Document 5b0G0Z7j7moyBZydre9MObkxD

I OCCUPATIONAL EXPOSURE AND BRAIN TUMORS Andrew R, Moss University of California. San Francisco, California Epidemiological evidence of an occupational risk of brain cancer has been reported in four industries where chemical exposures are likely, most recently in a senes of prospective studies in the petrochemical industry. However, only in the esse of vinyl chloride exposure has an occupational centra/ nervous system carcinogen been iden tified. This report reviews the convergence of epidemiological and laboratory evi dence that established the occupational carcinogenicity of vinyl chloride, and dis cusses in detail the current evidence for an occupational risk of brain turnon in (he petrochemical industry. INTRODUCTION Primary brain tumors of the glioma series, which make up about half of all brain tumors, show a male-to-female ratio of about 1.5 to 1, Brain tumors in general appear to be increasing in incidence among older cohorts (Waxweller et al., 1983). Thus brain tumors In general, and glioma series tumors in particular, are sometimes considered as a priori likely suspects in the search for occupationally related tumors. Furthermore, gliomas can be produced experimentally in rats by at least four groups or experimental chemicals: aromatic hydrocarbons, N-nitroso compounds, triazenes,and hydrazines. More recently, brain tu mors including gliomas have been shown to be produced by inhalation in rats by three industrial chemicals, bis(chioromethyl) ether, vinyl chloride, and acrylonitrile (Maltoni et al., 1982). These experimental results, together with the established history of brain-tumor carcino genesis in vinyl chloride workers, have led many researchers to lend at least preliminary credence to reports of brain tumors occurring in other industries where there are chemical exposures. Most recently, a series of studies have suggested excess risk of brain tumors in the petrochemical industry. Evidence suggesting an excess of brain tumors has now been re ported for at least four occupational groups: rubber workers (begin ning with Mancuso's investigations in the 1950s), chemists, vinyl chlo- This research was supported in part by grant 1 R01 OH 01557 (ram the National Institute tor Occupational Sa/ety and Health. Requests for reprints should be sent to Andrew R. Moss. Epidemiology and International Health, UCSF, San Francisco General Hospital, 995 Potrero Avenue. Sldg. SO. Ward 84. San Fran* daco, California 94110. 703 Journal of Toxicology and Environmental Health 18:703- 711, 19BS Copyright t I98S by Hemisphere Publishing Corporation AP00019169 . 704 A. R. MOSS ride workers, and petrochemical workers. However, the convergence of epidemiological and laboratory investigations that is usually ac cepted as the informal criterion of proof in such situations has been hard to demonstrate (see, e.g., Cole and Merietti, 1980; Doll and Peto, 1981). Thus, Mancuso's original observations in the rubber industry were not followed by the identification of any specific carcinogen, and the existence of an occupational risk of brain tumors in rubber workers is currently open to doubt (Mancuso, 1982; Symons et al., 1982). Braintumor risk in chemists has been shown only in a series of Swedish studies, and as yet no likely carcinogen has been identified (Ofin and Ahlbom, 1982). The convergence of an epfdemiologically Identified risk and a suspected carcinogen has been demonstrated only among vinyl chloride workers. Thus Cole, in his surveys of occupational carcino genesis, recognizes only vinyl chloride as an established industrial brain carcinogen (Cole and Merietti, 1980). Doll and Peto, more con servatively, admit the brain only as a ''possible'' site of occupational carcinogenesis (Doll and Peto, 1981). The history of vinyl chloride as a central nervous system (CNS) carcinogen Is discussed in the next sec tion. Historically, the establishing of occupational or environmental risks has often been a complex and drawn-out process, sometimes requiring decades for the development of a consensus on a particular exposure. Thus although the issue of excess risks of brain tumors in rubber workers is currently in debate, the persistence of reports of brain tumor excesses in