Tobacco Documents Online

Exp. Pathd. ( 989: 37: t77-180 vt:B Gusuv Fischer verlag Jena PqCr The role of histopathology in the evaluation of risk of lung cancer from environmental tobacco smoke By J. M. FACCINI Adiress to[ ooe+espo.desce: Prof. J. M. FAcctrn. Robens Institute of Lndustrial and Enviton- mental Health and Safety, University of Sutrey, Guildford. Surrey GU2 SXH, UK Kay woedu lung cancer. risk evatuation: risk evalua[ion. lung cancer. cancer. lung, risk evaluation: evironmennt to6acco smoke (ETS): ETS. lung cancer S11111111111111111ry Current clinical practice for treating lung cancer does not provide adequate histopathological evaluation for the clear distinction between primary and metastatic neoplasia. Reliance on such clinical diagnosis jeopardises the scientific validity of many epidemiological studies designed to assess the risk of inhalation of environmental tobacco smoke (ETS). Iehrodt.ctfotr The debate concerning a possible relationship between lung cancer and ETS has intensified since the publication in 1981 of studies from Greece and Japan reporting a positive association. Subsequently, other cohort and case control studies have supported this while others have found no relationship. In 1986 the U.S. Department of Health and Human Services published the Surgeon General's (S.G.) Report on the health consequences of involuntary smoking which analysed 13 such studies (6). The authors acknowledged the difficulties involved in reaching conclusions from the published epidemiological studies but concluded that taken as a whole they still provided evidence of a positive effect although not all of them showed a positive association. The major patt of the debate on whether these studies do provide evidence of a significantly increased risk has centred on the methods employed to obtain evidence of exposure to ETS and, in particular. the danger of misclassification of active smoking status. The misclassification of the lung as the primary site of neoplasia and the absence of histological confirmation have been acknowledged as giving cause for concern (6) but this has not prevented the inclusion of many cases where this confirmation has beO lacking. Because the lung is such a frequent site of metastases. however. correct medical practice demands microscopic confirmation. Futtheataie, even when there is histological or cytological evidence of adenocarcinoma involving the lung in nonsmokers. the possibility of other extrapulnanary tissues being the ptimary site of origin. puticululy the breast, needs to be seriously considered. Given the importance of histopathology in the debate on lung cancer risk and ETS and, the general lack of puhologisu' direct involvement in the andyses of the cases, it is appropriate to re- examine the 13 studies analysed in the United States S.G.'s Report. A total of 1.104 cases of cancer of the lung in people classified as lifelong nonsmokers have been identified in the 13 studies analysed in the S.G.'s Report. As the majaity of cases are females (1,041) - 8 studies concerned only females - and the debate centres round the risk to nonsmoking wives of sawkers. only diagnoses in women will be considered. 12 Exp. Pu6ol. 37 (t9a9) 1-a 177

Table L Pathological diagnoses in the 13 studies. Study nonsmokers % with histopath. % adenoca. source HIRAYAMA 200 10.5(0) 57.1 autopsy GARFINKEL 153 69(0) - death cert. GtLLIS 8 - - ca. registr. TtuCttopovt.os 77 18.1 (24.6) - hosp. recs. CottxEA 22 97 (?) 54 hosp. recs. CHAN and FUNG 84 82 (?) 45.2 - Koo et al. 88 97 (l6) 58.9 pathologist KABAT and WYNDEA 53 l00 (?) 62 hosp. recs. Wv 29 100 (?) l00 ca. rcgistr. GAttt1NKEL et al. 134 100(0) 65 pathologist LEE et al. 32 - - hosp. recs. AxtBA 84 53 (4) - registr. PERSHIGEN 77 9905.6) 57.1 registr. Figs. in brackets = cytology diagnosis; -= not stated 1~SL0-2--1 e7idlHC! Only 2 studies. GAXF'INKEL et al. and Koo et al. used microscopically confirmed lung cancers in their analyses by including a pathologist as co-author who reviewed the tissue diagnoses from microscopic slides of each case. Pathological diagnoses were obtained from hospital records in 5 studies and from death certificates, censuses or registries in 6. The percentage of wpporting microscopic diagnoses in the medical records varied (see table l). No more than half give any meaningful information about the types of lung tumour, adenocarcinoma being the most common type. Dtseaaaim The possibility of metastatic lung tumours masquerading as primary lung tumouts is not rcallv considered by the majority of authors. This is understandable because they are not pathologists and. therefore. are not familiar with the difficulties of histopathologital practice. Even the most experienced histopathologist on reporting a lung biopsy in which an adettocarcittoma is present will conclude that the diagnosis is only compatible with a primary lung neoplasm.. providing that a primary adenocarcinoma elsewhere in the body can be excluded. This is especially true in women. where the risk of a metastatic adenocarcinoma being mistaken for a primary lung tttmour is high: adenocarcinomas of the bteast, bowel, ovary and uterus are the most common tumours in nonsmoking women. This is not the case with primary adenoarcinoma of the lung in lifelong nonsmokers which is rare. GAtttINKEL provides examples of misdiagnosis among 203 nonsmoking women followed in the ACS study with a diagnosis of lung cancer on the death certificate. 