1 17 Mesothelioma and Asbestos Exposure J. Corbett McDonald and Alison McDonald Historical Background It was the Conference on the Biological Effects of Asbestos at the New York Academy of Sciences, organized by Irving Selikoff in November 1964 (1), that put both mesothelioma and asbestos on the map. Before that meeting, few people in the scientific or general community had much knowledge of either subject. There they learned the nature and numerous essential industrial uses of a group of naturally occurring mineral fibers, collectively known as asbestos, although in fact comprising at least five distinct materials, chemically, physically, and geologically. Of these, chrysotile, a serpentine mineral mined mainly in Quebec and the Ural mountains of Russia, made up over 90%. Of the remainder the two most important were crocidolite and amosite, produced mainly in South Africa and Australia, both amphibole minerals with distinctive qualities valuable for heat insulation, naval marine use, and the production of large-bore cement pipes. Two other amphibole mineral fibers were anthophyllite, of limited production in Finland, and tremolite, little used, though by far the most widespread geologically. Presenters at the conference stated that within some 20 years of the first industrial exploitation of asbestos in the 1880s, workers heavily exposed to airborne fiber and dust developed a distinctive, seriously disabling and sometimes fatal diffuse pulmonary fibrosis, later termed asbestosis. Little was done to limit exposure until the late 1930s, when after a well-conducted survey of four asbestos textile plants in North Carolina, Dreessen et al (2) and others of the U.S. Public Health Service recommended in 1938 that a workplace dust concentration of 5 million particles per cubic foot (about 15 fibers/ml) should not be exceeded. Mainly because of the Second World War, this recommendation was not implemented; and probably for the same reason it went unnoticed that there were case reports by some German pathologists (3) of malignant tumors of the pleura and peritoneum in men with asbestosis. Thus it was only in the 1950s that the causal association of asbestos exposure with lung cancer in the United Kingdom (4), and later with mesothelioma in South Africa (5), was recognized. 267
2 268 Chapter 17 Mesothelioma and Asbestos Exposure Until that time even the very existence of primary malignancies of the mesotheleum was questioned by reputable pathologists. Looking back, however, a review by Saccone and Coblenz (6) in 1943 had included the identification of over 40 cases in autopsies published since 1870, and referred to two cases of endothelioma reported in 1767 by Lieutaud in France among 3000 autopsies. That mesothelial cancers in low frequency probably occurred well before the industrial use of asbestos is discussed more fully later. Indeed, a low background incidence of unknown etiology has almost certainly continued, affecting both children and adults. The Link with Asbestos Two key events were undoubtedly responsible for organizing the New York Conference. The first was the report in 1960 by Wagner et al (5) of 33 cases of pleural mesothelioma in the northwest Cape Province of South Africa, 28 of which occurred in persons who had either worked in the crocidolite mines or lived close to them, even as children. The second was a study by Hammond et al (7) of 307 deaths in a small cohort of 632 New York insulation workers, including four from pleural and six from peritoneal mesothelioma. At the end of the conference, an expert working group of the International Union Against Cancer (UICC), after review of the evidence presented, made a number of recommendations for epidemiologic studies, in particular into the importance of asbestos fiber type, especially in mining industries rather than in manufacturing, to avoid the complication of exposure to mixtures (8). At the request of the Canadian Federal Government, and with support from the Research Institute of the Quebec Asbestos Mining Association, a comprehensive university-based scientific program was established almost at once (9). Elsewhere, only in the chrysotile mine of Belangero in northern Italy was anything done to implement these recommendations until over 20 years later. As a result, the lack of information on amphibole exposure, comparable to that for chrysotile, greatly delayed our understanding of mineral fiber carcinogenicity. Early Case-Control Studies Data from two important case-control studies in the United Kingdom were presented at the New York Conference and published more fully the following year. The first of these was by Elmes et al (10), who studied 42 male cases of pleural mesothelioma and 42 closely matched controls in Belfast, Northern Ireland, with detailed work histories taken from those still living or from relatives of those who had died. Thirtyseven cases had been exposed to asbestos, mainly in plumbing, insulation, and shipyard work, not all heavily and without mention of fiber type, compared with nine controls. The second study, by Newhouse and Thompson (11), was of 83 cases, half in men and half in women, who had died between 1917 and 1950, 56 from pleural and 27 from peritoneal mesothelial tumors, close to the large Cape Asbestos
3 J.C. McDonald and A. McDonald 269 Table Early case-control studies of mesothelioma giving definite or probable occupational exposure to asbestos First author Years Male Occupationally Relative (reference) Year Place diagnosed Cases/controls (%) exposed (%) risk Elmes (10) 1965 Belfast / Newhouse (11) 1965 London / McEwen (12) 1970 Scotland / McDonald (13) 1970 Canada / Rubino (14) 1972 Piedmont / Ashcroft (15) 1973 Tyneside / Hain (16) 1974 Hamburg / Zielhuis (17) 1975 Netherlands / Source: Based on Table 1 in McDonald and McDonald (3). Company s factory that opened in 1913 in London s East End. Their control series comprised patients admitted later to the hospital with other diseases, matched for sex and age. The authors acknowledged that neither these control measures nor the interview methods were ideal, but concluded that the case-control comparisons of occupational and residential histories were probably valid. Of 76 pairs, 18 cases (24%) had been employed at the Cape Asbestos