Document K9w5xL8o8v1VErrrJE36DEwr

Benzene and Leukemia: The 0.1 ppm ACGIH Proposed Threshold Limit Value for Benzene Peter F. Infante Health Standards Program, Occupational Safety and Health Administration, U S . Department of Labor, Room N 3718,200 Constitution Avenue, N.W., Washington, D.C. 20210 The American Conference of Governmental Indwuial Hygienists (ACGIH) has proposed a threshold limit value (TLV) for benzene of 0.1 ppm. Individuals representing the American Petroleum Institute (API) and the Chemical Manufacturers Association (CMA) have argued that 1) the risk assessment by Rinsky er al. which ACGIH panidly relied upon for its proposed TLV overestimates the risk; however, at the exposures levels of interest (e.&. 0.1 to 1.0 ppm) for establishing a benzene TLV, the Rinsky et al. assessment provides lower estimates of leukemia risk than most others; 2) ACGIH should not use the Dow study for direct observational evidence of leukemia risk associated with low-level benzene exposure because of confounding exposure; however, it is unlikely that confounding exposures played a d e in the excess of leukemia demonstrated in the study. and &e Dow cohort was exposed to an average benzene concentration of about 5.5 ppm benzene for 7.0 years (38.5 ppm-years), while some of the individuals in the study who died from leukemia were exposed to an average of only 1.0 ppm without the opportunity for highpeak exposures; 3) the Occupational Safety and Health Administration (OSHA) established an 8-hour timeweighted average ( W A ) of 1.0 pprn in 1987, and there is no new evidence chat would justify reducing the TWA below that level; however, the OSHA T W A of 1.0 ppm was based on economic feasibility and the level of excess risk remaining at 1.0 ppm, i.e., 10 excess leukemia deaths per IO00 workers over an occupariond lifetime (45 years) according to OSHA's preferred estimate leaves behind a risk considered significant by OSHA. In addition, chromosomal studies among workers and in animals exposed to benzene indicate chat low-level exposure. i.e.. 1.0 ppm. is associated with elevated cytogenetic damage. On the basis of adverse health effects data alone. in this author's opinion, it would be poor science and poor public health policy to establish a benzene TLV greater than 0.1 pprn. Infante, P.F.: Bamene and Leukemia: The 0.1 ppm ACGIH Proposed Threshold Umit Value for Benzene. Appl. Occup. En- viron. Hyg. 7(4):253-262; 1992. Introduction In July 1930,the American Conference of Governmental Industrial Hygienists (ACGIH) published(') a "Notice of Intended Change-Benzene," whereby it proposed a re- vision of its current threshold limit value (TLV)for benzene from an atmospheric concentration of 10 ppm to 0.1 pprn as a timeweighted average ( W A ) with 3 Skin notation and the designation as an A1 carcinogen-confirmed human carcinogen. The ACGIH document is well writlen and is based on a sound scientific evaluation of the iicerature. The proposed TLV of 0.1 ppm benzene is based upon: 1) the results of quantitative risk assessments of leukemia, with special emphasis on the Rinsky et af.") assessmect using the National Institute for Occupational Safety and Health (NIOSH) case-control data; 2) direct inspection of observational data pertaining to benzene exposure levels associated with leukemia cases/deaths from he Dow Chemical C o m p a ~ i f l ~c.o~h)ort mortalicy study; 3) benzene exposure levels associated with chromosomal breakage in epidemiologic and toxicologic studies. On March 26,1991, presentations made before the ACGIH Chemical Substances TLV Committee by individual^(^-^) representing the American Petroleum. Insticure (MI)and the Chemical Manufacturers Association (Chi.4) recom- mended that ACGIH establish a TLV higher than 0.1 ppm based upon the following arguments: 1)the risk assessment b!: Rinsky et CZZ.(~) gives too high a risk when cor,. pared with the assessment done by Bren et a f . ( l ou)sing the same case-control data source and model, but with different benzene exposure assumptions than chose used by Rinsky etaf.;2) ACGIH should not use the Dow smd+3~+) for direct obsetvational data indicating low-level benzene exposure and leukemia because of speculation that benzene exposure levels may have been higher char, chose cited in the report and because of possible confounding exposures; and 3) the Occupational Safety and H a l t h Admini. ration (OSHA) established an 8-hour l X A of 1 ppm in 1987, and there is no new evidence that would justrfy reducing the 'N(A below that level. The parties mentioned above also submitted written opinions to ACGIH prior to the meeting in support of heir views. The MI and CMA presentations and wrinen commenu did nor address the toxicologic and epidemiologic - Theviews xpressed herein are those of the auchor and do nor necesssrih represent nose of the Occupauonal Safety and Health Adminisrrarion. .APPL OCCUP.ENVIRON. HYG. 7/4/ APRIL 1992 253 studies of low-level benzene exposure and chromosomal breakage. This article addresses each of the major commen& listed above in the broad categor~esof 1)quantitative risk of cancer associated with benzene exposure; 2) direct observation of low-level benzene exposure and leukemid Ivmphoma; and 3) health evidence justifies an exposure limit below 1 ppm benzene. In addition, a section on lowlevel benzene exposure and chromosomal breakage that was addressed by ACGIH, but not by API or CMA, is also addressed. Quantitative Risk of Cancer Associated with Benzene Exposure @urnem char the leukemia nsk based on rhe carec o m l st@v @ Rinsc(?y et al.(2)giies too high a nsk w h compared to the messment done by Brett et al.(lo)on the same data set. Rinsky et estimated the odds of death from leu- kemia in relation to cumulative benzene exposure based on a nested case-control study. They selected nine leu- kemia dmths from their cohort and matched them each with ten controls based on year of birth and year first employed in Pliofilm@.They selected a logistic regression model to estimate the odds ratio for leukemia in relation to cumulative benzene exposure. Their analysis indicates that 45 years of exposure to 1 ppm benzene (45 ppm- years) would result in an odds ratio (relative leukemia mortality rate) of 1.76. This odds ratio translates into a leukemia rate that is 76 percent greater than chat of the unexposed population and would be equivalent to 5.3 ex- tra leukemia deaths per lo00 workers exposed to 1 ppm benzene over an occupational lifetime (45 years). AS in- dicated in the ACGIH justification(') for its proposed re- vision to a TLV-TWA of 0.1 ppm, the odds ratio for oc- cupational exposure to 0.5 ppm benzene for 45 years is 1.3 according to Rinsky et af.