Document KRKv2ex9GdMgY6aZ6vQJ60R5Q

CHAPTER 1 CARCINOGENICITY OF BENZENE, TOLUENE AND XYLENE: EPIDEMIOLOGICAL AND EXPERIMENTAL EVIDENCE A.J. McMichacl Depurtment of Communiry Medicine University of Adelaide Adelride, South Au.s*rralia lNTRODUCTI( IN Benzene was tirst isolated i n 1825, and cominercially recovered from coal-tarderived light oil in 1849 and from petroleum in 1941 (IARC, 1982). By the late 19505, most commercial benzene was petroleum-derived. Chronic human exposure' to benzene may result in adverse haematological effects, characterized by a variety of blood dyscrasias, including leucopenia, thrombocytopenia, pancytopenia and anaemia. Benzene is also considered to be a human carcinogen and is well established as a cause of leukaemia (IARC, 1982). It is recognized iis being carcinogenic by the governments of Finland, the Federal Republic of Germany, Italy, Japan, Sweden, Switzerland and the U.S.A. By comparison with benzene, there is no direct epidemiological evidence concerning the carcinogenicity of toluene and xylcnes, and only limited experimental cvidence from i i n i d studies. / EVIDENCE OF UISNZENE CARCINOGENICITY IN HUMANS ' Lorrkoiwici mid orher l y n t p l ~ o - h a ~ ~ r n ccm~wt ~rs~: ~ea~rl~y is~tu~diie~s There have been many case reports. case scries and epidemiological studies associating exposure to benzenc with leukaemia in humans. However, very few of the studies have provided quantitiltive information about the relative risk of leukaemia under specified conditions of exposure. -'Exposurc CoiIccntrations arc givcii throughout in parts per million ppm. To convert to IS0 rccornincndcd units, for bci17znc. I ppm = 3.25 mg/m3 (-20 'C) 4 McMlCHAEL The lirst citse report suggesting an association between exposure to benzene and leukaemia was given by Delore and Borgomano ( I 928). An acute lymphoblastic leukaemia was observed in a worker in France, who was exposed to benzene for five years. Since then, numerous reports describing leukaemia cases among persons exposed to benzene have appeared, predominantly from Italy, France and Turkey (Goldstein, 1983). Several pioneering epidemiological studies investigating the relationship between exposure to benzene and leukaemia, as well as other lympho-haematopoietic cancers, were published in the mid-1970's. Aksoy published a number of reports on the risk of' leukaemia among Istanbul shoe makers exposed to benzene-containing solvents (Aksoy et d.,1972,1974;Aksoy, 1980). He estimated that the annual crude incidence rittc of ICukiicmiti among the shoe niakers was 13.5 pcr 100 000, comparcd with an overall annual incidence rate in the general population of 6 per 100 000. Vigliani ( 1976). reviewing earlier studies (Vigliani & Saita, 1964). estimated the increase in risk of leukaemia to be 20-fold for workers exposed to benzene in shoe liictories in Italy. As with Aksoy's studies, the apparent levels of exposure to benzene wcrc pl.cdo111iniintly within the 200-500 ppm range. However, these cxposure data were too imprecise for further quantitative analysis. Thorpc ( 1974) reported iin epidemiological study of leukaemia incidence atnorig workers from eight Esso Petroleum alfiliates in Europe. Employees were divided into two groups: lhosc with il potential exposure to 1 % or inorc benzene over ii five-year period and lhose not exposed or only occasionally exposed. Among the exposed, 8 leukaemiii deaths were observed, compared to the expected 6.6. yielding an ageSLiinditrdizcd mortdity ratio (SM R) of 121 (not stiitistically signilicant). Among lllc non-exposed. the number of deaths observed was I O and the expecled figure was 16.7. tlic corresponding SMR being 60 (not stntistically signilicant). These figures indicate thiit thc cmployccs exposed to benzcnc experienced an approximutely two-f'old risk 01' letlkileI11iii when compared to non-exposed employees. No benzene exposure levels wcrc iiviiilitblc from the study. . I:j~irlt~niiolo~~ics.itiridl ics wddiiig qumiiitilivc) csiiimiic*.s of' risk $)I* lc.irkticniiti 'k W S U I ~ S of' ii U.S. Ni1tit)Iliil liistitutc of Occul>itti<)niilSiifety iind HcLilih (NIOSH)s ~ u d yol' 748 white I11iIle workers exposed to benzene during the I11iillufiicttircof'rttbhcr hydrochloridc have been piiblislicd 011 three occiisions ( I n f i i n k 1'1 til., 1977. 1970: Itinsky ci d.,1981). The lirst two papers were bit& on the results W i I h 75'%,corllplctc I O I I O W - U ~of vital stitlus, while the Rinsky c'i til. (1911I ) piipcr WiiS basc.ct on diitiI with 98% complete follow-up. The findings were essentially the saiiie in Ciich of' the three piipers: therc wits a threc-fold increusc i n the risk of death f'roin l y ~ ~ ~ ~ h ~ ~ - h i i ~ ~ ~ ~ciiirllcocrpsoaincdl i ciii1 iilmost six-fold increase for leukiiemiii spt.cilic:illy (7leukae111iil deaths I'~~I:w.v 1.25 expected). Among workers employed for 1110rcIhii11 5 yOi1I-s. 5 hiid dicd from Icukitemiit, comparcd to 0.23 expcctcd. indicating ;I ?l-ll)ld increiise in risk. Rinsky ('1 til. (198I ) dcvoted considerable discussion to the industrial hygiene llleitSurell~elltSavailiible for one of the two study locations in Ohio. However. the invcstigiitors 111itde 110 link bctwccn the iiviiiliible beiwcnc nieasurcments and thc CARCINOGENICITY OF BENZtNt, I VLUCIYC MIXU 6 I Table 1. Estimate of excess risk of leukaemia deaths per 1OOO workers exposed to benzene. Number o( years exposed 46 ' 30 15 5 1 Exposure lovd 10 pPm NlOSH study. 44-152 30-104 15-54 5-18 1-4 OOW study' 48-136 32-93 16-48 5-1 6 1-3 1 PPm NlOSH 8Ndy 5-16 3-1 1 1.53 0.5-2 0.1-0.4 DOW Study 5-1 5 3-10 2-5 0.5-2 0.1-0.3 'From While el PI. (19821. 