers has been only lately discovered and its meaning is not precisely known. Finally, effects have so far been found only in somatic rather than germinal cells, so a relation of benzene to hereditary damage is not known. Mpst relevant to the human effects of benzene are cyto genetic studies of occupational exposure at subacute dosages. So far; however, these tend to involve small samples with de tailed, medical .reports that support the qualitative association of benzene exposure and chromosomal change. Hartwich (1972) DO 06,8} 01 OONFTDFNTTAl YZOO15938 examined nine seemingly healthy refinery workers three to seven years after exposure to benzene. They found an aberration rat of eight to 12 percent with a mean of 10.4 percent, significantly higher than the 5.1 percent in the matched controls (p<0.1). They noted that the incidence of alterations was at the outer limit of normal rates but pointed to the low exposure level, well under 25 ppm. Porni (1977b) examined cytogenetic effects in patients with various degrees of benzene poisoning. Of the twenty-five sub jects who had suffered benzene poisoning with bone marrow impair ment one to 18 years before the chromosomal study were almost all found to have normal blood counts indicating recovery. An examination of peripheral blood, however, showed that 18 patients had more than one percent unstable chromosomal changes and 17 had one percent or more stable changes.il/ This compared with only one patient with more than one percent unstable and one with one percent stable changes among the controls. Pollow-up studies showed that the chromosomal effects persisted long after exposure and recovery. The rate of breakages had decreased in many cases while changes in chromosome number showed a definite increase in some cases, a finding that Porni compared to the symptoms of individuals with past exposure to ionizing radiation. It is hard to compare this occupational study to others due to the differing degrees of hematotoxicity in the subjects. / Mutational or "stable" change occurs in chromosome number; change in structure is toxic or "unstable". DO 06810? conftdfnttai YZOO15939 38 Tough (1965, 1970) studied three factory cohorts. In the first, with exposure levels between 25 and 150 ppm of benzene, Tough found a significant increase in the percentage of aberrant cells.While he found no correlation between the duration of exposure and the number of cytogenetic changes, he did deter mine an increased proportion of effects in older workers. Twenty men from factory #3 were exposed intermittently to around 12 ppm for two to 26 years (exact and relative exposures are not known). They were compared to a matched control group comprising gen ral population and on-site unexposed workers. The survey found about the same aberration rate in the exposed and unexposed workers, which did not significantly differ from that of the general population.11/ Dr. Dante Picciano (1979) did a cytogenetic evaluation of 52 workers exposed to low levels of benzene in a chemical plant, the first occupational study of chromosomal effects to try to measure dose. The benzene concentration was calculated on the basis of personnel sampling, area air samples, and urine levels of phen 1, a benzene metabolite considered a good measure of exposure. The time-weighted average exposure was estimated to be 2.1 ppm and H7 He found 1,5 percent stable and two percent unstable changes compared to .6 percent stable and .4 percent unstable changes in the controls. 33/ The sample from factory #2 comprised 12 men exposed for six to 25 years to about 25 to 150 ppm and on-site unexposed workers. The frequency of aberrations was high in the exposed group compared to the general population but also in the controls, possibly due to ambient air exposure. DO 068103 OONFTDFNT TAl YZ0015940 39 * the duration waa one month to 26 years (average exposure 56.6 months with an average of 39.3 months). The control group comprised 44 pre-eroployraent individuals with an average age of 26.6 years. Picciano examined 200 cells per subject. He found the same number of abnormal cells as in the controls but twice the percentage of chromosome breaks (.67 percent compared to .35 percent in controls), three times the percentage of marker chromosomes (1.6 percent compared to 0.6 percent in controls) and 10 times the percentage of controls showing both aberrations. Analysis for the age difference found no significant difference in aberrations. Picciano18 attempt to correlate dose with chromosomal response is subject to uncertainties. Picciano acknowledges that the discovered damage may be due to a worker's prior exposur to higher levels of benzene which recent phenol samples would not indicate. He had access to records of only five years of prior employment.25/ Picciano's cohort was the combined benzeneexposed . populations of three plantss22/ it is unclear why only 52 workers were available during an employment period of 26 years. His control group is small and its compatability with his benzene-exposed subjects is uncertain, for example as regards age. Despite his finding the age difference insignificant, other 34/ Picciano, personal communication. 35/ Picciano, personal communication. DO 068104 CONFTDFNTIAl YZ0015941 40 studies (Tough, 1970) have associated age with an increased incidence o benzene-induced chromosomal aberrations. The significance of cytogenetic studies of occupational ex posure is limited by lack of data on dosage, the generally small size of populations, an inability to control for possible varied degrees of latency in the subjects, and by the often low number of aberrations in workers (around one percent). Court Brown . (1965) has noted that the incidence of chromosomal aberrations in the general population is 2.6 in males and 3.0 in females. This evidently spontaneous incidence of aberrations might also result from the complexities of laboratory techniques. The association of benzene with cytogenetic damage is subject to further uncertainty due to the still early stage of research. Some non-carcinogens and non-mutagens (e.g., caffeine) produce chromosomal damage. Conversely, carcinogens do not necessarily damage cultured chromosomes Jji vitro. Finally, the significance of such aberrations is uncertain: it is not cl' ar whether chromosomal alterations are causative or whether they are perhaps accidental or exceptional occurrences such as accompany the attack of any disease. Evidence shows that benzene causes cytogenetic damage. The research is new and the dose-response relationship is unclear. More importantly, the meaning of chromosomal effects are not known, though these effects bear some resemblance to those of / Goldstein, personal communication. DO 068105 CONFTDFNTTAl. YZOO15942 41 documented carcinogens, particularly ionizing radiation, and may have a role in the development of leukemia. Chromosomal damage has been shown to persist long after exposure. There is no known correlation between the persistence or extent of chromosomal \ aberrations and benzene dose. III. Agencies/Quantitative Risk Assessment!!/ OSHA The Occupational Safety and Health Act requires that in r g- ulating toxic materials, the agency "set the standard which most adequately assures, to the extent feasible, on the basis of the best available evidence, that, no employee will suffer material impairment of health or functional capacity" ($6(b)(5)). Under this authority, OSHA promulgated its benzene regulations based on a review of the scientific evidence summarized above and a feasibility study of the 1 ppm standard. 