Tobacco Documents Online

Special papers • Health effects of Historical Exposures to Asbestos ~ Exposure-Response : Asbestos and Mesothelloma Prof. Douglas Liddell

Uddell D. Heslth effects of hisioricil exposures to asbacto.t. In: Gibbs GW, Dunnigan J, Kido M. HigLChi T. (Eds) Henbh Riska Jrom F.rposyre to Mineral Fjbrts: an lnternrttiotial Pcrspertive. North York. Ontirio: Captus University Press 1993: 49-65. (1992.01.10; with minor typographical corrections 1993.03.09) Health Effects of Historical Exposures to Asbestos Douglas LFddell De,patlmeat of Epidemiology and Biostatistics, MoGill University. 1020 Piae Avenue West, Montreal, Canada H3A 1A2 ABSTRACT ,ifter the 1977 New York ?lcadaay of Sciencer sy.nposiuaa on HeaLh Hazords ojAsbertos Fxpasare, it was c+anurronly beiiesrd, particularly in the United States, that asbertos, in all iu fornas, was sruh a hozardous mineral that 'one fibre could kill'. ?hree sus of faus had been ignored. Frrst, the rndoubted orcesses until then of mortaliry, from puLnonary ftbt»sfs (asbestosis), lung aincer, and maothetiorna, had been due to crpasure 20-60 years ptrviously, and for most of this period the 'avrmge' concentration of asbtstos fibrrs in the occupational environnient had been greater than 100 frbm/milliGtne (/m4 Lewels were being brought down, to around S ffinl in 1975, and nearer I fJnt1 by 1980, so that the risk of asbestos-related d'ueare must atro have boen falling rapidly. Secondly, oaer 90% of aarnmercially arrrila.ble asbestar always has been c:iirysotilt ?hitdly, the only taajor study of duycodte workers showed that rxaxs lung ccncw was Wrtuatly confined to thase who had been a.xposeid to ,nore than IQoO (fbrerlml) x years (e.g. m+er 20 years at 3o fhn1J. TYdss paper reviews all the historical eWdena to date, and the following cnnclusions are drawn. ?)ie risk of nscrothelionra rftir expostre to dtryrotile is one order of magnitude less than frorn amasite, and two orders less than fnanr avtidoiite- Tlu risk of ezoess lung cancer as a result of zsposure to d:rysotile (ercept in textiks) is less, by more than one order of tnagnitude, than that frnvn rnocidolite r.Wsum On real}stlc assunrprionr, the risk of tutig canoer fram inhaiing one drrysotlk f:bre Ls obout one ia a hundrrd-thousand-billion; the rislts of inesothelionta or asbutasis are even lower. TJu highest euimate of lung cancer risk frwrr a single frbre - of a+oddolue - is kss than one in a thousand-liiUlon. The bellefs that all fotms of asbesxos are equally hazardous and that one asbestos fibre can kill are seen to be sryths - untrue and comp?etely withoru foundation. INTRODUCTiON It is fu11y established that severe exposure to respirable asbestos fibres has led to respiratory fibrosis, now called asbcoxis, excess lung can=, and maotheliomi of the pleura, and sometimes of the paitcnaum. Unffoctuaazely, the word 'severe' in the preceding sentence is often over]ooked., aad in so= countries there is widespraad 'fibra-phobii', propagated by a number of myths' including thosa in Table 1. The most iaportant lessons to be learnt fc+om h,istory ana those; concerning exposuro-rr.spoase rrlationships, and this paper emphasises the knowledge rva have gained about the strength, shspe and gradient of the relationships between asbestos exposure and the risks of inesothelioma and of lung cancer. ALSTORICAL BACKGROUND Arbertos produaion The modern asbestos industry started in 1878, when about 50 tons of chrysotile xwe produced in Quebec. By 1977, when world production was at its peak, about 5.5 talllion short toas of chrysotile were prodaaed, 24 % in Quebec, 4795 in the USSR, and 29% elsewhere, usd in addition about oae-quarter of a mMon tons of the other - amphibole - forms of ssl~os, mainly amosite and crocidolite. It is of great importaace that throughout - from 1878 to nowadays - chrysotile has always been by far the predominant fiime. -1-

For example, a total of 764 million kg of asbestos was u.sexi in the United State,c in 1974, of which over 94% was chrysotile; the other 5.696 was amphibole, of which 4.49s was crocidolite and 1. 1 `Xo amosite. 1878 - 1930 As early as 1898, one of HM Inspectors of Factories for England and Wales drew attention to lung injury in asbestos sifting and carding (as well as in silk opening and combing, and hemp spinning). Eight years later, a departmental conmmittee heard of the death in 1900 of a man of 33 with lung fibrosis who had spent 14 years working with asbestos, 10 of those years in the carding room. Ten men who started the same work at the same time had already died of a similar illness. No action was taken until after the rtport= in 1930 by E R A Merewether, HM Medical Inspector of Factories, and C W Price, HM Engineering Inspector of Factories. They showed a high prevalence of pulmonary fibrosis in the 363 workers examined, clearly related to duration and intensity of exposure to the extent that almost alll men reaching retiring age after a working life in asbestos textiles or the manufacture of asbestos insulating matetials were expected to