Document 449veZOQEn5Rk3MJQp312vk3j

SBufeNALjQEifcJD-USTRIAL 8 {1997} Radiograpjnra^iiiga qslslsNot a Prerequisite for AlsIftEsttaSS '"** Lung Cancer Cement Workers fljrtkelstein, PhD, mdcm* In recent years, *c5rit} r'^mBUL of a prospectiv^Wd^M, relation toraSia^p^^* were mterp/etpWiwjSij^ an 1L0 code$$lf$ asbestosis letter, fhelmgxmlf years latency $`d$WM with asbesiosis^mm or later. THeftiiMm (95%CI 2,7%Sillmfj ashestos^exp'df^^MppWi. dernonsmitea^SmSm they have.a' lsBmoyp* (isbestosisfiimfi Wwifpedfaboat whether pre-existing asbestosis is a Uspejtias-rflated lung cancer. This paper presents the results Jm^ntanoijzsbestos-cement workers, of lung cancer in wJiofarWiSyearsfromfirst exposure to asbestos. Radiographs mwBrfified "B!`reader, and asbestosis was defined to mean '^vezj0.43,.subjects (123 without asbestosis, 20 with Yfipr- mterpretaiiort at 20 years from first exposure or mtortdliiyraiio(SMRJ among men without asbestosis at 20 were 128 subjects (114 without asbestosis. 14 mitablefioFdnterpretation at 25 years from first exposure ffieznentwithout asbestosis at 25 years latency was 5.81 mmfiamconsistent with those of epidemiologic studies of ggwKwHfty of exposure situations. These studies have uekvated in the presence of radiographic asbestosis. but ffeer^jsk.may be elevated in the absence of radiographic ?|F8?i997. 1997 Wiley-Liss. Inc. KEY WORDSfeptcfeituo tdy;. asbestosis; lung cancer; SMR INTRODUCTION a: W Mfc Hughes and Weill [1991], In 1984, a colleague and I published a study of radiographic abnormalities in a longitu in recent years a con si derabl'e.icon&^pwv [ has' He1, el dinal investigation of 181 production workers from this oped about whether pre-existin&lstjesiMM8p-brereoujsite ' plant [Finkelstein and Vingilis, 1984], With follow-up to for the diagnosis of asbestos^related^^^^^St_er_^_ljones'er;. 1980, it was observed that men with radiographic abnorrrmii- at.. 1996; Egiimaa and Reinerlj|f99_ ^JS^j^^yrdies had higher lung cancer mortality rates than men with Abraham [ i 994J the controvefsy||ao|$ SU- . normal radiographs. Because a ``best evidence" ciassifica- ence, but in litigation" and ther acceptance! s,he other can mak, e a ,la r.ge d..i-tate. r4eSntf cedivi cation. This paper presents newSftdtnat: ytm-?;0vi-t`.-0n or ^^jS5kpi,ssue` ^0E1 f cause of death was used, no comparisons were made t0 *^e tnortaiity experience of the general population and it was no possible to say whether the lung cancer risk of a In 1948 an American rnultinationaludd^^fatiorf'^opened; factory for the manufacture o^g^emenfpfoducts;-- . mwoermkbeersrs woifthiho7ugtenaesbraelstosis was increased in relation to population. Since information front in Scarborough, Ontario. This was the same corporation that this population is directly relevant to the question of owned the Louisiana asbestos^emenfepiantspistudied .by asbestosis as a precursor to lung cancer, 1 have updated the -T mortality experience of the workers in this cohort to the end Department of Family anti Community Medlcine^Srtment of Preventive of 1985 and have compared their mortality experience to Meflic.ne and Siostatistics, University of Torontoi:Toronti>(.iCinada ''lt-.,j;;-fC-;-... that of the general population of Ontario. A clear and 'Correspondence to: Dr, Murray Finkelsternf40Q University Avenue,-7ttt Roar, unambiguous finding is that lung cancer mortality was Toronto, Ontario, Canada M?A 1T7; E-mail;mufrayfir!,ke!Stefn@utoro'nto.ci Accep' ted for puMcafa1i21 6 Fetjruary ' V '-. ^ * '' ' significantly increased among workers without asbestosis on their rad [graphs, c 1997 Wiley-Liss, Inc. 342 Finkelstein et al. METHODS <`he Factory and Exposure The workers in this cohort were employed at a factory that manufactured asbestos-cement pipe and board from chfysotile and crocidoiite asbestos, silica, and cement. Cumulative exposures to asbestos were calculated using a model that extrapolated measurements made by the per sonal membrane filter, a method that came into use 21 years after the plant opened in 1948 [Finkelstein, 1984]. Estimates of exposure to the silica component of the dust cloud are not available. Eighteen-year cumulative expo sures were calculated by summing annual exposures during the 18 years from first exposure; later accumulations were generally well less than 10% of the 18-year totals. 