Document 0gggOXgD093No8NN594yq809b

British Journal ofIndustrial Medicine 1989;46:537-540 Relation between asbestosis and bronchial-cancer in amphibole asbestos miners GKSLUIS-CREMER, B N BEZU1DENHOUT From the Epidemiology Research Unit. Medical Bureau ofOccupational Disease. PO Box 4584, Johannesburg 2000, Republic ofSouth Africa abstract In a necropsy scries of 339 amphibole asbestos miners heavy smoking, age, and the presence of asbestosis were significantly associated with the presence of bronchial cancer. Of the 35 cases of bronchial cancer, 24 were associated with asbestosis. Eleven cases of bronchial cancer occurred in men without asbestosis; all were smokers. Standardised proportional mortality rates indicated no excess of bronchial cancer in 302 exposed men without asbestosis whereas these rates were progressively raised in men with slight or moderate/severe asbestosis. Of the four exposure variables introduced separately into a logistic regression model, "years ofexposure" made a small bul significant contribution; "residence time" marginally failed to achieve a 5% level ofsignificance. Two other exposure variables tested including cumulative fibre exposure (fibre years) made no significant contribution. In the absence ofasbestosis at necropsy a bronchial cancer in a man exposed to asbestos is unlikely to be due to asbestos. The question of whether the presence of parenchymal asbestosis is a* necessary prerequisite to attribute a bronchial cancer to inhalation of asbestos is con troversial. Some compensation authorities insist that asbestosis must be shown before compensation is granted. Furthermore, some authorities believe that the dose of respirable asbestosis dust that will cause bronchial cancer is about the same1 or even greater2 than that which will cause asbestosis. The object of the present study was to determine if any parameters ofexposure to asbestos dust exert any additional risk of developing bronchial cancer after allowing for the presence of asbestosis. Materials and methods A mortality study on all white men identified from company records who were employed by amphibole asbestos mines (crocidolite and amosite) in South Africa is being analysed. Records regarding employees were preserved by different companies from between 1945 and 1956. At the latter date the keeping of records became compulsory. Black employees were not considered in this study because ofinadequacies of registration of causes of death and poor smoking and exposure data. Altogether 7318 men were identified and death certificates sought at the Department of the Interior. Of these, 1165 (15-9%) had died by 31 Aoxpicd ft August 1988 December 1980 and their death certificates were obtained. Necropsies had been performed on 427 (367%) of the men and the reports of pathologists (staffofthe National Centre for Occupational Health) on macroscopic and histological findings in the lungs were available. These reports indicate whether asbes tosis was present, its grade, and whether bronchial cancer was present (table 1). Asbestosis was graded as follows: slight asbestosis implies organisation of alveolar spaces with some interstitial thickening due to collagen orreticuiin with areas of normal lung. In moderate asbestosis changes are more extensive but less so than in pronounced asbestosis. Pronounced asbestosis implies organisa tion ofalveolar spaces and fibrosis ofalveolar walls to the extent that there is no normal lung or massive fibrosis even without the above. In all cases asbestosis bodies or fibres must be present. An unmatched case referent design was adopted for the study. Thecases were 35 individuals with bronchial cancer as identified by necropsy. Ten of the 35 cases had no mention of bronchia! carcinoma on the death certificate, yet, in eight of these necropsy clearly proved that bronchial cancer was the cause of death. In two cases the cause ofdeath was not so certain. In one there was a bronchial adenocarcinoma with meiaslases but evidence of extensive non-asbestotic fibrosis of the lungs leading to severe cor pulmonale. In the second case obstructive airways disease appeared to be the cause ofdeath but a small oat cell HWBUI0010182 538 Sluis-Cremer. Besuidathmtt