Document 10B65qxGbyrmyENwwvjdJ5EkE

European Journal ofEpidemiology 16: 411-417, 2000, 2000 Kluwer Academic Publishers. Printed in the Netherlands. Environmental exposure to asbestos and risk of pleural mesothelioma: review and meta-analysis Environmental exposure to asbestos and mesothelioma Valerie Bourdes1, Paolo Boffetta1 & Paola Pisani2 1 Unit ofEnvironmental Cancer Epidemiolgy; 2Unit ofDescriptive Epidemiology, International Agencyfor Research on Cancer, Lyon, France Accepted in revised form 22 February 2000 Abstract. A number of epidemiological studies have addressed the risk of pleural mesothelioma from environmental (household and neighborhood) expo sure to asbestos, but no overall risk estimate is available. We reviewed the epidemiological studies on risk of pleural mesothelioma and household or neighborhood exposure to asbestos. We identified eight relevant studies; most were conducted in pop ulations with relatively high exposure levels. We combined the risk estimates in a meta-analysis based on the random-effects model. The relative risks (RRs) of pleural mesothelioma for household expo sure ranged between 4.0 and 23.7, and the summary risk estimate was 8.1 (95% confidence interval [Cl]; 5.3-12). For neighborhood exposure, RRs ranged between 5.1 and 9.3 (with a single RR of 0.2) and the summary estimate was 7.0 (95% Cl: 4.7-11). This review suggests a substantial increase in risk of pleural mesothelioma following high environmental exposure to asbestos; however, the available data are insufficient to estimate the magnitude of the excess risk at the levels of environmental exposure com monly encountered by the general population in in dustrial countries. * Key words: Asbestos, Meta-analysis, Non-occupational exposure. Pleural mesothelioma Introduction Asbestos and asbestiform fibers are naturally occur ring fibrous silicates with important commercial use mainly in acoustical and thermal insulation. They can be divided into chrysotile and the group of amphiboles, including amosite, crocidolite, anthophyllite, actinolite and tremolite fibers. Chrysotile is the most widely used type of asbestos. While all types of as bestos are carcinogenic to the lung and the mesothelia, the biological effects of amphiboles on the pleura and the peritoneum seem stronger than those of chrysotile [1]. The use of asbestos has been re stricted or banned in many countries. In contrast to the large amount of epidemiological studies of asbestos exposure at the workplace, the evidence concerning the health effects of environ mental exposure is rather limited. Two main types of environmental asbestos exposure are distinguished: domestic or household and residential or neighbour hood exposure. The most common source of house hold exposure is the installation, degradation, removal or repair of asbestos-containing products. An additional source of household exposure, concerning family members of asbestos workers, is the asbestos dust brought home from the workplace on the clothes. Neighborhood exposure, which results from out door air pollution, is mainly due to asbestos mining and manufacturing close to the place of residence. It may also result from release of fibers from buildings or other sources, like vehicle brake linings, and from re-aerosolization of fibers following indoor or out door sedimentation. In addition, a natural source is the erosion of asbestos or asbestiform rocks. The assessment of environmental asbestos exposure is difficult since levels are generallylow and the duration and frequency of exposure and the type of fiber are seldom known with precision. Table 1 summarizes the results of selected studies on non-occupational expo sure to asbestos. In general, levels of indoor exposure show a greater variability than outdoor levels. In some cases, values above 10 fibers/ml have been reported. While there is a general acceptance that environ mental exposure to asbestos may cause pleural mes othelioma, the magnitude of such an effect is not known with precision. We conducted a review of epidemiological studies of pleural mesothelioma and environmental (household and neighborhood) expo sure to asbestos; our review also included a quanti tative meta-analysis. Methods We carried out a literature search on articles included in Medline and Current Contents that appeared 412 Table 1. Results of selected measurements of non-occupational exposure to asbestos Source of exposure Country Level (fibers/1) (range, mean) Reference