be diseased. As in those days asbestosis (the name now given to this disease) was frequently fatal, a substantial proportion of the workforce must have sutxumbed before reaching retirement age: the risk must have been unimaginably high. 1930 - 1964 The first environmental control measures were devised in the light of Price's recommeadations2 on dust suppression and embodied into the Asbestos tndustry Regulations 1931; they came into force in specified parts of the British asbestos industry in 1933. The earliest studies of asbestos-related disease in the United States appear to have taken place in the mid-1930s, and a Threshold Limit Value, based on the findings of Dreessen u aL,3 was set by the American Conference of Governmental Industrial Hygienists; it was endorsed by the ACGIH in 1964. As John Gilson wrote in 1982: "Most occupational diseases, including asbestos-induced lung cancer and trtesotheliotna, were first detected by astute observation by clinicians or pathologists; it has usually taken many years before their severity and extent have been revealed by systematic surveys.' From 1934, primary carcinoma of the lung, then a quite rare condition, was sometimes recorded in asbestos workers, usually in association with asbestosis. Then, in 1955, Doll' was able to show that the average risk of lung cancer in workers who had boen exposed to asbestos dust for 20 years or more was of the order of ten times that experienced by the general population. Several later occupational cohort studies of asbestos workers neported excess lung cancer, and it is now generally accepted as a possible effect of asbestos exposure. Although first suggested only in 1960, by Wagner and his colleagtxs,° in South Africa, within a few years it was generally accepted that asbestos dust could cause mesothelioma, i.e. malignant mesothelial ttunotus of pleura or peritoneum. Although from the 1960s, there have been suggestions of associations between exposure to asbestos and gastro-intestinal carcinoma, and possibly malignancies at other sites, none have been generally substantiated. By 1964, there was widespread concern about the ill-effects of asbestos. The Naw York Academy of Sciences sponsored a scientific conference, and a Working Group on Asbestos and Cancer was convened, also in New York, under the auspices of the U1CC. This Working Group's reports stated that asbestos-associated lung cancers were not limited to exposure to any one type of asbestos fibre, but further investigations were urgently needed to establish whether the degree of risk depended on the type of fibre inhaled, although research in several countries had suggested that exposure to crocidolite might be of particular importance as a cause of mesothelioma. Insulcuion wonYers; 'Convuui'onal Wudoen' Peri" the most influential studies since 1964 have been by the Mount Sinai group in New York, particularly those on a very large cohort of insulation woticers. TItCy followed 17,800 men employed in the US and Gnada in 1967; there had been 2,271 deaths by 1976, a massive excess of 37% over what would have been expected on the basis of the vital statistics of the US.' Excesses were reported of deaths due to nxsotheliotna (104) and of lung cancer (323, or 4-fold); also of malignancies at each of the following sites: oesophagus, stomach, colon-rectum, larynx, pharynx & buccal cavity, kidney, pancreas, liver & biliary passage, pros<ate, skin, and brain (in all, 258 observed where only 144.4 were expected). Although only 8 cancers of bladder and of testes were observed (2.0 fewer than the expected number), a fitrtlter 123 cancers at sites not listed above were reported, against 59.7 expected. The most likely explanation for this enormous "blanket' of cancer excesses is that the causes of death 'as recorded from death ceRificate information only' were treated as having been due to cancer if there were any indication to that effect on the death certificate, whether or not the nosological code used in the official vital statistics indicated cancer. But even if they wdne not inflated in some such way, the excesses of asbestos-related malignancies must be put into perspective: the working methods were such that the intensity of exposure was particularly high; and it must be appreciated that most of the men had been exposed to amphibole asbestos (amosite and/or crocidolite) as well as to chrysolite. No other investigations have indicated excesses approaching such a magnitude even for the malignancies now generally accepted as asbestos-related, leave alone for the very long list of other cancers. -2-