37/ This and subsequent sections were based in part on inter views with: Dr. Peter P. Infante, OSHA; Dr. Jim Vail, Chemical Specialties Manufacturers Association, former project director for benzene at OSHA; Grover Wrenn, Clem nt Associates, former Health Standards Director OSHA; Dr. Robert E. McGaughy, Deputy Director, CAG; Diane Berkley, Solicitor's Office, OSHA; Ed Klein, Solicitor's Office, OSHA; Morton Corn, Johns Hopkins University, former Adminis trator of OSHA; Richard Johnson, Environmental specialist, EPA, formerly project manager for benzene; Dr. Bernard Goldstein, Rutgers University; Dr. Dante Picclano; Barbara Bankoff, Special Assistant to the Administrator for Air Qualtiy, EPA; Robert Kelham,'BPA, RTP; Michael Dusetzina, EPA, RTP. DO 0f>81 06 OONFTDFNTTAI YZOO15943 - 42 OSHA did not base its standard on any quantitative risk assessment for benzene exposure. The primary basis of OSHA's standard was the evidence of benzene leukemogenicity. Its revi w of the scientific literature showed the range of effects associ- ated with benzene and pointed to evidence of effects at low levels. OSHA acknowledged unanswered questions about benzene's malignancy but noted that under the OSH Act it could not wait for answers while workers were exposed to a life-threatening sub stance; Because of benzene's chronic toxicity and the unknown range of human susceptibility, OSHA could not "conclude that 10 ppm provide[d] sufficient protection ... to all workers" (43 FR 5925). Thus the limiting criterion was feasibility. OSHA commissioned a consultant, A.D. Little 6 Co., to study the feasibility of the 1 ppm standard. The report considered in detail the use and the estimated levels of occupational exposure to benzene in various industries. Its study of the standard's economic impact assessed effects on price, productivity and market structure. It also listed as benefits of the standard an expected reduction in potential leukemia cases and blood abnor malities; a better-informed workforce resulting from a training programs required by the standard; and increased medical know ledge due to detailed recordkeeping. The report concluded that the 1 ppm standard was- feasible and that compliance would impose capital investment costs of around $266 million, first year operating costs of about $187 million and annual costs of $34 million (43 FR 5934). DO 068107 CONFTDFNTTAl YZOO15944 43 In answer to legal challenges that it should perform a quantitative risk assessment, OSHA argued repeatedly that ther was insufficient data to engage in risk assessment. It stated that it was not in principle opposed to quantitative risk assess ment, particularly in setting priorities, but it would not use it in setting exposure levels. In the absence of contrary evid nee, the Agency adhered to its policy that there is no safe level for a carcinogen and that protection of the worker dictated reduction of exposure to the lowest feasible level. Indeed, OSHA interpreted the OSH Act as precluding a balancing of benefits and costs.li/ Relying on the one-hit dose-response model that assumes there is no safe level for a carcinogen, OSHA understood its mandate to be setting a standard assuring "to the extent feasible" that "no employee will suffer material impairment of health" (29 U.S.C. 5655(b)(5)); thus, the standard for a carcinogen could not be set higher than the constraints of feasibility, technology and economics would allow. OSHA perceived as well that interpreting its mandate to allow consideration of different achievable levels of exposure would involve dealing with the methodological uncertainties of risk analysis. Prior to the benzene case, courts had upheld OSHA's interpretation of feasibility as economic cost and as the engi neering capability of the Industry as a whole..!?/ 5s/ Industrial Union Dept.. Brief for the Federal Parties. 39/ B.g., D.C. Circuit in Industrial Union Dept. v. Hodgson, 499 F2d 467 (1974). ----------------- ------- [^0 068108 OONF TOFNTT AL YZ0015945 - 44 - OSHA argued before the Supreme Court that in making judg ments about specific hazards, the agency has the duty to set a standard at the level which assures the greatest protection (Brief for Federal Parties, 1979). Given its policy that there is no safe level for a carcinogen, feasibility would be the only limit to OSHA's standard. Reasons for selecting 1 ppm as the lowest feasible level" include limits to the technical capa bility of measuring low concentrations (OSHA, 1977). Also, OSHA claimed that 1 ppm approaches the level at which benzene occurs in nature; an even lower- standard would require cleaning the ambient air (OSHA 1977). Furthermore, OSHA felt that the 10 ppm standard had no Intrinsic foundation. It was adopted in 1971 from the American National Standards Institute concensus standard which was established in 1969 when the scientific data on benzene was less than that available. The past standard was also based on the association of benzene with non-malignant blood disord rs, not on its leukemogenicity (43 FR at 5918). Finally, OSHA was motivated in its choice of a 1 ppm standard by its conviction that that level was readily achievable. The A.D. Little study, in fact, concluded that the majority of 98,000 refinery workers were exposed to levels actually under 1 ppm. OSHA asserted that if A.D. Little had found i ppm in fact not feasible, it would have explored other standards. OSHA did offer a quantitative risk assessment in its Supreme Court brief after the Fifth Circuit had criticized its failure to DO 068109 CONFTDFNT TAl YZOO15946 45 engage in a balancing of costs and benefits.i2/ This exampl seemed intended primarily to show the crude nature of risk assessment and how different assumptions could produce a signifi cantly different result based on the same data. OSHA's current position on risk assessment is enunciated in its generic policy for the "Identification, Classification, and' I Regulation of Potential Occupational darcinogens" (45 FR 5001). i After extensively reviewing the uncertainties of risk assessm nt and proposing models for extrapolating from animal studies, OSHA concludes that the uncertainties of extrapolating from high d ses in animals to low doses in humans are jto*great to be useful to a risk-benefit balancing and Setting exposure limits. Quantitative i i risk analysis has a role in setting priorities by comparing I relative risks, and it can be used to estimate the reduction of risk that has resulted from a regulatory action. OSHA furth r suggests that risk assessment should not rely on single numbers but state the range of risk based on cautious assumptions about the uncertainty. j EPA On April 4, 1980, EPA proposed an emission standard of 97 percent control for existing maleic anhydride plants as a first step in reducing benzene in the ambient air. EPA has authority under $112 of the Clean Air Act to regulate "hazardous air pollutants" which, "in the judgment of the Administrator may 4b/ See Appendix B for a short review: of the OSHA risk assessment. DO 06811o CONF tdfnttal YZOO15947 - 46 - cause or contribute to an increase in mortality or in serious irreversible or Incapacitating reversible illness" (42 PR 29332). The proposed standard was based on EPA's report "Health Effects of Benzene Germane to Low Level Exposure (1978a), an "Assessment of Human Exposure to Atmospheric Benzene" (1978b) and calculation of a dose response curve by its Carcinogen Assessment Group (1978c). The scientific basis for EPA's