this and other industries has, as it were, promoted brain tumors to the top of the list of tumors that may be associated with important occupational risk (along with lung cancer, leukemias and lymphomas, primary liver cancer, and, more recently, melanoma). This was the situation in 1978 when a cluster of brain tumors was reported in petrochemical workers in Texas, initiating a major wave of research into the subject. The petrochemical research is also reviewed in this article. VINYL CHLORIDE AS A BRAIN CARCINOGEN Cole observes that occupational carcinogens are usually recognized because they generate an unusually high rate among a small exposed population, and because they produce tumors that are otherwise un usual (Cole and Goldman, 1975). Vinyl chloride (VC) is the classic case of a carcinogen that has such an effect, producing in workers exposed to high levels of the gas the very rare angiosarcoma of the liver (ASU. The first reported case of A5L in a vinyl chloride worker was diagnosed at the B. f. Goodrich company in Louisville, Kentucky, in May, 1970; the second in March, 1973; and the third, at autopsy, in December, 1973 (Heath et al., 1976). Since the incidence of ASLin the United States AP00019I OCCUPATIONAL EXPOSURE AND BRAIN TUMORS 705 was then 20-25 cases/yr, the appearance of 3 cases among a small group of workers prompted considerable concern. Furthermore, concurrent studies by Viola (1971) had shown skin, lung, and bone tumors in rats exposed to VC by inhalation. [Viola's carcinogenesis studies were re portedly an outgrowth of his work on acroosteolytis in VC-exposed workers. Hepatitis-like liver changes had been reported as early as 1949 (Nicholson, 1977).]. Following Viola, Maltoni demonstrated angiosarcomas in rats ex posed to VC in 1972 {Maltoni etal., 1982). This result was communicated to the Manufacturing Chemists Association, which following the dis covery of the third case at B. F. Goodrich, made public the finding of three cases of angiosarcoma in VC-exposed workers in the Wall Street journal. With this announcement, a major effort began to study vinyl chloride workers, including a National Institute of Occupational Safety and Health (NlOSH) follow-up study of 1294 workers with extensive VC exposure in four plants. A concurrent histopathology study reviewed tumors reported on death certificates of all workers in the four plants. The NlOSH study, published in 1976, showed very high relative risks in the category `'biliary and liver cancer." The relative risk was 11 for workers with 10 or more years of exposure and 16 for workers with 15 or more years of exposure. Furthermore in the concurrent histopa thology study, 11 of 14 reported liver and biliary cancer cases on death certificates were ASL (Waxweiler et al., 1976). These observations have remained the primary epidemiological basis for the assertion of the carcinogenicity of vinyl chloride. However the study also observed a significantly increased relative risk for brain tumors, with 3 observed and 0.6 expected among workers exposed more than 15 years. At the same time Maltoni (1976a) reported that brain tumors as well as ASL could be produced in rats exposed to VC by inhalation. Furthermore the NlOSH histological study showed that 9 of 10 brain-cancer deaths in the VC worker cohort were glioblas toma multiformes (the most commonly occurring tumor in the glioma series). This proportion was thought to be unusually high. (However, it has been shown that the expected proportion of glioma-series tumors in such a series depends heavily on the number diagnosed at autopsy (Schoenberg et a)., 1978)1. These results were sufficient for the deveiopment of a reasonable consensus that brain tumors as well as ASL were associated with VC exposure. The primary evidence in the devel opment of the consensus was the correspondence between the human data and the laboratory inhalation studies, in which the exposure levels were thought to be close to human occupational exposures, (see, e.g.. Wagoner and Infante. 