it is, therefore, difficult in the extreme to see any relevance of these findings to any of the problems faced today. Nevertheless, these findings have been the basis of what may be called the 'Conventional Wisdom'. Chrysorik txposure Meanwhile, the impact on Conventional Wisdom of the resu!ts from Quebec was negligible. Over 10,000 male chrysotile production workers (miners and millers) in Quebec, born 1891 thmagh 1920, have been followed; the first report on their mortality was in 1971,= the latest in the 1988s.}~ This cohort is especially important for several reasons: a) its size - there were almost 4,500 deaths by 1975, already reported;' and nearly 3,000 more by 1988, recently traced; " b) the exposure of the great majority of the cohort was to chrysotile alone; c) quantitative estimates have been made of dust concentrsUions; and d) it includes workers with very mild exposure as well as those severely exposed. From the start, excess lung csncer mortality has been clearly related to total dust exposure. Such relationships have now been shown in severil other cohorts of asbestos workers; however, the slopes of the relationships vary greatly with fibre type and process. Berause the slope for Quebec miners and millers was so shallow, the Conventional Wisdom has beea1= that there were major atethodological shortcomings and that the comparatively low reported health risks from exposure to chrysotile were anomalous. More recently, there have been several investigations in workforces exposed to chrysotile alone, and, except in the textile industry, these confirm the low health risks from cluysotile. In 1980, McDonald and collearues' wrote of the Quebec cohort: If the only utbjects studied had been the 1904 with at i east 20 years' employment in the lower dust cancentrations, averaging 6.6 ntillion particles per cubic foot [mpcfj, i.e. about 20 f/ml, excess mortality would not have been considered statistically significant, except for pneumoconiosis. • For the complete cohort, there had been about 6 S6 more deaths (all causes) than expected on the basis of mortality in the Province of Quebec, despite exposure aueraging roughly 700 (fibreslml) X years. This 6% excess is in complete contrast to that of 37 96 among the 17, 800 insulation workers referred to earlier. Cleariy, the consistent findings in humans exposed to chrysotile, by far the most common form of asbestos, provide a much asore realistic basis from which to evaluate health effects than do any unique and extraorditLatry results among amphibole asbestos workers. EtperFrrtental findings Meanwh'sle, experimental studies were suggesting that chrysotile was at least as dangerous as amphiboles. A heuristic explanation of the rrasons for differential effects in animals and humans, based on the argument of E1mGC," follows. Massive doses of chrysotile fibres, once rd,ained in the lung of a rat, split up into finer and finer bundles, and eventually into fibrils (the basic crystal units, which are ksa than 0.11cm in dianseter), and so provided a much increased surface area for biological reaction. Within the animai's life-span, ckzrance from the lung was not great, and doses were so enormous that svbstantial proportions of animals developed tumours. However, the sune dose of ansphibole fibres was retained virtuzily unchanged, with minitmtl clearaoce. Thus the effective dose of chrysotile was much greater than that of amphibole. In man, the retained doses per gram of lung were very much smaller, even after the most severe of historical exposures. 37se initial build-up of surface area was much grtater, exposure for exposure, with chrysotile than with anphibole, but the chrysotile was largely cleared from the lung before tumours could develop, whereas amphibole was retained indefinitely. The effective dose of cfuysotile in the human lung was thus very much less than of amphibole. It is a teaet of scientific metifod that an experiment should be designed to test a specific theory. Progress is made when eaperimental findings sre not expLsined by the theory, which has to be discarded in its specific fona: another theory is thea developed, to be tested in its turn by a new experiment. Progress is also made when epidemiologic findings are in conflict with those from experiment: the theory apparently supported by the experimeat has to be discarded and replaced. It is contrary to the principles of scientific method to maintain the tlk+ory in the face of disproof, whether that disproof is demonstrated by experiment or by epidemiology. Therefore, when the theory that chrysotile wss at least as dangerous as amphiboles was disproved in man, although supported in animal studies, it was essentisl on scientific grounds that the human evidence should pceva3l over that from animals, and hence over theory. However, the unprincipled approach of ignoring the httmsn evidence was incorporated into Conventiooal Wisdom, thus perpetuating the first of the asbestos myths of Table 1. Larency One of Merewether's findingsz was that: 'with continued exposure to high concentrations of dust, the fibrosis tnay be fully developed in from 7 to 9 yesrs, and may cause death after about 13 years of exposure ...' This implied inverse association between severity of exposure and 'latency', i.e. the interval between first exposure and manifestation of disease, was in conformity with general toxicological knowledge. It is, however, important that this was probably the last study in humans to show such an association. In the one study of 3