'Alnrky d a/. (1Wl). '011 ef a/. (1978). '45 years occupational litelime. etnploymenl histories of the individual cohort members. (Subsequently, Crump and Allen (1984) and Rinsky et ul. (1985) did link these data in unpublished risk nssessments. Cumulative benzene exposures of individual workers in this cohort were estimaled from job-exposure matrices. This approach indicated a heightened, 45-foM, increase in risk of leukaemia in the highest cumulative exposure category (Rinsky et d.,1985). See also Appendix, p. 18.) Ott e! til. (1978) repqrted the mortality of a cohort of 594 Dow Chemical workers exposed to benzene in the production of chlorobenzol, alkyl benzene and ethyl cellulose from 1938 to 1970. Their mortality experience was observed from 1940 through 1973. With expected deaths based 011 U.S. white male mortality, no statistically significant increases were found in any cause of death for the cohort as a whole. even when mortality was analysed by estimated cumulative benzene dose (ppin-months) and latency. However, further review of the three cases of leukaemia showed all three to be myelocytic, with two being classified as acute. Based on incidence data reported in the U.S. Third National Cancer Survey, the expected incidence of niyelocylic leukaemia was only 0.8 cases. Because the Occupational Safety and Health Administration in the U.S.A. is legally required to consider quantitative risk assessments whenever possible from available data. such an assessment has been attempted (White et al., 1982) using the abovementioned two cohort stud6s. Uncerlainties rcgdrding the levels and lengths of exposure to benzene were incorporated into the analysis and the results are shown in Table I . Based on the one-hit model, and subject to the typi'cal uncertainties of epidemiological data, the assessment indicates that a working fifetime exposure to bcnzene at an exposure level of IO ppm poses a substantial excess risk of death from leukaemia. While et til. (1982) selcctcd the one-hit, or linear, model because the biological niechanism of' benzene-induced leukaemia is not understood well enough to justify a niore coinplicated model, and because information from epidemiological studies regarding the relative risks of leukaemia at dikrent levels of exposure to benzene was 6 McMlCHAEL * Table 2. Analyses of relative risk and excess risk of death from leukaemia per I 000 workers exposed to benzene for an occupational lifetime. I Authors Relative risk Eslimaled cohofl exposure ~ ~ ~~~ ~ Excess rink per 1 OW workers _ ~ _-~~ ~~ Excess rink per 1 OOO exposed l-o 10 ppm 1 ppm 011 et a/. (1978) Infante et a/. (1977) Rinsky et a/. (1981) Vigliani (1976) Aksoy (1977) 3.75 5.6 21 20 25 1-30ppm for 8-9 years 10-100 ppm, 8.5 years average 10-100 ppm, 2 5 years exposure 20&500 ppm, 9 years average 150-210 ppm, 9.7 years average 72 170 140 475 534 24-720 17-170 14-140 9.5-23.8 25.4-35.6 2.4-72 1.7-17 1.4-14 1.0-2.4 2.5-3.6 `Table adapled lrom Inlanle and White (lU35).C.kulalionr in the Iarl three columns were performed uring the method applied by IARC (1982) 10 dam lrom Inlanle et at. (lQ77)and Rinsky et J/. (1961). inadequate for testing more complex models. Furthermore, as Infante and White . (1985) have subsequently noted, several indications of benzene toxicity to bone inarrow and peripheral blood cells have demonstrated a .linear dose-response relntionship. This linear, non-threshold model assumes that every increment of dose is accompanied by a commensurate increment in the excess cancer risk. The use ot`this toxicologically plausible model allows extrapolation of risks from relatively high dose levels, where cancer responses can be measured, to relatively low dose levels, where such risks are too small to be measured directly through animal or epidemiological studies (WHO, 1981). The IARC (1982) also conducted a quantitative cancer risk assessment for benzene irnd leukaemia using the results of four studies (Vigliani, 1976; Aksoy, 1977; Ott et ul,, 1978; Rinsky el d.,1981). IARC assumed a linear dose-response relationship, with cumulative benzene dose predicting the relative risk of developing leukaemia. I t also made. slightly dinerent assumptions from those of White e/ d. (1982) regarding the length of exposure for the NIOSH (Rinsky) and Dow (Ott) studies. For purposes of comparison with the estimations of,White e/ ul. (1982). a summary of the IARC (1982) assessment of benzene risk, with extrapolation to levels of 10 ppm and I ppm is shown in Table 2. The calculations in the last three columns were performed by Infante and White (1985). using the method applied by the IARC (1982) to the NIOSH study and taking account of the IARC's opinion that "risk calculations .(should) reflect the degree of uncertainty in the estimates of dose rate. . by citing upper and lower bounds of such estimates" (IARC, 1982).The estimated excess risks of leukaemia associated with a working lifetime (45 years) of exposure to 10 ppm and I ppm of benzene are similar in Tables 1 and 2. Recognizing the need for further quantitative investigations into the relationship between benzene exposure and lymphatic and haematopoietic cancers, Wong (1983) CARCINOGENICITY OF BENZENE, TOLUENE AND XYLENE 7 carried out a large industry-wide mortality study of U.S. chemical workers occupationally exposed to benzene. The population in this historical cohort study consisted of 4 602 male chemical workers from seven plants, who were occupationally exposed to benzene for at least 6 months between 1946 and 1975, and a comparison group of 3 074 male chemical workers from the same plants who were employed for at least 6 months during the same period of time, but who were never occupationally exposed to benzene. Exposure to benzene was divided into two categories: continuous exposure and intermitted exposure. Continuously-exposed