regulation, exhaustively re viewed in its Health Effects Document, comprised the same studi s on which the OSHA benzene standard was based. EPA recognized that this evidence primarily Involved research into occupational exposure at levels higher than those found in the ambient air, but noted the Administrator's "generic determination that, in view of the existing state of scientific knowledge, prudent public health policy requires that carcinogens be considered for regulatory purposes to pose some finite risk of cancer at any exposure level above zero" (45 PR 58646). Finally, EPA used the quantitative risk assessment in a general balancing of the risks and costs of options to arrive at its final standard (45 PR 58646). The comparison of options considered environmental impact (the degree of reduction in benzene emissions), the energy impact of particular methods of controls (such as carbon adsorption and thermal incineration), and aspects of economic impact. The option chosen, 97 percent control with best available technology (and 100 percent control DO Of>B11 1 CONFIDFNTIA YZ0015948  , .r 47 in new plants)* is expected to result in only .03 to .19 deaths per year.Ai/ it was estimated to cost $6.6 million in capital* $2.5 million in additional annual costs* possibly one plant closing* and a 1.2 percent price increase for maleic anhydride. EPA's June 1977 decision to list benzene as a hazardous air pollutant under Section 112 followed OSHA's promulgation of an Emergency Temporary Standard. During 1978* the agencies had numerous meetings. OSHA evidently considered EPA's risk assess ment but did not formally respond to it. After the decision of the Fifth Circuit requiring a risk assessment* OSHA's couns 1 appended CAG's risk assessment to its brief but evidently nly for the purpose of showing the crudeness of the technique. OSHA modified its regulation to exempt gasoline stations (and indus tries using liquids containing less than one percent benzene) in recognizing service stations as a borderline case between occupa tional and ambient air exposure and in expectation of appropriate coverage by EPA's action. Neither agency has yet moved to regulate service stations. EPA has specified the general role of risk assessment in setting ambient air standards in its "National Emission Standards for Identifying* Assessing and Regulating Airborne Substances Posing a Risk of Cancer" (44 FR 58642). Citing the uncertainties of ris)f assessment* EPA limits the use of risk assessment to 4l/ 6ac estimated 3.39 deaths would result from emissions from chemical manufacturing in general (EPA* 1979). fr 06811? OONF TDFNT TAI YZ0015949 48 providing supplementary evidence in support of a finding of significant risk: the assessment cannot be ignored if it shows risk, but is too crude to be the basis for not listing a sub stance. Thus, EPA will conduct a quantitative risk assessm nt for agents listed as carcinogens under Section 112. The tisk assessment will be used to set priorities for regulations of particular source categories and to determine the degree of control required in final emission standards for those source categories.2/ B. Critique: OSHA The Occupational Safety and Health Act does not require a quantitative risk assessment. While the Act may mandate quanti tative methods in the setting of priorities (Industrial Union Department), the basic limit it sets to OSHA standards is one of "feasibility" the meaning of which is loose and ambiguous. OSHA argued in court that there was too little data on >rhich to base a risk assessment for benzene. It has also stated that it did use 4 risk assessment generally in setting priorities and in showing * the benefits of a standard once the lowest feasible level had been chosen. (Industrial Union Department, Brief.) Other elements in its rejection of risk analysis included the policy * concern that the Secretary of Labor, who has a highly defined . constituency -- a benzene or vinyl chloride standards affects 42/ Appendix D briefly reviews the background of the initial EPA tegulation of benzene. DO 068113 CONF TDFNT T At YZOO15950 - 49 known individuals -- could not be associated with any quantitative method that approached cost-benefit analysis and the s tting of a dollar value on life.Al/ There were also significant political pressures for a new benzene standard.14/ Finally, the agency, in 1977, was regulating in an atmosphere of judicial deference to administrative discretion.^/ While OSHA may have made a less substantial showing of the standard's basis than it had in past regulations, no legal precedent provided any basis for expecting that once it had detailed its safety concerns it would have to show more. Courts reviewing Emergency Temporary Standards have be n concerned with the actual probability that an asserted hazard poses an actual hazard and with the gravity or severity of th harm threatened. The Court of Appeals for the Third Circuit has held that OSHA must show more than "some probability" that a 3^/ This is in contrast to EPA which has a diffuse constituency as large as the United States. Noted by Morton Corn, personal communication. 44/. Background to the benzene standard is reviewed in Appendix C. 45/ Courts have required that OSHA be explicit in its rationale for a particular standard. See, Dry Colors Manuf. Assoc. v. Dept, of Labor, 486 P2d 9lTT3d Cxr. 1973). See also, Industrial UnionDept. v. Hodgson,. 499 F2d 467 (D.d. Cir. 1973): "where that choice [of a regulatory course) purports to be based on the existence of certain determinable facts, the Secretary must ... find those facts from evidence in th record. By the same token, when the Secretary is obliged to make policy judgments where no factual certainties exist ... he should so state and go on to identify the considerations he found persuasive." DO 068114 OONFTDFNTTAI YZ0015951 50 substance is carcinogenic and that OSHA need not show actual harm, but simply that harm would result.i/ While OSHA has consistently argued against the use of benefit-cost analysis in the actual setting of a benzene stan dard, the 1 ppm standard represents an implicit weighing of cost and benefit. It is unclear from OSHA's promulgation of the standard and from interviews with agency personnel, how 1 ppm was arrived at as the "lowest feasible level," rather than 5 ppm or, in view of the A.D. Little conclusion that 1 ppm was easily feasible, .5 ppm.AZ/ Instead of comparing a number of levels to find the "lowest feasible", OSHA evidently selected a level that would be reasonable in terms of economic cost and political feasibility. OSHA's protestations to the contrary, this elem nt of balancing appears inevitable in standard setting.45/ OSHA has argued that it could have applied a safety factor of 10 or 100 to the demonstrated effects of benzene at high 46/ Dry colors Manuf. Assoc, v. Dept, of Labor, 486 F2d 98 (3d Cir. 1973) (reviewing an ETS for 14 carcinogens). 47/ Discussions with OSHA personnel indicate that A.D: Little either considered the feasibility of 10, 5 and 1 ppm levels or such an examination was discussed in the contract stage. But the final A.D. Little document gives no hint of this nor a reason for not discussing other levels. 48/ The Supreme Court noted the balancing implicit in the benzene standard: "... OSHA adopted an action level [of 0.5 ppm below which monitoring was not required], largely because the Insignificant benefits of ... performing the necessary monitoring did not justify the substantial cost." Industrial Union Department, 00 S.Ct. 2844 at 2868. 