5-year intervals (determined by film quality) were coded. Radiographs were interpreted without knowledge of the worker's cumulative exposure category, the last year he had worked in asbestos exposure, or his smoking habit. Smoking Information Beginning in the early 1970s, pulmonary function technicians obtained smoking information at the annual examinations. Missing information was sought by interview and from physicians' records. Men were classified as never smokers, as ex-smokers if they had given up cigarettes within 10 years of first exposure, or as current smokers if they smoked cigarettes beyond 10 years from first exposure. Statistical Analysis The Study Population and Follow-up The study cohort consists of production workers hired prior to 1960, who were employed for 9 years or more, and who worked for at least 12 months in asbestos exposure. There were 186 men who were eligible for inclusion. At least one radiograph was available for interpretation for 151 men; these are the subjects of the present study. Mortality in the cohort was ascertained to the end of"1985 by record linkge to the Canadian Mortality Database at Statistics Canada. The Radiographs and Their Interpretation While employed, workers received annual chest radio graphs as part of their routine medical surveillance by the Occupational Chest Disease Service of the Ontario Ministry of Labour, and most of these films were available. Many long-service employees returned for ongoing surveillance after separation, and others who had filed claims with the Workers' Compensation Board were also followed by the Chest Service. In 1980, former workers who had not had a recent radiograph were invited to return for an examination, and we sought radiographs of deceased workers from hospitals. Radiographs (posteroanterior projections) were inter preted by a single National Institute for Occupational Safety and Health (NIGSH)-certified "B" reader using the 1971 TLO Classification of Radiographs of the Pneumoconioses as reference standard. Film quality was somewhat variable (most films had been obtained during routine surveys), but it was believed that reasonably consistent interpretations could be made. File folders were retrieved and stacked in unor dered sequence. Individual files were examined in known emporal order. To avoid coding many normal films, the files ere usually scanned quickly to find the earliest abnormal film, and this film and later ones taken at approximately The radiographic abnormalities selected for analysis were small irregular opacities and bilateral pleural thicken ing, both associated with asbestos exposure. Radiographic asbestosis was defined as the presence on the radiograph of smalt irregular opacities, ILO category 1/0 or more. It was previously reported that asbestosis was a progressive disor der in this cohort [Finkelstein and Vingilis, 1984], I thus selected three evaluation times: 20 years; 25 years, 30 years from first exposure as times at which to classify the workers with respect to asbestosis and to begin mortality foilow-up. Comparison was made to the mortality experience of the general population of Ontario using the person-years com puter program [Coleman et al., 1986]. Confidence intervals (95%CI) were computed assuming a Poisson distribution for the numbers of observed deaths. Cox proportional hazards regression analysis was used to examine risk while control ling for confounders [SPSS, 1994], RESULTS The 1984 [Finkelstein and Vingilis, 1984] study demon strated that the risk of developing radiographic asbestosis depended on the time from first exposure and on the cumulative exposure to asbestos. The time of assessment is thus important, since a worker without asbestosis at 20 years may have progressed to clinical asbestosis at 25 years from first exposure. Mortality in Relation to Radiographic Asbestosis {Small Opacities >1/0) Mortality in Relation to Radiographic Coding at 20 Years from First Exposure There were 143 subjects (123 without asbestosis, 20 with asbestosis) with a radiograph available for interpretation at 20 years from first exposure or later. Table I shows the standardized mortality ratios (SMRs). All Cause mortality Asbestosis Is Not a Prerequisite for Lung Cancer 343 TABLE I. Standardized Mortality Patios in Relation to Latency and Radiographic Asbestosis: Asbestosis = ILO l/0 Causa of death