workers were classified ilccording to both an eight-hour time-weighted average (TWA) and a peak exposure level, whereas intermittently-exposed workers were characterized by a peak exposure level only. The eight-hour TWA's were estimated by company industrial hygienists, using uniform criteria; nevertheless, it was not possible to quantify accurately the exposure of workers to benzene. The major findings of this study were as follows. When compared to the U.S. populution, the SMR's in the exposed group for all lymphatic and haematopoietic cancer combined. for leukaemia, and for np-Hodgkin's lymphoma ( m hosarcoma, +reticulosarcoma itlld other lymphoma) were slightly, but not significantly, higher t an thc national norm. When the grobp with no occupational exposure was used for comparison. the cxposcd group (continuous and intermittent) experienced a lympho- -hIah.cismeaxtcoepsosiwetaisc ciincer relative risk ( R prim;irily due to seven R ) of 2.99 leukaemia (borderline statistical de-iths in the exposed significance). group, versus noiic in the comparison group. For the continuously exposed group. the R R fol- lympho-haematopoietic cancer \viis 3.20 ( p<0.05). The association bctwecn expowre to benzene, and, specifically, kukiie111ii1W;IS ol' borderline significance when those exposed continuously and those cxposcd intermittently were grouped together, but the association between colltinuous cxposure illone iII1d lcukiteiniii was statisiically significant(p <0.05). None of thc seven hkaemiii deiiths was of the acute myelogenous cell type. The R R for non-Hodgkin's lymphoma il11long white males in the continuously exposed group was 5.02. buI wiis not signilicant ( p = 0.12). Furthcrmorc. the dt1tit dcnionstriltcd stiitistically significant dose-response rcli1tionships between the estiiiiiited cumulative exposure to benzene and mortality I'm11 :dl l y ~ i ~ p l i o - l i i i ~ ~ ~ ~ ~ i tCoiplnoceir~st iccombined and from leukaemia. Those chilici11 workers with I I cuniulativc exposure of at 1e:lst 720 ppm-months experienced ;in ItR of 3.93 for lyl~iphi~ticind huematopoietic cancer, when compared to the workcrs with no occupationid exposure. However, as Wong (1983) acknowledges, t h t liriiitations in exposure estimates preclude useful dose-response analysis at low levels of exposure. I n ii iiiore recent altciiipt to derive quiinlitative risk assessments of leukaemia in subjects cxposcd to bcnzenc. Crunip and Allen (1984) used the data from Ott et al. (107X). Rillsky PI d.(1981) and Wong (IYX3), along with animal experimental data I'roiii inhalation studies. The estimiites of leukaemia risk derived by Crump and Allen ( 1984) rroni iiIiililal inhalation studies (Table 3). while quite consistent, are all col1siderilbly less than those made from human data (Table 2). Accepting the :lssumptions underlying these estimates, this indicates that rodents may be less susceptible than humans to benzene-induced leukaemia. McMlCHAEL Table3. Estimates' of lifetime human risk per 1000 exposed persons"from animal inhalation studies` Data aet All squarnous-cell carcinomas, male rats, NTP (1984) . All squamous-cell carcinomas, male mice NTP (1984) Leukaemia, both sexes, rats, Maltonl et 81. (1983) Leukaemia, male mice, Goldstein el 81. (1982) 1 10 1 10 1 10 1 10 0.77 7.65 2.01 19.90 0.15 1.46 0.04 0.38 -*Aaaumlng riak to animala la the u m e u rlok to hummi whrn doae la meuured in mglkg body weighUday. .Length 01 human erpowro to benzenb 40yaara. `Table baaed on Crump and Allen (1980. Crump and Allen (1984) consider that their estimates from human studies, based upon cumulative or weighted exposure, are the most reliable. The attendant estimates of additional leukaemia cases per I 000 workers exposed to IO ppm of benzene range from 0.5 to 2.6 cases from one year of exposure beginning at age 20, to I5 to 88 ctses from 40 years of exposure. The "window exposure" variable (defined as a shorterterm exposure occurring at the putative, latency-related,cancer induction time) tends to give lower estimates than either weighted or cumulative exposure. However, Crump and Alien ( I 984) judge that the estimates based on either cumulative exposure or weighted exposure are preferable to those based on window exposure. Non-linear models would predict about the same results as linear models a1 high doses (IO ppm for 40 years). A person exposed to IO ppm for 40 years would attain a cumulative exposure of 400 ppm-years, which, in the Rinsky el al. (1981) cohort, would place him in a category for which a significant excess of leukaemia occurred. Indeed, the subsequent re-analysis by Rinsky et al. (1985) gave an estimated 45-fold increase in risk of leukaemia for that category of exposure. Thus, no extrapolation to low doses is required to estimate risk from lifetime occupational exposure to IO ppm. However, the assumption of linearity plays a more important role at lower exposure levels, and the estimates of risk are more uncertain. This is important, since, under the benzene exposure limit of IO ppm prevailing in the U S A . , the risks estimated from continuous exposure to IO ppm represent a theoretical maximum, as it is unlikely that any current worker's exposure will approximate this if the 10 ppm standard is strictly enforced. Crump and Allen (1984) point out that their estimates of risk from human data were limited to leukaemia, as this appeared to be the only type of cancer conclusively linked to exposure to benzene in humans. This approach could underestimate cancer - .. >. CARCINOGENICITY OF BENZENE, TOLUENE AND XYLENE 9 risk if, in fiict, benzene causes other cancers, as it does in rodents. The recent epidemiology literature (DecouflC et ul., 1983; Rinsky et al., 1985) suggests that benzene