factory, five (7%) at other asbestos plants, and eight (11%) as insulators or laggers, compared with one (1%) and four (5%) controls respectively. A further nine cases (12%) were in persons who had lived in the same house as an asbestos worker and were indirectly exposed, compared with one control (1%). Only crocidolite from South Africa was used in the Cape Asbestos factory until 1926, when small quantities of chrysotile and amosite were introduced. During the next few years, six more case-control studies were published in addition to the two studies cited above, all with a substantially increased relative risk associated with occupational asbestos exposure (Table 17.1). Little attention was given to fiber type explicitly, but it can be seen that five of the eight studies were in shipyard areas where amphibole use was common. Rather different from the other seven, and perhaps more generally informative, was the study across Canada by McDonald et al (13). This entailed an approach to all 423 of the country s pathologists, whereby 165 known deaths from mesothelioma in 1959 to 1968, diagnosed by autopsy or biopsy, were registered (i.e., 1 per million population per annum). Of this total, 65% were in men; 70% were pleural, 27% peritoneal, and 3% pericardial. Detailed occupational and residential histories of exposure to asbestos and six other materials used industrially were obtained blind from relatives and friends in 90% of the cases, and from two matched case-control series, one of primary and one of secondary lung cancer, selected from the same autopsy records. An association with definite or probable occupational exposure to asbestos was clearly demonstrated indeed with the highest relative risk (7.0) of all eight case-control studies but only 20% of men and one woman had any such contact. Almost all the excess was in the manufacturing and industrial application of asbestos, rather than in mining or milling. No association was found with lesser
4 270 Chapter 17 Mesothelioma and Asbestos Exposure degrees of occupational exposure or residence in asbestos-mining areas, but there was a small excess of possible domestic exposures. The smoking histories in the mesothelial tumor and male control groups were almost identical, and considerably lower than those in cases of primary lung cancer, implying that unlike asbestos-related lung cancer, smoking did not contribute to this disease. In 1972 the survey was repeated, with extension to all pathologists in North America (some 7000 of them), almost all of whom agreed to contribute cases identified at autopsy or biopsy for the one year only (18). A control subject with death from metastatic lung disease from a primary tumor outside the chest, matched for date, sex, and age, was selected from the same pathology file as the case. Relatives were interviewed usually by a public health nurse who was not informed about the case-control status, and detailed residential and occupational histories were recorded. Jobs were also coded blind, using a list compiled by four expert international groups, according to the probability of asbestos exposure. Of 344 male cases of mesothelioma, 188 (55%), compared with 78 (23%) controls, fell into one of the five exposed categories: insulation work, an infrequent occupation in controls, had the highest relative risk (46.1); asbestos production and manufacture were next in line (6.1), followed by employment in heating trades (excluding insulation work), shipyards, and construction (3.4). Subjects from this survey were subsequently used as the base in a case-control analysis of lung burden with the fibers identified and concentrations estimated by electron microscopy. The results of this study are summarized later, together with those from other lung burden analyses. In the Canadian surveys, 1960 to 1966, the average annual incidence was one case per million persons about 1.5 for males and 0.8 for females; however, there may have been underreporting during these early years. In 1966 to 1972, the incidence in Canada was 2.9 per million males and 1.4 per million females, and in the United States in 1972 the corresponding rates were 2.7 and 0.8 per million. These estimates were used in 1975 in a geographical analysis of all known cases of mesothelioma worldwide in areas where reported cases could be linked to population estimates. By applying age- and sex-specific rates found in Canada, the number of mesotheliomas expected on this basis was compared with the numbers observed (19). High observed-to-expected ratios were found in many European shipyard cities, notably Walcheren, the Netherlands (23.3), Wilhelmshaven, Germany (21.5), and Plymouth, UK (14.3). In two locations with large asbestos product manufacturing industries, there were also high ratios: Dresden, Germany (16.8) and the Manville-Somerville area of New Jersey (26.5). In the early 1970s, mesothelioma mortality in North America was already two or three times higher in males than females. This pattern soon became apparent in most industrialized countries and was followed by a steady upward trend in male mortality, which still continues. The steep rise in males, which probably began in the 1940s, reflects a parallel increase in the industrial use of asbestos, from about 1910,
5 having taken account of a 30- to 40-year latency. As a result of this steady increase, mesothelioma is now responsible for some 20 deaths per million male population in Western Europe and North America compared with an estimated 1 to 2 per million 30 to 40 years ago. In early studies, only a minority of male cases were attributable to occupational exposure, whereas now up to 90% are. J.C. McDonald and A. McDonald 271 Occupational Risk Apart from the early case-control studies just described, the main body of information on risks from exposure to airborne asbestos fibers in the workplace is contained in over 40 historical cohort mortality studies reported during the past 30 years. These studies have varied in quality and in the extent to which exposure has been assessed in duration, intensity, or asbestos fiber type. In fewer than 10 cohorts had there been any attempt to estimate exposure intensity for each cohort member, and in very few cohorts indeed was there exposure to only one type of