Asbestosis 20 years latency ------------------------------------------------------------------------------------ Obs Exp SMR 9s%ei Obs 25 years latency "........... Exp SMR 95%C1 30 years latency ................................................... .. ................. Obs Exp SMR 95%C1 Ait causes ' Ai! malignancies Digestive cancer IC0150-159 Lung cancer Pleural mesothelioma Circulator disease Respiratory disease pneumoconiosis No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes 45 22.2 2.02 1.47-2.70 36 14.8 2.43 1.70-3.36 1 2.25 0.44 0.01-2.45 15 4.4 3.43 - t.92-5.66 9 2.24 4.01 1.83-7.61 11 2.66 4.13 2.06-7.39 28 6.03 4.65 3.09-3,72 23 4.23 5.44 2.68-6.35 0 0.68 0 -- 5 1.14 4.30 1.42-10.2 3 0.61 4.92 1.01-14.4 6 0.75 7.98 2.93-17.4 8 1.84 4.34 -- 7 1.27 5.50 -- 0 0.20 0 -- 1 0.36 2-76 -- 1 0.19 5.32 -- 3 0.23 13.20 -- 12 2.17 5.53 2.86-9.66 9 1.55 5.81- 2.66-11.0 0 0.25 0 0-8 4 0.40 9.96 2.71-25.5 2 0.22 9.20 1.11-33 2 0,27 7.48 0.90-27 3-- 0-- -- -- -- -- 2-- 0-- -- -- -- _ 0-- 1-- -- -- -- 8 11.0 0.73 0.3-1,4 7 7.26 0.96 0.4-2.0 1 1.03 0.93 0.02-5.2 5 2.30 2.18 07-5.1 3 1.28 2.35 0.5H5.9 3 1.18 2.59 0.5-7.6 2 0,89 2.26 0.3-8.2 2 1,33 1.50 0.2-5,4 0 0,14 0 -- 4 0.29 13.80 3.76-55 3 0.16 19.30 4.0-56 1 0.19 5.20' 0.1-29 1 0.06 18.00 -- 0 0.04 0 ~ 0 0.01 0 -- 2 0.01 -- 1 0.01 -- -- 1 0.01 -- -- was significantly increased among both groups; 3.43 among the men with asbestosis, and 2,02 among the men without asbestosis. Mortality from digestive cancer (including peritoneal mesothelioma) was significantly elevated in both groups. Overall, there were three carcinomas of the stomach, one of the rectum, four peritoneal mesotheliomas, and one gastrointestinal carcinoma with uncertain primary. With respect to lung cancer, the lung cancer SMR among men without asbestosis at 20 years latency was substantially and significantly elevated at 5.53. The lung cancer SMR among the men with asbestosis was almost twice as high but, because' of the small sample size, the confidence intervals'overlapped, and there was no signifi cant difference between the two groups of workers. Workers with asbestosis had an increased SMR for circulatory disease, but the increase was not significant. Both groups of workers had increased mortality from respiratory disease, including pneumoconiosis. Mortaiity in relation to radiographic coding at 25 years from first exposure lung cancer SMR among the men with asbestosis was almost twice as high, but, because of the small sample size, the confidence intervals overlapped, and there was no signifi cant difference between the two groups of workers. Workers with asbestosis had an increased SMR for circulatory disease, but the increase was not significant. Both groups of workers had increased mortality from respiratory disease. Mortality in relation to radiographic coding at 30 years from first exposure There were only 33 subjects (15 without asbestosis, 18 with asbestosis) who were still alive and had sufficient time from first exposure to have a radiograph available for interpretation at 30 years from first exposure or later. Table I shows the SMRs. All cause mortality was significantly increased among the men with asbestosis, but there was only one death, versus 2.25 expected among the men without asbestosis. There were two deaths from lung cancer, both among men with asbestosis. The number of subjects was so small that the SMR, 7.46, among men with asbestosis was not statistically significant. There were 128 subjects (114 without asbestosis, 14 with asbestosis) with a radiograph available for interpreta tion at 25 years from first exposure or later. Table I shows the SMRs. AH cause mortality was significantly increased among both groups; 4.01 among the men with asbestosis and 2.43 among the men without asbestosis. With respect to lung cancer, the lung cancer SMR among men without asbestosis at 25 years latency was significantly elevated at 5.81. The Investigations of Other Factors Potentially Associated With Lung Cancer Mortality Cumulative exposure Workers were divided among the five exposure categories used in the 1984 asbestosis paper [Finkelstein and Vingilts, 344 Finkelstein et a). TABLE It. Lung Cancer Standardized Mortality Ratios in Relation to Cumulative Exposure: Follow-up from 20 Years After First Exposure Cumulative exposure Obs Exp SMR (95%C.I.) <50 f-y/ml1 50-99 f-y/ml 100-149 f-y/ml 150-199 f-y/ml 200 f-y/mt . M-y/ml = fifief-years/tsii. 2 0.35 5.76 (0.7-21) 5 0.91 5.51 (1.8-12.9) 7 0.73 9.60 (3.9-20) 3 0.44 6.87(1.4-20) 3 0,33 . 