00 068115 oonfiofnttai YZ0015952 >`' 51 levels to justify a standard of 1 ppm.!!/ The safety margin would be justified by the uncertainties of benzene research and the nonsensitive nature of epidemiological studies, as well as by the potentially important factor of individual susceptibility which research has not tried to measure. However, even this roughly quantitative approach was not used in its promulgation of the standard. OSHA's review of the research on benzene's health effects was, with the exception of its initial reliance on the Infant study as quantitatively significant, thorough and accurate.!!,/ OSHA issued the new exposure standard based on benzene leukemogenicity because the Infante study represented strong qualitative evidence of this link which was not considered in setting past standards!!/ and because in contrast to leukemia, blood disord rs caused by low level exposure are apparently reversible. OS HA also considered the slight evidence of benzene's cytogenetic effects, but little research on these effects and on fetotoxicity was available at the time. W Industrial Union Dept, v. American Petroleum Institute, Brief for the Federal Parties, p.21, n.25. 50/ OSHA's assessment that the results were significant or indicative or risk is a matter of policy-based judgment. 51/ OSHA, however, issued its ETS under the impression that the Infante study offered data about low benzene concentrati ns, and the association of benzene and leukemia was reasonably well-established prior to the Infante study. DO 068116. CONFTDFNTTAL YZOO15953 52 Finally, the uncertainty associated with OSHA's argument that there is no safe level for a carcinogen should be noted. The linear no-threshold theory assumes that one molecule of a carcinogen hitting a cell at the right site can cause cancer. It is a subject of ongoing debate but is, in any case, accepted government policy.(Nelson* 1978? Lave, 1970). In sum, OSHA did not' use risk assessment even to set priori ties before deciding to regulate benzene. To add quantitative substantiation to its case, OSHAvcould have pointed to evidence , of benzene leukemogenicity at exposure levels around 100 ppm and used a safety factor based on the recognized scientific uncer; tainties about benzene's effects and about extrapolation. Since the 1 ppm standard ostensibly based on feasibility itself in- v volved some sort of balancing (known familiarly as turning the screws until the screaming gets too loud), it could have ex plicitly considered other levels of feasibility or cost, includ ing S and .5 ppm. Finally* it could have adapted to its own showing of health effects a quantitative risk assessment like that produced by EPA finds health effects at levels in the low parts per billion. In the wake of the Supreme Court's ruling on benzene, OSHA will have to confront quantitative means of sub stantiating health effects. EPA CAG's risk assessment model for benzene is plausible and CAG generally made reasonable assumptions in the face of limited DO 068117 C0NFTDFNTTA1 YZOO15954 >* 53 evidence. A review of the risk assessment, as detailed in Appendix A, indicates that CAG might have been more explicit about the uncertainties of its assumptions. It would have been helpful for CAG to have considered several values, for example in the estimation of dosage, to indicate a range for the calculat d risk. Also, its estimated confidence interval (95 percent from 34 to 235 assuming a precision of within two fold in the exposure estimate) must be interpreted carefully; it refers only to th standard error in the estimate, calculated using standard assump tions. No account is taken of the uncertainties that arise from the unknown exposure levels in each case, misdiagnosed cases (both false positives and false negatives), exposure to oth r leukemogens, and failure to take into account of latency periods. The CAG quantitative risk assessment estrapolates down to low exposures on the basis of the linear non-threshold model. As the Table below indicates, the highest response to exposure and effectively all the expected leukemias occur at levels below the lowest concentrations, that is, below 4ppb. Less extreme func tional forms for a dose-response relation would, by contrast, indicate almost zero deaths below 10 ppb. 00 068118 oonftdfnttai YZOO15955 *4 54 Annual Average Benzene Concentrations (PPb) Expected Number of People Million of ppb- Benzene-Caused Exposed______ Person-Years Leukemia Deaths/year 10.0 4.1-10.0 1.1- 4.0 0.1- 1.0 80,000 200,000 115,000,000 7,000,000 1.2 1.4 287.5 3.85 0.41 0.47 97.50 1I .30U 99T;70 Total This table is a modification of Table 1-2 in the "Assessment of Human Exposure to Atmospheric Benzene" (EPA* 1978) with the expected leukemias calculated following the CAG quantitative risk assessment (see Table 5, CAG risk assessment). The "Assessment of Human Exposure to Atmospheric Benzene" (1978b) commissioned by EPA from the Stanford Research Institute, is subject to uncertainty due to the paucity of information n which it was based. Data from actual samples of benzene cone ntrations have subsequently become available.-^/ while any updating of the risk assessment should utilize these more accur ate measures of benzene emissions, it appears that the differ nee between the old and new estimates does not significantly affect the risk assessment. As a final point on the scientific basis of the risk assess- 4 ment, it should be noted that due to the long association of ben zene and hematotoxicity, this chemical is one of the roost ext n- sively researched. There have been several epidemiological studies of benzene. While dosage data are scarce, such studies I 52/ A detailed critique of the Exposure Assessment was submitted by. API in its comments on the Draft Health Assessment, December 9, 1977. YZ0015956 55 are a better indicator of human effects than any research on animals^!/ (Risk Assessment document). Deputy Director of CAG, Dr. Robert KcGauhy, has noted on the basis of CAG's significant experience with risk assessment (on over 60 chemicals so far) that the data on benzene are nearly as extensive as that avail able for any substance.il/ Despite the amount of research on benzene, the unsettled un certainties, the cost and problems of providing large samples, and the possible lack of occupational settings in which benzene can be isolated as the only contaminant, will hamper the produc- I tion of more precise, quantifiable information in the futur . Given, in addition, the low incidence of leukemia (still rela tively low even when induced by benzene), it is unlikely that a well-designed study could show a significant increase in the incidence of leukemia from low-level exposure. It would be impossible to find thousands of workers exposed to 5-15 ppm of benzene over several decades, being sure that exposure was never higher and that exposure to other leukemogens was controlled at 10 ppm. For a quantitative supplement to its finding of health effects, OSHA must turn to extrapolation from epidemiolog ical and toxicological studies. Such extrapolations cannot be 53/ See the Quantitative Risk Assessment Document. 