(*)which translates into 2.3 extra leukemia deaths per lo00 workers-a level of excess risk considered signdicant by OSHA(11)and other regu- latory agencies. Based partly upon this analysis, ACGIH proposed a benzene TLV-NlA of 0.1 ppm, which Rinsky et al.C2)determined to be associated with an odds ratio of 1.05 that can be translated into an excess risk of 0.4 leu- kemia deaths per lo00 workers exposed over an occu- pational lifetime. Brett et al.(lo) conducted a similar type of analyses to that of Rinsky et al. using the same model and study pop ulation as the latter group. For each of the leukemia deaths in the Rinsky et af.study, Brett et af. selected slightly dif- ferent sew of controls. The preferred set of controls se- lected by Brett et al. were matched on date of birth, date of entering Pliofilm, and employment at the same plant. (Rinsky et al. did not match on employment at the same plant.) In addition, Bren et af. did not use the exposure assumptions of Rinsky et d.,but rather used the highest benzene exposure estimates as determined by Crump and Allen.(12)The preferred estimate of risk by Brett et ul.(loi suggests h a t occupational exposure to 1ppm benzene for 45 Years would result in 0.5 e!ma leukemia deaths per lo00 workers. Exposure to 0.5 ppm benzene Over +jvearS would result in 0.3 extra leukemia deaths per 1000 work. Thus, in relation to benzene exposures of inreresc (0.1 to 1.0 ppm) in establishing an ACGIHTLV, one could argue, as did API and C A Nthat the estimates provided bv Rinsky et af.(*)are too high. [Their estimates of risk (10) at the lower end of the range of benzene exposures of interest, however, are Slightly lower than those provided by most other risk assessments.] Conversely, one could argue that the estimates of leukemia risk by Brett et 10) are too low, particularly in light of the results of &e other risk assessments on benzene and leukemia as discussed below. Wbyare there d@wmces between the &messmen& by R i m et alJ2) and @ Bren et al.('O) when the authors ured the same database? The major difference in leukemia risk asessment between the results of the Rinsky et aLc2)and those of Bren et uf.(lo)stems from a difference in exposure estimation for periods when no benzene exposure data are available. Which set of assumptions is closer to the "truth"will never be known. Rinsky et af. made reasonable exposure assumptions. The Crump and Allen(12)exposure assessment used by Brett et &.(lo) is also reasonable. Whichever estimate one chooses to use is a matter of personal preference chat is not based upon any available data analysis. Funhermore, regardless of the number of additional estimates ofbenzene exposure to the Rinskyet af. cohort 2, thatmay be developed in the future, it is unlikely that they will be any more precise than those already available. For example, Paustenbachc6) made a preseneation during the ACGIH TLV benzene deliberations whereby he claims to have made "recent advances in undersmding the exposures of workers in the Pliofilm cohort,"ThLSefforr included: *Interviews with three former Pliofilm workers from one of the facilities studied that was located in St. M a r y , Ohio. [Note:In 1976, a NIOSH team of investigators interviewed employees and members of rnanagernenc and reviewed the Pliofilm process while it was still in operation.] Review of the Kipen et af.t13) study discussed below. Review of a 1942 labor conference transcript that dis- cussed exposure situations and acute toxicity among rubber workers, but not necessarily those who were studied by Rinsky e?d.(2) All these aspens of exposure were evaluated ;IS pan of the OSHA hearing. Notfung presented by Paustenbach(6' is new in terms ofa& data or information. The only new aspect is the different assumptions made in his report about exposure for timeswhen no acrid exposure dam are amiable. me p o t e n d effecton the risk estimates ofunaccounred -ne exposure was discussed t h o r ~ u g h3ls~pan of h e OSHAh-ng a d is discussed in detail in the final OSHA benzene smdard(11) The analyses by Brett et af.(")were presented at the OSHA benzene hearing. OSHA concluded(11) "that there was some additional e.uposure. but 254 APPL OCCUP. ENWRON. N6. 7/41 . APRIL 1992 I the data indicate that it did not make any subsrantial dif- ference in the risk estimates . . . internal analysis of the data show that any extra dose received by the cohort was not large enough to make a difference in the estimates." For example. in the prospective malysis of the Rinsky et ai. cohort performed by Crump and Allen,' additional, unreported, atmospheric benzene exposures in the Plio- film operations and unaccounted for atmospheric benzene exposure in non-Pliofilm jobs or from skin absorption while workers were employed either in the Pliofilm op- eration or on jobs outside of Pliofilm that occurred more or less uniformly across the study cohort would manifest themselves in estimates of the intercept from the risk anal- ysis xs being larger than 1.0 and/or in an excess (relative to the control population) of leukemias in the lower ex- posure groups. In fact. neither of these conditions oc- curred. Estimates of the intercept in the five analyses by Crump and that used only the Rinsky et uf.data were 0.87, 0.82. 0.88, 0.18, and 1.11. All but one of these estimates of intercept is less than 1.0, and the exception is from the analysis that was considered the least reliable by Brett et af.,(I0i.)e., the "window dose analysis." Thus, the knsky et ai. data provide evidence against uniform exposures from other sources at a magnitude to affect leukemia rates in the study cohort.(l*) X paper published in 1989 by Kipen et uf.(l3) purports to show the benefits of using peripheral blood counts for the reconstruction of previous benzene exposure levels experienced by the Rinsky et uf.