asbestos. These serious deficiencies, given the potentially great difference in carcinogenicity between chrysotile and the amphiboles, render almost uninterpretable the results of the many cohorts exposed to mixtures where one type, usually chrysotile, was said to predominate. Despite this lack of specificity, cohort surveys, by being prospective, have provided more useful information on occupational risk than was usually obtained from retrospective case-referent studies. This is more true, however, of lung cancer and nonmalignant respiratory disease (NMRD) than of mesothelioma. Standardized mortality ratios (SMRs) can usually be calculated by age, sex, and exposure variables for the former but not the latter, as even now mortality rates for mesothelioma in the general population are seldom available, let alone reliable. Investigators have had to fall back instead on measures of proportional mortality, though these depend enormously on age composition of the cohort and length of follow-up. For example, in the large cohort of Quebec chrysotile miners and millers, described more fully below, the proportion of all deaths ascribed to mesothelioma rose steadily from 3 of 2413 (0.12%) in 1966 to 38 of 8009 (0.47%) in 1992 (20). Even so, in the absence of any better index, proportional rates interpreted with care can be useful. Thus, in Table 17.2, where the salient findings from the 43 main cohort mortality studies published before 1999 are summarized, the differences in proportional mortality for mesothelioma are large and systematic. This is especially so when differences between chrysotile and amphibole exposure within the same industrial sector are considered in more detail. The largest and most complete of the mining cohorts was of all 10,918 men born in 1891 to 1920 who, in 1966, had served for 1 month or more in the Quebec chrysotile production industry, either as miners or millers in the town of Asbestos or region of Thetford Mines, or in a small asbestos products factory (23). Excluding losses, almost all before 1935 and in men with very short employment, 8009 (82%) of 9780
6 272 Chapter 17 Mesothelioma and Asbestos Exposure Table Cohort mortality studies of asbestos-exposed workers Deaths All First author No. of causes Lung cancer Mesothelioma (reference) Year Country Subjects n % n SMR Excess PMR/1000 n PMR/1000 Relevant exposure Mining and milling Piolatto (22) 1990 Italy Liddell (23) 1997 Canada Chrysotile Meurman (24) 1974 Finland Brown (25) 1979 U.S McDonald (26) 1986 U.S Armstrong (27) 1988 Australia Various amphiboles Sluis-Cremer (28) 1992 South Africa Sluis-Cremer (28) 1992 South Africa Manufacturing Cement products Weiss (29) 1977 U.S Thomas (30) 1982 U.K Ohlson (31) 1985 Sweden Chrysotile Gardner (32) 1986 U.K Hughes (33) 1987 U.S. (plant 1) Finkelstein (34) 1984 Canada Alies-Patin (35) 1985 France Hughes (33) 1987 U.S. (plant 2) Chrysotile and Magnani (36) 1987 Italy crocidolite Raffn (37) 1989 Denmark Albin (38) 1990 Sweden Neuberger (39) 1990 Austria Textiles McDonald (40) 1983 U.S Dement (41) 1994 U.S Chrysotile McDonald (42) 1983 U.S Chrysotile and Peto (43) 1985 U.K crocidolite
7 J.C. McDonald and A. McDonald 273 Friction products McDonald (44) 1984 U.S Newhouse (45) 1989 U.K Chrysotile Insulation products Seidman (46) 1979 U.S. Unclear Acheson (47) 1984 U.K Amosite Levin (48) 1998 U.S Filter assembly McDonald (49) 1978 Canada Jones (50) 1980 U.K Talcott (51) 1989 U.S Crocidolite Acheson (52) 1982 U.K. (a) Acheson (52) 1982 U.K. (b) Chrysotile Various products Newhouse (53) 1985 U.K Enterline (54) 1987 U.S Liddell (23) 1997 Canada Chrysotile and crocidolite Product application Insulation work Elmes (55) 1977 U.K Selikoff (56) 1979 U.S. and 17, Canada Newhouse (53) 1985 U.K Järvholm (57) 1998 Sweden Chrysotile, crocidolite, and amosite Dockyard work Rossiter (58) 1980 U.K Chrysotile, crocidolite, Kolonel (59) 1985 U.S and amosite PMR, proportional mortality ratio; SMR, standardized mortality ratio. Source: Based on Table 5.1 in McDonald (21).
8 274 Chapter 17 Mesothelioma and Asbestos Exposure traced had died before 1993; 38 (0.47%) probably from mesothelioma, distributed as follows: Average exposure Mesothelioma Location Traced Deaths (f/mil.y) n PMR (%) Thetford Mines Asbestos Factory (in Asbestos) f/mil.y, fibers/million per year; PMR, proportional mortality ratio. At both the town of Asbestos and the region of Thetford Mines, the average duration of employment for mine, mill, and factory workers was about 10 years, and of those traced, 76% had worked for more than 1 year. No case occurred among some 4400 men employed less than 2 years; eight were in men employed 2 to 5 years, and the remaining 30 in men employed 20 to 49 years. All employees in the factory were also potentially exposed to crocidolite and amosite, mainly used in cement and friction product manufacture. Among these were two cases in men who in 1939 to 1942 worked on the carding of pure crocidolite for military gas mask filters. The exposures, shown in f/mil.y, are approximate and were obtained by conversion from cumulative exposures expressed in million dust particles per cubic foot.y (mpcf.y), accumulated by workers to age 55. The cohort of 7317 white South African amphibole miners was established in 1981 from records of men employed since 1925 (28). Only some 45% of the cohort were employed for more than 1 year, and 8% for more than 10 years. Excluding 167 lost to follow-up, 1225 of the remaining 7150 had died; 30 (2.4%) probably from mesothelioma, distributed as follows by type of exposure: Average exposure Mesothelioma Exposure Traced Deaths (f/mil.y) n PMR (%) Crocidolite only Amosite only Mixed nk nk, not known. The cohort of Australian crocidolite miners and millers comprised 6505 men first employed between 1943 and 1967, with 4653 (72%) traced to the end of 1980; by then 820 were known to have died, 32 (3.9%) from mesothelioma (27). Their median cumulative exposure was estimated at 6 fibers per cc/year; their median duration of employment was only 4 months, but their exposure was described as intense. Despite the far lower cumulative exposure experienced by South African and Australian workers than the men in Quebec, their proportional mortality from mesothelioma was about 10 times higher.