Mining and industrial sources Asbestos cement plant Asbestos cement plant, mines Residence of miners Near asbestos cement plant Near asbestos deposits Near asbestos cement plant Geological sites Near chrysotile mine Near tremolite mine Near erionite sources Near chrysotile mines Mining area: 1974 Mining area: 1984 Rural and urban pollution Large city: Heavy traffic Large city: Expressway Sacramento and San Francisco Los Angeles San Jose Important urban traffic Urban areas Urban areas Residence outside mining area Urban area Rural areas Urban areas Urban areas Household Sweeping houses Colour washing (tremolite) Residential Residential Residential Sprayed buildings Schools Offices and plant buildings Schools Schools: Chrysotile Schools: Amphiboles Schools Buildings USA Canada South Africa Austria Austria Italy Italy New Caledonia Turkey France Canada Canada USA USA USA USA USA Austria Canada Germany South Africa Italy Japan Japan France New Caledonia New Caledonia UK UK USA USA Austria UK USA USA USA Canada 0.6-2.2 7.8 2-11 0-2.2 0-0.5 48.4 2.5 59-670 6 1-17 46 10 0.9 3.3 1 9 150 4.6 2-4 0.2-5 0.2-0.8 5.6 4-91 4-111 0.47 78000 558 0.4 0.28 0.1 1^10 <22 0.5 8.3 0.65 0-24 0.42 [2] [3] [3] [3] [3] [4] [4] [1] [5] [1] [6] [6] [2] [2] [7] [7] [7] [3] [3] [3] [3] [4] [8] [8] [9] [1] [1] [10] [11] [12] [2] [13] [10] [14] [14] [15] [16] between 1966 and 1988. In order to identify addi tional studies we systematically searched the contents of key journals and the lists of references of all identified studies. We analysed those studies which provided results on pleural or peritoneal mesothelioma from house hold or neighborhood exposure by inhalation. We retained only studies comparing clearly defined ex posed and unexposed groups, including those using an ecological approach. The interpretation of ecological studies is, however, limited by the lack of individual information on asbestos exposure and potential confounders. We excluded a small number of case reports and case series because of the lack of a definition of the population at risk and of a comparison group. Finally, only studies were included for which a risk estimate and its variance or confidence interval (Cl) either were reported or could be obtained from the raw data. When the same population had been studied several times, we used only the most recent report. We included studies presenting results on either mortality or incidence since the average survival from mesothelioma is low. We focused on adult neoplasms, since mesothelioma is very rare in children [17]. We extracted from each study the main charac teristics of the design, the definition of asbestos 413 exposure, the predominant type of asbestos fiber, the risk estimate (rate ratio, odds ratio, standardized mortality ratio, thereafter denoted as relative risk [RR]) and its variance, back calculated from the Cl if not available. In addition to a narrative review of the studies, we performed a meta-analysis on a randomeffects model [18], which takes into account addi tional inter-study variability into the calculation of combined risk estimate and its variance. We considered separately household and neigh borhood exposure to asbestos. Since results in women are less likely to be confounded by concomitant oc cupational exposure to asbestos, we focussed on re sults specific to women. If a study provided separate results for men and women, we combined them and considered also the combined estimate. We also stratified the studies according to the de sign of the study (cohort, case-control and ecological) and to the predominant type of asbestos (chrysotile, amphiboles, and mixed or unspecified), as defined by the authors of the original reports. Results Only two studies [19, 20] presented results on peri toneal mesothelioma following environmental asbes tos exposure. Newhouse and Thompson [20] reported two cases of peritoneal mesothelioma in women with household exposure, none with neighborhood expo sure and seven cases, including three in men, without evidence of occupational or environmental exposure to asbestos. Vianna and Polan [19] reported eight cases of peritoneal mesothelioma among women without occupational exposure: household exposure was reported for seven of them, and residential ex posure for three, including the case without the household exposure. We retained eight studies providing results on pleural mesothelioma [20-25] (Table 2). Two studies were from USA or Canada, two