(2c)ohort in the 1940s. The authors concluded that there was a significant inverse relationship between benzene exposure (as determined by their preferred estimate of exposure, i.e., the Crump and Allen maximum exposure estirnate,'l2) but not by the IZlnskv et uf.estimate) and white blood cell (WBC)count among benzene-e.uposed workers in the NIOSH study. Based upon this analysis, Kipen et uf.concluded that their pre- ferred estimate of benzene exposure to the NIOSH cohort (as performed by Crump and Allen(12))was better than that done by NIOSH. If such an analysis was correa, it would support use of the exposure assumption preferred by Brett et uf.tlO)for their logistic regression analysis of the Rinsky et uf.case- control study.(2)Although the Kipen et ul.analy~is(~r3ep) - resents a creative approach, there is potential for tremen- dous selection bias in the study as the investigators based their analyses upon 128 (11%)of about I200rubber work- ers included in the cohort study. It is not known whether this small portion of workers' blood analyses represents the total population that was studied. However, it is likely that the results are not representative because the average W C count for the group studied was 9300Imm3. The nor- mal range of WBC count for the general population is 5000/mm3 to lO,OOO/mrn3, with an average of about 7500/mm3.Thus, it is perplexing that a group of individuals exposed to such a potent bone-marrow depressant as ben- zene would have an average count so much higher than normal. In the report, Kipen et af. stated,' l3) " . . . we do not have an explanation for this aberration in the data set." A possible e.uplanation is that workers whose blood counts were on the low side of normal were removed from the job II their count did not improve. Hence, workers with low counts were not available for succeeding blood tests.Thoce who remained on the job in the mid- to lare-I9+0s,a period when most of the secular increase took place, and had at least five WBC counts taken (a criterion for selection into the Kipen et d .smdy(I3))were those who had blood countS that were either normal or on the high side of normal. Thus, the Kipen et uf.(l3) findings may be the result of selection bias and they cannot be used to supporr any particular exposure estimate. In response to the Kipen et uf. paper,'13) Hornung et d c 1 4 ) of NIOSH presented analyses of W B C counts among the NIOSH cohon. They concluded that the temporal in- creases observed among the NIOSH cohort cannot be attributed to a reduction in benzene exposure because preemployment blood counts showed the same remporil increase in WBCs. In the opinion of Hornung er af.,( l.0 the temporal increase was more likely due to changes in laboratory practice during the 1940s. Should the Rim@ et al. or the Brett et al. anafysis be relied upon by ACGIH to determine the nsk of leu- kemia relaredto benrae -&- rhe nv3 During the OSHA benzene hearing, the Rinsky et u f . ( 2 ) study was characterized by witnesses representing both industry and labor as one of the most thoroughly studied occupational cohorts in existence.('') Based upon all of the evidence submitted to the OSHA record, OSHA was also of the opinion that the study was carefully conducted. OSHA concluded that both the Rinsky et ai. risk assess- menb2) and the Brett et uf.risk using the NIOSH case-control data and a logistic regression model, were well conducted. O S W however, did not rely upon either of these anal- yses for its best estimate of leukemia risk for several rea- sons.(ll) First, the model was not developed for cancer dose-response analysis. The assumption of a log-linear rehionship between dose of benzene and relative odds of developing leukema will cause any variation in the es- timate of dose to result in an exponential change in the associated risk This type of model seemed to be biolog- ically less plausible than a linear model which also fit the case-control data set. Second, the use of a model resulting in an exponential change in risk with a change in dose introduces more uncertainty. This is of panicular concern in light of the argument about different estimates of ex- posure by API, CMA, and Rinsky et uf.for members of the Rinsky et uf.studv.(ll)Third, the studies by Rinsky et af.' ? ) and Brett et uf.(l0) were based only on one of the three available studies that had good information on benzene dose. Thus,it seemed preferable to base estimates of risk using all three of the high-quality epidemiologic studies chat contained data in a forma chat allowed for reasonable estimates of benzene d0~.(2-*15) ACGIH should refy upon the Cru& and Ailen as- .APPL OCCUP W W R O N . HYG. 7/41 APRIL 1992 255 sement to estimate leukemia riskfiom benzene epsure. OSHA relied upon the risk assessment by Crump and Allen,' which used the Crump and Allen maximum exposure assessment, For its preferred estimate of risk for the following reasons. Crump and Allen combined data From all three of the highquality epidemiologic studies with good benzene exposure dam;' 2-+.15) thus, the best available data were used to determine excess risk. They demonstrated that the data from all three studies fit a linear model well. Furthermore. the linear model projected excess leukemia risks that were more mid-point estimates as compared to the logistic regression model when considering both low and high ranges of cumulative dose. There also appeared to be a biological basis For a linear model because benzene had shown a linear relationship for chromosomal breakage and cancer in anima1s.(l1) As shown in Table I, the Crump and Allen risk assessprojects a risk of 10 excess leukemia deaths per 1000 workers as a result of 1ppm benzene exposure for 45 years. The risk at 0.1 ppm For 45 years would be 1 per 1000 (ten times lower) since the relationship is linear at low doses. These findings are similar to those reported by others who used a linear model to estimate leukemia risk associated with benzene exposure.' l2*l6-l9) TABLE 1. Estimated L e u k ~ ~ iDOWh loa0 worksrs Exposed to Bwmfor 4-45 Y m (1 ppm or 10 ppm) Using Relathre Rlskand CwnulaUva Exposun M O W Study 40 ppm-yrs (95%C.l.3 400 ppm-yrs (95%C.I.) Rtnsk-p wong('51 Rinsky.lZ) W ~ n g . " ~ ) and O W 6.6 (2.1-15) 13 (01-31) 10 (4-22)' 63 (21-129) 121 (1-243) 95 (37-186)' 'Source 29 CFR Pan 1910'"1 FI. = confidence mvai Clndicates 45 pwn-yrs md 450 ppm-yn. resgectivdy. The only risk assessment, besides Crump and Allen,' which used all three of the highquality studies with good dose data available to estimate risk is that of Aukin et al.t1') This study estimated an excess of 53 leukemia deaths per 1000 workers as a result of 300 ppm-years of benzene exposure. Since these authors used a cumulative dose concept and a linear model, as has the majority of the risk assessments For benzene{12*16-l9) the risk associated with 0.1 ppm e-xposure For 45 years would be 0.8 excess leukemia deaths per lo00 workers. This estimate is virtually the same as the 1 extra leukemia death per loo0 workers estimated by Crump and Allen.