latency I am aware of," the average interval in 244 cases of lung cancer exposed to chrysotile was almost 40 years, quite independent of severity of exposure or even of smoking habit. The UICC Working Group's report' had stressed that the interval between first exposure to asbestos dust and detection of related tumours was many years, usually 20 and up to 60 years, so that further cases must be expected to occur for many more years (after 1965), even although dust exposures had been greatly reduced. Temporal patterns of r.xposure and drsersse A recent chapter's reviewed historical levels of asbestos exposure. It was, of course, impossible to be at all precise about exposures 'avenged' across workforces at different periods, but the attempted summary is shown in Table 2. The late 1960s and early 1970s appeared best documeated: an evaluation of the data suggested that the 'average' might have been of the order of 20 flm1. The other 'averages' were even more speculative, but all were thought to indicate the right orders of magnitude. It is undoubted that today's asbestos-related disease was initiated by exposures between, say, 1935 and 1965; without claiming any accuracy for Table 2, it is surely reasonable to state that average occupational exposures were then about two orders of magnitude worse than in 1980. Further, the incidence of disease that must be anticipated from today's exposuns must surely be much less. However, it will not be until well in to the 21st Century, in the year 2025 or thereabouts, that the benefits of reducing the intensity of exposure to around S f/ml will be reaped; and it will probably be a further 10 years before the lowering of occupational exposure limits to below 1 flail will be reflected. EXPOSURE-RESPONSE RELATIONSHIPS Mesorhe!'ioma In their report to the Health and Safety Commission in 1985, Doll and Peto" stated that, although the predicted risk of mesothelioma increased in approximate proportion to duration of asbestos exposure for exposures of up to about 10 years, 'thero was very little difference between the predicted effects of stopping or continuing exposure after 20 years'. However, close examination of their data suggests a much more positive conclusion." A model of risk, endorsed by Doll and Peto, which incorporated a term for duration of exposure, provided a good fit to these data, whereas the fit by a model without such a term was very poor. A formal test of the degree of improvemeat yielded aX2 statistic of 9.8, which, with I degree of freedom, is of enormous statistical significance. While it is true there was no death from mesothelioma among those with 30 or more years of service, the expected number of deaths (on the model incorporating duration) was so low the shortfall was quite insignificant, while for men with 20-29 years of scheduled service, there were nearly twice as many cases as expected. For each of the 12 cohort studies in which examination of exposure-response is possible from the reports, a test for trend of risk of inesotheiioma in relation to duration of exposure has been carried out, and the relevant x2 statistic calculated. All 12 statistics indicated a direct relation, seven of them being associated with P-values between 0.051 and 0.0005; even the most conservative approach to the test of the overall null hypothesis yields an extremely low P-value. This evidence that the risk of mesothelioma increases with duration of asbestos employment is overwhelming. For ten cohorts (nine of the 12 above, and one other), information was provided on short exposures: in each, there was a period of service, as short as 90 days or as long as 25 years, without tumour manifestation. The numbers of inesotheliotaas that would have been expected in these periods on the basis of proportionality are of course small, but they total 6.1. Sub-linear response to short durations of exposure may, at worst, be taken as hypothesized by McDonald and McDonald," in the light of their Table 4. Excluding those findings provides a legitimate test, which yields P= 0.0247. This does not, of course, constitute a proof that exposure for at least three months is a necessary condition for the development of isbestos-related mesothelioma; however, it is evidence that, for short durations, the exposure-response relationship is sub-linear. At the other end of the exposure continunm, there is insufficient evidence for fitll evaluation of the shape of the exposure-response relationship at long periods of employment. However, the indications are that it does not rise exponentially, and is not even supra-linear, but rather sub-linear. On toxicological grounds, it would have been anticipatad that the relationship would be sigmoid, with the point of inflexion somewhere near the EDs. However, the findings in man are for 'doses' very much below the ED.,, so it might have been anticipated that the risk would increase more-or-less exponentially, however, this is not the epidemiological finding. . Two other pieces of evidence that appear to run counter to the findings just discussed here must be mentioned. First, G Berry (personal communication) states: "Ihere was one case [at Wittenoom] with only three days exposure (a man who started work there when aged 39 and was diagnosed at age 70; he worked in the mill)'. This exposure was so extremely short that the finding is bizarre even in the context of the other cases at'Wittenoom, leave alone in the general context. If accepted, it weakens - but does not destroy - the hypothesis -4-