Table I Coxes ofhronehitd cancer and referents by smok ing status and decree ofasbestosis Smoking suiux Never Ex tight Moderate Heavy Total Asbestosis None Slight Moderate/pronouncod Total C'dtfl No ir.i 407-1) 504-51 HOI-4) 11 <31-4) 55 11(31-4) 15(42-9) 9(25-7) 3S Referents No (Hi 35(9-4) 36 |9 9J 7009 2) 175(48 1) 4*<!3'2) 364 291 (794) 54(14 8) 19(5-2) 364 bronchogenic carcinoma was also found at necropsy. The latter two were included among the cases. Data ofsmoking habits were not available in 28 men (none with bronchial cancer) and they were excluded from the analysis. The group for the study therefore comprised 399 men (table I). The referents were the 364 men without bronchial cancer who had smoking data. The files of the cases and referants kept by the Medical Bureau for Occupational Disease (MBOD) were searched for details ofexposure to asbestos and other mining dust and for smoking habits. Smoking habit was taken as the number of cigarettes or equivalents for dgais or pipe recorded in the last MBOD clinical examination before death. Smoking status was categorised as never a smoker, ex-smoker (cessation more than two years before date ofexamin ation), current light smoker (1-9 cigarettes a day), current moderate smoker (10-I9cigarettes a day), and current heavy smoker (20 or more cigarettes a day) (table 1). Years of exposure to asbestos and residence time (time of first exposure to date of death) were known from the MBOD files. Dose ofasbestos was computed by multiplying the measured fibre/ml levels for each mine, occupation and calendar period by time a man had worked in each occupation, and summing the various exposures over his career and expressing the result as fibre years. The residence time weighted dosage (residence TWD) as described by Finkelstem' was also computed (tabic 2). To mode! the effect of exposure to asbestos on the risk ofdeveloping branchial cancer when adjusting for the effects of smoking habits, age, and asbestosis, an unconditional logistic regression analysis was perfor med using the program "RISK."' As exposure to asbestos was expressed in four ways (years of exposure, residence time, fibre years, and residence TWD), four models were created to analyse each exposure variable uniquely with smoking, age, and asbestosis. For the categorical variables smoking and asbestosis. design (indicator) variables were gen erated to contrast each category with the baseline category (no smoking and no asbestosis respectively) by the "partial" coding scheme.' To test how adequately the models fitted the data, smoking, age, asbestosis, and the respective exposure to asbestos variables were entered in a stepwise manner in that order, and the goodness of fit j? was tested for significance using the change in the log likelihood T ratio.4 Odds ratios were determined from the regression coefficients as estimators of increased risk of bronchial cancer. To test how the prevalence of bronchial cancer in the three groups, no asbestosis, slight asbestosis, and moderate pronounced asbestosis, compared with the age, race, and sex standardised rate in the white male population of South Africa, three standardised proportional mortality rates (SPMR) were estimated. Results The fit ofthe model to the data improved as each ofthe variables smoking, age, and asbestosis was entered (table 3). The most significant contributor in predict- Table 3 Goodies* offt f valuesfor the stepwise logistic regression analyses Table 2 Means andstandarddeviations (SD) for age and exposure to asbestos by cases ofbronchial cancer and referents Cases (rt " 35! Mean (SD) Referents (n 364) Mean ISD) Age Years exposure Residence time Fibre years Residence TWI> 59-5 11*2 21*3 73-7 1202-2 (84) (8-2) (74) 053-7) (2808-0) 514 41 14-6 23-7 385 7 <12-13 (6-0) (8-2) <88-3) (2161-6) TWD -- Time weighted dosage. Model Slep Term entered Improve' Degree ment in LRtf 0/ freedom p Value I Smoking 10-13 4 2 14-83 1 3 Asbestosis 30-1 2 2 4 Years exposure 6-36 s 24 Resident* time 2-78 J 3 4 Fibre years 0-11 t 44 Residence TWD 130 1 0-04 0000! <0-000! 