9.06(1.9-26) 1984], The results for lung cancer are presented in Table H. There is the suggestion of increasing lung cancer SMRs with increasing cumulative exposure, but the confidence intervals are very wide, and there are no significant differences among the exposure groups. In this small group of men, the influence of competing causes of death such as mesothelioma, gastrointestinal cancer and asbestosis has a substantial influence on the observed mortality rates for lung cancer. Figure J shows the mortality rates for lung cancer, as well as for "all asbestos diseases" in relation to exposure. The pattern for "all asbestos diseases" is more monotonic than is the one for lung cancer. Smoking ered: age, asbestosis, smoking, and cumulative exposure, only age was significantly associated with lung cancer risk. For asbestosis as a risk factor, the relative risk for lung cancer for men with asbestosis, compaT&d to men without, was 1.4 (P - 0.58). This compares with a ratio of 1.8 for the SMRs for lung cancer in Table I. Time-dependent Cox analysis The probability of development of radiographic asbesto sis increases with time from first exposure. This issue was addressed previously by examining risk in relation to three evaluation times at 20, 25, and 30 years from first exposure. Another approach is to follow the cohort through time and to take note of the time at which asbestosis is diagnosed. This approach was canned out with a time-dependent analysis. Of the four variables considered: age, asbestosis, smoking, and cumulative exposure, only age was significantly associated with lung cancer risk. For asbestosis as a risk factor, the relative risk for lung cancer for men with asbestosis, compared to men without, was 2.3 (P = 0,15; 1-tailed P = 0.07). This compares with a ratio of 1.8 for the SMRs for lung cancer in Table I. Pleural thickening There were only three workers at each of 20,25, and 30 years latency who had pleural thickening as the only Beginning in the early 1970s, pulmonary function technicians obtained smoking information at the annual -examinations. Missing information was sought by interview and from physicians' records. Men were classified as never smokers, as ex-smokers if they had given up cigarettes within 10 years of first exposure, or as current smokers if they smoked cigarettes later than 10 years from first exposure. For analysis here, the former and current smokers were combined. The only deaths from respiratory disease occurred among the smokers (Obs = 7, Exp 1.32, SMR 5.32). However, three men who reported that they had never smoked died of lung cancer, and there was no difference in lung cancer SMRs between the smokers (Obs: 14, Exp: 2.18, SMR: 6.44) and nonsmokers (Obs = 3, Exp = 0.49, SMR = 6.18). 1 23 Exposure Group 4 5 Multivariable analysis Cox regression analyses were performed in order to examine the simultaneous effects of all the lung cancer risk factors', it must be emphasized that the Cox analyses provide only internal comparisons, and do not permit comparison with the general population. Analyses were conducted using the asbestosis classifications at 20 and 25 years latency. The results essentially reproduce the relationships in the SMR calculations shown previously. Of the four variables consid B Lung Cancer -y,,- All Asbestos Disease FIGURE 1. SMRs for lung cancer and alt asbestos--ssociated disease in relation to estimates of 16-year cumulative exposure. Group 1: <50 f-y/ml; group 2: 50-99 f-y/fflf; group 3:100-149 f-y/ml; group 4:150-199 f-y/mf; group 5: 200 f-y/ml (f-y/mf = fiber-years/ml). mm Asbestosis Is Not a Prerequisite for lung Cancer 345 radiographic abnormality. All three workers survived to the asbestosis but, in a number of instances, the pathologists end of follow-up. reported histologic asbestosis while the radiologist did not 1 gs?1 DISCUSSION code radiographic asbestosis. Kipen and colleagues [1987] reported on pathologic-radiologic comparisons in 138 lung cancer cases from the American insulation workers study. There has been substantial controversy in recent years as to whether asbestosis is a prerequisite for the develop ment of asbestos-related lung