(12) Thus, the quantitative risk assessment that OSHA relied upon for its best estimate of risk,(*2)or that by Austin ef provides estimates of risk associated with 0.1 ppm benzene exposure for 45 years that OSHA considers s i g mficant, Le., 1 per 1.OOO.For the reasons a& above,ACGIH should use the Crump and Allen(12)risk assessment based upon the relative model For its preferred estimate ofdose- response For benzene and leukemia. OSHAdid not set ;t permissible e . p s u r e limit ( PEL) below 1.0 ppm because it was noc economically Feasible to do SO. Somefactors thar may mult in an Lcnderesrimatimateof clztzcer risk asoctiated with benzene e.wsure Lcbicj, are not accountedfor in any of the nsk mmenrs, 1. Benzene-related diseases other than leukemia have not beencounted in m a t quantitative risk Smq-nenE. The risk of disease from benzene exposure based on leukemia only serves to underestimate risk. However, dis- eases other than leukemia have not been included mob[ risk assessments of benzene-exposed workers. Benzene exposure may cause other, "nonmalignant" blood dise.aes and lymph0ma.(2-iJ5~20-2B1)y considering just the ratio of leukemia to multiple myeloma(23:9) in the four major cohort studies of benzene-exposed workers where such data were p r ~ v i d e d , ' ~ . ~oJn~e. c~o~ul)d add an additional 39 percent of excess cancer deaths to the quantitatiir risk. 2. The use of a cumulative dose concept to estimate benzene exposure in relation to bone marrow COXicity, chromosomal aberrations, and cancer. All of the risk assessmenrs for benzene and leukemia have used a cumulative dose concept For estimating risk. Given the type of exposure data that are available, it would be difficult, if not impossible, to develop what may be considered a more meaningful dose concept for benzene, based upon toxicologic study results. Several toxicologic studies, as indicated in Table 11, demonstrate that the mode of exposure to benzene has a significant effect on benzene's toxic response. Thus, the risk of disease From benzene exposure to a group of individuals might be underestimated, depending upon the manner in which dose is received on a particular job in relation to the manner in which the dose was received in the epidemiologic study used to estimate quantitative risk. For e?cample, 1ronsc23)has shown that the administration of an important benzene metabolite, hydroquinone. can c a w bone marrow depression as a result of intermittent expasure to 45 percent of a dose from which the animals' bone marrow was refractory.Likewise,the study by Dempster et d . ( 2 Q demonstrates that anemia results from 33 percent of a dose chat did not cause a reduction in red blood cells and depends upon the mode of exposure. Luke et have demonstrated a greater suppression of polychromatic erythrocytes (PCEs), a reflection of recently induced bonemarrow damage, as a result of a lesser benzene dose given intermittently. Cancer response in experimental animals also has been relareed to mode of exposure. Snyder et al.(z6)have demonstrated that 64 percent of a benzene dose given intermittently results in the same cancer response as the full dose given on a more continuous basis. These smdy results again demonstrate that the mode of exposure to benzene is related to the severity of response for seven1 toxic endpoints including cancer. In these studies, less cumuhtive 256 APPL OCCUP. ENVIRON. HYG. 7/41 . APRIL '992 11. Effect of ode of Becueno Exposwe on Type and Severity of Toxic Rssponse lrons:(a' Benzene metahlite (hydroquinone)results in bonemrrow depression as a result of 45% of a dose from whid animals bone marrow was refractory Dempster ef dl 44 Anemia results from 33% of a dose that did not cause red blood cell reduction. 10 days x 100 ppm (loo0 ppm-days) = anemia 3 days x 300 ppm (900 pgm-days) = anemia 3 days x 1000 ppm (3000 ppm-days) = refractory 1 day x 3000 ppm (3000ppm-days) = refractory Luke et a/ 4251 A greater suppression of PCP as a result of a lesser benzene dose given intermittently Exposed DBA mice for 13 weeks to 300 pprn benzene Regimen 1 = 5 dayvweek x 6 hourslday Regimen 2 = 3 daywe& x 6 hourYday Signilicantly > suppression of PCEs with Regimen 2 Snyder el a/ Cancer in animals: CD-1 mice. 1195 ppm x 50 days Come = 59.750ppm-days 298 ppm x 129 days Int.c = 38.442 ppm-days 64% of dose given Int., but at lower levels. results in Same tumor response as the full dose given at higher levels, but continuously. C57B1 mice 1 195 pprn x 50 days Cont. = 59,750 ppm-days 300 ppm x 181 days Int = 54,300ppm-days 90% of a dose given 'lnt" results in a significant increase in tumors. whereas larger dose given Cont. does not The studies indicated abm dsmwtrate that a cumulltfve dosa model may undacatimale risk 01 bone marrow toxicity or cancer WE = polymmmat~~erymrocyIes 'coni i. dosa administared 5 @stweak lor 10 wmks Clnt = dose administaed 5 O a v s A d ~lollovrad bv 2 w k s no dosk benzene dose caused more disease and the same cumulative dose caused different cancer response rates depending upon the intermittency of the exposure. With regard to epidemiologic data, the analyses by Wong"5) might lead one to conclude that only cumulative dose is important in determining a relation berween benzene exposure and lymphopoietic cancer. His analyses demonstrate a significant dose-response for lvmphopoieuc cancer and benzene when the data were analyzed by cumulative benzene dose, but not when the duration of exposure or peak exposure were used as surrogates of dose. However, duration of exposure can only be a meaningful measure of dose when all cohort members had the same level of exposure during their entire emplovment period. Since benzene exposure levels have generally declined over calendar time periods, duration of exposure cannot, therefore, be a meaningful measure of benzene dose. Although the benzene peak exposure analysis did not demonstrate a signficant dose-response in che Wong study, the relative risk was 3.38 for lymphopoietic cancer at the lowest-peakexposure level (< 25 ppm) in the report.