0-02 0-10 0-74 10 'Each model contained smoking. age. asbeslosii, aad tftc exposure mutable in Slept. . _ . .. tThe maximised tog likelihood ratio f <t.R f) statistic at STEP 0 wish only a constant in the model was -- 118-59 with i DF. The LR g3 reported is the improvement in the fit of (he model In the data over the model fitted at the previous step. ) HWBUI0010183 Relation between asbestosis and bronchial cancer in amphibole asbestos miners Table 4 Estimates ofrelative risk (odds ratios)for each ofthefour models with 95% confidence interrals (95% Cl) $39 Model 1 Oddsratio 19S% Cl) Model 2 Odds ratio (95% Cl) Models Odds ratio (9SM.CH Model d Odds ratio (95% Cl) Smoking; Ex Age Asbestosis: Slight Moderate/pronounced Exposure variable* 14 3-06(0-52-18-04) 2-99(047-1888) 339(063-1* 19) 11-23(241-62-98) 1-07(1-03-1-13) 1-0 4 45(1-64-12-08) 5*20 0-60-16-88) 1-08(1-02-1-14) t-0 245(043-14-05) 2-46(0-40-45-02) 2-43(0-47-12-48) 9 94(1 *3-53 91) 1-06(1-01-1-11) 1-0 5-63(2-15-14-73) 8-53 (2-92-24-91) 145 (099-1-11) 14 2-60(0-45-14-88) 247(040-1522) 2 51 (0-48-13-08) 940(1-72-51-36) 147(142-1 12) 1-0 740(2 98-1840) 9 85(3-28-2*58) 1-0 1-0 263(046-1546) 2-53(042-1566) 2-57(049-1342) 9-53(1-74-52-16) I-07(H)2-I>t2) 1-0 7-60(306-18-84) 10-40 (3-57-30 32) 1-0 Exposure variable relen to years of exposure lo asbestos, residence time, fibre years, and residence TWD for models I, 2, 3, and 4 respectively. irtg (he likelihood of being a case of bronchial cancer was asbestosis with an improvement in the likelihood alio t (LR^) of 301 with 2DF(p ~ 0-0001) after adjusting for the effects of smoking habits (LR jf* = 10-1; 4 DF, p = 0-04) and age (LR j? = 14-83; 1 DF, p = 0-001). In model I years of exposure to asbestos emerged as an additional contributor to the likelihood of being a case (LR y* 6-36; 1 DF, p = 0-01). No other measurement of exposure emerged as an additional risk factor. The pattern of odds ratios remained consistent for each of the four models (table 4). Raised odds ratios were found for heavy smokers, age, both categories of asbestosis, and years of exposure. In model 1, for example, the estimated increase in risk of developing bronchial cancer was 11-25 for heavy smoking (95% Cl, 2-01; 63-0), for each increasing year of age 1-07 (1-03; M3), for slight asbestosis 4-45 (1-64; 12-08), moderate pronounced asbestosis 5-2 (1 -60; 16-88), and for each additional year of exposure to asbestos 1-08 (1-02; 1-14). The latter extrapolates to 2-08(1-20; 3-59) for every additional 10 years ofexposure to asbestos. The SPM Rs shown in table 5 are calculated both for the 35 cases of bronchial cancer proved by necropsy and the 25 cases of bronchial cancer as certified on the death certificate. Table 5 SPMStsfor the number ofcases ofbronchial cancers observedby grade ofasbestosis Asbestosis No Expected Observed* SPMFI Observed! SPMR None 302 124 Slight 69 3-6 Moderate/ pronounced 2* 3-6 II IS 9 887 8 4167 11 562-5 6 645 305-6 3754 "Cases of bronchial cancer proved by necropsy. tCascs of bronchial cancer as certified on death certificate. SPMR. Standardised proportional mortality, rates. Discussion This analysis indicates that of the variables ofinterest, the presence of asbestosis and indices of exposure (years of exposure, residence time, fibre years, and residence TWD), the presence ofasbestosis was by far the most significant risk factor to being a case of bronchial cancer, p < 0-0001 (table 3). The only index of exposure that remained sig nificant after the effect of the presence of asbestosis had been taken into consideration was years of exposure p = 0-01 (table 3). The odds ratio, 1-08, was small but the 95% Cl were narrow, 1 -02-1 -14 (table4). Cumulative asbestos respirable dust exposure expressed as fibre years had no significant effect and the effect of residence time (p = 0-1, OR 1-05 (0-991-11) remained doubtful. Kipen et al recently reported that all 138 cases of lung cancer for whom they were able to obtain a tissue specimen had asbestosis on histology.7 The cases arose out ofthe large prospective mortality study ofasbestos insulation workers studied by Seliko'fT et al* Unfor tunately no information on duration or intensity of exposure was given in this paper but exposure was probably prolonged. In rats Wagner et al found that there was a positive association between asbestosis and lung tumours in a large