exposure may also cause multiple mycloma in humans. Other less yuuntitutiw or less specific studies In a cohort study of 35000 workers at 8 petroleum refineries in the U.K., the leukaemia SMR was found lo be 94 (Rushton & Alderson, 1981a). Using data from this cohort study, 36 cases with leukaemia mentioned on the death certificate were the subject of a subsequent case-control study (Rushton & Alderson, 1981b). Benzene measurements were not available, but each worker was assigned, based on work histories, to either low, medium or high levels of exposure to benzene. A two-fold increase in risk was detected for workers with medium or high exposure, relative to those with low exposure to benzene, when length of service was taken into account. This two-fold increase was of borderline statistical significance (p = 0.05). There was no evidence that the risk increased with length of service. DecouflC t ~ tril. (1983) conducted a historical cohort mortality study of 259 male employees of a chemical plant (converted from a petroleum refinery) where benzene was used in lilrge quantities. Unfortunately, n o quantitative measures of exposure were available from the plant. Four deaths from lympho-haematopoietic cancers were observed, when only 1.06 deaths were expected; the SMR of 377 was statistically significant. Three of these four deaths were from leukaemia: one was chronic lymphocytic. one acute monocytic and one acute myelomonocytic. This last Icukilemiii cilSe also had a history of multiple myeloma. Since the fourth lymphohuematopoietic cancer death was due to multiple myeloma, the authors suggested a possible link between benzene and multiple myeloma. A number of epidemiological studies have found significant mortality excess from Icukuemia m o n g workers exposed to solvents i n the rubber industry. However, the relationship between this observed leukaemia excess and, specifically, benzene is less clear. I n one cohort, benzene was not the predoininant solvent used and the authors wcrc hcsi tan1 in ascribing to benzene the sole responsibilityfor the excess of leukaemia (Monson & Nakano, 1976). In another cohort of rubber workers, the excess risk was limited to lymphocytic leukaemia and was not associated with either monocytic or myelocytic leukaemia (McMichael et d.,1975, 1976; Wolf et ul., 1981). However, workers in this cohort were exposed to Other solvents as well as benzene, although the latter often occurs as a contaminant df other solvents (e.g., petroleum naphtha). To further investigate in this latter cohort the relationship between the observed cxccss of' lymphocytic leukaemia and exposure to various solvents, Arp et al. (1983) examined the solvent exposure histories of IS caws of lymphocytic leukaemia and 30 matched controls within this second cohort. An individual was classified as exposed to ilsolvent if the cumulative length of exposure exceeded 12 months. The relative risk estimates were elevated for exposure to benzene (RR = 4.50) and to other solvents ( R R = 4.50). Neither finding reached statistical significance at the 0.05 level. A much more detailed analysis of this same population was reported by Wilcosky et al. (1984). Details of estimated individual exposure to each of 25 solvents were obtained. Lymphosarcoma and lymphatic leukaemia were strongly associated with exposure to 10 McMlCHAEL carbon tetrachloride and carbon disulphide, whereas weir association with benzene and xylene was of marginal statistical significance only. Several epidemiological studies of petroleum refinery workers who are potentially exposed to petroleum products, including benzene, have also found an excess of lympho-hacmatopoietic cancers (Tabershaw/Cooper Associates, 1974; Thomas el ul., 1982; Haiiis e/ d.,1982). However, none of these refinery studies provided any direct link between the observed mortality excess in lympho-haematopoietic cancer and benzene exposure. Eniliitriioir o f Ilie c~/~i~l.iniolo~icc.vriirkleiic.eJ On the bitsis of these studies there is substantial epidemiological evidence that persons exposed to benzene (or an associated impurity) experience an elevated risk of certain types of lymphatic and haematopoietic cancer, particularly leukaemia. Reports providing such epidemiologicalevidence include those of Aksoy el (11. (1974). Ott iv d . (1978). Vianna and Polan (1979). Aksoy (1980). Rinsky e'/ crl. (1981). Rushton and Aldcrson (I98lb). Decoutlt et u/. (1983) and Wong (1983). The most frequently reported form of leukaemia associated with benzene exposure is tlie acute. myclocytic form. However, research on rubber workers (McMichael CI d . , 1975. 1970) indicatcs thiit, i n chronic low-lcvcl exposurc situations, lyniphocytic Icukucmia 1iiiiy predominate. Perhaps this is a more likely manifestation of non-iicute bone , iiiiirrow insult. (Chronic foriiis of Icukaemia itre dso less likely to be rcported on deiith certificiites as the underlying ciiuse of death.) On Ilic othcr hand, there are reports indicating no increased risk of` lynipholii~e~i~i~topociacnticcer as a result of benzene exposure. A recent cohort study (Tsai (`I til., 19x3) ol'454 niille relinery workers exposed to benzene during 1952 to 1978 k i t ii Tcxiis rcfincry did not find any lyinpho-haciiiiitopoicticcilliccr deaths; howcvcr. thc cohort size was sniull and the statistical power to detect il modest risk i n either lympho-lincm~~topoictciicincer or leukiiemiii wiis limited. The case-control stiidics of lymphocyticleukaemia iiinong rubber workers (Arp 1'1 d.,1983; Wilcosky 1'1 d . ,1984) indicuted thiit the lymphocyticleukaemia excess i n the rubber industry ciiiiie primarily from carbon tctruchloridc and carbon disulfidc, rather than from xylenc or, cvcn less likely. l'roiii