cancer. Much weight has been given in these discussions to the results of Hughes and Weill [1991] from a study at a New Orleans asbestos-cement manufacturer. Hughes and Weill used the results from a 1969 radiographic survey of the workforce as input to a prospec tive study of mortality. By coincidence, in 1984 I had published the results of a prospective study of the relation ship between asbestosis and mortality among Ontario em ployees of the same multinational corporation [Finkelstein and Vingtlis, 1984], The present analysis updates the mortality experience to the end of 1985 and carries out comparisons with the general population of Ontario, Several important characteristics of radiographic asbes tosis must be borne in mind, First, asbestosis is a disorder that usually takes 15 or more years to become evident and which often progresses in severity. Second, the probability of developing asbestosis is related to the amount of asbestos dust inhaled. Third, workers who smoke have a higher probability of being diagnosed with asbestosis than workers who do not smoke [Finkelstein and Vingilis, 1984], These All 138 cases had histological evidence of parenchymal fibrosis, but in 25 (18%) there was no radiographic evidence of fibrosis. There was no significant difference between Vingtlis and Kipen and colleagues with respect to the proportion of subjects who had histologic asbestosis but whose radiographs were coded < 1/0. Analysis of the Ontario data shows that asbestoscement workers without radiographic asbestosis had a significantly increased risk of lung cancer in comparison to men in the general population. An advantage of this study over those which use cross-sectional data is that the employ ees of the plant were under annual surveillance by the Occupational Chest Disease Service of the Ontario Ministry of Labour, so that temporal changes in their chest x-ray status could be assessed. Increased lung cancer risk among workers without asbestosis was apparent in relation to classifications of the radiographs at 20 and 25 years from first exposure to asbestos. There were no lung cancer deaths observed among workers without asbestosis at 30 years latency, but the number of surviving workers was so few that only 0.25 deaths were expected. The 95% confidence interval on the lung cancer SMR at 30 years latency ranged from 0 to 8.0. factors are consistent in the prediction that workers without Surprisingly, we found no relation between lung cancer asbestosis will have a lower lung cancer risk than will risk and smoking habit. Three of 20 men with lung cancer workers with asbestosis, since workers without asbestosis reported that they had never smoked. 1 suspect that these will, on average, have inhaled less dust and have smoked men were misclassified as never-smokers. In any event, the less than workers with asbestosis. The question at issue is lung cancer risk in the cohort was so large that any whether workers exposed to asbestos, but without asbesto misdassificacion of smoking habits wil l not alter the conclu sis, might have a risk of lung cancer increased above the sions about lung cancer risk in comparison with the general baseline risk In the general population, population. A higher prevalence of smoking among blue In this Study, radiographic asbestosis was defined to collar workers might be expected to increase the lung cancer mean an ILO code for small irregular opacities of 1/0 or SMR by only about 20% [Finkelstein et al,, I987J. ' more. The radiographs in this study were interpreted by a Figure 1 shows that lung cancer risk tended to increase single NIOSH-certilied "B" reader (Dr. J. Vingilis). Work with exposure, but there was no significant relation between : ers without asbestosis were found to have an increased risk estimates of cumulative exposure and lung cancer risk in this of lung cancer, but might the reader have coded as normal cohort. This is probably related to the small population size those films that other readers would have coded ILO (which resulted in wide confidence intervals oh the estimates s- category 1/0 or more? One piece of evidence that the of SMR) and the effects of "competition?from other-'." yW ' ;, h. radiographic coding was valid and consistent is the dose- response relationship observed between the risk of asbesto How do the results of this stuaymDnrano.asbestos-rSwfette sis and estimates of cumulative exposure [Finkelstein and cement workers compare with those'&f-^Hughes and Weill .