from UK, two from Italy, and one each from South Africa and Turkey. Four studies had an ecological design, one was a co hort study and three were case-control studies. Two studies investigated populations predominantly ex posed to amphibole asbestos, while the main exposure was chrysotile in two studies and in the remaining four studies there was no clear indication of a predominant type of asbestos. Three studies presented separate analyses on neighborhood and household asbestos exposure for a total of six studies reporting risk esti mates for neighborhood exposure and five studies reporting risk estimates for household exposure. Four studies reported estimates of standardized mortality ratios of mesothelioma based on ecological assessment of asbestos exposure. Residents of three crocidolite mining districts in South Africa had a ninefold increased mortality as compared to the population of the region [21]. Residents in three villages in South-eastern Turkey used whitewash and stucco contaminated with tremolite fibres: the standardized mortality ratio of mesothelioma was 21, as compared to non-contaminated villages [5]. Other studies reported an excess of mesothelioma following similar exposure circumstances to reunite in Central Turkey [26]. Finally, a study from Quebec, Canada, reported seven deaths (0.92 expected) among women resident in villages with chrysotile mines, with an estimated prevalence of occupational exposure of 5% [6]. The incidence of pleural mesothelioma was 5.8fold higher among residents of an Italian city with an asbestos cement factory, after exclusion of occupa tionally related cases, as compared to the region [24]. An additional study reported three deaths from mesothelioma among women in New Caledonia during 1978-1987, which represents a fivefold excess when compared to rates in New South Wales, Aus tralia [27]. In a cohort study of non-occupationally exposed wives of asbestos workers from Italy, three deaths from pleural mesothelioma were reported, which represented a 8-fold excess risk [23]. Three case-con trol studies analyzed the risk of pleural mesothelioma from household or neighborhood exposure to asbes tos. Risk estimates were elevated in one early casecontrol study from the UK [20]. In the other two studies, the RR from household exposure was in the order of 5; the risk from residential exposure, on the other hand, was not increased in one of these studies, from the USA and Canada (exposure defined as liv ing within 20 miles of chrysotile mines) [22], while it was increased in the other study, from London (ex posure defined as residence within 0.5 km of indus trial sources) [25]. Table 3 presents the results of the meta-analysis. The combined RR for neighborhood exposure was 7.0 (95% Cl: 1.8-7.0). There was a non-significant increased risk in the two studies considering mainly chrysotile exposure. The design of the study did not seem to influence the combined risk estimate; simi larly, the results did not change depending on whether the studies considered mortality or incidence of cancer or according to gender (results not shown in detail). The combined RR of pleural mesothelioma from household exposure was 8.1 (95% Cl: 5.3-12). All but one study were conducted in areas at either pre dominant or concomitant amphibole exposure, a fact that limited the analysis according to fiber type. Discussion The main result of this analysis is a strong rela tionship between pleural mesothelioma and high environmental exposure to asbestos, whether the source of exposure is domestic or neighborhood. The results also suggest a higher risk from exposure to Table 2. Studies on environmental exposure to asbestos and pleural mesothelioma included in the meta-analysis number o f exposed cases; RR, relative risk; C l, confidence interval; Ref., reference. * Including cases o f ashesto sis. **Pleural tumors. 