Tobacco Documents Online

PA,SSIVE SMOKING AND LL'\G ADFJapCSRCINUMA K,ht. L.C, Ho, LH C., SAw, 1). and Ho. C.Y., Measurements ofp+ssive smoking and estimutcs of lung cancer risk ~unong non•smnk;ng Chinese femake. Jnr. J. Cnncee 39, 162-169 (19871. LAM, T.H., KaNC,1.T.M. WoNO, C,M., LAA+. W.K., KLEEVENS, J.W.L., 9Aw, D., Hsu, C., SErEVIRATNr, S., LAM, S.Y., Lo, K.K. and CNAN, WC., Smoking. pacaive smoking and histological types in lung cancer in Hong Kong Chincse women. erit. J. CanceA 86, 673-678 ((987). LEE, P.N., MixluSsificarion of smoking hahip and Ixcuivr rmnkingr a rewew y rhe nidenee. lm. Arch. Occup. Envuon. Health Suppl. Springer, Berlin (1988). VYanNG, F., AGREMus, V., SVARTENCkEN, K., SVENSSON, C. and PEKfHAGBN, C., Environmental tobacco smoke and lung cancer in nonsmokers--0ocs time since expnnire play arole? Epideminingy, 9, 301-308 (1998a), NYnERtn. F., Agnldo, A., Doffeua R. Funes, C, Gontakz. :.a. and PERStucEN, G., A F.uropean validation study of smoking and envirnnmenral tobacco SmerAc expnsure in nonsmoking lung cancer rnscc and controls. CancrrCnuses Control, 9,173-182 (199gh). PLNtiIIAI:lN, G., {)RVHEC, Z. and $CF.NYaM, C., Passive 5moking and lung oancerin Swedish women. Amer J. EpidemWl.. 125, 17 24 (1987). 639 Rsnou. E., and 16 others. ExfKtsure of nonsmoking women In environmen- tel to6ucco smokce a 100.country collahoralive study. Canee. Causes Canrmf.1,24}-252(1990). STOCKN'eLL, H.G., GOLDMAN. A.L., LYMAN, O.H., NOSS, C.I., AFMSTRONO, A.W, PmKKAM, P.A„ CAnvFx.oxA, E.C. and BRUSA. M.R„ Environmental tobacco smoke and lung cancer risk in nonsmoking womea J. nar. Cancer Jnse, 84, 1417-1422 (1992). Wu. A.H., HENUaxsoN, B.E., pIXF, M,C. arrl Yu, M.C„ Smoking and other risk faetors for lung cancer in women. J. nor. Cancer Inst., 74, 747751 (1985). Wo-Wn.staMS, A.H- and SAMET,I.M., Ltmg cancer and ciFuette amuking ln:1.M,Samet(ed.),EPtrJrmiologyoflungcanrrsLung BmlogymHealth aud Disese, Vol. 74, pp. 71-108, MzrLel Dekker, Now York (1994). ZARIUTE, U., MAxIMGVrrCN, D., Z9MLYANAYA, G.. AfIAKOV, Z.N. and NnFrr.rrA, P, Exposure to envrrontnemal tnhncno smoke and risk of lun8 canccr In non-smoking wnmen from Moscow, Russia. Jnr. J. Cnnrrr, 7R, 33>-338 (1998). RJR0000001025129413 -70016 5707

r 431187Z rnr.l. Cancer: 83, 635-639 (1999) 0 1999 Wiley-Liss, Inc. tj~:~ L_ PuellCarbn a, me InrlmarinMl IMinn AaAlnal CanCer ~ Pudicalioa de I'U~m, idsmalbnal. CoMU Ie Cercer EXPOSURE TO ENVIRONMENTAL TOBACCO SMOKE AND RISK OF ADENOCARCINOMA OF THE LUNG Pauln HOrAF-TtAl*, Wolfgang AHRFN$2, Fredrik NYBRRGt3 AGUsh MllKRR1A', Irene ftRU.SKR-HOHLFELDS, Cristina FoRTES°, Vali CONSTANTINESCUr• Lorenzo SIMONATOS, Helina BATURA-GABRYELY. Suzanne LEA' 10, Valdrie GABORtEW and Slmone BGNhAA10Cl I 'International Agency for Research on Cancer (fARC). Lyon, France 2Bremen Institute for Prevention Research and Social Medicine (BlPS), Bremen, Germany }lnstirute of Environmentat Medicine, Karo(ins'ka Institute, Stockholm. Sweden 4lnttitxre ofCarcinogenesis, Cancer Rrsrmch (.'entre, Mo.srox; Rnecinn Frderrsfinc sNarionai Research Cenrer for Environment and f/ealrh (GSF)• Munich, Gerrnany 'Fpidemiology Unir fntium Region, Rome, Italy 'National Institute of Public Ilea(th, Bucharest. Rurtrania sVenero Cancer Regisrry, Padua, Italy 9Deparmtent of Lung Diseases. Medical School. Poznan, Poland t°SrhtxA of Puh/ic Neatrh, Ilnivrrsity uf California at Rrrkefeyl Rerkeley, CA, USA °Narional Institute of llealrh and Medical Research (INSERM) Unit 351, LiUejuif, France We conducted a case-control study of adenocarcinnma of the lung and exposure to environmental tobacco smoke (ETS) in 7 countries. We interviewed 70 cases of adenocarci- noma of the lung and I TO population or hospital controls. All subjects had smoked fewer than 400 clgarettes in their Iifetimes, Ever exposure to ETS from the parents during childhood was associated