34 (16.7 9'0 ) were reported to be cancers of other sites on the final medical report - about ~5 died from breast canoer. In his second study. GAttFINKEL had the advantage of a pathologist as co-author who examined the slides of each case. The pathologist. OSCAR AUEBBACH, found thu out of 283 nonsmoking women. 36 (12.7 %) proved not to have primary lung catker on slide review. In addkion. 113 (39.9 96 ), were found to be smokers upon interview, leaving only 134 (47.3%) who ware lifelong nonsmokers with histologically proven primary lung cattcer. The same authors cite previous work: in a study of 774 noasmokers reviewed for asbestos expostue, 49 had a discharge diagnosis of primary lung cancer: following review of hospital records, histological sections and intetviews, only 10 cases of ptimary lung cancer in lifelong 178 Etp. Patqt. 37 ( t 9ti9) 1-4

womokets remained - 50% of the others had been smokers and 50% had a primary cancer in 1„other site metastasising to the lung. This problem is dealt with at length by another author of one of the l3 studies. WYNDER. in a ,zp,vate paper on the histological patterns of lung cancer in nonsmokers (7): he quotes examples of ntisclassification from the literature occurring even when histopathological diagnoses have been ,,w&. For example. TRttvtnwn et al. who described 10 cases with a clinical diagnosis of bronchial ,,-jrcinoma confirnted by biopsy and cytology that on autopsy examination proved to be primary ,Koplasms of the pancreas. kidney, adtettal or retropetitoneal lymphosarcoma. The lymphosarcoma had been classified as an oat cell carcinoma and the other cancers as KRt:YSERO group 1 ttunours i.e. ;quanwus or small cell, despite the fact that they had originated from glandular epithelium. WYNDER and BExG (7) highlight the problems which stem from current clinical practice. "A plug or two of cancer in a submucosal lymphatic of a bronchial biopsy like positive cytology is enough evidence for initiating treatment but may well not serve for full classificatiott." Even in epidemiological studies supported by histopathology, thetefore, there is still a risk of misclassification. This risk is magnified if hospital records only are used, even though these may contain histopathological reports, and it becomes still greater if no histopathology is available. On weighing the 13 studies used for the S.G. 's Report for adequacy of histopathological evidence, only 2 of tltem. GAmrtuEt. et al. and Koo et al.. can claim to have sttmtg evidence of a primary pulmonary neoplasm. For the majority of the studies, the most that can be said is that they provide moderate evidence, while the studies of HIRAYAStA and TRICHOPOULOS et al. provide only weak evidence. It is of interest that GARF1NtuaL ct al. found a weak association and Koo et al. no associatiott, while the studies reporting the highest risk ratios (R.R.) at= HIRAYAMA and TRtceoPOUt.os et al. The importance of this criticism is underlined by the fact that each study contains only a few cases of suspected primary lung cancer. If, therefote, any of them are misclassified, it means that the true numbers ar still smaller. While recognising this problem, the S.G.'s Report (6) claims that bias would tend to dilute a true effect. From the point of view of correct scientific methodology. it would seem preferable to rely on hard data. espiciaily as there is also a related problem of misclassification of smoking status. Subsequently to the publication of the studies quoted in the S.G.'s Report (6) other epidemiological studies have been reported which show a similar spectrum of findings and which also highlight the need for further research. For example. Huutst.FE et al. (4) reported 28 nonsmoking women with lung cancer, for 17 of whom the histopathological diagnosis was reviewed by a panel who concurred with the original histopathological diagnosis in only 8 cases. BttowNsoN et al. (1) from 102 cases of adenocarcinoma of the lung (50 m and 52 f) identified 19 nonsmoking women - all controlled by histopathology: they concluded that prior cigarette use remained the most significant predictor of risk of adenocarcinoma in both sexes and after adjustment