414 a tTi tN On 00 <N iN r^> oo Z5 CN tN 3 -- 4i-n 4<n 7 C4 <N <N -- r- sn CN " 00 SO vi 'O ^ C*T V"1 a. S a, 0. [1, ++ SS LU E Z oo o u. a Uh + a Cl, Cl, ++ aa EZ Cl. Cl. ++ XX M oW3 U.. V5 ts 5 ^ (J a *-GO3 0233 &a <G sv <dT cco Lw .32 JO S * o8 o E 05 -I 8 S S' X0) 305 (g 2P -O 8 fe a 8I! o'oI Co *a 8 Jt:i Z os oo os fet0 .C5 iO-v* u 0O0 -g *a S 3-f O ,2 o -2 S 4> W J3 - CG s o | 3 *0 8 C OJ O *5 " I14 e CG * CG c S 2 +o- *co -'eS>-5HO uj5f |<85 .I2 *pC3 ^" S~s I 5 C 8 * M o c v o c os > i78 5G --C aC giS9 o--IB " Q, C Ea 8I ""si "fCl oEofa'oBo tQCwl, ^0 .S5S 6 s ,ico <G s -- E oG d* 8 ISc &Eo 3. Si -SS2 vvsc G rtN I? .i a So c .2 ?P 2 o\ o tt ^cf, S" >. C? id3 ^ -E T CT"C3 dcj 1 Q. TS aAP. oE IS '5 a .2 ^- ^-3 8 ^ *3 -- 3 U S2 .2 5 o *5 CG O CV GO 0 ^7 SSi 3 ri 12 S 3 w >f -d 3w e Q 00 *E - V *e "1 -a o f 3 4 - 8 05 V|3coo x *3 8 fal ^ O o> 3 i! co ` ^3 nm E^a(ufl ^sPOi v<"U ' Je *0 .s % 3oo .23 --j C "l-< O ? w P , **>53 ."2y? eo 2 I *5 WU -^uS J^aJ '3 *55 ^ S It o a^.S o<u wl O *- i 3SS".,5g=* *|?0e51) s03o"5 2 E= S c 2 -o ^ 52^ e oGS. tu 5536S2 ."i^GJ uo aa PP as 3D u cu U o <oo D rt Tl cMcd o o O u a u u 3 H3 o o P >S S' *1 Table 3. Results of the meta-analysis of studies on environ mental exposure to asbestos and pleural mesothelioma Neighborhood Household exposure exposure N RR 95% Cl N RR 95% Cl All studies 6 7.0 4.7-11 5 8.1 5.3-12 Type of fiber Chrysotile Amphiboles 2 1.5 0.04-53 1 4.0 0.8-20 1 8.7 6.7-11 1 21 2.8-157 Mixed/unspecified 3 6.7 4.4-10 3 8.2 5.2-13 N = Number of studies; RR = relative risk; Cl = confi dence interval. amphiboles than from exposure to chrysotile. Most of the available studies addressed the effect of long term environmental exposure. A problem in the interpretation of our results de pends from the fact that the studies investigated dif ferent sources of environmental asbestos exposure. Natural sources were included in the study from Turkey, while mining represented the main source of environmental exposure in studies from Canada, South Africa and the USA, and asbestos cement plants were the main source of exposure in Italy and the UK. However, the studies included in the meta-analysis addressed circumstances of exposure to relatively high levels of asbestos. No epidemiological studies are available on more common situations such as exposure in buildings, in schools, or in the general urban environment. The results of our meta-analysis are therefore likely to overestimate the risk of envi ronmental asbestos exposure experienced by residents of industrialized countries without a specific source of exposure; they are, however, useful to indicate a plausible upper range of the risk from environmental asbestos exposure. One possible limitation of our review and meta analysis is publication bias. Publication bias may occur when studies are performed but not published, a reason not to publish a study being the fact that it did not show a strong association. Ignoring these results may lead to overestimate the effect of expo sure. A second case of publication bias occurs when studies are published but not identified in the litera ture review. We tried to minimize this potential bias by using various source of references. A third source of publication bias depends on the inclusion criteria of our meta-analysis. We excluded one study that did not provide the number of exposed cases [28] and a few studies which did not provide a comparison with a reference population (e.g., the studies by Goldberg et al. [27 and Anderson [29]). The size of a study has an impact on the precision of the resulting risk esti mate, thus affecting the level of statistical significance of any increase or decrease in risk. It is therefore plausible that results of small studies that did not 415 show a significant result, are less likely published than results of larger studies. In order to assess publication bias, the risk estimates from individual studies can be plotted against the inverse of their variance [30]. The plot for pleural mesothelioma (Figure 1) did not suggest that small negative studies were missing for the set of results we used, speaking against publica tion bias. Heterogeneity across studies may result from chance, from different strategies of adjustment for potential confounding factors, and from differences in definition and assessment of exposure, in level and duration of exposure, and in ascertainment of diagnosis. A meta-analysis can help to elucidate the impact of such factors. A potential confounding factor in this study is represented by concomitant occupational exposure to asbestos, particular in studies collecting information on asbestos exposure from proxies. In some studies, household exposure of women was assessed based on questionnaires on the occupational history of patients and their relatives [19, 20]. In some cases the authors excluded subjects with occupational exposure [23, 24]. Respondents in a study from Canada and the USA were asked whether any member of the house hold had brought home some dusty clothing and, if so, the nature of the work [22]. The proportion of subjects included in the meta-analysis who might have been exposed to asbestos in the workplace is very small (Table 2). Another possible source of bias deals with outcome misclassification and is relevant in particular when the diagnosis of mesothelioma is derived from death certificates [31]. However, only one of the studies on mesothelioma included in the present meta-analysis was based on death certificates [23], and its exclusion would not affect the overall results. Underdiagnosis will not lead to bias as long as both the exposed and the unexposed populations are concerned in the same proportion. In contrast, bias may arise when diag nosis is related to exposure, as in the case of a higher probability of diagnosis among subjects exposed to asbestos. As an example, in a mining area of Quebec weight Figure 1. Plot of logarithm of relative risks [ln(RR)] and weights used in the meta-analysis (inverse of variance of relative risk) of studies on mesothelioma risk from envi ronmental exposure to asbestos. 416 with high rates of mesothelioma, overdiagnosis was suggested after review by a pathology panel [32]. Comparison of RRs between different study pop ulations is feasible if the populations experienced similar levels of exposure, as is probably the case in our meta-analysis. However, even if there were a common underlying dose-response curve of risk from asbestos exposure, the RRs reported in each study would be modified by the mean cumulative exposure in each population. Unfortunately the required in formation (e.g., average lifetime cumulative asbestos exposure) was not available from most studies. A higher risk of mesothelioma from exposure to amphiboles than to chrysotile has been suggested from studies of occupational cohorts: [33] our meta analysis indicates that this pattern may also hold for environmentally exposed populations, although the difference is not statistically significant and it should be interpreted with caution. The excess risk of mesothelioma is lower than that reported among workers occupationally exposed to asbestos: most cohort studies, however, refer to exposure circumstances in the order of 1000 100,000 fibers/1 [1], that is, two or three orders of magnitude higher than the exposures encountered in most non-occupational circumstances (Table 1). In conclusion, our meta-analysis showed an in crease of several folds in mesothelioma risk among subjects at high levels of environmental exposure to asbestos. The available data are insufficient to esti mate the magnitude of the risk at the levels of envi ronmental exposure encountered by the majority of the population in industrial countries. It is plausible, however, that such low levels are responsible for a sizable number of mesothelioma cases, although a precise quantitative estimate of the risk cannot be based only on the results of this meta-analysis. Acknowledgements This study was partially supported by a grant from the European Commission's Europe Against Cancer Programme (Contract No. SOC 96-200504-05F02). Dr Bourdes worked on this study under the tenure of a Special Training Award from the International Agency for Research on Cancer, We thank Dr J. Siemiatycki for comments and advice. References J. Institut National de la Sante et de la Recherche Medicate. Effets sur la Sante des Principaux Types d'Exposition a l'Amiante. Paris: INSERM, 1997. 2. Bignon J. Mineral fibres in the non-occupational en vironment. In: Bignon J, Peto J, Saracci R (eds) Nonoccupational Exposure to Mineral Fibres (IARC Sci entific Publications No. 90). Lyon: International Agency for Research on Cancer, 1989, pp. 3-29. 3. International Program for Chemical Safety. Asbestos and Other Natural Mineral Fibres (Environmental Health Criteria No. 53). Geneva: WHO, 1986. 4. Chiappino G, Sebastien P, Todaro A. L'inquinamento atmosferico da amianto nell'ambiente urbano: Milano, Casale Monferrato, Brescia, Ancona, Bologna, Fire nze. Med Lav 1991; 82: 424-438. 5. Yazicioglu S, Ucayto R, Balci K, Sayli BS, Yorulmaz B. Pleural calcification, pleural mesotheliomas, and bronchial cancers, cancers caused by tremolite dust. Thorax 1980; 35: 564-569. 6. Camus M, Siemiatycki J, Meek B. Nonoccupational exposure to chrysolite asbestos and the risk of lung cancer. New Engl J Med 1998; 338: 1565-1571. 7. Howitt DG, Hatfield J, Fishier G. The difficulties with low-levels asbestos exposure assessments in public, commercial, and industrial buildings. Am Ind Hyg Assoc J 1993; 54: 267-271. 