benzene. The large inoriaiity study of `rhorpe ( 1974) of38 000 petroleum workers polenlially exposed to benzene hiis often been cited a s showing no evidence o+gu&uil leukaemia excess as a result of exposure to benzene. This interpretat?on. howevcr. ignores the oomparison'of the Q&osed workers with' the control workers I(l*ronitho snnie facilities, but not exposed to benzene); thc exposed group had ii two- fold risk for leukaemia relative to the non-exposed group. Thcrc iirc iI number of outstanding issues regarding the relationship bctwccn exposure to benzene and increased risk of lyiiipho-haeiiiatopoietic cancers. First, none of tlie published studies provides adequate exposure information for satisfactory dose-response iiliiilysis. Second, the choice of iiti appropriate comparison is ill1 iiiiportunt issue. I n ;I number of studies, analyses based on coniparisons with the gener:i I pop11Iii tioii titiled to detect a11 increased rihor iyni pho-haeniiitonoiehc C i G C r or i c ~ i ol` refinery workers in the U.S.A. (Tabershaw/Cooper Associates. 1974) and the y :ARCINOGENICITY OF BENZENE, TOLUENE AND XYLENE 1 11 Table 4. Studies of chromosomal breakage In workers exposed to "low" atmospherlc concentrations of benzene Author Level of exporure lo benzene (TWA) Picciano' (1979) Sarto ot a1.c(1984) Controls 3% (workers)' 1.2% (cells)b 1 PPm 1-2.5 ppm 2.5-10 ppm 22% 26% 33yo Range of short-term samples = 0.2-12.4 ppm 2.7% ' -'Benreno m u u r e d in b h TWA'r;chromoaomo breakage = breaka. dicenlriw. tranrlocatlonr and exchange figurer. K H of workerr with chromwme breakago. 'Benzene mearured in 5 and W m i n ampler. cohort study of eight oil refineries in Britain (Rushton & Alderson, 1981a), the initial calculations of the leukaemia SMR, based on a comparison with the general population, revealed no significant excess. Howcver, sybSequent analyses that took account of variations in benzene exposurk between categories of workers re(ealed ?? m.,positive associations with leukaemia. Third. although some studies (Aksoy, 1980;Ott 1978;Rinskyertil., 1981) indicate that acutc myelogenous leukaemia is the type l*requentlyassociated with benzene exposure, other reports show associations with dilrerent leukuemia cell types (Tabershaw/Cooper Associates, 1974;McMichael et ul., 1976;Linos e/ ti/., 1980;Schottenfeld ('I a/..I98I). The studies of Decouflt et a/. (1983) illid llinsky CI til. (1985) have also suggested an increased risk of multiple myeloma. h . / i r ~ t i , ~ c t i i c . iti ~t i.d i~lirotnosoti~t&ilc/s iti Iiictiiiitu Numerous studies have been carried out on the chromosomes of bone-marrow cells and pcripheriil lymphocytes from people known to have been exposed to benzene (I>ciln. 1078; Picciano. 1070; Wutanabe e/ d.,1980; Sarto et al., 1984). The populations included in these studies fall into two general categories: (I) patients with either it current or a past history of benzene-induced blood dyscrasias ("benzene hi1elnopathies"). often associated with extensive exposure to benzene; and (2) workers with known current or past exposure to benzene, but with no obvious clinical effects. In some of' thesc studies, significant increases in chromosomal aberrations have been seen; i n some c w s these have persisted for years after cessation of exposure (IARC, 1082). The liiidings from two studies are summarized iii Table 4. Picciano (1979) reported that 52 workers exposed to average benzenc: concentrations below 10 ppm demonstrated a statistically significant two- to tnree-fold excess of chromosomal dill11iige. compared to that observcd among 44 unexposed workers. Furthermore, the percentage of workers exhibiting both chromosome breaks and marker chromosomes wx 20. 25 and 32% for workers categorized as exposed to average benzene coriccntrations below I, I .0-2.5 and 2.5-10.0 ppm. respectively. Less than 3% of the iiiiexposed workers exhibited both chromosome breaks and marker chromosomes. Morc rcccnlly. Sarlo et til. (1984)have reported supportive, albeit weaker, evidence of chromosomal breaks i n workers exposed to benzene. 12 McMlCHAEL Oilier cmwrs in liuinaris Although induction of types of cancers other than of the lympho-haematopoietic system has been suspected, these have not been adequately evaluated from epidemiological studies of workers exposed to benzene. The studies have not had sufficient statistical power to detect low excess relative risks. However, since 1978, experimental studies have demonstrated that benzene administered either by oral gavage or by inhalation induces cancer at multiple sites in experimental animals. This research is reviewed further on. Possible carcinogenicity of toluene and xylene No epidemiological studies of differentiated exposure to toluene or xylene have been reported. I t is possible that some or all of the above-mentioned increased cancer risk associated with non-specific exposure to solvents is attributable to these, or other, non-benzene solvents. However, evidence of any such effectwill be difficult to obtain, particularly since benzene occurs as an impurity in toluene and xylene. ANIMAL EXPERIMENTAL STUDIES OF BENZENE CrrrcinogetI icily lntragastric administration of benzene dissolved in pure olive oil to female rats' induced Zymbal gland carcinomas, mammary gland carcinomas and leukaemia. With the exception of leukaemias in the high-dose group, no such tuniours occurred in male rats (Maltoni & Scarnato,' 1979). Experimentid studies of lifetime benzene inhalation at 300 ppm im mice show the widespread occurrence of anaemia, lymphocytopenia, neutrophilia and bone-marrow hypo- or hyperplasia (Snyder et ul., 1980). Statistically signiticant increases in the incidence of lympho-haematopoietic tumours (lymphocytic