54/ Conversation per telephone, July 24, 1980. This evaluation is based on the availability of several epidemiological studies on the effects of benzene while data on most toxic substances is limited to animal tests. DO 068170 OONF TDFNTTAI YZ0015957 56 mathematical straitjackets, as the Supreme Court recognized. But used #ith explicit awareness of their limitations, they offer a significant basis for regulatory decision as EPA's use of risk assessment suggests. The subsequent history of EPA's quantitative risk assessment indicates the ultimate limit on the use of quantitative methods. Commenters at EPA's maleic anhydride hearings, having criticized specific assumptions underlying EPA's calculations, concluded that EPA's risk assessment would show on the same data, but und r somewhat modified assumptions, that there is in fact no risk associated with atmospheric exposure to benzene.-^/ However, once the quantitative showing of risk has been factored into th declsibnmaking process, the decision to regulate is a matter of policy. H7Appendix A suggests modifications of EPA's interpretations or the data, emphasizing objective rather than policy-based criteria for modifying the calculation. 0 0681?i OONFTDFNT TA( YZOO15958 A APPENDIX A Review of CAG Risk Assessment Mathematical Model Employed The CAG assumes that for low exposures, the lifetime proba bility of leukemia may be represented by the linear equation P A + B x, where A - the leukemia rate in the absence of benzene, x - the average lifetime background benzene exposure In ppm, and B the change in the leukemia rate per 1 ppm benzene. 0 Assuming that the relative leukemia risk, R, of benzene- , exposed workers compared to' the general population is independent of the length or age of exposure but depends only on the total exposure, it follows that r . 2 1 or ...1 *21 A + Bx 1 RP1 - A + B (x1 + x2) Pj A + B X1 so that B where Zt-LR-Z-U *2 ambient air level of benzene x2 " industrial level of benzene P1 - lifetime probability of dying of benzene with no or negligible benzene exposure. ^0 o&sip? CONFTDFNTTAI YZOO15959 u A-2 Estimates of R and Xj are obtained from the epidemiology ical studies. The exposure values for are obtained from the Assessment of Human Exposure to Atmospheric Benzene. The estimation of the lifetime probability, P^, of death due to various forms of leukemia for a member of the general population is based on data from the "Vital statistics of the United States," 1973, Volume 11. Age-specific rates are estimated by dividing the number of deaths due to a particular leukeiqia by the total number of people in that age class. CAG estimates these probabilities on a "constant segmented model" and gives the results in a table. Infante Estimation of Relative Risk CAG accepted Infante's recalculation, based on his more com plete followup, of the expected incidence of leukeumia from 1.38 to 1.25 per 100,000. CAG also added to Infante's seven leukemia cases two additional ones reported by Dr. Marvin Sakol which wer "probably* in Infante's cohort. (See Infante et al., Letter to The Lancet, October 22, 1977, p.868).-/ CAG calculates R 1775 " 7*2 Since the additional leukemias were not discovered in the framework of the original study (indeed Infante omits them from his V/ CAG was persuaded by Dr. Sakol'a conviction that leukemia deaths were consistently under-reported for reasons associ ated with workman's pompensation, according to Bob McGaughy of CAG. ^ 06fti ooNrrDFNTTA( YZ0015960 r V A-3 own calculation of the leukemia incidence ~ in his letter to the Lancet), we omit them from the calculation of relative risk: R " T7I5 * 5*6 Estimation of Average Occupational Exposure There are no air monitoring records covering 1940 to 1946. A new ventilation system installed in 1946 resulted in a reported range of zero to 15 ppm in "most areas" of the plant. All areas were reportedly within the maximum safe limit of 100 ppm (Baier Testimonyr OSHA hearings). CAG assumes on the basis of this information that exposure before 1946 was at least 15 ppm and less than or equal to 100 ppm. In the absence of any data on pre-1946 exposure levels, though, it is likely that they were higher before 1946 than afterwards. There is no data on man hours spent at particular locations. There was also testimony of occasional excursions above 1,000 ppm and of uncertainty about the actual use of respirators during duty in areas of unusual concentrations. Also, CAG assumes that actual concentrations were within the recommended levels during the 36 year period: there is no guarantee of this. In addition to the uncertainty about pre-1946 exposures, there is also doubt about technical accuracy in measuring low doses of benzene in the 1940s. HO 0681P4 OONFTDFNTTAI YZ0015961 t A-4 4 Based on the assumption that the recommended levels were the actual levels for the years of the study,-2/ CAG estimates a continuous lifetime exposure level of 40.36 ppm for workers starting before 1946, when the level was reduced by improved ventilation, and an exposure duration of 35 years. CAG assum s an exposure of 23.7 ppm for workers starting after 1946, and an exposure duration of 25 years. This and the following calcula tions hssume, non-conservatively, that a worker starting in 1940 or in 1946 stayed in Pliofilm during his whole working life. The group includes workers who may have been employed as little as one day. CAG produced the following two estimates: Employed prior to 1946 Employed after 1946 40.36 x xlx* 4.4 ppm 365 3 70 23.7 x x ^ x M 1.8 ppm 365 3 70 CAG calculates the geometric mean of these continuous exposures, 4.4 x 1.8 - 2.81 I, however, will use both high and low estimated lifetime averages in subsequent calculations to indicate the range of exposures. T7 dAG tabulates the recommended exposure-limits for the particular years, e.g. 1940-46, 100-15 ppm; 1947,' 50 ppm; 1948-56, 35 ppm. CAG arrives at a time weighted averaqe of 23.3-39.3 ppm. y OO 068105 OONF TDFNT TAi YZ0015962 Estimation of Lifetime Probability of Leukemia Per Unit Exposure Using the calculation of relative risk (R), the lifetim l probability of dying from a particular type of leukemia (P), and the model discussed previously, CAG estimates: B - i 13x--_.JL1 B - .006732 x Using the lower number of deaths (7) found by Infante (resulting in the relative risk estimate of 5.6) and using both the high and low estimates of lifetime exposure arrived at above, I calculate estimates 1/4 to 1/2 those of CAG: High: B .006732 x .00738 Low: B - .006732 x .00374 Aksoy Estimation of Relative Risk In calculating the leukemia rate per unit of benzene xpo- sure using Aksoy*s results, CAG first estimates the incidence of leukemia using the figure of 25 cases out of a population of 28,500 shoeworkers at risk during the 6-2/3 years of observation. CAG, eliminating the cases of lymphoblastic leukemia, "not thought to be associated with benzene exposure" (1978c, p.15), estimates the incidence among shoeworkers to be 13.15 per YZ0015963 I 4 A-6 100000. Modifying the available estimate of the incidence of all leukemias in the Turkish population (2.5 to 3.0 per 100/000), CA6 calculates an expected incidence of non-lymphoblastic leu kemia of .66 per 100,000. There is uncertainty about the actual incidence of leukemia in the general Turkish population; industry witnesses at the OSHA hearings pointed to higher European inci dences. But the cancer rate in less developed Turkey is pr sumably lower and CAG's estimated number of expected deaths is reasonably conservative. CAG arrives at the incidence of leukemia associated with occupational benzene exposure as _ 13.15 6S~ 19.92 Estimation of Lifetime Average Exposure CAG uses Aksoy's estimate that the benzene concentration in the shop outside of working hours was 15 to 30 ppm and the level during working hours was 150 to 210 ppm. CAG takes the geom trie mean of these exposures: x2 (15 + 30) _ (150 + 210) 5* 2 63.6 ppm (a) AsBurning a ten hour working day, 300 day working year, aver age Age at the end of observation of 50, and average period of exposure of 9.7 years, CAG calculates lifetime average exposure *v2. 