8. Kohyama N. Airborne asbestos levels in non-occupa tional environments in Japan. In: Bignon J, Peto J, Saracci R (eds) Non-occupational Exposure to Mineral Fibres (IARC Scientific Publications No. 90). Lyon: International Agency for Research on Cancer, 1989 pp. 262-276. 9. Martinon L, Billon-Galland MA, Valentin F, Brochard P. Etide qualitative et quantitative de la pollution de fond urbaine par les fibres d'amiante et par les fibres minerales synthetiques - LEPI (Contrat de Recherche 93131). Paris: Ministere de I'Environnement, 1996. 10. Burdett GJ, Jaffrey SA. Airborne asbestos concentra tions in buildings. Ann Occup Hyg 1986; 30: 185-199. 11. Gazzi D, Crockford GW. Indoor asbestos level on a housing estate determined by transmission electron microscopy- Ann Occup Hyg 1987; 31: 429-439. 12. US Consumer Production Safety Commission. Report on the First Round on Air Sampling on Asbestos in Home Study. Washington, DC: USCPSC, 1987. 13. Altree-Williams S, Preston JS. Asbestos and other fiber levels in buildings. Ann Occup Hyg 1985; 29: 357-363. 14. Corn M, Crump K, Farrar DB, Lee RJ, McFee DR. Airborne concentrations of asbestos in 71 school buildings. Reg Toxicol Pharmacol 1991; 13: 99-114. 15. Health Effects Institute. Asbestos in Public and Commercial Buildings. A Literature Review and Syn thesis of Current Knowledge. Cambridge, MA: HEI, 1991. 16. Pinchin DJ. Asbestos in Buildings (Study No 8). Toronto: Ontario Ministry of Government Services, 1982. 17. Fraire AE, Cooper S, Greenberg SD, Buffler P, Lang ston C. Mesothelioma of childhood. Cancer 1988; 62: 838-847. 18. DerSimonian R, Laird N. Meta-Analysis in clinical trials. Controlled Clin Trials 1986; 7: 177-188. 19. Vianna NJ, Polan AK. Non-occupational exposure to asbestos and malignant mesothelioma in females. Lancet 1978; i: 1061-1063. 20. Newhouse M, Thompson H. Mesothelioma of pleura and peritoneum following exposure to asbestos in the London area. Br J Ind Med 1965; 22: 261-269. 21. Botha JL, Irwig LM, Strebel PM. Excess mortality from stomach cancer, lung cancer, and asbestosis and/ or mesothelioma in crocidolite mining districts in South Africa. Am J Epidemiol 1986; 123: 30-40- 417 22. McDonald AD, McDonald JC. Malignant mesotheli oma in North America. Cancer 1980; 46: 1650-1656. 23. Magnani C, Terracini B, Ivaldi C, et al. A cohort study on mortality among wives of workers in the asbestos cement industry in Casale Monferrato, Italy. Br J Ind Med 1993; 50: 779-784. 24. Magnani C, Terracini B, Ivaldi C, Botta M, Mancini A, Andrion A. Pleural malignant mesothelioma and nonoccupational exposure to asbestos in Casale Monfer rato, Italy. Occup Environ Med 1995; 52: 362-367. 25. Howel D, Arblaster L, Swinburne L, Schweiger M, Renvoize E, Hatton P. Routes of asbestos exposure and the development of mesothelioma in an English region. Occup Environ Med 1997; 54: 403-409. 26. Baris YI, Sahin AA, Ozesmi M, et al. An outbreak of pleural mesothelioma and chronic fibrosing pleurisy in the village of Karain/Urgua in Anatolia. Thorax 1978; 33: 181-192. 27. Goldberg P, Goldberg M, Marne MJ, Hirsch A, Tredaniel J. Incidence of pleural mesothelioma in New Caledonia: A 10 year survey (1978-1987). Arch Envi ron Health 1991; 46: 306-309. 28. Teta MJ, Lewinsohn HC, Meigs JW, Vidone RA, Mowad LZ, Flannery JT. Mesothelioma in Connecti cut, 1955-1977: Occupational and geographic associa tions. J Occup Med 1983; 25: 749-756. 29. Anderson HA. Family contact exposure. In: Proceed ings of World Symposium on Asbestos, Health and Society (Montreal, 25-27 May 1982). Montreal: Canadian Asbestos Information Center, 1982: 349-362. 30. Light RJ, Pillemer DB. Summing up: The Science of Reviewing Research. Cambridge, MA: Harvard University Press, 1984. 31. Siemiatycki J, Boffetta P. Invited Commentary: Is it possible to investigate the quantitative relation between asbestos and mesothelioma in a community-based study? Am J Epidemiol 1998; 148: 143-147. 32. McDonald JC, McDonald AD. Epidemiology of mes othelioma from estimated incidence. Prev Med 1977; 6: 426-442. 33. Stayner LT, Dankovic DA, Lemen RA. Occupational exposure to chrysolite asbestos and cancer risk: a re view of the amphibole hypothesis. Am J Public Health 1996; 86: 179-186. Address for correspondence: Dr Paolo Boffetta, Interna tional Agency for Research on Cancer, 150 cours AlbertThomas, 69008 Lyon, France Phone: 33-4-727 38485; Fax: 334-727 38575 E-mail: boffetta@iarc.fr