(27) Hence, the relative risk was already elevated at the lowest maximum peak evaluated in the study. The inability of Wong to demonstrate a dose-response by maximum-peakexposure level may be a reflection of the peak levels he arbitrarily chose to evaluate for dose-response. For ex- ample, if it had been possible to have chosen peak levels of C 5 ppm, < 10ppm, and < 25 ppm. etc..Wong may have been able to demonstrate a dose-response by pertk a- posure level. However. his data set was too small t o conduct such a retined analvsis. Direct Observation of Low-Level Benzene Exposure and Leukemia/Lymphoma The Dow st@v 4 OnlBond provides direct et*iife)ice ofleukemia asa result ofLou-level beurene espostrre. and nmnnaliglMtll blood dkases as a result oj*slight!)* higher Lwek of exposure. As mentioned in the hCG[H benzene documentation.' 1 the risk of low-level benzene exposure does not need to be estimated from formal quantitative cancer risk assessment. The D0d3.4) study provides direct evidence thdt lowlevel benzene exposure is associated with an incresseci risk of leukemia. As shown in Table 111. the average benzene exposure received by the cohort was 5.5 ppm for -.O years, or 38.j ppm-vears cumulative dose. Thus. a 0.1 ppm exposure limit for 45 years (4.5 ppm-years) would provide slightly less than a tenfold protection factor The average exposure for the five leukemia cases, G ppm for 12 )ears or 72 ppm-years of exposure, was higher than that for the entire cohort. As shown in Table 111, leukemia cases #? and #j were only exposed to an average of 1 ppm. while cases #l and #3 were only exposed to an average of 5 ppm. Case #4 was exposed to an average of 18 ppm. Written comments provided to the Chemical Substances TLV Committee by Dr.Ekmd@)suggest that the Dow cohort<3-*)may have been exposed to benzene levels higher than those provided in the published reports of the study. However, the authors of h e Dow cohort have previously published the best available data on benzene exposure to the cohort.As appropriately stated by Bond, exposure prior to 1%0 "may have been considerably higher than we 35sumed for purposes of our investigation. . . . we have no -data that would validate or invalidate this assumption."'*' TABLE 111. Myebgmous Leukemia Deaths Among Workers Exposed to Benzene by Exporrun Level, Y e a n Exposed, Cumulative Dose, and cas0 No. 1 2 3' 4 5 Avaraoe (Cases) Enbre Cohort Aveaage Benzene EI(p0IRICII Ippm) 50 10 50 180 12 60 55 Years Exposed 10 7 15 50 19 5 23 3 12 0 70 Cumulabve Benzene Dose (PPm-Yrs) 54 0 15 25 4 351 0 20 0 72 0 30 5 Latency clears) 11 15 15 37 39 0trrsvad~ogarouslaulranudssmr= 4 ~ ~ o p a r o w l a u l w w d a m= s09 SMA = 444 'U-ing CMS d dsah clarjclied as pneumonia AML lisled unaef other signillant conditions SolHca oa u alm and Bond 8tal'4 APPL OCCUP. ENVIRON. HYG. 7(4) * APRIL 1992 257 nus,data that would invalidate the estimates of benzene exposure in the published reports do not exist. The question of whether low-level exposure to benzene could cause leukemia was also a concern of the GMAduring the OSHA rulemalung on benzene. The Association asked Or. Brian Mac,,lahon to evaluate the literature.Dr. iMacMahon concluded that it is more probable than not that benzene exposures as low as 10 ppm and below increase the risk of leukemia in ;i meaningful He was m a t im- pressed with the Dow study as showing increased leukemia risk as ;i result of exposures below 10 ppm in rendering his opinion. His report was submitted by CMA to the OSHA benzene docket and considered in the Agency's delib- erations. In Dr. Bonds statement to the ACGIH(8' and in the orig- inal On et ai. repo1-63) of the Dow cohort study, it is men. tioned that one of the individuals who developed leukemia as a result of only 1.0 ppm benzene exposure for 1.5years also worked between 1948 and 1950 in a sawmill that manufactured veneer. Bond@)and On et a1.(3)cite Mil- ham(29'as a source indicating that this occupational ex- posure is associated with an increased risk of myelocytic leukemia. The implication is that this leukemia death as observed among the Dow workers exposed to benzene may have been caused by the individual's working in the sawmill. The supportive documentation for this statement is simply incorrect. Data from the cited Milham r e p ~ dfo~r ~the) category of exposure titled "sawmill and other mill workers" do not indicate an excess of mortality from leukemia. The only cancers shown to be in excess according to Milham are cancers of the pancreas and the testis. In addition, the International Agency for Research on Cancer (IARC) re- viewed epidemiologic studies of workers in "The Lumber and Sawmill Industries (Including Logging)." IARC con- ~ l u d e d ( 3th~a)t "The epidemiological data are not sufficient to make a definite assessment of the carcinogenic risks of employment in the lumber and sawmill industries." The only cancer risks of possible concern mentioned by IARC(3O) in association with this industry were nasal cancer, soft tissue sarcoma. and histiocytic lymphoma. Thus, the im- plication that one of the leukemia deaths associated with low-level benzene exposure. as observed in the Dow study, Data from the Bond et d.")study also indicate a sig- nificantly elevated risk of death (3 observed versus 0.7 expected, p < 0.05) from nonmdignant bl& ~~~ diseases, The individuals who died from these diseases e;uperienced slightly higher levels of benzene exposure than those who died from leukemia. One death was observed from aplastic anemia (`IWA = 4.6 ppm benzene for 8.2 years of expo- sure); one death was observed from "pernicious anemia;' which is megaloblastic anemia` ) according to tissue etval- uation (TWA = 30.1 ppm benzene for.15.3 years of ex- posure); and one death was observed from myelofibrosis ('IWA = 19.3 ppm benzene for 26 years of exposure). Lmek of benzene eqmsw amchted with I a r k e m h and b`loma in recent occupational cohort mor. tal@studies Fifteen or more years ago, many Of the cases of leukemia reported in the literature were associated with relarlvelv high levels of benzene exposure. For example, the series of leukemia cases reported by Viglianit3l) indicate that workers were exposed to average benzene levels of about 200-500 ppm, while the leukemia cases reported in the earlier days by Aks0$32) indicate chat workers experienced benzene levels in the range of about 150-210 ppm.