of sub-linearity of response to short exposures. Secondly, Browne and Smither," reporting 144 cases of confirmed mesothelioma among former employees of Cape Industries Ltd, noted nine cases (six pleural, three peritoneal) with no more than three months occupational exposure. Because this was a case series, no 'denominators' are available, and so these results cannot be evaluated: it is by no means impossible that the distribution, by duration of employment, of all Cape Asbestos employees over the years were so highly skewed that there is no real contradiction with the present findings. Two further factors must be borne in mind: the Cape Asbestos experience dates back many decades when intensities of exposure were unthinkably high; and a substantial proportion of 'short-servicx cases' msy in fact have arisen from neighbourhood exposure, which was known to have been far from negligible near the Barking plant in which the great majority of the cases had worked. It must also be borne in mind that the whole of this paragraph concerns employees exposed to Australian crocidolite alone or to 'mixturas' comparatively rich in t,.ape crocidolite. There have been very few mesotheliomas in the three cohorts exposed to chrysotile alone or with only a small admixture of amphibole, and there is no evidence to contradict in any way the conclusion that the risk of aesotheliocnz after only a few months exposuae, at least to chrysotile, is vanishingly low. Lung eaxcer It has long been accepted that the risk of lung cancer is closely related to aa:umulated asbestos exposure. In 1985, i.iddell and Hanley'0 showed that, in those cohorts for which adequate data were available, the relationship between SMR and exposure, x, measured as (mpcf) X years, could be modelled by a straight line of the form SMR - a + b.x. However, they recognized that the choice of a linear model was a matter of convenience, and pointed out that it was not biologically plausible. More recently, Vacek and McDonald have investigated the shape of the relationships in five cohorts of asbestos woricers.2' For the Quebec study, they defined six levels of the intensity of exposure as follows: (1) < 1 mpcF (2) 1, <3 mpcf; (3) 3, < 10 mpcf; (4) 10, <25 mpcf-, (5) 25, <5a mpcf; and (6) 50 mpcf or more. Then for each worker in the cohort they counted the number of years of exposure at each of these six levels. Conditional logistic regression analyses were carried out on the 230 cases of lung cancer occurring after 1950, and at least 20 years after first employment, and their referents, comprising all male workers born in the same year as the case, employed in the same mining area, and alive at the end of the year in which the case died. In the full analysis, the 'indepetdent' variables were: (1-6) years spent at each intensity level; (7) total duration of gaps in employment; (8) years since first employment; and (9) smoking habit. The regression coefficients for variables (5), (6), (8), and (9) were all of extremely high statistical significance, but those for variables (1), (2), (3), (4), and (7) were quite small relative to their standard errors. When these last five variables were excluded, the goodness-of-fit statistic was reduced by a quite insignificant amount. Similar patterns were found in tiie three other cohorts which had yielded apparently linear exposure-response relationships for cumulative asbestos exposure. (In the fifth, there was little evidence of excess risk by any method of analysis.) In a later neportz' on one of these other cohorts, the same authors pointed out that their results were consistent with a relationship in which risk of lung cancer a) was absent at [relatively] low concentrations; b) increased rapidly as concentrations increased; and c) levelled off at high concentrations. Comments a) and b) could also be made about their reailts on the other three cohorts, and comment c) for the Quebec miness and millert, the only other cohort in which a substantial proportion of members had been exposed to intensities over 10 mpcf (i.e. over approximately 35 f(ml). These authors also ststed however that, when risk was modelled as a function of cumulative exposure, the fit was almost as good as just described. Although there is insufficient evidence to postulate a threshold level of intensity below which there is no risk of lung cancer, the exposure-rtsponse relationship does appear swb-line:r at levels of intensity up to at ieast 3 nipcf (roughly 10 flml) for textile workers and versniculite miners, or up to 10 mpcf (or ca. 35 flnil) for chrysotile miners and millers. Sumfiary