with a decreased risk [odds ratio (OR) 0.6,95% confidencelnterval (CI) 0.3-1.2], and there was a suggestion of a decreasing trend in risk with increasing duration of exposure. Ever exposure to ETS from the spouse was not associated with an increased risk (OR 1.0, 95% CI 0.6-1.8), while the OR of ever exposure to ETS at the workplace was 1.5 (95% Cl 0.8-3.0). For both exposure sources, an Increased risk was observed among the highly exposed, and the OR among those with the highest duration of exposure to ETS from the spouse or at the workplace was 1.6 (95%CI 0.5-6.2). A similar risk was estimated fur current exposure to ETS from either source. Our results confirm previous reports of a weak effect of adult ETS exposure on risk of adenocarcinoma of the lung. Bias and confounding cannot be excluded as explanations for the apparent decrease in Nsk from childhood exposure. Ina J. Cancer S3"S-839, 1999. O 1999 Wiley-Lisa. 7nc. MATERIAL AND METHODS During 1994 1996, we enrolled nan-smoking cases of lung adenocarcinoma from 9 centres in 7 countries: Stockholm (Swe den), Paris (Prance), Bremen and Munich (Gormany)• Padua and Ronle (Italy). Poznan (Poland), Moscow (Russia) and Ducharest (Rumani:). 9hese subjects were purt of a larger study aimed at assessing the role of markers of individual susceptibility to lung cancer among non•smokers (data not shown). The subjects from Sweden were alm part nf a previously reported analysis (Nyberg er nt., 1998a). In Sweden and Germany, cuntrols were selected amnng non-smokers from the underlying population, while in Padua and Romania, they included both subjects from the underlying popula- tion and hospital patients. In the other centres, they were selected among non-smoking healthy individuals or patients admitted to the same hospitals as the cases. Controls were frequency matched to cases on age and gender. Patients admitted to the hospitals for tobacco-related diseases were not considered as potential controls. Cases and controls were administered a standardized questionnaire, which included detailed sections on occasional smoking, ETS exposure during childhood, ETS exposure during adulthood from the spouse and at the workplace, occupational exposures, diet and family history of eaneer. The section on ETS exposure had been used in an earlier study conducted in Europe (Boffetta rt a1., 1998) and previously validated against urinary cotinine measurements (Riboli et al., 1990). We defined as non-smokers those cases and controls who during their lives had smoked fewer than 400 cigarettes or the equivalent amount of tobacco from cigars. cigarillos or pipe. This corresponds to about I cigarette a day during I year. Non-smokers included in the study were classified according to ever exposure to ETS from the mother or the father during childhood (age 0-18) and ever exposure during adulthood to ETS from the spouse, ETS at the workplace or both. Further ETS exposure variables that we developed Include: 1, childhood: ever exposure to father's and mnther's smoke, tmal duration of exposure and duration of exposure weighted for the type of smoker (mother: 1; father: 0.75); 2. spouse: cumulative exposure (number of cigarettes smoked per day by the spouse in the presence of the index subject multiplied by the number of years of exposure), duration of exposure (number of daily hours of exposure multiplied by the number of years of exposure); 3. workplace: duration of expusure (number of daily •Conespondtnce to: Unit or Environmentel Cancer Epidemiology, International Agency for Research on Cancer, 150 cours AlbenThomas. F69008 Lyon, pranee. Pax: +331,-72738342. E-mail: boffetta@larc.fr to provide additional evidence on the association between ETS exposure and lung adenocarcinoma in European populations. Received 23 April 1999: Revised 28 June 1999 Exposure to environmental tobacco smoke (ETS) has