6033.06 VX (!-!>jj> XX (2-<2) X (T9r.y7-) - 4- .2,,2 ,,ppm ^0 0681?7 ^ONrr dfNTtai YZ0015964 A-7 Computation (a) above assumes that the duration of the non-working hour concentration was equal to that of the higher working hour concentration. This restilts in an estimate of average exposure, 63.6, that is lower than the mean of the \ working hour exposure, which would be 180 ppm. In the calculations that follow, CAG does not take into account this non-working hour workshop concentration. But considering that the shoeworkers' shops are likely attached to their abodes, this exposure level represents a background expo sure that must be distinguished from the background ambient air exposure of American workers, which is on the order of a few parts per billion. Instead of taking the mean of these exposures and weighting them for 10 hours of work per day, we assume a 10 hour working exposure at the high level and add a 10 hour non working hour exposure to the background level. Also, the figures used by CAG are low. Aksoy has noted the poor ventilation of th shoeworkers' shops? the low boiling point of benzene would 1 ad to higher concentrations in a warm climate. In any case, Aksoy himself uses the range of 210 to 650 ppm.1/ we use these modified assumptions and figures in calculating lifetime av rage exposure: . ,(15 30) 5 10 + ( -2-1--0--- +5--6--5--0- ,io, .300, .9.7, (3or> 30.06 ppm J7 Aksoy, "Leukemia in Shoeworkers Chronically Exposed to Benzene." oo OQ/Vfr 0687 TDfNTT4, YZ0015965 I 4 A-8 To indicate the range of possible* exposure# we also calcu late lifetime exposure on the basis of CAG's more conservativ * figure for working hour exposure. t (15 + 30) ------ 2------ ) ;i50 + 210 + < "2-------- ,10. (74) <> ,9.7. {^r> 30**06 ppm Estimation of Lifetime Probabilities of Levels per Unit Exposure CAG applies its average exposure figure to the equation given earliert B - li_iLzJLI *2 8 - .04517 x 1} - 0.20252 1 calculate with the, higher exposure figure and accounting for . ** t the relatively high background exposure B 04517 x (19.92 - 1) 15.MSS 00342 ... : * '1 Estimation of Relative Risk Ott 3 deaths with .8 expected (p<0.047) R - - 3.75 .# t Estimation of Lifetime Average Exposure ptt estimated the ppm/months of exposure of each worker from fecords and plant benzene measurements (in ranges from high > V. -- , . DO 088179 OONFTDFNTTAl YZ0015966 A-9 (25+ ppm TWA) and low (2 to 9 ppm TWA) exposure) (high (1,000+ ppm months) low (zero to 499 ppm months)) observed and expected deaths and tabulated them in terms of observed and expected deaths corresponding to duration of exposure (see Ott, Table 7). CAG took the average of each exposure level and weighted averages by expected deaths. (250 x 65.1 + 750 x 16.2 + 1250 x 32.8) ----------------- 65.1 + VSlT + 'SiS.'S------------ 608.46 ppm months CAG then calculated the average lifetime exposure on the assumptions of an 8-hour day, a 240-day working year, and an average age at the end of observation period of 65 years. (608.46) (IS) (240) (753) 171 ppm The object of weighting exposure by expected mortality valu s is to introduce an element of duration and proportionality (deaths correspond to level of exposure and time) but exposure can be cor related more directly with time by weighting the mean exposure level in a production area by duration of exposure in that area. In Table 5, Ott gives the number of workers who spent a c rtain number of years in a particular production areai <1 year 1-9 years 10-19 years 20+ years I9 II 21 III 96 36 64 139 20 23 39 43 27 24 no c.onf 068130 tDFNTTW YZ0015967 A--1 0 *4 The average number of years spent in a particular production area is derived from the table by weighting the mean number of years in a range by the number of workers in that range I (9x1 4 36x5 4 20x15 + 43x30) (9 4 36 4 20 4 43) - 16.47 II (3x1 + 12x5 + 1x15 4 5x30) (3 4 12 4 1 4 5) - 10.85 III (1x1 + 13x5 4 13x15 + 10x3) (1 + 13 + 13 4 10) - 15.16 These figures for the average number of years spent in a produc tion area are UBed to weight the mean exposure level for each production area, to derive the average occupational exposur . Ott gives the ranges of exposure level for the production areas on pp.3-4: I Low (2-9 ppm TWA), very low (<2 ppm TWA) II High (25+ ppm TWA) moderate (10-24 ppm TWA), low, v ry low III High (3 x 16.47 13 x 10,85 t 30 x 15.16) 16.47 + i07$1"+15. 16 15.19 ppm averag occupational exposur Using CAG's assumption of an 8-hour day and 240-day working y ar. and estimating the average duration of years on the job to be 15.5, we calculate -a ,, 240 1 (15.50) 15,19 x 75? x 1 x "TO-- .737 ppm Estimation of Lifetime Probability of Leukemia per unit Exposure Using the equation B - .-_L> X2 DO 068131 CONFTDFNTIAL YZ0015968 A-1 1 CAG calculates the lifetime probability of leukemia per unit ex posure to be B - .002884 x {3` J7r-1-) *04638 Substituting our value for the lifetime average exposure, we estimate B - .002884 x 1} - .01076 Finally, CAG takes the geometric mean of its values for B to compute the total probability of deaths due to 1 ppm of benzene in the air breathed over a lifetime: 3 B - /.011854 x .020252 x .048180 "W .'024074 However, because the studies vary considerably in population siz and thus should not be accorded equal statistical weight, we weight our estimates of the lifetime probability of leukemia per unit of benzene (BJ by the square root of the population on which the probability is calculated: 8 748 +-----287506 +----- 584------------------------- -00471 Expected Number of Leukemia Deaths Due to Environmental Exposure to Benene g With D representing units in 10 ppb person years, the expected number of leukemia deaths per year may be estimated approximately by the relationship >0 DOhlfr D&fii 3p TDFNTT4i YZ0015969 ,, _ .024074 x D x 103 MD * ----------- VFtt----------- .33962 D where .24074 is the geometric mean of the slope parameters of the three studies and 70.96 is the average life expectancy in the U.S. By this equation and using our weighted average of th slope parametersr we calculate .0047193 x D x 103 "d .066506 D Using exposure figures for sources from the "Assessment of Human Exposures" document, CAG calculates expected deathst--4'/ Sources of Exposure Exposure in 106 ppb-peraon yean years CAG Expected Number of Benzene Leukemia Deaths Our recalculation per year of expected deaths Chemical Manufacturing Coke CVens Petroleum Refineries Automobile Emissions Gasoline Service Stations Self-Service Gasoline 8.5 .2 2.5 150.0 19.0 1.6 2.88 .07 .85 50.89 6.44 .54 6 f 67 .57 .01 .17 9.98 1.26 .11 12.10 CAG also uses exposure figures based on a second model that instead of assuming a static population tries to follow the typical individual through a typical day, a primary factor of which is the location of his residence. CAG then calculates deathBj5-'/ 77 This iff Table 4 in the CAG risk assessment, with our calcu"" lations added. 