In the light of all the evidence, it is clear that there are very strong exposure-response relationships for asbestos and both mesothelioma and lung cancer. The risk of nxsothelioma after only a few months exposure, at least to chrysotile, is vanishingly low. The risk after long exposure to asbestos seems less than linearity would predict, and therefore much less than exponential. Thus the relationships appear neither linear nor exponential, but probably sigmoid. The risk of lung cancer is closely related to cumulative exposure to asbestos, but probably not in linear fashion. The risk seems to be even more closely related to the duration of exposure to high concentrations; it appears very slight at concentrations up to 10 flnd or more, increasing with severity up to around 25 flm3, and then levelling off. 'lbe relation with cumulative exposure would appear to be sigmoid, i.e. sub-linear at low and at high concentrations, but relatively steep in between. -5-

DIFFERENTIAL CARCINOGENICITY OF CHRYSOTILE AND AMPHIBOLE ASBESTOS Mesorhelioma Table 3 sumrnarises information on mesothelioma from 28 male asbestos-exposed cohorts, classified by fibre type and process, and arranged in ascending order of Proportional Mortality Rate (PMR), i.e. the number of deaths due to mesothelioma expressed as a percentage of the total number of deaths. The results are reported, as far as possible, from 20 years since first employment. It is fully appreciated that PMRS are affected by the age distribution of the cohort, the intensity and duration of employment, and the length of follow-up. These factors may account at least in part for the heterogeneity within the three classes marked by asterisks; within each of the other five classes, the PMRs were similar. The 30-fold difference in PMRs for cohorts exposed to the tlirae main types of asbestos (0.296 for chrysotile alone, 2.9 R for amosite alone, and 6.6 % for crocidolite alone) just canno+t be explained by variations in method, even those mentioned above. The differences are emphasised by the fict that there was no overlap between these three classes: the highest PMR for chrysotile alone was 0.3 9G, the lowest for amosite alone 1.556; the highest for amosite alone was 4.6 % and the lowest for crocidolite alone 4.956. Further, in 11 fernale cohorts, the pattern was aimilar: the PMR for three cohorts exposed to cbrysotile alone was 0.2% (1 mesothelioma among 414 deaths); for six cohorts exposed to mixtures it was (38/1049 -) 3.6%; and for two cohorts exposed to amphibole only was (19/179 -) 10.6%. The only comparisons that can take into account most of the factors mentioned above are made in Table 4, for miners and millers of chrysotile,' amosite and crocidolite,' the numbers of deaths here are since first employment (not after 20 years). The PMRs follow the same pattern, but the chrysotile cohort was followed for much longer than were the amphibole cohorts, and the average rumulative exposure was many times higher. The last line of this tibie gives heuristic estimates taking account of differences in exposure; it seems clear that the risk of mesothelioma from chrysotile exposure is at least one order of magnitude less than from amosite exposure, itself sn order less than from crocidolite. The chrysotile risk is substantially lower still when account is also taken of the length of follow-up. Lung cancer Table 5 gives Standardized Mortality Ratios (SMRs), with confidence intervals, for lung cancer in 32 cohorts, classified by fibre type and process; the SMR takes into account the age constitution of the cohort, but exposure and length of follow-up are not allowed for. In the nine cohorts exposed solely or predominantly to chrysotile (but not in textiles). the lung cancer SMRs were much lower than in cohorts exposed to amphibole. SMRs for insulation were very much higher again; and in textiles, the SMRs were higher than where the fibre was used otherwise, even although the intensity of expOSUre was usually less. The 'rdative slopes' of Liddell and Hanley,7D augmented by the few more recent findings, provide the best indicators of differential carcinogenicity; in the following sanunary, adapted from a recent review by Hugbea," the term SMR[i0] indicates the lung cancer SMR predicted from the relevant slope at a cumulative exposure of 10 (fibres/ml) x years. The steepest slopes (approximately 1.0) were in textile factories, using ;chrysotile or mixtures, and in an asbestos-cement plant regularly using crocidolite; the corresponding SMRj10] is 1.1. The slopes were probably also as high as 1.0 for amosite factory workers and for Wittenoom crocidolite miners and millers. A slope of approximately 0.2 (SMRj10] $ 1.02) was observed for manufacturing retirees many of whom had ltad mixed fibre exposures. Much shallowex slopes were reported for four cohorts exposed solely or predominantly to chrysotile, in mining and milling, friction product manufacture, and asbestos-cement manufacture. The steepest of these slopes was•0.05, for which the SMR[10] is 1.005. Summary