1977). Vinyl chloride is the only occupational chemical carcinogen for which a correspondence exists. Waxweiier et ai. (1983) also note that in the 9 glioblastoma multi forme cases identified in the VC workers histology study, the average 706 A.R. MOSS time since first exposure to vinyl chloride was 21 yr. Thus the epide miological observation of risk may have followed the oncogenic expo sure by a generation, and the circumstance of exposure could well have changed by the time the risk was recognized. This long latency period is characteristic of occupational carcinogenesis. The excess risk of mesothelioma in asbestos workers, for example is not generally ob servable until 20 yr after exposure (Selikoff, 1977). The long latency pe riod has important consequences >n the case of brain tumors, which are not rare tumors. The identification of VC-produced brarn tumors, currently the only accepted example of occupational CNS carcinogenesis, offers some important lessons for studies of other industries. First, vinyl chloride was identified as a CNS carcinogen because it produces, as its primary effect, the extremely rare ASL. The excess risk of brain tumors would probably not have been recognized if it had not "piggybacked" on the rarer tumor. Second, the brain was only recognized as an additional site of carcinogenesis because comparable ranges of tumors were seen in human and animal inhalation studies. There is no other occupational carcinogen for which such a correspondence is currently known. Third, the number of tumors reported in the primary study was very small: there were 3 cases in the high-exposure cohort, and 10 in the concur rent histopathology study. Fourth, the period between initial exposure and observation was so long that observed effects may in fact have been archaeology rather than epidemiology--the carcinogenic situa tion may well have no longer existed when the excess risk was ob served. BRAIN TUMORS AND PETROCHEMICAL EXPOSURE The intensive study of brain tumors in the petrochemical industry began in 1978 when as employee at the Union Carbide plant in Texas City, Texas, reported a newly diagnosed brain tumor to the local office of the Occupational Safety and Health Administration (OSHA). A joint study with the Oil, Chemical and Atomic Workers union (OCAVV) and company representatives identified a total of 10 brain-tumor deaths and cases among workers at the plant (not including the original com plainant, whose tumor proved to be metastatic). Inis number was suf ficient for the OSHA office to request a formal study by NIOSH. A joint study, with Union Carbide researchers, was initiated. A proportional mortality study of OCAW members in South Texas was already in prog ress, and additional studies in other refineries and chemical plants were initiated over the next year. However, in 1979 both the Three Mile Island incident and the Fed eral declaration of emergency status for the Love Canal area took place. AP00019172 OCCUPATIONAL EXPOSURE AND BRAIN TUMORS "07 and in the ensuing controversies over environmental carcinogenesis the atmosphere became polarized between industry and govern mental-union forces. As a result, competing industry and govern ment-union studies were set up in several of the exposed petrochem ical cohorts in Texas and Louisiana. When a meeting was convened at the New York Academy of Sciences in 1980 to review the studies, a total of eight were presented, four from each group. The four govern mental-union studies uniformly indicated a twofold risk of brain cancer among petrochemical workers; the four industry studies uni formly failed to detect any excess risk. "To the innocent observer/' Science reported, "the crisp pattern of falling chips appeared at first sight to display blatant gerrymandering" (Lewin, 1980). It was also true, however, that the range of methods and cohorts involved was consid erable and that there was extensive methodological disagreement. Reeve et al. (1985) reviewed the issues. The primary methodological questions were (1) the tendency of proportional mortality studies, as carried out by NIOSH, to inflate cancer risk, and (2) the tendency of industry-wide cohort studies, as carried out by the industry groups, to bury smaii groups of at-risk workers in very large cohorts. These issues were not resolved as, with the new Federal administration in 1981, NIOSH capacity to continue the epidemiology studies was lost. With no pressure from the regulatory side, the