9 Most of the other industry-specific comparisons present much the same pattern. In friction product manufacture, for example, although 11 mesothelioma deaths were observed by Newhouse and Sullivan (45) in a large cohort of over 9000, virtually all of whom were exposed only to chrysotile, 10 of the 11 were definitely, and one possibly, members of a small group who worked for a short time on a special crocidolite contract. In the textile industry, the two cohorts studied by McDonald et al (40,42) were in plants owned by the same company and similar in every way, except that in the one with 14 deaths from mesothelioma, crocidolite was also used. Even more dramatic results were obtained from the two cohorts employed briefly during the early years of the Second World War in Canada (49) and in England (50) on the manufacture and assembly of filter pads made with pure crocidolite for military gas masks. Cases of mesothelioma began to appear in both cohorts 18 years later, with PMRs reaching 16% and 17%, respectively. A disaster of similar severity affected a small group of employees in the United States engaged in the manufacture of cigarette filters, of all things, from crocidolite. In contrast, a study of 570 British workers employed in manufacturing civilian gas masks using chrysotile filters produced only one case, an employee previously exposed to crocidolite (52). In summary, of 11,538 deaths in the chrysotile cohorts, 44 were from mesothelioma (PMR per thousand 3.8), whereas in the amphibolerelated cohorts of 19,622 deaths, 590 were from mesothelioma (PMR per thousand 30.1). While the carcinogenic potency of crocidolite thus seems clear, that of amosite, particularly in mining, is less so. High rates of mesothelioma were observed nevertheless in the manufacturing of insulation materials and among insulation workers, both groups heavily exposed to amosite. In none of the reports on cohorts in Table 17.2, however, is there any reliable indication of risk in relation to estimated intensity of exposure to any specified fiber type; the interpretation of the PMRs, therefore, entails several assumptions. A further point of interest concerns the general parallel exhibited in Table 17.2 between levels of mesothelioma and lung cancer excess mortality in all industries except textile manufacture. In the four cohorts shown, two of which were in the same plant, there were only two mesothelioma deaths in all, whereas for lung cancer the general pattern was reversed (40,41). In the other two textile plants in which crocidolite was also used, there were 24 deaths from mesothelioma, but with only modest SMRs for lung cancer (42,43). This anomaly, which is confined to the use of chrysotile in this particular plant, has never been satisfactorily explained (60), and is all the more important in that it did not apply to mesothelioma. J.C. McDonald and A. McDonald 275 Other Causes Mention has been made of the occurrence of possible cases of mesothelioma in autopsy series at the end of the 19th century. Given the long latency of this disease (20 to 50 years or longer), it is unlikely that they
10 276 Chapter 17 Mesothelioma and Asbestos Exposure could have resulted from the industrial exploitation of asbestos, which began in the 1880s and then only on a small scale. This evidence in itself is not conclusive, but there are stronger grounds for both the existence of a low background incidence of mesothelioma of unknown etiology, and for causation by at least one environmental mineral other than asbestos. Fibrous Erionite The evidence against fibrous erionite is virtually confined to the disastrous occurrence of unequaled mortality from these tumors in a localized area of central Turkey. The population of three villages in Cappadocia had been exposed since birth to airborne fibers of erionite, a zeolite mineral quite unrelated to asbestos, though with fibers having some physical similarities to crocidolite. The fibers are derived from volcanic rock, or tuff, which is used in the area for construction of houses and other buildings. In the three affected villages, 29 of 125 deaths during a defined period were from pleural mesothelioma, and four others from peritoneal disease (61). Investigation showed that airborne fiber concentrations were higher in the villages affected than in one that was not; also sputum samples from residents in the former contained ferruginous bodies with an erionite core. In most cases exposure was from birth, with death occurring 27 to 40 years later. In common with amphibole fibers, electron microscopic studies of lung tissue showed erionite to be highly biopersistent. Deposits of fibrous erionite occur in volcanic areas elsewhere, for example in some Rocky Mountain states; there has also been some synthetic production for industrial catalytic purposes. No clear link with mesothelioma mortality has been shown with either of these possible sources; perhaps because nowhere other than in Cappadocia has volcanic tuff been used for domestic buildings, resulting in exposure of young children. It is worth adding that deaths from mesothelioma have also been recorded in Scandinavia and other parts of Turkey among former residents of the affected villages. Apart from asbestos and erionite, no other environmental agent has been incriminated with any certainty. Some suspicion has rested on the long, thin silica fibers created by the burning of sugar cane in India (62) and in the southern states of the United States (63). A similar process might also arise in forest fires, but none of these suggestions has been supported by experimental or lung burden evidence. Mesothelioma in Children Despite diagnostic uncertainties, greater even than in adults, it is evident that mesothelioma does occur in children. In a review of 80 cases, reported from 1969 to 1986 (64), the ratio of males to females, aged 1 to 19 years (mean 9.7 years) was 1.4:1 and only two had a possible exposure to asbestos. Of these, 68% were pleural, 25% peritoneal, and 8% pericardial. Given that the latency of mesothelioma is very seldom less than 20 years, and in the Turkish cases at least 27 years, it is most unlikely that childhood cases could be due to asbestos expo-
11 sure. The incidence of mesothelioma in persons younger than 20 years of age can be roughly estimated from three studies (65). In the Canadian survey from 1960 to 1968, four fatal cases were ascertained by systemic inquiry from all pathologists a rate of about 0.7 per 10 million per annum. A very similar figure can be derived from 13 cases identified among death certificates in the United States from 1965 to Finally, data from the Surveillance, Epidemiology, and End Result (SEER) program in the United States, estimated the case incidence from 1973 to 1984, at 0.5 per 10 million. As mesothelioma in children may well be underdiagnosed, a conservative estimate for the annual case incidence in North America may well reach 1 per 10 million, but even so, appreciably lower than any comparable estimate for adults. Statistical Extrapolation In Canada, annual incidence rates based on cases of mesothelioma ascertained through pathologists, when extrapolated backward, suggest that male and female rates were similar at or before 1950, at a level of about 1 per million population. This pattern is similar to trends found in the SEER cancer surveillance program in the United States, and