experiment in which the animals had been exposed to clouds of various mineralogical types of asbestos for periods varying from one day to two years' The animals with tumours had significantly more asbestosis than those without (p < 0001). Even rats with minima! or slight asbestosis bad a significantly higher rate oftumours than those without asbestosis ifthey had survived at least 600 days. On the other hand, rats with no asbestosis had an incidence of tumours no higher than unexposed controls. There is some evidence that the presence and HWBUI0010184 540 Sluis-Crtmer. Bezuidenhoui profusion of irregular opacities on the radiograph correlate with the risk of bronchial cancer, though there is probably still a small risk when the radiograph is normal." Liddell and McDonald have reported that men with 20 years or more employment in the lower dust concentrations showed no statistically significant excess mortality for any condition except pneumocon iosis and conclude that most but not necessarily all cases of lung cancer attributable to asbestos would show opacities before death." A histological diagnosis ofasbestosls as presented in this paper should give a more accurate estimate of the association between exposure to asbestos, asbestosis, and bronchial cancer. The SPM R for both the 35 cases proved by necropsy and the 25 cases certified as bronchial cancer on the death certificate show no excess bronchial cancer in the group without asbestosis whereas there is an excess in those with asbestosis. The excess increases with the severity ofthe asbestosis found at necropsy (table 5). In conclusion this study suggests that asbestos caused bronchial cancer is almost always associated with some degree or histologically demonstrable asbestosis. It must be emphasised that these results should not affect compensation bodies dealing with living subjects exposed to asbestos as slight asbestosis is commonly, and moderate asbestosis occasionally, undetected radiologically." We thank Dr W P D Logan and Dr P A Hesscl for advice on the analysis. We acknowledge the work of the pathologists of the National Centre for Occupational Health which generated the data analysed in this paper. References 1 The Royal Commission on Matters of Health and Safety arising from the use of isbestos in Ontario. Report. Vol I. Ottawa: Ontario Ministry of the Attorney General. 1984:283. 2 Wcin H. Asbestos. A summing up. In: Wagner JC. td. Biological effects ofmineralfibres. Vol 2. Lyon: International Agency for Research on Cancer, 1980:867-73. (Pub! No 30.) 3 Finkclstem MM. A study of dose response relationships for asbestosis associated disease. Br J Usd Med 1985;42:319-25. 4 Thomas DC. "RISK'" program. Montreal: McGill University, 1980. 5 Lcmcshow S. Hotmcr It DW. Estimating odds ratios with categorically scaled covariates in multiple logistic regression analysis. AmJ Epidemiol If84:119:147-51. 6 Breslow NE Day HE Statisticalmethods as cancerresearch l. The analysts ofcase controlstu&s. Lyon: International Agency for Research on Cancer, 1980:193-249. 7 Kipan HM, Libs R. Suzuki Y. Valciukas JA, SeGkofT 13. Pulmon ary fibrosis in asbestosis insulation workers with lungcancer: a radiological and histopathological evaluation. Br J ini Med 1987:44:96-100. 8 Sclikoff IS, Hammond EC, Scsdman H. Mortality experience of insolation woriters in the United Stales and Canada 1943-1976. Am Hit AcaiSei I979-J30SI-II6. 9 Wagner JC, Berry G, Skidmore JW, Timbrelt V. The eilccts of she inhalation ofasbestos in rats. BrJCaneer 1974;29:252-69. 10 Bohlig H, JacobG. Muller H. Ok Asbestostder Longest. Stuttgart: GeorgThiemc Vertag. 1960:60. 11 Liddell FDK. McDonald JC. Radiological findings as predictors of mortality in Quebec asbestos miners. Br $ tad Med 198ftJ7:2S7-67. 