(33) These relatively "old" case repom have lead somt sci- entists to conclude that only high levels of benzene exposure can cause leukemia These reports, however, have no bearing on direct observations indicating an association between leukemia and low-level benzene e-xposure. As benzene exposures in the workplace have been tre- mendously reduced over the past 50 years, individuais exposed to low-level benzene, in the range of only a few parts per million and less, are now known to have devel- oped leukemia and lymphoma Table IVshows the average benzene exposure levels associated with leukemia cases as derived from the epidemiologic cohort studm of benzene-exposed workers where such information was rep0rted(~-~J5~3A4s) indicated, 34 percent of the leukemia deaths were observed among workers whose averagt exposures were characterized as being below 6 pprn; 48 percent were reported to have been exposed below 16 Level Studles 0.1-5 6-15 16-30 31-60 60+ All levels 7 4 7 6 6 30 1 1 3 4 0 9 `Batsd on 6casesWhan dataavaiW 4 1 0 0 0 5 ParCent of 5 17 34 1 7 14 0 10 20 0 10 20 0 6 12 6 50 APPL OCCUP. WVIRON. HYG. 7/41 . APRIL @pm) 0-5 6-15 16-30 31-60 60+ All levels Ylnor 7 4 7 6 6 30 Rinskyow 1 4 4 4 0 13 Bone 5 1 2 0 0 0 wong'= 10 2 3 0 0 15 Percent of Total Total 23 3_5_ 11 17 16 24 10 15 69 66 -PPm. The experimental studies cited above, the results of which Table V presents the same type of analysis for leukemia, are shown in Table 11, support the conclusion that inter- other "nonmalignant blood diseases,"and lymphoma were mittent exposure to a lesser total amount of benzene can reported by exposure level. The distribution of diseases cause relatively more bone marrow toxi~iy,'~3.?c-h+r)n- is about the same; 35 percent were exposed to benzene mosomal aberrations/25)and cancer.'26)For example. the concentrations below 6 ppm, and 52 percent were exposed study by Luke et al.(25)indicates that 3 days of benzene to average levels below 16 ppm. exposure followed by 4 days of nonexposure caused more In its justification for a 0.1 ppm TLV,ACGIH cited the suppression of polychromatic erythrocytes than j days of Dow study for direct-observational evidence of benzene's exposure to the same daily dose followed by 2 days of ability to cause leukemia as a result of low-level exposure. nonexposure. Likewise, the study by Snyder et dem- The Chemical Substances TLV Commiaee,Cl)however, failed onstrates a greater cancer response in CjTBl mice as a to include similar evidence from the Wong study.(15)In result of 181 days of interrupted exposure at 300 ppm as the Wong study,(l5)where dara were provided for workers compared with 51 days of more continuous exposure at who were continuously exposed, 10 of 15 deaths from about 1200 ppm benzene (Table 11). lymphopoietic cancer were exposed to average benzene While few epidemiologic data are available to evaluate concentrations of 5 ppm or less. Data in Table 16 of the the influence of intermittent and peak benzene exposure Wong repod'5) indicate that three of the individuals who on subsequent disease, the study by Wong'27-'shows that died from lymphopoietic cancer were only exposed to peak exposures above 100 ppm are .not associated wirh average benzene concentrations of 0.5 ppm. any greater odds of developing leukemia than peak ex- The data presented above collectively indicate that low- posures below 25 ppm. In the more detailed version of level exposure to benzene carries with it a risk of devel- his paper submitted to OSHA,"-)the data in Table 41 oping leukemia, lymphoma, and nonmalignant blood dis- indicate that the odds ratio ( 0 . R ) for lymphopoietic cancer eases,although the latter group of diseaseswere associated for those who experienced peaks greater than 100 ppm with exposures at rhe high end of the low-exposure range. was 3.01,while the 0.Rfor those who experienced ben- Periodic peak exposure and mtenninent lower level egosure as related to the akelopaent of leukemia and otber diseases in war&mw d to benzene. zene peak exposure levels below 25 ppm was 3.38 Likewise, the study of refinery workers by Devine and Barr0n(3~)indicates no excess risk to the cohort overall: however, those who experienced jobs where intermittent Repons that have indicated elevated risk of leuke- exposure would occur (utility workers and pipefictersi tnia(*-+J53) or excessive chromosomal breakagd35) as a demonstrated signficantly elevated O.Rs (4.6 and 2.7, re- result of low-level benzene exposure have sometimes been spectively) for leukemia Data on the level of peak benzene followed by statements, without documentation, that the exposures that may have been associated with jcbs indi- elevated risks observed in the studies were the result of cating an elevated leukemia risk in the latter study, how- unaccounted for high-peak benzene exposure l e ~ e l s . ( ~ * * ~e)ver, were not reponed.In the latter study/3') it is possible These comments have been speculative from the stand- that these peak exposures were high, and as a result, those point of exposure level. They also do not take into account who received intermittent exposure may also have re- the influence of mode of benzene exposure as it relates ceived the relatively greatest cumulative dose of benzene. to hematopoietic diseases. Data now suggest that inter- Therefore, the Devine and Barron(3') study results raise mittency of exposure, whether the exposure be high or the possibility that intermittent exposure plaved a role in relatively low, may be a major factor in the induction of the elevated risk of leukemia observed; however, it is not leukemia and other diseases associated with benzene possible to separate the effect of intermittent exposure exposure. from the effectof total cumulative dose. APPL OCCUP. ENVIRON. HYG. 7/r! * APRIL l#n 259 &-jdi[iondl observations, however, indicate an association between intermittent, low-level, peak exposure