for mortality observed in Britain, Finland, Norway, and Denmark, where an increasing male excess appears due to the more frequent history of occupational asbestos exposure in men than in women. In the SEER program, regions with higher age-adjusted incidence rates, presumably attributable to work-related asbestos exposure, had higher ratios of male to female cases than regions with lower rates, and linear extrapolation would suggest that, at the point where the sex ratio is equal, the incidence might be as high as 5 per million, though with wide confidence limits. In Los Angeles County, equal numbers of cases in men and women were without history of exposure to asbestos, suggesting a background incidence of about 2 per million. In France, careful inquiry failed to identify any opportunity for asbestos exposure in younger subjects with mesothelioma, with equal numbers of males and females. In all these various studies, efforts to detect a cause other than asbestos have been largely unsuccessful (65). Lung Burden Analyses Valuable though cohort mortality surveys have been in assessing the health effects of asbestos in selected industries, they have not contributed much to knowledge concerning risk in relation to intensity of exposure to specific types of fiber. In diseases such as mesothelioma, the relevant exposures took place many years before adequate measurements of respirable dust particles, let alone fibers, had been made. Any such estimates remain at best a rough surrogate for what an individual worker inhaled and retained. Thus the development in the 1970s of electron microscopy with energy-dispersive analysis, to identify, count, and size mineral fibers in lung tissue, held great potential, though also with limitations. Apart from selective biases resulting from the availability and nature of lung samples obtained for analysis, fibers J.C. McDonald and A. McDonald 277
12 278 Chapter 17 Mesothelioma and Asbestos Exposure seen at autopsy or biopsy may not reflect what were present years earlier; much depends on the ability of fibers to penetrate the airways, and on their subsequent durability and biopersistence. As it is precisely in these latter qualities that chrysotile and the amphiboles differ, any epidemiologic study of lung burden must be carefully controlled and the results interpreted with considerable caution. Despite these difficulties, there have been several well-designed case-control studies of reasonable quality in Europe, Australia, and America that, though not conclusive, have provided consistent evidence implicating amphibole fibers rather than chrysotile in most cases of mesothelioma (Table 17.3). The two most recent studies from Germany (72) and the United Kingdom (73) are particularly informative in that they addressed risk in relation to fiber concentration in lung tissue, i.e., retained dose. In both these studies, a highly significant linear relationship was observed between odds ratios and concentrations of amphibole fiber, but not with chrysotile. In the German study, risk was greatest with fibers longer than 15mm, and in the British, short, medium, and long fibers were all associated with risk, but most closely with those in the longest category ( 10mm). In the latter study, the strong linear trend shown by crocidolite, amosite, and tremolite when combined suggested that their effects were probably additive (Table 17.4). Overall, these analyses indicated that some 80% of cases studied were attributable to amosite or crocidolite, and 7% to tremolite. The contribution of chrysotile could not be reliably assessed because of its low biopersistence, but as over 90% of all asbestos used is chrysotile, for which tremolite is a valid marker, it must be small. The British study just described was based on a larger number of cases reported by chest physicians in a national surveillance scheme in men younger than 50 years of age at time of diagnosis. It was thought that most, if not all, of the occupational exposures would have been since 1970 when the importation of crocidolite to the UK was virtually eliminated. In fact, it was found that almost all the cases were in men who had started work several years before that date. Of 37 occupations analyzed, odds ratios against expected values obtained from the census were significantly raised in only eight, of which five were in the construction industry: carpenters, plumbers, electricians, insulators, and unskilled workers. The remaining three categories at increased risk were workers in shipbuilding, cement, and mineral product manufacturing, all less important in this than in earlier surveys (74). The Tremolite Factor When we began in 1965, at the behest of the UICC Working Group, an extensive program of epidemiologic research in the Quebec asbestos mining industry, it was in the belief that we were dealing with exposure to chrysotile only. Clear evidence was found of a systematic relationship between quantitative estimates of airborne dust particle exposure and all measures of morbidity and mortality of primary interest, including lung cancer, radiographic change, lung function, and
13 J.C. McDonald and A. McDonald 279 Table Analyses of mineral fibers in lung tissue from mesothelioma cases and controls First author Odds ratio for (reference) Year Country Cases Controls amphibole fibers* Evidence on chrysotile Jones (66) 1980 U.K. 86 cases notified 56 cases (lung cancer 27, 7.4 Chrysotile present in two by coroners and cerebrovascular disease 29) of four cases without pathologists, 1976 matched for age, sex, and place amphiboles McDonald (67) 1982 Canada 99 cases from Secondary lung cancer; matched 3.8 In pairs where amphibole and U.S. survey of for age, sex, date, and hospital content was <10 6 F/g closely pathologists similar distributions of chrysotile Mowe (68) 1985 Norway 14 cases, county 28 cases excluding malignant 8.5 Fiber type not identified cancer registry, and chronic pulmonary disease (based on all types of matched for age, sex, year, and amphibole fiber) residence Gaudichet (69) 1988 France 20 cases from 20 each of adenocarcinoma and Amphibole fiber Similar concentration in Nantes district, squamous carcinoma, secondary concentration 2 3 times cases and controls lung cancer and cardiovascular higher than in controls disease, matched for age, sex, and hospital McDonald (70) 1989 Canada 78 cases from Nonmalignant nonrespiratory 6.6 for fibers 8 mm in Low-level risk in univariate survey of disease, matched for age, sex, length analysis and none in pathologists, date, hospital, and type of sample multivariate analysis Rogers (71) 1991 Australia 221 cases from 359 tissue samples from a 16.6 for fibers 10 mm in 7 of 25 cases and 3 of 31 national hospital in Sydney excluding length controls without amphibole surveillance, nonmalignant respiratory disease fibers had 10 5 F/g chrysotile and abdominal cancer; unmatched Rödelsperger 1999 Germany 66 cases from five 66 cases undergoing lung 4.5 for fibers 15 mm in No increase in odds ratio (72) German cities resection mainly for lung cancer, clear linear dose matched for age, sex, and region response McDonald 2001 U.K. 69 male cases 57 cases of accidental or sudden Related linearly to No significant increase in (73) aged 50 years at death of similar age, sex, and concentrations: odds ratio diagnosis region mg 8.8 ( ) mg 59.9 ( ) * Based on Table 5.5 in McDonald (21).