The risk of inesothelioma after exposure to chrysotile is at lasi two orders of magnitude less than from similarly severe exposure to crocidolite. The hazard from amosite exposure appears one order worse than from chrysotile and one order less than from crocidolite. The risk of excess lung cancer as a result of exposure to chrysotile (except in textiles) is less, by more than one order of magnitude, than that from crocidolite ratposure. Asbestos textile manufacture is associated with high risks of lung cancer, commensurate with those from crocidolite, and amosite may carry a similar baard of lung amcer. Em+ot That all forms of asbestos are equally hazardous is seeo to be a myth; this commonly-held belief is completely untrue. , F.STiMATION OF LUNG CANCER RISK Table 6 presents estimates of risks of excess lung cancer following asbestos exposure, based on the slopes discussed above and in the recent chapter by Hughes:-" the slopes are in the first line, followed by the methods -6-

of calculating SMR and excess risk. The risk estimates are soen to depend not only on the level of exposure but also on the background life-time risk of lung cancer, two examples are given in detail. The approximate nature of these estimates must be emphasised; also that the slope of 0.05 used for chrysotile is an upper bound. Further, all estimates are on the basis of proportionality, i.e. are considerably higher than would be given by the sub-linear exposure-response relationships that appear to hold at exposure intensities below about 10 fibres/ml. From basic considerations, a person exposed occupationally to 1 fibrelm] inhales:-1A X 10' fibres a minute, assuming minute voiume of 20 litres/min, which is 120 X 10 ` fibres an hour; i.e. 30 X 10 6 fibres in a S-day week of S hours a day; which, in a 48-week working year, amounts to 14 X 10' fibres a year. Thus, exposure at this intensity for 20 years implies the inhalation of 28 x 10' fibres. According to Table 6, such exposure to crocidolite entails an excess lung cancer risk of 2096 of the background. Therefore, it takes inhalation of 28 x 10' crocidolite fibres to produce a risk of lung cancer of 0.02 (on a 10 k background). So the inhalation of a single crocidolite fibre is associated - on these bases - with a lung cancer risk of less than (0.02)1(28 x 10') = 7.1 x 10 '" = 1/(1.4 x 10'=), i.e. much less than one in a tttousand-billion. Witb slope 0.05 but otherwise on the same bases, the risk from inhaling one cluysotile fibre would be 20 times lower. On mot+e realistic assumptions and a 5% background, the risk would be down to about one in a hundred- thousand-billion. F.nuoi That one asbestos fibre can kill is seen to be a myth; this commonly-held belief is completely without foundation. CONCLUSION The ill effects of very severe occupational exposure to asbestos have included respiratory fibrosis (asbestosis), lung cancer and mesothelioma. Until the 1930s, when there was little if any environmental control, asbestosis was a common outcome, often manifest within ten years of first employment, and frequently fatal. However, as BecklakO remarks, with the> implementation of progressively stringent environmental controls, this disease appears to clinicians and pathologists to be less frequent, less severe, and less likely to cause impiirnatt. Thus, in 1989, when a group of experts met in Oxford under the auspices of the World Health Organisation to recommend an occupational safety limit based on health grounds, they considerad that controls to reduce the risk of asbes7os-related eaacer to 'a«xptsble' limits would effectively eradicate asbestosis. The interval between first exposure to asbestos and the manifestation of related cancers is long - very seldom less than 20 years, and about 40 years on average - and is not related to degree of exposure. Thus, today's asbestos-related caacers were undoubtedly initiated by exposures 1935 - 1965, when the concentrations of respirable fibres in the occupational environment wene about two orders of magnitude higher than in the 1980s. Disrsse caused by today's exposuses will be much less - probably by at least two orders of magnitude - w}ten . it does develop, but that will not be until well into the 21st-Ceatury. The risks of cancer depend greatly not only on the degree of exposure but also on fibre type and industrial process. Although chrysotile is and has been the predominant asbestos fibre, with amphiboles always accounting for only 5-10% utilisation, the majority of disease has been caused by crocidolite and to a lesser extent amosite. The WHO experts in 1989 recommended that amosite and crocidolite should not be used if at all possible. As asbestos textile manufacture has been associated with a high risk of lung cencer, whatever the fibre type, and as it is not yet known with certainty how to manufacture textiles from chrysotile with adequate safety, this use should be proscribed, along with any use of amosite or crocidoiite. Expassue-response relationships are very strong, and are probably sub-lineu: for