-5$ Vingilis, 1984]. In addition, Table 5 of the 1984 paper from New Orleans? There are some difficulties'In making [Finkelstein and Vingtlis, 1984] shows a comparison of the the comparison. Hughes and Weill --included two New radiographic codes and histologic findings for 26 workers Orleans plants in their studies. Plant I was housed in a single from the cohort. The pathologic interpretations were made building and manufactured building products; Plant 2 con by a number of local pathologists without any uniform sisted of four separate buildings, each manufacturing differ protocol for grading the specimens. In general, the opinions ent products, including asbcstos-cenient^pipe. The lung of the radiologist and the pathologists were in reasonable cancer risk in these two New Odeanffplahts was verv agreement with respect to the presence or absence of different, being twice as high in PI ant'2 os iPwas in Plant 346 Finkelstein et al. TABLE 111, Comparison of New Orleans and Ontario Studies of Asbestos-Cement Workers Latency (Ontario) Ontario plant Obs Exp SNia 95%Ci New Orleans plant (all latencies) Obs Exp SMR 95%CI L Small Opacities >1/0 20 yr 25 yr 30 yr 4 0.40 9.96 2.7-25 9 2.1 4.32 1.98-8.2 2 0.22 9.20 . 1.11-33 ---- -- -- 2 0.27 7.46 0J0-27 -- -- -- -- ll. No radiographic asbestosis 20 yr 25 yr 12 2.17 5.53 2.86-9.96 16 13.3 1.20 0.69-1.95 9 1,55 5.81 2.66-11 ---- -- -- Total (no asbestosis): Ontario (20 yr) and New Orleans Total (no asbestosis): Ontario (25 yr) and New Orleans 28 15.5 1.81 1.20-2.61 25 14,9 1,68 1.09-2.48 [Hughes et al., 1987], Despite these differences in lung cement workers, in the combined populations, who did not cancer mortality rases, Hughes and Weil) [199!] combined have radiographic asbestosis. the populations for their analysts of asbestosis and mortality. The finding of increased lung cancer risk among They indicate [Hughes and Weill, 1991: pg 231) that all the asbestos-cement workers without radiographic asbestosis is excess mortality occurred in the second plant, but they do consistent with epidemiologic findings in ocher settings. In not specifically tell us about the relationship between fact, epidemiologic studies show that lung cancer risk may asbestosis and mortality at the second plant, We are thus be increased among workers with, and without, radiographic presented with a mixture of two workforces, one with an asbestosis, and among workers with, and without, histologic increased risk of lung cancer and the other without, The asbestosis. The major studies are discussed below, Ontario facility was similar to Plant 2, comprising a number Liddell and McDonald [1980] studied mortality, during of buildings for the manufacture of different asbestos- 1967-1975, in a cohort of 4559 Quebec chrysotile miners cement products, including pipe, in addition to rockwool and and millers, in relation to the coding of the most recent fibreglass insulations. Ideally, one would want to compare radiograph obtained before 1967. There were 119 deaths the Ontario workers with those from Plant 2 in New Orleans, from lung cancer, as compared to an expectation of 67. but this is not possible with the data presented by Hughes Among workers with small opacities ^0/1, the lung cancer and Weiil. SMR was 3.3 (33 observed; 10.6 expected). See the Table ill compares the results of the Ontario and New Appendix For the details of a computation that demonstrates Orleans asbestos-cement studies. For workers with asbesto that among workers without small opacities the lung cancer sis, that is, among men with small opacities of a 1/0, the SMR was 1.51 (95%CI; 1.21-1.87). lung cancer rates in both plants were very high. The rates ' Hillerdal [1994] conducted a prospective cohort study were higher in Ontario, but the confidence intervals overlap, of 1,596 men from the general population of Uppsala, and there is no significant difference between New Orleans Sweden, who were found to have pleural plaques on their and Ontario for workers with small opacities s 1/0, Now, radiographs. Interviews revealed that 95% of the subjects what about workers without asbestosis? As shown in Table had a history of exposure to asbestos. During follow-up from III, lung cancer