14 280 Chapter 17 Mesothelioma and Asbestos Exposure Table Distribution of lung fiber concentrations with grouped and continuous odds ratios (OR) Concentration Fiber type (per mg) Cases Controls Crude OR Adjusted OR a Crocidolite ( ) 4.6 ( ) ( ) 3.9 ( ) Linear model b 13.2 ( ) 40.0 ( ) Amosite ( ) 5.1 ( ) ( ) 17.9 ( ) Linear model b 11.4 ( ) 14.3 ( ) Tremolite ( ) 2.3 ( ) Linear model b 6.9 ( ) 29.6 (<0 340) All amphiboles ( ) 8.8 ( ) ( ) 59.9 ( ) ( ) Linear model b 19.4 ( ) 47.6 (6.0 >999) Chrysotile ( ) 1.9 ( ) ( ) 2.2 ( ) Linear model b 0.1 (<0 1.2) 2.2 (<0 >999) a Crocidolite, amosite, and tremolite are adjusted for each other. Total amphiboles and chrysotile are adjusted for each other. b Average increment in odds ratio per fiber/mg. Source: From McDonald et al (73). respiratory symptoms (75). Except at very high exposure levels, these adverse health effects were not severe, and even among 2413 deaths in a cohort of some 10,000 men, only three (0.12%) were ascribed to mesothelioma. This seemed in marked contrast to the findings of Selikoff et al (56) of 22 deaths (5.8%) from this cause among 380 deaths in a small cohort of 632 American insulation workers (76). It was observed at the outset, and by local physicians for many years, that pleural thickening and calcification were much more frequent among workers in the Thetford Mines region of Quebec than in the town of Asbestos, some 60 miles away. In a detailed study of pleural calcification Gibbs (77), who was responsible for the environmental aspects of our research program, noted in 1972 that these radiographic changes were considerably more prevalent in some mines than in others, suggesting to him that minerals other than chrysotile might be responsible. Over the next few years, a series of studies was published with results based on electron microscopic analyses of lung tissue at autopsy, which, taken together, indicated that the exposure experi-
15 enced by workers with Quebec chrysotile was much more complicated than had previously been supposed. First came the observations of Pooley (78), and then of Rowlands et al (79), who found that, in the lungs of former Quebec miners at autopsy, chrysotile and tremolite fibers were present in surprisingly similar concentrations (Fig. 17.1). Later, further analysis of data from these studies suggested that tremolite concentrations were perhaps two to three times higher in the region of Thetford Mines than in the town of Asbestos (80). There followed a much larger investigation by Sébastien et al (81) that, though undertaken for an entirely different purpose, added considerably to several aspects of the tremolite question. The primary objective of this study was to explain the much greater risk of lung cancer, though not of mesothelioma, in asbestos textile workers in Charleston, South Carolina, than in Quebec miners J.C. McDonald and A. McDonald 281 Figure Lung of Quebec chrysotile miner at autopsy. Ch, chrysotile fibers; T, tremolite fibers. (Source: Copy of photomicrograph kindly provided by Dr. Patrick Sébastien.)
16 282 Chapter 17 Mesothelioma and Asbestos Exposure and millers both exposed to chrysotile from the same source. One hundred sixty-one lung tissue samples from deceased cohort members (72 from Charleston and 89 from Thetford Mines) were collected for analysis by transmission electron microscopy. Altogether 1828 chrysotile and 3270 tremolite fibers were identified; in both cohorts tremolite predominated and fiber dimensions were closely similar. Analyses that took account of duration of employment, exposure intensity, and time from last employment to death concluded that none of these variables could explain the higher lung cancer risks observed in textile workers. The possible co-carcinogenic role of mineral oil used to control dust in textile plants was an alternative explanation, which has yet to be adequately tested. However, the findings from this large survey made it possible to address several other questions. The first of these studies sought to explain the remarkable predominance of tremolite fibers in the lungs of men overwhelmingly exposed at work to chrysotile, although in the city of Thetford Mines tremolite represented only 1.5% of asbestos fibers in the ambient air (82). That this enormous difference was probably due to the far greater durability and biopersistence of tremolite was demonstrated by examining mean lung fiber concentrations in deaths at varying times after the date of last employment. The lungs of even men who died while still employed contained twice as many tremolite fibers as chrysotile; at more than 10 years after leaving work, the ratio rose eightfold (83). In another analysis, the results of which are shown in Figure 17.2, it can be seen that in relation to cumulative exposure, the lung concentration of tremolite in Quebec miners increased linearly, whereas that of chrysotile did not (84). This pattern was virtually identical to that Figure Human and experimental data on the relationships between cumulative exposure to asbestos dust and lung retention. [Source: Based on Sébastien et al (84).]