to benzene and lymphopoietic cancer. The short period of low-level benzene exposure for some of the individuals in b e studies who developed leukemia"-+.li.3+' make it less likely that their exposure estimates would be in error. For example. case #2 in the O W n d study, as shown in Table 111. was only exposed for 1.5 years to an average of 1 pprn. This individual worked in an area of the plant categorized as "potential exposure to ven low concentra- tions of benzene < 2 ppm W A dosage: 18ppm-months )." With regard to the potential for intermittent, low-level exposure to benzene in the Wong study.tl5) his Table 16 indicates that case #8 (multiple myeloma) and case #11 (chronic myelogenous leukemia)were exposed to avenge benzene concentrations of 0.5 ppm for 2.3 years and 1.2 years, respectively. Their maximum peak exposures to benzene were categorized as below 25 ppm. Since some of the individuals in the cohort studies who died from leukemia and lymphoma did not experience high peak benzene exposures, they must have experienced low exposures. These exposures would be received on an intermittent basis as is usually the case in the occupational setting. Workers are not exposed continuously to the "average" level over an &hour shift. Also, their daily exposure fluctuates from one day to the ne=. Thus, the fourfold risk of myelogenous leukemia associated with low-level exposure in the OttlBond study43.*) and the low benzene exposure levels associafed with a number of the leukemia/lymphoma deaths in the hnsky et al.,c')Wong,t'5' and Yin et aL(3.i)studies may perhaps be enhanced by the intermittent nature of exposures as they occur on the job. Since benzene exposures have been reduced in the workplace over the past four decades, intermittent, low-level exposure to benzene may present a greater absolute risk of benzene-related diseases than brief, periodic, high-level exposure. Low-Level Benzene Exposure and Chromosomal Aberrations AS part of its justification for its 0.1 p p m ' n v , ACGIH cited studies in humans and experimental animals related to chromosomal damage as a result of low-level benzene exposure. Commentors representing MI and CAMAdid not address this concern during their presentations before the Chemical Substances TLV Committee. Nevertheless, chromosomal damage has been observed in workers exposed to low average levels of benzene. The data in Figure 1, taken from Picciano,(35)demonstrate an elevated frequency of chromosomal breakage at exposure levels down to 1 ppm. When the study results were reported to the U.S. Environmental Protection Agency (EPA),(x) it was stated that industrial hygiene data indicated that benzene peak exposures exceeded 100 ppm and that this pealung in exposure was presumably responsible for the observed clastogenic effects among the benzene-exposed workers. However. data later supplied by the company indicated that high-level benzene exposures to most of these work. ers could not be subsrantiated and could not have 3ccounted for the elevated frequency of chromosomal bre&age seen in the studV.(3*) While there can always be argument that unaccounted for exposures were experienced by the workers in the Picciano study'35' or any other study that shows adverse effectsfrom low-level benzene exposure, this mamer mav be resolved in the laboratory where exposure can be controlled. In this regard, experimental study has slso demonstrated that low-level benzene exposure can cause chrom w d aberrations. The Chemical Industry Institute of Toxicologystudy by Ere.xson et ~ f . , 'zs~ c~it'ed in the xCGIH benzene documentation, demonstrates a significant increase in chromosomal breakage as a result of exposure to 1 ppm benzene. For a substance known to cause bone marrow toxiciry and lymphohernatopoietic cancer, the chromosomal studies alone showing an increase in breakage at 1 ppm should lead one to consider setting an exposure limit based on health below 1 ppm. Hoalth Evidence Justifies an Exposure Limit Below 1 ppm In commenrs on the ACGM Notice of Intended ChangeBenzene received from W ( 9 ) the issue is raised about ACGIH using a "no increased risk' standard in recommending a 0.1 ppm 'Iz;V for benzene. This argument is moot in relation to occupational benzene exposure because the estimated excess risk associated with the proposed 0.1 ppm TLV is about 1 per 10oO workers for leukemia only. The risk would be higher if other benzeneassociated diseases were included in the risk assessments. Argument is raised that OSHA set a PEL of 1 pprn in 1987 and that there is no new evidence that would jut& reducing the TLV to a lower level. OSHA, however, set a PEL of 1.0 ppm because it was not economically feasible "I30 >2.s BIWS- -0LUI IM 991 FIGURE 1. Comparison of some breaks and marken the distribution as a function of obfeinnzdeinveiduexaplsowsumre.bSoothurCceh:roDm.oJ-. Picciano.'m to set 3 lower limit. The OSHA however, in- cludes other provisions to help lower the risk These in- clude: 1) an action level of 0.3 pprn as an incentive for manufacturers to go below this level in order to save on costs ofcompliance with the additional ancillary provisions of the standard: 2)medical surveillance to help identify some of the i n d i v i d d more susceptible to benzene-relaced blood diseases so they can be removed from exposure; 3 1exposure monitoring; and 4 ) training and education about the hazards of benzene and appropriate work practices to use when it is present xjpart of the manufacturing process. Recommendations ACGIH has done a rigorous job in its documentation of 3 0.1 ppm TLV-NCA for benzene. It should. however, consider the Crump and Allen risk assessment for estimation of excess leukemia in its final evaluation since these authors used all of the available data from the three epidemiologic cohort studies that included data on benzene dose. ACGIH should acknowledge that risk based upon leukemia underestimates total disease risk from benzene exposure because it does not include lymphoma, aplastic anemia, and other cytopenias. Likewise, the final documentation should include a discussion of factors that may be related to potential underestimation of risk, such as use of a cumulative dose concept, in the quantitative risk assessments for leukemia. Finallv,in this author's opinion, it would be poor science and poor public health policy to establish a TLV greater than 0.1 ppm based on the health data currently available. Therefore. the ACGIH should not establish a TLV for benzene above 0.1 ppm. References 1. Xotice of Intended Chmges-Benzene. Appl. Occup. Environ. Hvg. i('):453-+63 ( 1990). 