12 Sluis-Cremer GK, Hesse! PA, Hairdo E Factors influencing the readingofinegubropacities in# radiologicalsurvey ofasbestos miners in South Africa. Arch Environ Health (in press). HWBUI0010185 British Journal of Industrial Medicine August 1989 Vol 46 No 8 Contents Editorial Confounding from smoking in occupational epidemiology o axelson page 505 Cancer and other mortality patterns among United States furniture workers > a miller, a e blair. it L raynor, PATRICIA A STEWART, SHEILA HOAR ZAHM. J F FRAUMEM. JR page 508 Malignant lymphomas and occupational exposures 8 persson, anN-marue dahlander, m fridriksson, H NOORUND BRACE, C-C OHLSON, O AXELSON page 516 Ecological analyses and case-control studies of gastric cancer and leukaemia in relation to DBCP in drinking water in Fresno County. California o wong. r W MORGAN, M D WHORTON, NANOf GORDON. tJEEKA KKBFCT5 page 521 Cancer mortality in relation to measures of occupational exposure to crocidolite at Wiitenoora Gorge in Western Australia nhde klerk,b karegtrong, aw music, ms t hobbs page 529 Relation between asbestosis and branchial cancer in amphibole asbestos miners a k sluis-cremer, b n bezuidenhout page 5J7 Deaths from asphyxiation and poisoning at work in the United States 1984-6 a suruda, j agnew page 541 Variability in biological monitoring of organic solvent exposure. II Application ofa population physiological model PODROZ,MHWU.WG CUMBERLAND page 547 Excretion of 1,2,4-bexuenetrioi in the urine of workers exposed to benzene o inoue, k seui. h nakatsuka. t watanabe. S-H YIN, a-L U, S-X CAT. C RN, M IICEDA page 559 Diurnal variation in peak expiratory How rate among grain elevator workers P XEVSBECH, G ANDERSEN page 566 Correspondence between neurological symptoms and outcome ofquantitative sensory testing in the hand-arm vibration syndrome Lena ekenvall, costa gemne, r tegner page 570 Transitory postural vasomotor dysfunction in the finger after short term hand vibration N olsen, o U fetjuno, N ROSSINO page 575 Cancer mortality in the asphalt industry, a ten year follow up of an occupational cohort eva s Hansen page 582 Prenarcotic and neuraesthenic symptoms among Dutch workers exposed to organic solvents c van vuet, gmh swaen, JUH MEtfERS, J SLANGEN, T DE BOORDER, F SIURMANS page 586 Correspondence Influence of design characteristics on the outcome of retrospective cohort studies M jane teta page 591 Notices page 592 ASTM CODEN: BJIMAG 46(8) 505-592 (1989) British Medical Association Tavistock Square London WC1 ISSN 0007-1072 Published by Bn'tiih Medical AtsoeUtion. Tavistock Square. London WCt M 9JR and printed ia EflgUod by Eyre & Spouhwoodc Lid, London and Margate. HWBUI0010186 rional iolna, Id D, eruion. idy on >73;SI: i. Risk Engl J na$o/Natl ry JT, :cupa- f Natl !e and cional * tract *6;43: 1961udics ire to mihe Seme -69. tn the 7oad, 1981: : and mial and tahh British Journal of Industrial Medicine 1990;47:215-216 215 CORRESPONDENCE Relation between asbestosis and bronchial cancer in amphibole asbestos miners Sir--Sluis-Cremer and Bezuidenhout (1989;46:537-40) present data on the relation between asbestosis and bron chial cancer in amphibole miners but analyse the effect of dose of asbestos only after allowing for the effect of asbestosis grade. Since the risk and severity of asbestosis are themselves dose related much of the effect of dose on the risk of cancer had already been allowed for by the analysis of the effect of asbestosis grade. Nevertheless, years of exposure, probably the most reliable measure of dose because it is known more accurately than intensity, still had a significant effect. This is consistent with the dose of asbestos rather than asbestosis being the major determinant of the risk of cancer. It would be interesting to see the results of a further analysis of the data in which the various measures of dose were entered into the logistic regres sion before the asbestosis grade to determine the effect ofasbestosis after the effect of dose of asbestos has been allowed for. To shed further light on the relation between asbestosis and bronchial cancer the ideal analysis would compare the incidenceofcancer in subjects with and without asbestosis matched for dose ofasbestos received. I wonder whether their data include sufficient subjects to attempt such an analysis? Their analysis concerned subjects who had undergone necropsy, which occurred in a minority of deaths in the study population. The criteria for selection for necropsy were not men tioned. If necropsy had been more ; 103-ATC-13. t .90iikely to be carried out in. men with cancer ifthere had also been evidence ofasbestosis during life, as has been the case in the United