lymphoma, plasma- cytoma and myelogenous leukaemia) also occurred (Snyder el d., l978a. 1980). However, 110 such tumour occurred in similarly exposed (male) rats (Snyder el ut., I978b). Subsequent studies by Maltoni and co-workers have shown that benzene is a potent carcinogen in rats of both sexes, whether ingested or inhaled; benzene produces difkrent types of turnours in different organs and is therefore a "n~ultipotential carcinogen" (Maltoni et ul., 1985). . Since all the benzene inhalation studis reported to date entail concentrations in the 200-300 ppm range, there is no direct animal experimental evidence relating to low- concentration exposures that now typify human occupational exposures. Long-term carcinogenicity bioassays at 50, 25, IO, 5 and I ppm may therefore be helpful for scientific risk assessment in humans. An earlier comprehensive review of the experimental data (IARC, 1982)concluded that benzene does not induce specific gene mutations in bacterial test systems, in CARCINOGENICITY OF BENZENE, TOLUENE AND XYLENE 13 Dro.sup/tilu melmogusier, or in mammalian cells. However, there was evidence that benzene induces cytogenetic abnormalities (chromosomal aberrations in bonemarrow cells) (IARC, 1982). Furthermore, the micronucleus test in mice and rats has been consistently positive, and numerous studies have shown that exposure of experimental animals to benzene in vivo leads LO the induction of chromosomal aberrations in bone-marrow cells (IARC, 1982). While Van Raalte et (I/. (1984) have argued that only peak exposure to benzene above 100 ppm ciln cause leukaemia or bone marrow toxicity, the results of recent epidemiological and experimental research suggest otherwise. Several recent animal experimental studies have demonstrated adverse effects on chromosomes and bone marrow as a result of low-level exposure to benzene. I t has recently been demonstrated by Tice et af. (unpublished data quoted by Infante & White, 1985) that exposure to benzene a t a concentration of 10 ppm in air, 6 h/day for 9 days, induces a significant increase in micronuclei (a measure of bonemarrow chromosomal damage) in peripheral blood cells of mice. Mice exposed to 28 ppm benzene for a total of only 4 h had a two-fold increase in sister chromatid exchanges (SCE)in bone-marrow cells. The eflkct on mouse bone-marrow cells increased linearly with increasing levels of 4-h benzene exposure, indicating a linear dose-response. Previously, Kligernian had reported that mice exposed to IO ppm benzene by inhalation for only 6 h demonstrated a significant increase in SCE in peripheral blood lymphocytes and in micronuclei in bone-marrow erythrocytes (Kligerman et d., 1983). Gad-El-Karitn et d. ( 1984) have reported a dose-response increase in micronuclei in bone-marrow erythrocytes of mice administered a total of two oral doses of benzene at concentrations as low as 8.8 mg/kg body weight. This dosage would be equivalent to two 8-h atmospheric exposures of approximately 6 ppm. Collcctively, these studies denionstrate chromosomal damage in bone-marrow cells ;is a result of the equivalent of one to several working day exposure to 10 ppm of benzene. They also demonstrate a linear dose-response for benzene exposure and chromosomal damage to both marrow and periplieral blood cells. Since most leukaemias and related disorders in man seem to involve stem-cell i1nomnlies, the response of haematopoietic stem cells was investigated by Baarson et td. (1984) who exposed mice by inhalation to IO ppni benzene for 6 h/day, 5 days/week lor 32. 66 and 178 days. Progenitor cells from thc exposed mice showed markedly reduced abilities to form colonies, compared to ct.11~from control mice. I n mother recent study, shorkterni exposure (6 h/day for 6 days) to IO ppm bcnzenc in air significantly deprcsscd mitogen-induced blastogenesis of both B- and T-lymphocyies in mice (Rosen ei til., 1984). The authors concluded that benzene cxposure at or near LO ppm may affect immune function. Thus. less than one week of exposure to 10ppm of benzene is associated with chromosomal damage to bone-marrow cells, significant depression of the bone marrow and disturbances of immune system function. Since most leukaemias and related disorders in man seem to involve stem-cell abnormalities and immune system deficiencies, these. animal experimental findings may be highly relevant to leukaemia and exposure to benzene in humans. The findings provide further support I'or the conclusion that an elevated risk of a bone-marrow proliferative cancer 14 McMlCHAEL (leukaemia) could be expected as a result of exposure to IO ppm benzene (Infante PC White, 1985). ANIMAL, EXPERIMENTAL STUDIES 01'TOLUENE A N D XYLENE In a series of experiments, primarily directed at studying benzene carcinogenicity in rats, Maltoni et a/. (1985) have also studied the effects of other benzene-correlated aromatics (toluene, xylene and ethyl-benzene). At high concentrations, toluene and xylene (and, to a lesser extent, ethyl-benzene)caused an increase in the total number of malignant tumours. The increase i n tumour yield was approximately one-third that induced by an equal dose of benzene. No othcr experimental data on this question appear to have been published. SUMMARY Bivr:ivria Thc evidence for carcinogenicity of benzene in hunians was evuluiitcd by the IARC in I982 as follows: "It is established that human exposure to comniercial benzene or benxenecontaining mixtures can cause damage to the hacmatopoictic system, including pi\ncytopenia. The relationship between benzene exposure und the development ol' iicutc myclogenous leukacniia has been estiiblishcd in epideniiologicul studies. "Reports liiiking exposure to benzene with other maligniincies were considered to bc