5/ Table 5, CAG risk assessment (EPA 1976c). DO 068133 CONFTDFNT TAl YZ0015970 A-13 Vicinity of Residence Chemical Manufacturing Coke Ovens Petroleum Refineries Urban Areas Exposure in 106 ppb-person years years CAG Expected Number of Benzene Leukemia Deaths per year Our recalculation of expected deaths 10.0 .2 4.5 250.0 3.39 .07 1.53 84.80 .67 .01 .29 16.63 264.7 89.80 17.60 We thus calculate that about 18 leukemia deaths per year can be expected from ambient benzene exposure, compared to 90 found by CAG. 00 068134 C0NFTDFNTTAI YZ0015971 APPENDIX B Risk Assessment in Reply Brief' for the Federal Parties, Appendix, Industrial Onion Department v. American Petroleum Institute The appendix to the OSHA reply brief is a review of the uncertainties of risk assessment compiled in response to the Fifth Circuit decision requiring an agency risk assessment before setting a standard. OSHA presented a risk assessment in conclu sion to its critique as a counter example. In its example, OSHA uses only the Infante study, taking th figures of 5.6 excess leukemias in a population of 748 and an ex posure level of 10 ppm. OSHA assumes an exposure duration of five years (resulting from a five year turnover) and calculates that a follow-up over the lifetime of the cohort would find 17 leukemias resulting from the constant leukemia rate. Assuming that 30,000 workers are exposed to 10 ppm of benzene, then 690 leukemias would result every five years. OSHA uses only one study. It makes the unfounded assumption that the Infante study involved average exposures of 10 ppm. While it is correct in trying to account for turnover, duration of exposure is unknown and an average of only five years seems low. OSHA makes an exaggerated speculation that a follow-up of the remaining 25 percent of the cohort would result in a total of 17 leukemias assuming the rate of observed over expected leukemias of the first 75 percent remained constant over the lifetime of the cohort. While it is appropriate to account for latent effects and thus to expect that some leukemias will yet DO 068135 OONFTDFNTTAl YZ0015972 B-2 develop in the future, 27 years had already elapsed since the period of exposure involved in the study, an ample space for the development of latent leukemias. Finally, the calculation used here is simplistic. Whereas CAG factors in several elements to arrive at a relatively narrow range of slope parameters, OSHA's calculation (10 ppm * 7 deaths in 748 workers) will give one data point that cannot be .i readily compared to another derived from Aksoy or Ott.J/ OSHA's critique of risk assessment in its brief Is reasonable; its list of uncertainties should have been incorporated in its ri^k assessment. 1/ OSHA's risk assessment is, however, no more simplistic than tfiat of Richard Wilson (direct testimony before OSHA, In r_ Proposed Standard for Occupational Exposure to Benzene, OSHA Docket No. H-059, 1977), a risk assessment that evidently pfersuaded the Fifth Circuit that risk analysis was feasible. OSHA answered with the above assessment and pointed out that Wilson had not explained his calculation. OSHA, however, did not respond to the more detailed analysis prepared by CAG 068186 CONFIDENTTAI YZOO15973 APPENDIX C Outside Pressure on OSHA The following is a brief review of the background of the pSHA benzene standard to suggest the political pressures involved .*2/ Benzene had long been of concern to the Rubberworkets Union; a number of studies had found excess leukemia deaths associated with solvent use in that industry. But OSHA had rejected pres sures for a benzene standard for three years after the chemical was designated a "suspected leukemogen" in the 1974 NIOSB Cri teria Document in the absence of data to warrant changing the standard. In 1976, the Secretary of Labor had again directly rejected urgings of the Rubberworkers for a new standard. In 1976, the Assistant, Secretary for OSHA, newly designat d but not yet confirmed, began to immediately organize an emergency standard for benzene. Dr. Eula Bingham was a toxicologist and knew the scientific data on benzene. She strongly felt it should be regulated to the lowest feasible level. Dr. Bingham was also interested in the advocacy function of OSHA. OSHA's past refusal to regulate benzene in.the face of union concern, was seen as symbolic of its past failure to regulate toxic chemicals.--^ 2/ This account is based on several interviews that are n t cited individually. 2/ Union pressure aside, OSHA was criticized in 1977 in a GAO report to a House oversight hearing for its slowness in promulgating health, as opposed to safety, standards, as a result of new interest in OSHA health standards generated by Bingham's predecessor. Dr. Norton. Hearings before a subcomm., Comm, on Gov. Operations, House of Representa tives, 95th Cong., 1st Sess., April 27, 1977. DO 068137 OONFTOFNTTAl VZOO15974 C-2 Thus, a benzene regulation would be a sign of OSHA's concern about toxic agents. Promulgation of a new benzene standard could be sufficiently justified by evidence of benzene's association with leukemia: past regulations had been based on benzene's general toxicity and elation to non-malignant blood disorders and OSHA had a policy holding that there is holding that no level of a carcinogen is Bafe. An ETS, however, is a relatively extraordinary measur . To justify it, OSHA would have to find a new scientific basis for regulating benzene after having refused to do so for three years. Dr. Bingham, who knew Dr. Infante personally and was aware of his study of benzene then in progress, sought his results as a new basis for an ETS. The study was subsequently rushed int print in The Lancet with the follow-up only 75 percent complete. OSHA relied heavily on the Infante study in its ETS promulgation and apparently attributed to it quantitative significance con cerning the effects of low benzene dosage. As OSHA subsequently recognized-the basically qualitative focus of the study, it relied less on the study and in the proposed permanent standard emphasized the standard's foundation in the whole scientific record in the proposed permanent standard. OSHA'b scientific staff made clear the problems, primarily the lack of new data, with the proposed standard. It felt that it was not consulted to any extent because the decision had already been made to set the new standard. DO 068138 CONFTDFNTTAl YZ0015975 APPENDIX D Background on SPA Risk Assessment In 1975, benzene was listed among 10 high volume industrial chemicals for which limited budget, fast assessments were under- \I taken by EPA in prioritizing air borne pollutants. These assess ments considered' the sources of emissions, the populations exposed, and potential technological controls. Subsequently, EPA collated priority lists based on the biological hazards and environmental impacts of pollutants, of various agencies and bodies and identified benzene as a high priority. In 1977, th recommendation of NIOSH that an emergency standard should be issued for benzene added momentum to EPA's handling of benzen . Pinally, the Environmental Defense Fund's petition precipitat d the listing of benzene as a hazardous air pollutant. The scientific staff dealing with benzene at EPA did not no tice political pressure for the listing of benzene, though policymakers might have. EPA's constituency is not as focused as OSHA's and likely had less influence over specific programs. EPA's. compiling of a human exposure survey and quantitative risk assessment in addition to the usual health effects docum nt was an innovation intended to establish the parameters of ri k associated with benzene. The actual role of the risk assessment in regulating benzene emissions is unclear. The health effects and human exposure documents were evidently considered the substantial bases for action. The risk assessment, based on benzene leukemogenicity, does not cover the other effects to DO 0f>81 39 CONFIDFNTTAl YZOO15976 which the general population is exposed. Although the listing of benzene as a hazardous air pollutant in 1977 signifies only th identification of a potential hazard# no substance so listed has subsequently been removed from the list. Thus# the preliminary stages evaluating the health effects of benzene largely deter mined Its regulation. The quantitative risk assessment# devel oped after the listing of benzene in 1977# was used as suppl mentary evidence of a hazard. It evidently was a factor in priori tizing benzene sources and thus in regulating maleic anhydrid . It was an element in the risk-benefit balancing on which th choice of a regulatory option was based. DO 068140 OONFTDFNTTAl YZ0015977 REFERENCES Aksoy, H., et al. Hematological effects of chronic benzene poisoning in 217 workers. 