by logistic regression for age etc. that no significant adenocarcinoma risk for passive smoke exposure was found among females. Two studies in Chinese women which are important because of the numbers involved have appeared more recently. The fitst, LAm et al. (5). reported 44S Hong Kong Chinese women with lung cancer confiitned_by lung resection (129'0), pleural (99'0) or lymph node biopsy (8%) and cytology (35 %): this series included a high peteentage of claimed never smokers - 45.5 9'o in all and 62.4 % of these cases were adenocancinoma. Following a case control analysis, the R.R. for nonsmoking women whose husbands smoked l-10 cigarettes per day was of the same order of tnagnitude as women actively smoking 11-20 cigarettes per day. Given that a carcinogenic effect in both animals and tnan is time and dose dependent. and the risk of developing lung cancer from active smoking is also related to the duration of smoking and the number of cigarettes smt>ked. these results reported by LAx and his colleagues seriously question the metbodology used to assess risk from. ETS. This also recalls an enigma cited by GARFttaKu (3): the atypical cells in bronchial epithelium which are classically associated in a dose-response fashion to active smoking ate viRually absent from the bronchi of nonsmokers with lung cancer. The secatd. a Sino-American cooperative study (2) of 672 Shanghai women with lung cancer - 542 confirmed by tissue biopsy (43 9'0 ) or cytobgy (38 %): only 35 % were smokers; tbe R. R. for smokers was 7.2 for squaatous cell l2• Fxp. Pat6d. 37-(1489) 1-4 179 I

carcinoma• but only 1.5 for adenocarcittolna - figures similar to those of LAm et al. (5). Adenocarcinoma accounted for 61 % as against squamous cell carcinoma (22 %) and only 6% of adenocarcinomas were attributed to smoking. This raises a question voiced by several authors: if lung cancer is caused by ETS why does the majority of nonsmokers suffer from adenocarcinoma and not squamous cell carcinoma (long known to be associated with active smoking)? In this study (2), the R.R. for a nonsmoker living with a smoking husband was only of some importance if he smoked > 40 years i.e. 1.7 (48/542) and this increased if only squamous or oat cell carcinomas were concerned (R. R. = 2.9). The importance of this stud_v is that 61 % of the cases were diagnosed as adettocarcinotna and only a minority of these were active smokers - the extent of stnoking in China being far less than the Western World. The authors concluded there was little or no association with ever living with a smoker and suggested other possible causes should be looked for - the view voiced by some of the authors of the studies quoted in the S.G.'s Report. Most of the subsequent studies are still not adequately confirmed by histopathology. nonetheless. and until studies have been completed with appropriate attention to histopathology and especially tumour type, the association between ETS and lung cancer remains speculative. It is important moreover, to keep the conclussions in perspective: even if the inadequacies of the current epidemiologic investigations are ignored, and some increased risk from ETS exposure is assumed. this fact remains. as stated by KABAT and WYNDER, that the occurrence of primary bronchial carcinoma in lifelong nonsmokers - even when exposed to ETS - is tare. One can only agree with E. L. WYNDEA.- "To the epidemiologist, we note that the dominance of cigarette smoking as a risk factor has led to a focus on this variable to the relative neglect of other factors" (8). Rdelresces I. BRowNsoN. R. C.. REIF,1. S.. KEEFE.1.1.. FERGUSON. S. W.. PtrrzL. J. A.: Risk factors for adenocarcinoma of the lung. Am. J. Epidemiol. 1987, M 25-34. 2. GAo. Y. T.. BLOT. W. J.. LiEN. W.. ERsHow. A. G.. Hsu. C. W.. LEVIN, L. L. ZHANG. R.. FRAUMENI. J. F.: Lung cancer among Chinese women. In. J. Cancer 1987: IM: 604-609. 3. GARFINKEL. L.: Passive smoking and cancer - American experience. Prevent. Med. 1984: 0: 691-697. 4. HUMRLE. C. G.. SAMET. J.. PATHAK. R.: Marriage to a smoker and lung cancer tisk. Am. J. Publ. Health 1987: 77: 598-601. 5. LAM. T. H.. KUNG. I. T. M.. WONG. C. M.. t.AM, W. K.. KLEEVENS. J. W., SAw, D.. HSU, C.. SENEVIRATNE. S.. LAH. S. Y.. Lo. K. K.. CH_AN. W. C.: Smoking. passive smoking and histological types in lung cancer in Hong Kong Chinese women. Br. 1. Cancer 1987: 56: 673-678. 6. U.S. Department of Health and Human Services: The health consequence of involuntuy smoking: A report of the Surgeon General. Office on Smoking and Health. Rockvilk. MD. 7. WrNDER. E. L.. BEttG. J. W.: Cancer of the lung among nonsmokers - special eefetence to histologic pRtterns. Cancer 1967: 20: 1161 -1172. 8. - CovEY. L. S.: Epidemiologic patterns in lung cancer by histologic type. Eur. J. Cancer Clin. Oncol. 1987; 23: 1493-1496.