industry studies appear to have slowed down also. However, most of the proportional mortality studies were eventu ally converted to standardized mortality form, and a review of these studies is now possible. Savitz and Moure (1984) have recently reviewed the six prospective studies of refinery workers that were initiated by labor-governmental and industry groups in 1979-1980. Table I is adapted from Savitz and Moure, with the addition of results from two studies of petrochemical plant workers, those of Waxweiier et al. (1983) and Reeve etal. (1983). [One study, that of Thomas etal. (1982); remains a proportional mortality rate study (PMR): the others use standardized mortality ratios.1 All cohorts of petrochemical workers had low overall mortality, as is generally the case in studies of "healthy workers," and six ot the seven for which all-site cancer mortality or incidence were reported showed lowered total cancer rates. [The exception was the PMR study of Thomas et al. (1982).] However, brain-tumor rates were elevated in all but two of the studies, the exceptions being the rofinery-wide study of Wen et al. (1983), which included salaried workers, and the industry wide study of the British petrochemical industry by Rushton and Al* derson (1981). There was additional evidence of an association in more detailed substudies. Thus Savitz and Moure (1984) note that while Hants et al. (1982) report an overall relative risk of 1.02 for brain tumors in their study of the Baton Rouge Exxon refineries, a more highly exposed roa A. ft. MOSS TABLE 1. Relative Risk for Brain Cancer and Melanoma in Prospective Epidemiological Studies at Refinery and Petrochemical Workers** All cancer* Study RR /V Theriault and Couiei, (1979) Rusbton and Alderson, (1981) Schoitenfeld et al. (1981) (mortality) Schottenfeld et al. (i98i> (incidence) Thoma* et al. (1982) (PMR) Hanfi et al. (1982) Wen et ai. (1983) Waxweiler et al. (1963) Reeve et ai. (1983) 0.49 0.89 0.76 0.96 1.19 0.92 0.96 0.81 N(A 23 1147 127 240 474 249 839 131 N/A 1 Adapted from Savitz and Moure <19341. * RR, relative risk; V, number. Brain HR Melanoma RR ,V 3.90 0.80 1.63 1.29 2.28 1.02 D.99 1.81 1.09 3 36 2.16 8 9 1.32 27 1.61 S 30 - 1.22 13 25 -- 14 13 11 16 -- subgroup had a higher risk, in the Reeve et ai. study of Dow chemical workers, while the overall relative risk was 1.09, the risk among workers hired before 1945 was 1.83 (Reeve et al., 1983). On the basis of the first six studies, Savitz and Moore comment that there was reasonably consistent evidence of a potential association between petrochemical exposure and brain cancer. The two chemicalplant studies appear to strengthen the association. In an independent review of all NIOSH investigations in eight in dividual refineries and chemical plants in Texas, Reeve et al. (1983) note that inconsistencies remain in the plant-by-plant results, and that casecontrol studies within the studied cohorts show no consistent patterns of exposure. However the consistently elevated brain-tumor risk in the prospective studies appears sufficient at least for this line of research to continue. Savitz and Moure note that melanoma risk is rather consistenrfv elevated in the studies of petrochemical cohorts, as well as brain-tumor risk. This result is of interest because of an emerging pattern in recent studies of nuclear fabrication workers. Hadjimichael et at. (1983), in their study of the United Nuclear facility in Connecticut, observed rel ative risks of 2.40 for brain tumors and 2.12 for melanoma, against an AP00019174 OCCUPATIONAL EXPOSURE AND BRAIN TUMORS 709 all-sites relative risk of 0.88. Furthermore, Wilkinson et a). (1983) ob served a tripled risk of brain tumors at the Rocky Flats nuclear plant, and Austin (1981) observed a tripled risk of melanoma at the Lawrence Livermore laboratory. Attempts to explain these risks by radiation ex posures have proved generally unsuccessful; the similarity with the pattern of risk among petrochemical workers suggests that perhaps the cause is in some other aspect of the fabrication process. CONCLUSION Although this review suggests that the issue of brain tumors in pet rochemical workers is at least worth further study, most of the pro spective studies appear to have been dropped. However, a second wave of large case-control studies of occupational exposure in brain tumors has now been set in motion (see Table 2). In these studies, all brain-tumor cases or deaths in a specific case-finding area over a spe cific period are interviewed, either in person or by proxy. These large studies may eventually resolve the issue of whether there is a genuine TABLE 2. Case-Control Studies Currently in Progress of Occupational Exposure in Glioma Patients investigator Thomas, T. L., NCI Environmental Epidemiology Branch Preston-Martin, S.