17 reported by Wagner et al (85) in laboratory rats after inhalation of amosite and chrysotile. In his earlier study of pleural calcification in the region of Thetford Mines, Gibbs (77) had noted that these changes were more common among miners than millers, and particularly in men who had worked in a localized group of mines near the center of the town rather than in other mines located peripherally. He concluded that the cause might be related to some mineral closely associated with chrysotile, possibly mica, talc, or brünnerite, but he did not include tremolite, which even Riordan (86) had rarely mentioned in his comprehensive description of the geology of the region in Much later, when, by 1992, 38 probable cases of mesothelioma had been identified in the Quebec cohort among over 8000 deaths from all causes, it became clear that even at Thetford Mines they too were unevenly distributed. As mentioned in the previous section, among 4125 deaths in miners and millers at Thetford Mines, there were 25 (0.61%) from mesothelioma; at the town of Asbestos among 3331 deaths, the corresponding figure was 8 (0.24%). At Thetford, however, the cases were more common in miners, whereas at Asbestos the few cases were all in millers. A further detailed examination of work histories of the cases at Thetford showed that man-years of employment were concentrated in a localized area of five mines centrally located (area A), compared with 10 mines located peripherally (area B) (20). These were much the same as those observed by Gibbs for pleural classification. A more detailed analysis was then made of the data for the 83 subjects in Thetford Mines from the study of Sébastien et al (84), using available records of the specific mines in which each man had worked. This showed that the concentration of tremolite fibers, but not of chrysotile, were some four times higher among 58 men in area A (32/mg) than among 25 men in area B (7/mg) (p =.0002). A strictly controlled study of deaths from mesothelioma and other cancers, with analysis by logistic regression, was therefore undertaken (87). This showed that the odds ratios (OR) for work in the central mines (area A) were raised substantially and significantly for mesothelioma [OR = 2.55; 90% confidence interval (CI) ] and lung cancer (OR = 1.98; 90% CI ), but not in area B or for cancer at other sites in either area. Reanalysis by Sébastien of fibers from his earlier study (81) also confirmed that there was no important difference in their dimensions or composition between the two areas. None of these findings would necessarily have incriminated tremolite, as opposed to some other mineral with similar geographic distribution, in the absence of independent evidence of the carcinogenicity of fibrous tremolite. The strongest indication of this has been the experience of vermiculite miners and millers in Libby, Montana, exposed to contaminating amphibole fibers in the tremolite series, but to no other form of asbestos. In the early 1980s, parallel but independent studies of mortality and morbidity among the employees of the Libby plant were undertaken by us and the National Institute for Occupational Safety and Health (NIOSH). The results obtained by the two groups provided very similar evidence of high excess mortality from nonmalignant respiratory disease, lung cancer, and mesothelioma (88), J.C. McDonald and A. McDonald 283
18 284 Chapter 17 Mesothelioma and Asbestos Exposure Table Mortality in Libby cohort of vermiculite miners exposed to fibrous tremolite (n = 406) (reference: US white males) (90) Deaths to July 1983 Deaths since July 1983 a Total ICD-9 Observed SMR Observed SMR Observed SMR Respiratory cancers All other cancers , , NMRD , Circulatory disease External All causes (incl. 4 (PMR = 2.4%) 8 (PMR = 6.7%) 12 (PMR = 4.2%) mesothelioma) ICD, International Classification of Diseases; NMRD, nonmalignant respiratory disease. a To January 1, and an increased prevalence of small radiographic opacities of between 6% and 10% per 100F/mL years (89). As these findings on mortality were based on very small cohorts, a further follow-up to the end of 1998 has recently been completed, allowing a more reliable assessment of risk in relation to estimated exposure. Total deaths to the end of 1998 were lung cancer 44 (SMR 2.40), NMRD 51 (SMR 3.09), all causes 285 (SMR 1.27); included among the total were 12 deaths attributed to mesothelioma (PMR 4.21%) (90) (Table 17.5). Adjusted linear relative risks (per 100 F/mL.y), estimated by Poisson regression, were lung cancer (0.36, 95% CI ), NMRD (0.38, 95% CI ), and all causes (0.14, 95% CI ). The 12 deaths from mesothelioma, though with a typical latency range of 22 to 47 years (median 35.5 years) showed only a limited relationship to estimated exposure. The all-cause linear model would imply a 14% increase in mortality for mine workers exposed occupationally to 100F/mL.y or 3.2% for a general population exposed for 50 years to an ambient concentration of 0.1 F/mL (90). Synthesis Over 40 years have passed since 1960, when 33 cases of pleural mesothelioma were described by Wagner et al (5), almost all from the crocidolite mining region in the northwest Cape Province of South Africa. In the same year, 10 cases of peritoneal mesothelioma were reported by Keal (91) among textile employees of the Cape Asbestos Company in London, exposed to crocidolite from the same source. Recognizing therefore the importance of asbestos fiber type, the UICC Expert Group in 1964 had put priority on epidemiologic studies of miners and millers engaged in the production of the three main types of asbestos rather than on employees in manufacture and industrial application, usually entailing chrysotile-amphibole mixtures. During the next 20 years, unfortunately, most research focused on the latter