tnesothelioma, in relation to durations of less than, say, six months; for lung cancer, in relation to accumulated exposures up to, say, 10 (fibreslml) x years. Estimates of risks made on the basis of proportionality, as almost all are, are therefore gross overestimates. Chrysotiie except in textile manufacture has caused very little disease except when exposures were astronomically high. When controlled to in intensity of I fibreJml, the risk of excess lung cancer after 10 years employment is only 0.5% of background, assuming proportionality with a slope of 0.05 (we Table 8). But both these assumptions are liberal; the true risk is very much lower. The risk of mesothelioma as a result of controlled exposure to chrysotile also appears very low indeed. N . ~.n O ~ - -r -v W ~~J

ENDNOTES 1. In Chambers Concise 2Qrh Ceniury Dictionary, myth is defined as: a figment; a commonly-held belief that is untrue, or without foundation. 2 Merewether ERA, Price CW. Report on effects of asbestos dust in the lungs and dust suppression in the asbestos industry. HMSO, London 1930. 3. Dreesen WC, Dalavalle JM. Edwards Tl, Miller JW, Sayers RR, Easom HG, Trice MF. A study of asbestosis in the textile industry. Pub Health Bull 241. US Public Health Servioes, Washington DC 1938. 4. Doll RS. Mortality from lung cancer in asbestos workers. Br J Ind Med 1955; 12: 81-86. 5. Wagner IC, SlePga CA, Marchand P. Diffuse pleural axaothelionn and asbestos exposure in the Nordrmestem Cape Province. Br J Ind Med 1960; 17: 260-271. 6. Union Internationale Contre le Cancer. Report and recommendations of the working group on asbestos and cancer. In: Biological Effects of Asbestos. Ann. N. Y. Acad. Sd. 1965; 132: 706-721. 7. Selikoff U, Hammond EC, Seidman H. Mortality experience of insulation workers in the United States and Canada, 1943-1976. Ann NY Acsd Sci 1979; 330: 91-116. 8. McDonald, J.C., McDonald, A.D., Gibbs, G.W., Siemistyclci, J., and Rossiter, C.E. 1971. Mortality in the chrysotile asbestos mines and mills of Quebec. Arch. E'nvfran. Health 22: 677-686. 9. McDonald IC, Liddell FDK, Gibbs GW, Eyssen GE, McDonald AD. Dust exposure and mortality in chrysotile mining, 1910-75. Br J lnd Med 1980; 37: 1 t 24. 10. Liddell FDK, Thomas DC, Gibbs GW, McDonald JC. Fibre exposure and mortality from pneumoconiosis, respiratory and abdominal malignancies in chrysotile production in Quebec, 1926-75. Ann Acsd Med Singapore 1984; 13 no 2 suppl: 340-344. 11. Liddell, F.D.K., McDonald, A.D., and McDonald, I.C. The 1891-1920 birth cohort of Quebec chrysotile miners and millers: a preliminary report on mortality to 1988. In: Proceedings of the 9th International Symposirrm on Epidemiology in Occupational Health, C'incinnatt, September, 1992. In press. 12. Schneiderman MS, Nisbet ICT, Breti SM. Assessment of risks posed by exposure to low levels of asbestos in the general environment. Bundesgesundheitsamt (BGA) Berichte 1981; 4 part 3: 1-28. 13. Elmes P. Conflicts in the evidence on the health effects of mineral fibres. In: Liddell D, Miller K. Eds. Mineral Fibers and Health. CRC Press Inc, Boca Rston FL 1991: 321-335. 14. Liddell FDK. Latent periods in lung cancer mortality in relation to asbestos dose and smolcing. In: Wagaer IC Ed. Biological Effects of Mineral Fibres. IARC Scientific Publications 30. Interaationai Agency for Research on Cancer, Lyon 1980: 661-665. 15. Liddell D. Asbestos in the occupational eavironmeat. In: Liddell D, Miller K. Eds. Mineral Fibers and Health. CRC Press Inc, Boca Raton FL 1991: 79-87. 16. Doll R, Peto J. Effects on Health of Exposure to Asbestos. Hesith and Safety Commission 1985. I,oadon: HMSO, pages 33-40. ~ ~ C~ 17. Liddell, a. [1992, in press]. Exposure-respoasa: asbestos and mesotheliomi. Eur. Respir. Rev. ~ ~ 18. McDonald JC, McDonald AD. Epidemiotogy•of aesotheiioma. In: Liddell D, Miller K. Eds. Mineral Fibers and Health. CRC Press Inc, Boca Raton FL 1991: 147-168. ~ '+D ~ -8-

19. Browne K, Smither WJ. Asbesios-related mesothelioma: factors discriminating between pleural and peritoneal sites. Br J Ind Med 1983; 40: 145-152. 20. I,iddell FDK, Hanley JA. Relations between asbestos exposure and lung cancer SMRs in occupational cohort studies. Br J Ind Med 1985; 42: 389-396. 21. Vacek PM, McDonald JC. Effect of intensity in asbestos cohort ezposurc-response analyses. In: Sakurai H, et al. Eds. Occupational Epidemiology. Elsevier Science Publications 1990: 189-193. 22. Vacek PM, McDonald JC. Risk assessment using exposure intensity: an application to vermiculite mining. Br d Ind Med 1991; 48: 543-547. 23. Stuis-Cremer, G.K., Liddell, F.D.K., Logan,W.P.D., I3ezuidathout, B.N. 1992. The mortslity of amphibole miners in South Africa, 1946-80. Br. J. lnd. Med. 49: 566-575. 24. Hughes JM. Epidemiology of lung cancer in relation to asbestos exposure. In: Liddell D, Mi1ia K. Eds. Minersl Fibers and Health. CRC Prcas Inc, Boca Raton FL 1991: 135-145. 25. Becklake MR. The epidemiology of asbestosis. In: Liddell D, Miller K. Eds. Mineral Fibers and Health. CRC Press inc, Boca Raton FL 1991: 103-119. 0 -9-