mortality was high and significantly in 1963 to 1985, the lung cancer SMR was 2.3 for men with creased among the Ontario workers. Mortality was increased parenchymal asbestosis, category 1/0 or more. The lung among the New Orleans workers without ashestosis, but not cancer SMR was also significantly increased for men with significantly so. In the last rows of the Table, I have asbestos exposure but with ILO code <1/0, that is, without combined the lung cancer experience of the New Orleans parenchymal asbestosis (SMR = 1.4;95%CI= 1.04--1.97). and Ontario asbestos-cement workers without asbestosis. deKlerk and colleagues [1996] reported on the risk We see that the combined SMRs are significantly greater factors for lung cancer in a case-control study among than 1.0 even when the experience of the New Orleans Plant Australian crocidolite miners. In a logistic regression model, I workers was included. We can thus conclude that lung both radiographic asbestosis (Category 1/0 or more) and cancer risk was significantly increased among asbestos- cumulative exposure to asbestos fibers were significantly Asbestosis Is Not a Prerequisite for Lung Cancer 34? ^associated with lung cancer risk. This suggests that lung histologic asbestosis, but also that Sung cancer risk may be cancer risk increased with cumulative exposure even among increased in the absence of histologic asbestosis. miners without asbestosis, so that some cases of lung cancer Data published in the last decade demonstrate that the attributable to asbestos exposure must have occurred among statement "radiographic asbestosis is a prerequteij.e for the miners without asbestosis. asbestos-attributable lung cancer'1 is logically untenable. i The epidemiologic studies are consistent in their esti The reasoning is as follows: mates of the additional risk attributable to asbestosis. In the - Ontario study, the time-dependent Cox analysis found a 1. Histologic asbestosis may be present when the radio , relative risk of 2.3 for men with asbestosis in comparison to graph is normal. Kipen and colleagues [1987] found that men without radiographic asbestosis. In the study conducted A : by Hillerdal {1994], the relative risk was i.6, In the mild, moderate, and severe histologic asbestosis could occur in the presence of a normal radiograph. Based on - chrysolite study reported by Liddell and (McDonald [1980], his extensive experience in South Africa, Sluis-Cremer the relative risk for miners with small opacities s 0/1 [1989, p. 540] wrote "slight asbestosis is commonly, and compared to men without small opacities was 2. Among moderate asbestosis occasionally, undetected radiologi Australian crocidolite miners [deKlerk et al,, 19%], the cally.'' relative risk for miners with opacities of &1/0 was 2.5. 2. Histologic asbestosis is a marker for increased lung Compared to workers without radiographic asbestosis, work cancer risk, Sluis-Cremer and Bezuidenhout [1989] found ers with small opacities have a risk of lung cancer that is that, in comparison to miners without histologic asbesto increased 1.5- to 2.5-fold, but asbestos-exposed workers sis, miners with histologic asbestosis had an increased without radiographic asbestosis have a risk of lung cancer risk of lung cancer of fourfold or greater, increased above that among members of the population 3. It follows that lung cancer risk may be increased among without asbestos exposure. asbestos-exposed workers even in the presence of a The only study of histologic asbestosis and lung cancer normal radiograph. risk in humans is the case-control study conducted by Sluis-Cremer and Bezuidenhout {1989} among South Afri REFERENCES can amphibole miners. They found that among miners with even slight histologic asbestosis, the relative risk of lung t-' Abraham JL (1994): Asbestos inhalation, not asbestosis, causes Sumtcancer. cancer was increased 4.5-fold. However, similar to the study Am find Med 26:839-342. " conducted by deKierk et al. [1996] in Australia, both the Coleman M, Douglas A, Hermon C, Petto J (1986): Cohort study analysis asbestosis term, and an exposure term (years of exposure) with a Foreran computer program, im J Epidemiol 15:134-137, were each significant in a logistic regression model This deKierk NH, Musk AW, Clancy JJ. Peng SC, Limd HO. Olsen N, Hobbs again suggests that lung cancer risk increased