19 and, until the late 1980s, the only production workers studied were miners and millers in Quebec and Italy, both with similar and reassuring results for chrysotile. The first amphibole mine workers for whom there were comparable data were, in fact, the Libby vermiculite employees, exposed incidentally to contamination by fibrous tremolite (26,90). Only later were mortality data published in 1988 on crocidolite miners in Australia (27), and in 1992 on crocidolite and amosite miners in South Africa (28), but in the meantime a considerable number of cohort study results based on workers in the manufacturing industries or in asbestos product use were published. Exposure in all of these cohorts was mainly to chrysotile, but in most of them also to varying proportions of crocidolite or amosite. Investigators familiar with the disastrous experience of insulation workers in North America, where exposure had also been to chrysotile, and possibly amosite, found it difficult to believe that all types of asbestos were not equally harmful. This view was supported by experimental evidence, which showed that all fiber types were equally carcinogenic, without appreciating that biopersistence and durability would be far more important in humans, with a much longer life span, than in laboratory animals. Against a background of much suspicion and recrimination, the results of the several important cohort studies published in the 1980s failed to have much effect on entrenched and conflicting views. For those who saw chrysotile as a mineral fiber of low carcinogenicity, the findings summarized in Table 17.2 confirmed this opinion. Others, with legitimate concern for control rather than scientific niceties, found little difficulty in maintaining their disbelief. Uncertainties associated with mixed exposures, lack of information on exposure intensity, and statistical chance were often cited; other reasons were less flattering (3). Some resolution of this unpleasant and unhelpful controversy came with the use of lung tissue analyses in epidemiologic research. Despite difficulties in interpretation of results and the absolute need for properly selected controls (92), these studies demonstrated two things and revealed a third. First, was the clear evidence of an overwhelming predominance, with dose-response, of amphibole fibers in mesothelioma cases; second, that amphibole fibers persist in lung tissue, whereas chrysotile does not. The short life span of laboratory animals could not deal adequately with tumors in humans of long latency. Third, it has only been by analyzing lung tissue that the varying presence of fibrous tremolite has been demonstrated in commercial chrysotile. Although the recent update on mortality in the Libby vermiculite cohort has indicated that the ability of fibrous tremolite to cause mesothelioma is on a par with crocidolite, it remains almost impossible to estimate the contribution it makes to the carcinogenicity of commercial chrysotile, which greatly varies in level of tremolite content, both geographically and in time. There are two main reasons for this. First is our ignorance of how best to assess exposure to a carcinogenic agent that is biopersistent. Cumulative exposure clearly underestimates the potential effect of a retained carcinogen. The expo- J.C. McDonald and A. McDonald 285
20 286 Chapter 17 Mesothelioma and Asbestos Exposure sure index that appeared to do best in the Libby cohort was one in which the estimated fiber concentration at any year was weighted by residence time. Although conceptually reasonable, such an index was largely determined by estimated airborne fiber concentrations 30 to 50 years before death, for which only the most crude approximations can be guessed in this or any other mortality study. With only 12 mesothelioma deaths in a cohort of 406 men, a statistically significant discrimination between risk and any type of exposure index was not possible. Second, there is the analogous problem that results from a total ignorance of the tremolite concentrations to which Quebec miners and millers were exposed, as far back as 1918, when men who later developed mesothelioma were first employed. These cases were miners rather than millers, and in the relevant period there were almost 30 different mining companies, in few of which were any dust measurements made. Exposure to tremolite would certainly have been intermittent, as evidenced by the fact that mesothelioma risk in the Quebec cohort was related to duration of employment but not to intensity of dust exposure. Thus, to obtain any kind of answer to the question, we must take account of other types of evidence. For example, there was no case of mesothelioma in the Quebec cohort in men employed less than 2 years, and in the 21 of 38 men with mesothelioma whose lungs were examined at autopsy, amphibole fibers mostly tremolite and in high concentration were present in them all (20). In the eight studies of lung fiber burden in mesothelioma cases and controls (Table 17.3), little or no evidence of risk was observed with chrysotile only; indeed, amphibole fibers were present in most cases. Finally, the recent casereferent study in the United Kingdom of young adults with mesothelioma showed that crocidolite and amosite, singly or additively, could account for about 80% of cases, and tremolite for about 7%, leaving very few for chrysotile alone. Compared with amphibole fibers, pure chrysotile is removed much more rapidly from human tissue, but it is not without some biopersistence. It would be unreasonable, therefore, to conclude that when inhaled in sufficient quantity it carries no mesothelioma risk. It should be remembered, nevertheless, that the epidemiologic evidence reviewed in this chapter reflects exposure levels some 40 or more years ago, orders of magnitude higher than those that prevail today or that should be readily achievable. Unfortunately, past failure to discriminate between the carcinogenicity of chrysotile and the amphiboles allowed the latter to be inadequately controlled too long. Conclusion In the discussion session on mesothelioma that followed the presentations by Selikoff, Wagner, Newhouse, Elmes, and others at the New York Conference in 1964, Scheepers (93), a principal discussant, raised two prophetic questions that it has taken 40 years to answer. First, with regard to 11 cases of lung cancer with which he was familiar and whose