( University of Southern California Moss, A. R., University of California, San Francisco Buffier, ?., University of Texas Ahlbom, A., Hudinge University Hospital, Sweden Musicco, M.. Neurological Institute C. fiesta. Milan Site CNS Number of Number of cases controls 600 600 CNS 300 300 Glioma 500 500 Glioma Brain 650 870 150 300 Meningioma/ glioma HQO 1,600 Source of cases Death certificate Location New Jersey/ Louisiana Incidence Los Angeles County UC5F clinical cases and death certificates Northern California Incidence Coastal Texas and Louisiana Clinical cases in 2 hospitals Hundinge Hospital cases Milan * From Muir and Wagner 11982) and P. Buffler, personal communication AP00019175 7 TO A. ft. MOSS excess risk in petrochemical workers. [However, Cole notes that such large case-control studies are only effective when the exposed group makes up at least 2-3% of the population studied (Cole and Goldman, 1975).] It should also be noted that while large case-control studies of brain tumors may identify industrywide risk, they cannot, by their na ture, identify the pattern of tumor sites associated with a specific ex posure. This latter information is generally accepted as important in establishing risk. In conclusion, the prospective studies of petrochemical worker co horts reviewed here suggest that there is an increased risk of brain tumors associated with petrochemical exposure, possibly accompanied by an increased risk of melanoma. No evidence of a specific carcinogen has been established, and while the current generation of large casecontrol studies will probably identify any overall risk in the industry, these studies will not identify the pattern of sites at risk. In the light of the consistency between the petrochemical worker studies and the recent nuclear-fabrication worker studies, there is an argument for a coherent program of prospective research in these cohorts. Other studies that could usefully be coordinated with such a program include a review of the specific-exposure inquiries in the petrochemical worker cohorts, an examination of the change in the incidence of glioma series tumors among all brain tumors, and an exploration of other associa tions between brain tumors and melanoma. REFERENCES Austin, 0. F. 1981. Population-based tumor registries in the identification of occupational carcin ogens. Proceedings of the 1980 International Symposium on Cancer. Sept. 14-18, I960. In Career Achievements, Challenges and Prospects for the 1980s. eds. |. H.BurchertalandH. F. Oettgen, vol. 1. pp. Z91-298. New York: Crune and Stratton. Cole. P-. and Goldman. M. 8. W7S. Occupation, in Persons at High Risk or Concen An Approach to Cancer friotogy and Control, ed. L F. Frawment, pp. 167-184. New York: Academe. Cote, P.. and Merletti. F. 19. Chemical agents and occupationat cancer. In Cancer and the Bnvironment, eds. H. 8. Demapouies and M. A. Mehlman, pp. 399-4(7. Park Forest South. It.: Pathotox. Doll. ft., and Peto, ft. 1991. The Causes of Cancer, p. 1243. Oxford: Oxford University Press. Hadjimichaet, O. C.. Ostteid. a. m., D'AtH. 0. A., and Brubaker. R. E. 198i. Mnrulitv and cancer incidence experience ot employees in j nuclear fuels labrication plant, /. Occup. Med. Hams, N. M., Holmes. I. M.. Shatlenberger. L. G-. and (ones. K. . 