1971. Br. J. Ind. Med. 28: 296-302. Aksoy, M., et al. Details of blood changes in 32 patients with pancytopenia associated with iong-term exposure to benz ne. 1972a. Br. j. Ind. Med. 29: 56-64. Aksoy, M., et al. Acute leukemias due to chronic exposure to benzene. 1972b. Am. J. Med. 52: 160. Aksoy, M. et al. Acute leukemia in two generations following chronTc exposure to benzene. 1974a. Hum. Hered. 24: 70. Aksoy, M., et _al. Leukemia in shoe-workers exposed chronically to benzene. 1974b. Blood. 44: 837. Aksoy, M., et al. Types of leukemia in chronic benzene poison ing. A study in 34 patients. 1976. Acta Haeroat. 55: 65. Aksoy, H. Leukemia in workers due to occupational exposure to benzene. 1977. New Istanbul Centrls. Clin. Sci. 12: 3-14. : Aksoy* M. Letter "Benzene and leukemia." 1978a. Lancet. 1: 441. Aksoy, M., and Erdem, S. Follow-up study on the mortality and the development of leukemia in 44 pancytopenic patients with chronic exposure to benzene. 1978 b. Blood. 52: 285. Chang, 1m Won. Study on the threshold limit value of benzene and early diagnosis of benzene poisoning. 1972. J. Cath. M d. Coll. 23: 429-434. Cited in EPA, 1978a. Committee on Toxicology, NAS. Health effects'of benzene, a review. 1976. Consumer Product Safety Commissi.on Report. Review and analysis of the literature pertaining to the reproductive effects of exposure to benzene. 1980. Court Brown, W. Human Population Genetics. 1967. Deichmann, W., et al. The hemopoietic tissue toxicity of benz n vapors. 1963. Toxicol. Appl. Pharmacol. 5: 201-224. EPA, Assessment of Health Effects of Benzene Germane to Low-1 v 1 Exposure! September 1978a. EPA, Assessment of Human Exposure to Atmospheric Benzene. Jun 1978b. 00 068141 CONFTDFNTIAI YZ0015978 2 EPA, Cancer Assessment Group, Report on population risk from atabient benzene exposure, 1979. EPA, Initial evaluation of Exxon's 8(e) submission of benzene teratology, 1980. . t * Forni,A.M., et al. Chromosome changes and their evolution in subjects with past exposure to benzene. 1971. Arch. Environ. Health. 23: 385-391. ,*v . Forni, A.M., et al. Chromosome studies in workers exposed to benzene, toTuene, or both. 1977. Arch. Envrion; Health. 22: 373-378. Girard, R., et -al. Hydrocarbures, benzeniques et hemopathies graves. 1979. Arch. Naiad. P rof. 311 625-635. Gofmekler, V.A. Effect on embryonic development of benzene and' formaldehyde. 1968. Hyg, Sant. 33: 327^-332. * Goldstein, B.D. Hematotoxicity in humans. 1977, in Laskin, 69-107. Green,. J.D., et al. Inhaled benzene fetotoxicity in rats. 3978.// -T--o---x--i-c--o--l-.---A- "pp-l-.-T---P---h--a--r-m---a--c-o--l. 461 9-18. , v.i 'C~. Hartwich, G*r Schwanitz, G. Chromosomenuntersuchungert nach " - Chronischer Benzol-Exposition. 1972. 'Deutsche Med. Woch. 97: 45-51. ------------ ----- ---;----- -- -V; . Razelton Laboratories Study, 1977. Hett, J.r Maak, H. 1938. Rlinische Wochenschrift. 17: 1376. -Cited in Consumer Products Safety Commission. Industrial Union Dept, v. American Petroleum Institute. 100 S.Ct. 2844, 1980. Industrial Union Dept, v. American Petroleum Institute. Brief for the Federal parties in the Supreme Court, 1979. Infante, P.F., et al. Leukemia in benzene workers. 1977. Lancet. 2: 76-78. -------- -- Infante, P.F. Reply. 1977. Lancet. 2: 868. Jenkins, L.J., Jr., et al. long-term Inhalation screening' studies of benzene, toluene, o-xylene and cumene on experi mental animals.. 1970. Toxicol. Appl. Pharmacol. 16i 818-823. --:-------------- "------------------- -- no 06814? CONFIDFNTTAI YZ0015979 -3- Kissling, M., Speck, B. Chromosome aberration in experimental benzene intoxication. 1972. Helv. Med. Acta. 36s 59-66. Laskin, s., Goldstein, B.D. Benzene Toxicity, A Critical Evalu ation. 1977. J. Tox. Env, Health. Supp. i. Lave, L. Health benefits of abating air pollutants. 1978. Bull. N.Y. Acad. Med. 54s 1235-1244. Leong, B.L. Experimental benzene intoxication. 1977, in Laskin. 45-63. A.D. Little & Co. Economic Impact Statement. Dept, of Labor, 1977. Litton Bionetics, A teratology study in rats. 1977, 1978. McMichael, A.J., et _al. Solvent exposure and leukemia among rubber workers, an epidemiologic study. 1975. J. Occ* Med. 17: 234-239. Maltoni, C., Scarnato, C. The first experimental proof of th carcinogenic action of benzene. 1979. Med. Lavoro. 5s 332-357. Mitelman, P., Levan, G. Clustering of aberrations to specific chromosomes in human neoplasms. III. Incidence and geo graphic distribution of chromosome aberrations in 856 cas s. 1978. Hereditas. 89s 207. Murray, F.J., et al. Embryotoxicity of inhaled benzene in mic and rabbits. 1979, Am. Ind. Hyg. Assoc. J. -40: 993-998. Nawrot, P.S., Staples, R.E. Bnbryofetal toxicity and teratogen icity of benzene and toluene in the mouse. 1979. Teratology. 19 (2) s 41A. ---------------3LL Nelson,, N. Discussion of paper by Vaun A. Newill, R. Wyzga and James R. McCarrol. 1978. Bull. N.Y. Acad. Med. 54: 1245-1256. NIOSH, Occupational Exposure to Benzene. 1974. Occupational Safety and Health Administration. Transcript of Proceedings, Informal hearing on proposed standard for exposure to benzene. 1977. Ott, M.G., et jal. Mortality among individuals occupationally exposed to benzene. 1978. Arch. Occ. Health. 31s 3-10. DO 088143 CONF T OF NT T Al YZ0015980 r*V U 4 Picciano, D. Cytogenetic study of workers exposed to benzene. 1979. Env. Res. 19: 33-38. Snyder, R., Kocsis, J.J. Current concepts of benzene toxicity, June 1978. CRC Critical Reviews in Toxicology. 265-288. Stallones, R.A. Report on mortality from leukemia, Shell Oil Co. 1977. Tabershaw^ X., Mortality study among refinery workers. 1974. Tabershaw, X., Lamm, S., Letter, Lancet 2, 867 (1977). Tareeff, E.M., et al. Benzene leukemias. 1963. Acta. Unio. Int. Contra. ancrum. 19: 751-755. Thorpe, J.J. Epidemiological survey of leukemia in persons potentially exposed to benzene. 1974. J. Occ. Med., 16: 375-382. Tough, X.M., Court Brown, W. Chromosome aberrations and exposur to ambient benzene. 1965. Lancet. 1: 684-689. Tough, I.M., et al. Chromosome studies in workers exposed to at mospheric, benzene. . 1970. Europ. J. Cancer. 6: 49-55. Usery, W.J., Jr. Letter to P. Bommarito, May 14, 1976. Vara, P, and Kinnunen, 0. 1946. Acta Obstet. Gynecol. Scand. 26: 433-452. Cited in Consumer Products Safety Commission. Vigliani, E.C. Leukemia associated with benzene exposure. 1976a, Ann. N.Y. Acad. Sci. 271: 143-151. Vigliani, E.C,, Occupaitonal Animal factors, Proc. XX Intern, Congr. Cancer., 1976b. Vigliani, E.C. Benzene and leukemia, 1976, Env. Res., 11: 122-127. Vogel, P. and Motulsky, A.G., Human Genetics, 1979, 320-322. Watanabe, G.I., Yoshida, S. The teratogenic effect of benz ne in pregnant mice, 1970, Acta Med. Biol., 17(4): 285-291. Wolmi, S.R. Cytologic and cytogenetic effects of benzene, 1977, in Laskln, 63-66. DO 0RR144 CONFIDFNT T At YZ0015981