with exposure, even among men without histologic asbestosis. The authors fitted additional regression models using cumulative expo MST (1996): Crocidolite, radiographic asbestosis and subsequent iung Cancer. Presented at "Inhaled Particles." VISL Cambridge. August 1996. AnnOccup Hyg (in press). sure (f-y/m!) and residence-time-weighted exposure as the W" Egilman D, Reinert A (1996): Lung cancer and asbestos exposure: Asbestosis is not necessary. Am J (nd Med 30:398-406. .. exposure metrics. In neither of those models was the exposure term significant. A reasonable interpretation is that Finkefstein MM 0984): Mortality among employees of un Ontario the conversion of particle counts to fibers was so crude that asbestos-cement factory. Am Rev RespirDis 129:754-761. the exposure measures bore little relation to lung dose. Finksisitilt MM. Vingiiis JJ (1984); Radiographic abnormalities among asbestos-cement workers. Am Rev Rexpir Dis 129:57-22, CONCLUSIONS Ftnketein MM, Lias CM, Krammer F, Kusrak RA fl98?y Mortality among workers receiving workers' compensation awards for silicosis in Ontario; i940-l985. Br J fnd Med 44:588-594. Epidemiologic studies of asbestos-exposed populations l. Hillerdal O (1994): Pleural plaques and risk for bronchial carcinoma and in a variety of exposure situations have demonstrated that mesothelioma. Chest 105:144-150. iung cancer risk is elevated in the presence of radiographic - Hughes JM, Weill H (1991): Asbestosis as a precursor of asbestos related asbestosis, but they have also shown that lung cancer risk lung cancer. Results of a prospective mortality study. Br J Ind Med may be elevated in the absence of radiographic asbestosis. 48:229-233, These studies of North American asbestos-cement-workers, Hughes JM, Weill H, Hammad Y Y (1987): Mortality of workers employed Quebec chrysolite miners and millers, Australian crocidolite in two asbestos cement manufacturing plants. Br / fnd Med 44:J6i-|74, miners, and Swedish citizens demonstrate clearly that radio graphic asbestosis is not a prerequisite for asbestosassociated lung cancer. In addition, the study of SluisCremerand Bezuidenhout [1989] found that lung cancer risk is significantly elevated in the presence of even slight Jones SN, Hughes JM, Weill H (1996): Asbestos exposure, asbestosis. and asbestos-attributable lung cancer. Thorax 5 l(supp) 2):S9-5i5. Kipen HM, Lilis R, Suzuki Y, Valciukas JA, Selikoff U (1987): Pulmonary fibrosis in asbestos insulation workers with lung cancer: A radiological and histopathological evaluation. Br J tnd Med 44:96-100. 348 Finkeistein et al. Liddell FD, McDonald JC (1980): Radiological findings as predictors of mursaiay in Quebec asbestos workers. Br. J. Ind Med 37:257-267. Sluis-Cremer CK, Bezuidesihout BN (1989): Relation between asbestosis and bronchial cancer in amphibole asbestos miners. Br J tnd Med 46:537-540!-. SPSS Inc (1994), "SPSS for Windows Advanced Statistics Release 6.L" Chicago. APPENDIX: CALCULATION OF THE SMR FOR MINERS WITHOUT SMALL OPACITIES IN THE STUDY OF LIDDELL AND MCDONALD [1980] in the total cohort, there were 119 lung cancer deaths and 6? expected. The SMR was thus t ,77 and there were 52 excess lung cancer deaths. On page 266, the authors state that among 118 cases for which data were available, there were 33 lung cancer deaths among workers with small opacities. There were thus 118 - 33 = 85 lung cancer deaths among workers without small opacities. Table V of the paper cited above shows that among workers with completely normal films, there were 49 lung cancer deaths and the SMR was 1,08. Therefore, the number expected was 45.4. Table VI of the paper cited above shows that the relative risk for men with small opacities in comparison to men with normal films was 2.88. Now relative risk = SMR, /SMR, = Qbs/xp/1.08 = 33/Exp/I.08 and the number of expected deaths for men with small opacities was 2 0.6. Now, the total expected was 67, while the numbers expected for miners with normal films was 45.4 and for men with small opacities was 10.6. Therefore, the number expected for men with all other abnormalities was 67 - 45.4 - 10.6 IL Therefore, the number of lung cancer deaths expected among all men without small opacities was 45.4 + 11 ~ 56,4. The lung cancer SMR for miners .without small opacities was thus Obs/Exp * 85/56.4 = 1.52(95%CI: 1.21-1.87).-