Kingdom until recently, this would have tended to augment the apparent effect of asbestosis on the risk of cancer. It is not indicated whether the presence of asbestosis was assessed without knowledge of whether or not cancer was present. Pathologists com monly look harder for asbestosis when they know bronchial cancer is present. Table 2 of the paper shows that there were large standard deviations in the various measures of dose and sug gests that subjects with only brief exposure were included. The observa tion that there appeared to be no excess risk of cancer in the group without asbestosis may indicate no more than chat many of them had relatively slight exposure. RMRUDD London Chest Hospital, Bonner Road, London E2 9JX Dr Sluis-Cremer and Mr Bezuidenhout reply: We have carried out the two analyses suggested by Rudd. It must be men tioned, however, that what he first suggests is the assessment ofthe risk of developing bronchial cancer in the presence of asbestosis after controlling for the effect of the various measures of dose. In our paper we state our objective .. to determine if any parameter ofexposure to asbestos dust exert any additional risk ofdeveloping bronchial cancer after allowing for the presence of asbestosis." After adjust ing for the effects of smoking and age then entering the various measures of dose in separate analyses followed by asbestosis grade the latter still emer ged as a significant risk factor for bronchial cancer. Admittedly, years of exposure accounted for most of the variation in model 1 (table 1) but notwithstanding this, the grade of asbestosis still emerged as a highly significant risk factor for bronchial cancer. Following Rudd's next suggestion we were able to match to each ofthe 35 cases two to four referents with year of exposure within 20% of that of the respective case. A conditional logistic regression analysis was performed on the matched case-referent set of 157 subjects to assess the effect of asbes tosis absent/present on risk of bron chial cancer after adjusting for the effects of age and smoking (table 2). The improvement in the fit of the model to the data with the introduc tion of asbestosis was significant (LR / - 5-55; 1 df, p = 0 02) with the estimate ofthe relative risk by the odds ratio (OR) significant at 3-4 (p = 0 03; 95% CI:1T--10-46). As in the Table / Goodness offit valuesfor the stepwise unconditional logistic regression analysis Model 1 2 3 4 Step 1 2 3 4 3 4 3 4 3 4 Term enured Smoking Age Exposure (y) Asbestosis Residence time Asbestosis Fibre years Asbestosis Residence time weighted dosage Asbestosis Improvement in LR f Degrees offreedom 1013 1483 2544 1101 12-23 20-64 3-95 26-6 1*65 28 44 4 1 l '2 1 2 1 2 l 2 p Value 004 o-oooi <0-001 0-004 00005 <0-0001 0-05 <0-0001 0-20 <0-0001 Table 2 Goodness offit valuesfor the stepwise conditional logistic regression analysts matching cases and referents by years exposure Sup 1 2 3 Termauered Smoking Age Asbestosis Improvement in LR g? 9-29 0-39 555 Degrees ojfreedom 4 1 1 p Value 0054 0*53 0-018 216 Correspondence unconditional logistic analysis in the published paper, heavy smoking emerged as a significantcontributor so the risk of bronchial cancer (OR 7-9; p * 0 02; 1-4-43*2). The necropsy rate in the study population was low (37%) by South African standards; 65% in gold mis ers. The necropsy cases were not selected by any authorities but by the doctors and family of the dead ram in the hope of obtaining compensation. Men with advanced disease would probably have already obtained max imum compensation and there was thus no need for necropsy. The families of workers with short or for gotten service or ignorant of the regulations may not have insisted on necropsy. Lung cancers if diagnosed in life would have obtained maximum compensation. We agree that there are probably selection factors for necropsy ofwhich we are ignorant. The mean net duration of exposure of cases of cancer without asbestosis was six years arid their referents with out cancer or asbestosis was right years--this is hardly "slight" exposure. Among those with asbes tosis, patients with cancer had 13 years of exposure and their referents with out cancer .14 years. ) HWBUI0010188