iiiiidequate for evaluation. "There is si!ficiwif cvidmcc. that benzene is carcinogenic to man." This cvi\luation now warrants some elaboration and updating. While the cpidcmiologicid cvidence concerning benzene carcinogcnicity is strongest for acute myelocytic leukiiemia, there is some limited evidence of increased risks of chronic myeloid and chronic lymphocytic Icukacmia. In addition, rccciit studies havc suggested it11 increased risk of multiple myeloma, while others indicate a dose-related incrcuse for total lymphatic and haeinatopoietic neoplasms. Corroborative evidence t'or such ;I gcneralized elfcct comes Vrom expcrimentul studies showing that cxposurc .to benzene depresses all lyiiipho-haeiiialopoielic cell lines. Whilc only limited cvidencc of bcnzene carcinogcnicity i n experiincntul ;inimals exists, the recent findings of tha National Toxicology Program (NTP. 1984) in the U.S.A. iind Maltoni CI (I/. (1985) strongly indicate that benzene is an experimental carcinogen. 7bl1101~(1~11d .V~*/O1e1 Whilc no direct human evidence is available, there is recent evidence of ciircinogcnicity of toluenc and xylcnc tit high conccntrutions in cxperimcntal iinimals. It should also be noted that any future epidemiological observations of ciincer risks nssocioted with toluene or xylene would havc to take account of the suspccted efrects of benzene impurities, REFERENCIS Aksoy, M.(1977)Leukemia in workers due to occupational exposure to b ~ ~ ~Nenwe . lsrunbul Contrib. din. Sci., 12, 3-14 Aksoy, M. (1980)Different types of malignancies due to occupational exposure to benzene: a review of recent observations in Turkey. Environ. Res., 23, 101-190 Aksoy, M.,Dincol, K., Erdem, S.& Dincol, G. (1972)Acute leukemia due to chronic exposure to benzene. Am. J. Med., 52, 160-166 Aksoy, M., Erdem, S. & Dincol, G. (1974) Leukemia in shoe-workers exposed chronically to benzene. Blood, 44,837-841 Arp, E.W., Wolf, P.H. & Checkoway, H. (1983) Lymphocytic leukaemia and exposures to benzene and other solvents in thc rubber industry. J. Occup Med., 25, 598-602 Baarson, K.A., Snyder, C.A. & Albert, R.E. (1984)Repeated exposure of C57B1 mice to inhaled benzene at IO ppm markedly depressed erythropoietic colony formation. Toxicol. Lett.. 20,337-342 Crump, K.S.& Allen, B.C. ( I 984)Quuntitutive Estimates of Risk of Leukaemiafrom Occwp~itioncil E.vpo.wrij to Bcwzcne. Unpu'dished report prepared for US. Occupationcil Safely and Health Administration Dean, B.J. (1978)Genetic toxicology of benzene, toluene, xylene and phenols. Mutat. Rox.,47, 75-97 Ilecoufle. P.. Blattner, W. & Blair, A. (1983)Mortality among chemical workers exposed to benzene and other agents. Environ. Res., 30, 16-25 [lelore, P. & Borgoniano. C. (1928)Acute leukemia following benzene poisoning. J . Mid. Lj'Ott, 9, 227-233 Gad-El-Karim, M. M., Harper, B. L. & Legator, M.S. (1984) Modifications in the myeloclastogenic efl'ect of benzene in micc with toluene, phenobarbital, 3iiiethylcholanthene, Aroclor 1254 and SKF-525A. Mutat. Res., 135, 225-243 Goldstein, B. (1983)Benzene is still with us. Atn. J. Ind. Med., 4, 585-587 Goldstein, B.D.,Snyder. C.A., Laskin, S., Broniberg, I., Albert, R.E.& Nelson, N. ( 1982) Myelogenous leukaemia in rodents inhaling benzene. Toxicol. Leff.,13, 169- I 73 Hanis. N.M.,Holmes,T. M., Shallenberger, L.G. &Jones, K.E. (1982)Epidemiology study ofrefinery and chemical plant workers. J. Occup. Med., 24, 203-212 IARC (1982) /ARC Motiogrtiphs on the Evulirulion of the Carcinogenic Risk of' . Clwniiculs to Hutntins. Vol. 29, Some Industrial Chemicals and Dyestufls, Lyon, International Agency for Research on Cancer, pp. 93-148 Inhnte. P. White, M. (1985) Projections of leukemia risk associated with occupational exposure to benzene. Am. J. Ititl. Med., 7, 403413 Infante, P.F., Rinsky, R.A., Wagoner, J.K. & Young, R.J. (1977) Leukemia in bcnzene workers. Luncel, 2, 7618 Infilnte, P.F.. Rinsky, R.A., Wagoner, J.K. gL Young, R.J. (1979) Leukemia in benzene workers. J. Etiviron. Purlid. Toxicol.. 2, 251-257 Kligerman, A.D., Erexson, G.L.& Wilmer, J.L. (1983) Development of rodent Peripheral blood lymphocyte culture systems to detect cytogenetic damage in vivo. 76 McMlCHAEL (Abstract) lnr. S j ~ pS.ister Clirornutid Excltar~ge,Brookhaven Natl. Lab., Dec. I983 Linos, A., Kyle, R.A., O'Fallon, W.M. & Kurland, L.T. (1980) A case-control study of occupational exposures and leukemia. Inf. J. Epidemiol., 9, 131-135 Maltoni, C. & Scarnato, C. (1979) First experimental demonstration of the carcinogenic effects of benzene. Med. Luv., 5, 352-357 Maltoni, C., Conti, B., Cotti, G. & Belpoggi, F. (1983) A multipotential carcinogen. Results of long-term bioassays performed at the Bologna Institute of Oncology. Ant. J . Itid. Med.. 4, 589-630 Maltoni, C., Conti, B., Cotti, G.& Belpoggi, F. (1985) Experimental studies on benzene carcinogenicity at the Bologna Institute of Oncology: current results and ongoing research. Am. J . Ind. Med., 7, 415-446 McMichael, A.J.. Spirtas, R., Kupper, L.L. &Gamble, J.E. (1975) Solvent exposure among rubber workers: an epidemiological study. J. Occup. Mcd., 17, 234-239 McMichael, A.J., Spirtas, R., Gamble, J. R. & Tousey, P. M. (1976) Mortality among rubber-workers: relationship to specific jobs. J. Occiip. Mcd.. 18, 178-185 Monson, R.R. & Nakano, K.K. (1976) Mortality among rubber workers. I. White male union employees i n Akron, Ohio. Am. J. Epitletniol., 10.3, 284-296 NTP ( 1984) National Toxicology Program. Toxicology titid ccircirto~~in~is.s(iu.srliw of hcwzcw iii F344/N ruts cmd B6C3F ttiicij (geivrrge slritlii~s).NTP-83-072, N IH Publication 84-2545. U.S. Department of Health and Human Services 011. M.G.,Townsend, J.C., Fishbeck, W.A.