1. Rinsk.Li:Smith. AB.; Hornung. R; et al.: Benzene and Leukemia. .+n Epidemiologic Risk hssessmenr. N. Engl. J. Med. 316(17):10441050 ( 198'). 3. On. G.M.: Towmend, J.C.; Fishbeck. W A ; Langer. RA: Monaliry among Individuals Occuparionallv Exposed to Benzene. Arch. Environ. Health 33( I ):3-10 ( 19'8). 4. Bond, G.G.: Mdaren, E A ; Baldwin, C.L; Cook. RR: An Update of &tonality znong Workers Exposed to Benzene. Br. J. Ind. Med. +3(10):685-691 (1986). 5 . DiPerna. C.: Overview of the APPVWSPA Presenrarion to the Chemical Substances TLV Committee of the ACGIH. Pmented at the ACGIH Chemical Substances TLV Committee Meeting on Benzene, iMarch 26, 1991. Cincinnati. OH. 6. Piustenbach. D.: Review of Recent Advances in Understanding the Exposures of Workers in the Plilofilm Cohort. Presented at the ACGIH Chemical Substances TLV Cornminee Meeting o n Benzene. March 26, 1 9 9 1 . Cincinnati. OH. -. Rodricks. .I.\'U.p:date of Leukemia Risk Potentially Associated with Occupational Exposure to Benzene. Presented at the ACGIH Chemical Substances TLV Cornminee Meeting on Benzene. March 26, 1991. Cincinnati. OH. 8. Bond. C.G.. Testimony before the American Conference of Governmental Industrial Hvgienists Regarding a Recommended Change in the Benzene TLV. March 26. 1991. Cincinnati. OH. 9. Cox, C.V , Stack. C.R: Comments of the Chemical Manufacturers AS- sociuion Benzene Panel on the Notice of Intended Chanae in Chc ThrrsholdLimit Value for&nene. Presented 3t the ACGIH Chemical Sukwvlca TL.V Committee Meeting on Benzene. March 16. 1991 Cincinnati. OH. 10. B r a . S.M.; Rodricks.J.V.: Chinchilli. V X : Review and L'phte ot it.<h i a Risk Potentially .Associated with Occupational Exposure to Ben- m e . Environ. Healrh Perspect. 82267-281 I1989). 11. U.S. Department of Labor. Occupational Safety and Health Administration: Title 29. Code of Federal Regulations. Pan 1910.Occupational Exposure to Benzene: Final Rule. Part 11. Fed. Reg. 52(1?6):3+tM- 34578 (September 11.1987). 12. Crump. KS.; Allen, B.C.: Quantitative Estimates of Risk of Leukemia from Occupational Exposure to Benzene. OSHh Docker H-0599.EUh. 152 (1984). 13. Kipen. H.M.;Codv. R.P.; Goldstein. B . 0 : Uses of Longitudinal .Qal!ws of Peripheral Blood Counts to Validxe Historicil Reconstmaions of Benzene Exposure. Environ. Halth Perspect. 82.199-206 1989) 14. Homung, RW.: Ward. E.: Morris.JA:Rinsky. R-i. Letter to the Editor. Toxicol. Ind Health 5(6k1153-1155 (1989). 15. Wong. 0.:An Industl). Wide Monality Study of Chemical Workers Occupationallv Lxposed to Benzene. 11. Dose-Response ..\nal!~is. Br. J. Ind Med. 446):382-395 (1987). 16. White, M.C.; Infante. P.F.: Chu. KC.. A Quantiutive Estimate of Leu- kemia Modir). Associated with Occupational Exposure to Benzene. Risk Anal. 2:195-204 (1982). 17. Austin, H.; Delzell. E.: Cole. P : Benzene 3nd Leukemia. A Review oi the Literature and a Risk Assessment. Am. J. Epidemiol. 12'13):+19- 439 (1988). 18. International Agency for Research on Cancer: MRC Monographs on the Evaluation of Carcinogenic Risk of Chemicals to Humans, VoI19. Some Industrial Dyestuffs. pp. 395-398. IARC. Lvon. France ( 1982). 19. U.S. Environmental Protection Agency. Office of Health and Environ- m e n d ksusrnent. Carcinogen ilSsessment Group: Interim Quanu- tatiwCancer Unit Risk Estimates due to Inhalation of Benzene. OSHA Docket H-0598. Exh. 175. US. EPA Washington. DC (1985). 20. key,M.: Benzene Cycinogeniciry. CRC Press.B c a Raton. FL 11988). 21. Goldstein. B.D.:Is Exposure to Benzene a Cause of Human Multiple Myeloma?Ann. N.Y. .&ad. Sci. 609:225-230(1990). 22. Decoulfe, P.;Blamer. W h ; Blair,A: Monalir).among ChemicalWork. ers Exposed to Benzene and Other Agents. Environ. Res. 30:16- 25 (1983). 23. Irons,R: Benzene: Metabolism and Mechanisms.OSHA M e t HQj3B. Exh. 159-41A .24. Dempster, AM.: Evans. H.L.; Snyder. C-Q: The Temporal Relationship between Behavioral and Hematologml Effecu of Inhaled Benzene. Toxicol. Appl. Pharmicol. ?6:195-203 (1984). 25. Luke, CA: Tice. RR; Drew, RT.: The Effect of Exposure Regimen and Duracion on Benzene-Induced Bone-Marrow Damage in Mice. Murat. Res. 203251-271 (1988). 26. Snyder, CA; Sellakumar. AR;James.DJ.; Albert. RE.: The Carcino- genicity of Discontinuous Inhaled Benzene Exposures in CD-1 and C57B1/6 Mice. Arch. Toxicol. 62331-335 (1988). 27. Wong. 0.:An Indusuy-Wide M o d i r y Study of Chemical Workers OcCupauonalIy Exposed to Benzene. Submined to Chemical Manu- facturers Association (1983). OSHA Docket H-059B. Exh. 151-h 28. -on. B.: Epidemiologic Evidence on the Possibilirythat Risk of Leukemia may be Increased by Exposure to Benzene Y Loa. Concentrations. Report Prepared for Chemical .ManufacrurersAssociation (April 23, 1985). OSHA Docket H-0596. Exh. 201-33 29. Milham,Jr.. S.: Neoplasia in the Wood and Pulp Industr).. Ann. N Y. &ad. Sci. 2-1:294-300 (1976). 30. International Agency for Research on Cancer: MRC Monographs on the Evaluation of rhe Carcinogenic Risk ofChemicals to Humans. Yo! 25, Wood, Leather and Some Awxiated Industries. pp. 99-19, MRC. Lyon. France (1981). 31, Vidiani, E.C.:Leukemia AEsOciated with Benzene qosure. .Ann. N Y. w ,Acad Sci. ?71:143-151(1976). 32. M.: Leukemia in Workers due to Occupational Evposure to Benzene. New Istanbul Contrib. Clin. Sci. 12:3-l* ( 1 9 7 ) . .APPL OCCUP. ENVIRON. HYG. 714) APRIL 1992 261 33. Infante,PI.;White. M.C.: Rojedonsof LeukemiaRLSkASSCChed with O m r p o d d J3pmuc to Benmu.Am J. Ind Med 7AO3-413 (1985). 34. Y ~ . S . - N . ; L i . G . - L ; T l i R F . - D . ; d a l . : A ~ C o h o r r S c u d y o f Leukemia and Orher Cancers in Benzene Workers. Environ. Health Pen- 82207-213 (1989). 35. Picciano. D.J.:Monitoring Indwrial Popuktionsby Cyro~eneticPro. cedures. In: ofa WOfkShOponMerhodo@' forlwsshg ~cproductiveHyards in the Workplace, pp. 293-306. P.F. Inhnte and M.S. legator. EQ.U.S. Cavemment Printingoffife. Washington. Dc (1980). 36. Holder. B.B.. Dimor of Biomedical ReJevdl Dow Chernid USA: mer to Ernst Linde. office of Adminisarm, U.S. Environmental .*prr OCWP. EUVIIVON.m.n41 APRIL 1992