1992, Epidemiologic studs oi refinsrv .md chemical plants workers. /. Ocrup. Med. 24:203-212- Heath. C. N.. and Falk, H. 1*176, Cfuracttvistict nt cases tw angiosarcoma ot the liver among uiM chloride svnrfcerx in the United States. Ann. ,V. Y. Acad. So. 267:231 -2Jf. Lewin, M. tVfto. Research News: Ciovernmem'mriustry dispute brain tumor risk. Science 210:906997. NUMoni. C. 19*r>a. Predictive value of cjrcirtoqenoMs biuassavs. Ann. N.Y. ,\cad. Sei. 271:431-441. Madont. C. W76b. Precursor lesions in exposed populations as indicators ot occupational cancer risk. Ann. N.Y. Ae.id. SCJ. 271:444-447. Maltom, C.. Cifiberti. A., and Caretti. D. 1982. experimental contributions in identifying brain fxitenlial carcinogens to the petrochemical industry. Ann. N.y. Acad. Nti. 181:216-249. OCCUPATIONAL EXPOSURE ANO BRAIN TUMORS 711 Mancuso. T. F. 1982. Epidemiological study of tumors of the central nervous system in Ohio. Ann. N.Y. Acad. Sc/. 381:17-39. Muir. C. S., and Wagner. G. eds. 1982. Directory of On.Coing Research in Cancer Epidemiology 1982. Heidelberg. Federal Republic of Germany: Laub. Nicholson, W. 1977. Cancer following occupational exposure to asbestor and vinyl chloride. Cancer 39:1792-1801. Oltn. G. R.. and Ahlbom, A. 1982. Cancer mortality among three Swedish male academic cohorts: Chemists, architects, and mining engineers/metallurgists, Ann. N.Y. Acad. Set. 381:197-201. Reeve, G. R.. Bond. G. G., Lloyd. I. W.. Cook. R. R.. Waxweiler. R. and Fishbeck, \V. A. 1983. An investigation of brain tumors among chemical plant employees using a sample-based cohort method. /. Occup. Med. 25:387-396. Reeve. C. R,, Haiperin, W. E.. and Landngan. P. (. 1985- Brain tumor investigations in Texas. Paper presented at Workshop on Cancer Prevention, sponsored by Texas State department ot Health, January 25-26. 1984. San Antonio, Texas. (Cushion, L, and Alderson, M. R. 1981. An epidemiological survey of eight oil refineries in Britain. Br.). Ind. Med. 38:225-234. Savitz. D. A., and Moure, R. 1984. Cancer risk among oil refinery workers. A Occup. Afed- 26:bb267Q. Schoenberg, B. S., Christine. B. W., and Whisnant. J. P. 1978. The resolution of discrepancies in the reported incidence of primary brain tumors. Neurology 28:817-823. Schottenfeld. D.. Warshauer. M. E.. Zauber. A. G., Meikle, |. C.. and Hart, B. R. 1981. A prospec tive study of morbidity and mortality in petroleum industry employees in the United StatesA preliminary report. Banburry Report No. 9, Cold Spring Harbor Laboratory. In Quanrincat/on of Occupational Cancer, eds. R. Peto and M. Schneiderman. pp. 247-265. Cold Spring Harbor. N.V.: Cold Spring Harbor Laboratory. Seltkoff. I. I. 1977. Cancer risk of asbestos exposure. In Origins of Human Cancer--Book C-- Human Risk Assessment, eds. H. H. Hiatt. J. O. Watson, and J. A. Winsten, pp. 1763-1784. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory. Symons. M. F., Andjelkovich, O, A., Spirtas. R., and Kerman, O. R. 1982. Brain and central nervous system cancer mortality in U.S. rubber workers. Ann. N.Y. Acad. Scl. 381:146-159. Theriault. G.. and Goulet. L. 1979. A mortality study of oil refinery workers. A Occup. Med. 21:367370. Thomas, T. L., Waxweiler. R. J., Moure-Eraio. R.. Itaya. S. and Fraumeni |r,, J. F. 1982."Mortality patterns among workers in three Texas oil refineries. A Occup. Med. 24:135-tJl. Viola, P. L-, Bigotti. A., and Cuputo, A. 1971. Oncogenic response of rat skin, lungs, and bones to vinyl chloride. Cancer Res. 31:516-322. Wagoner, j. K., and infante, P. F. 1977. Vinyl chloride: A case for the use of laboratory bioassav in the regulatory control procedure. In Origins of Human Cancer--Book C-- Human Risk Assessment, eds. H. H. Hiatt, ). D. Watson, and I. A. Winsten. pp. 1797-1805. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory. Waxweiler, R. J., Stringer, W,, Wagoner, J. K.. lones, I,. Falk, H., and Carter, C. 1976. Neoplastic risk among workers exposed to vinyl chloride. Ann. N.Y. Acad. Sci. 271:40-48. Waxweiler. K. I., Alexander. V., Lertingwell. S. 5.. Haring. M., and Lloyd. |. W. idSj. Mortality from brain tumor and other causes in a cohort or petrochemical workers, A Natl. Cancer Inst. 70:75-81. Won. C. P., Tsai. S. P.. McClellan. W. A., and Gibson, R. L. 1983. Long-term mortality study oi oil refinery workers: I. Mortality oi hourly and salaried workers. Am. A Epidemiol. 118:526->42. Wilkinson. G. 5., Voell. C. L.. Acquavella, I. F. 198). Mnrialitv among plutonium and other workers at a.nuclear tocifiiv. Proceedings ot the Midyear Topical Meeting'oi the Health Phvsics Society. In Epidemiology Applied to Health Physics. CONf-8)01C1. tan. 9-13. 1*183. pp.328-337.