& Langner, R.A. (1978) Mortalityamong individuals occupationally-~ exposed to benxne. Arch. Et~virort.Hcwlrlr, 33, 3-9 Picciuno, D. (I979) Cytogenetic study of workers exposed to benzene. Eiiviroit. I<os., 19. 33-38 Rinsky. R.A.. Young, R.J. & Smith, A.B. (1981) Lcukemiii in bcnzene workers. Ant. J. Iiitl. Mcrl.. 2, 2 17-45 Rinsky. R.A.. Smith, A.B.. Hornung. R., Filloon. T.G.. Young. R.J., Okun. A.H. & Land rigan, 1'. J. ( 1985) Be~irzc~nmcid kwkcvnirr: (111 c~j~i(1~~riiioriIsok~tircsscs.sntcnt. Unpublished report, U.S. Departinent of Hcalth and Human Services, Public Health Service Roscn, M.G.,Snydcr, C.A. & Albcrt, R.E. (1984) Dcprcssions in 13- and Tlyniphocyk mi togen-induced blastogenesis in mice exposed to low concentrations of' bcnzcnc. Toxicol. Lett., 20, 343-349 Rushton, L. & Alderson, M.R. (1981a) An epidemiological survey of eight oil rclincrics in Britain. Br. J . hid. Mal.,38, 225-234 Ilushton. L. & Alderson, M. R. (1981b) A case-control study to investigate the association bclwecn exposurc to benzene and deaths froin leuke111iil in oil rclincry workers. Hr. J. Ctiiicer, 43, 77-84 Siirto. F.. Comintito. I..Pinton. A. M .. Broveduni. P. G..Merler, E., Peruizi. M ., I3iunchi. V. & Levis, A.G. (1984) A cytogenetic study or workers exposed to low concentrulioiis of benzene. Ctrreino~encJsis5,, 827-832 Schottenfeld. D.. Wnrshauer, M. E., Zauber, A.G., Meikle, J.G. & Hart, U. R. (1981) A prospcctivc study of morbidity and mortality in petroleuni industry employees in Lhc Unitcd Slotcs - a preliminury report. In: Qucitit$ctition o/ Occupcitiorttrl C ' t i r i i w (Sad~iwjR-iiport 9 ) . Cold Spring Harbor, NY, CSH Press, pp. 247-265 i. CARCINOGENICITY OF BENZENE, TOLUENE AND XYLENE 17 Snyder, CA., Goldstein, B. D., Sellakumar, A., Albert, R.E. & Laskin, S.(1978a) The toxicity of inhaled benzene (Abstract No. 106) Toxicol. Appl. Pharmucol., 45,265 Snyder, C.A., Goldstein, B.D., Sellakumar, A,, Wolman, S.R., Bromberg, I., Erlichman, M.N. & Laskin, S. (1978b) Hematotoxicity of inhaled benzene to Sprague-Dawley rats and AKR mice at 300 ppin. J. Toxicol. Environ. Health, 4, 605-6 18 Snyder, C.A., Goldstein, D.D., Sellakumar, A.R., Bromberg, I., Laskin, S. & Albert, R. E. (1980) The inhalation toxicology of benzene: incidence in hematopoietic neoplasms and hematotoxicity in AKR/J and C57BL/6J mice. Toxicol. Appl. Pliurmucol., 54, 323-3 I Tabershaw/Cooper Associites (1974) A mortulity srudy of petroleum refinery workers. The American Petroleum Institute, Berkeley, California Thomas, T. L., Waxweiler, R.J., Moure-Erasso, R., Itaya, S. & Fraumeni, J.F., Jr. (1982) Mortality patterns among workers in three Texas oil refineries. J. Occup. Med.,24, 135-141 l'horpe, J.J. (1974) Epidemiologic survey of leukemia in persons potentially exposed to benzene. J. Occup. Mcd.. 16, 375-382. 'I'sai. S.P., Wen. C.P., Weiss, N.S., Wong, O., McCellan, W.A. & Gibson, R.L. ( 1983) Retrospective mortality and medical surveillance studies of workers in benzene areas of refineries, J. Occirp. M d ,25, 685-692 V a n I h l t e , H.G.S.. Grusso P.& Irvinc, D. (1984)'Tackling a very dillicult problem. Letter lo the Editor. Risk h i d . , 4, 1-8 Viiinna. N.J. & Polan. A. (1979) Lymphomas and occupational benzene exposure. L w w ! . 1394- I395 Vigliani, E.C. (1976) Lcukemia associated with beiizene exposure. Ann. N . Y. Acad. Sci.. 271, I43 -5 I Vigliani, E.C. & Saita, G. (1964) Benzene and leukemia. New Eng. J. Med., 272, 872-876 Wukinabe.T.. Endo. A.. Kato, Y.. Shima, S., Watanabe, T. & Ikeda, M. (1980) Cytogciietics wid cytokinetics of cultured lymphocytes from benzene-exposed workers. h t . Ardi. Owrp. Enviroir. H d h , 46, 31-41 Whilc. M.. Infante. P,& Chu. K. (1982)A quantitative estimate ofleukemia mortality iissociated with occupational exposure to benzene. Risk A n d , 2, 195-203 Wilcosky. T.C., Chcckoway. H.. Marshall, E.G. and Tyroler, H.A. (1984) Cancer mortality and solvent exposures in the rubber industry. Am. Ind. Hyg. Assoc. J., 45, 809-81 I Woll'. I'.I-l.. Aiidjelkovich, 11.. Smith. A. & Tyroler, H. (1981) A case-control study 01' Icukcmia in the U.S. rubber industry. J . Occ-rrl~Med., 23, 103-108 Wong. 0.( 1983) Ail iiitlirstr)~-~r~ii~oir~tdIi~tj~study (!/'chemicalworkers occupationally c ~ . v / ~ o s cto~ l h a i x n c . Wiishington. D.C.. Chemical Manufacturers Association, December, 1983 W H O ( 198 I ) ~ii~~iroiiiiic~itiIieNcIilili Crilericr 18, Arsenic. Geneva, World Health Organization. pp. 144-145 ?i? 18 McMlCHAEL . APPENDIX (addcd in proof) b l'hc cxtcndcd rollow-up or thc NlOSH cohort (Rinsky et ul.. 1985) has now bccn published (Rinsky et id.. 1987)'. Thcsc data appmr to providc a bcttcr basis for quantitativc risk assessment than do carlicr studic?r.Thc authors cstimatcd thc cumulativc cxposurc or individual workcrs to bcnzenc by usc or u jobcxposurc matrix to link dctailcdjob historics with industrial hygicnc mcasurcmcnl (historical air sampling ofbcn7cnc conccntrations). Thc ovcrall SMR for lcukaemid (9 dcaths) in bcn7~nc-cxposcdwarkcrs was 337 (95% confidence intcrval. 154-641). and that for multiple mycloma (4 dcaths) was 409 ( I 10-1047). For Icukacmio. u dosc-rcsponscrclutionship ww cvidcnt, with SMRs incrcasing from IO9 (2 .ascs.) to 322 (2 cam). I186 (2 mscs) ind 6637 (3 as&),with i n c r a m in cumulative bcnzenc cxposurc from lcss thitll 40ppm-ycars. to 40-199. 200-399 and 400 or morc. rcspcctivcly (Soappm-ycurs is cquivtllenl LO 11 mciln irnnuul cxpsurc: of IO ppm ovcr a 40-yar working lifctimc). Thc authors concludcd that an cxponcntial function b c s ~dcscribcd thc dow-rcsponsc rclationship, and that il substantiirl dccrcusc in thc risk for dcuth rrom lcukacmia could thcrcforc bc achicvcd by lowcring cK.cupiltil>nidcxposurc to bcnxnc. including (importantly)cxposurc within thc riingc IO ppm-0. I ppni.