been associated with an increased risk of lung cancer in epidemiological studies conducted in North America, Asia and Europe (Hackshaw er al., 1997). We have reported the results of a multicentre case-control study of lung cancer conducted in 12 European centres (Boffetta et af, 1998), In that study. which is the largest such investigation conducted in Europe, we detected a moderate in- crease in risk of lung cancer from ever exposurc to ETS from the spouse [relative risk (RR) 1.16, 95% confidence interval (CI) 0.93-1.44) and at the workplace (RR 1.17, 954o CI0.94-1.45) with a suggestion of an increasing trend in risk with increasing duration of exposure to either source of ETS or a combination of the two. Additional important results of our study included a higher risk for squamous-cell and small-cell earcinoma than for ademxarcinuma, a decrease in risk with increasing time since stopping ETS exposure, and lack of lung cancer risk associated with ETS exposure during childhood: T'he RR for any childhood exposure was 0.78 (95% CI 0.64-0.96) with a suggestion of decreasing risk with increasing duration ofexposure. We cOnducted an additiunzl case-cnntroi study amrmg non- smoking lung adenocarcinoma cases and controls to assess whether our results could be reproduced in an independent population and RJR0000001025129413 70016 5703

6}8 ROFF7iTTA LTAL. workplace exposure with a higher nsk among subjects at longest centres with hospital controls (OR 0.4, 95% Cl 0.1-I,q, exposure, and the suggestion of a decrease in risk following while centres with population controls had an OR of 1.2 (95% cessation of ETS exposure. All these findings parallel those of a CI 0,4-3.4), suggesting a possible bias from the use of hos- larger investigation we conducted in 12 Western Europeanlthtres, pital controls. On the other hard, we found no covelation between which included subjects Iroro 6 centres participating in this ETS exposure during childhood and during adulthood (both analysis (Boffetta et al., 1998). Mhrnr discrepancies can be from the spouse and at the workplace), suggesting a lack of primarily amibuled to statistical instability of the results of the reciprocal confounding. The lack of consistency with the results of current study. previous studies, with the exception of the 2 investigations We interpreted our previnus findings of a decreased risk of lung mentioned above (Wu et nl., 1985; Pershagen er aL, 1987), cancer among subjects exposed to F:I'S during childhood as points toward a non-causal interpretation of our findings. If the possibly due to chance (Boffetta et al., 1998). A previous study protective effect is real, its underlying mechanism remains to be reported an inere:u..ed risk following E7S exposure in childhood elucirJated, but not in adulthood (halerich el a/.. 1990). Most of the largest Our study suffers from a number of limitations. The use of a sludies, however, failed to detect an association tn all lung cancers. serles of hospital-based controls may have resulted in a bias toward We iue uware of 5 previous studies that, in addition to the tfie null if ETS exposure was associated with some ol the diseases multicenlre FumlR:an study mentioned above, have reported results of the controls. The main results of our study, however, held nlsu on risk of lung ademx:arcinonra following childhood ETS expo- when the 2 series ul' amlrnl.s were analysed separately. There was sure. Stockwell et al. (1992) reported a modest, non-significant no difference in response rnte between caces and controls, reducing inorea+e in the risk of adcnoearcinomn following exposure to ETS the likelihood of selection bias. In studies nr ha S and lung cancer, from the nmther but not from the father The results of the non-differential misclassification between cases nnd controls of remaining studies (Fontham er a1., 1994t Zaridze et al., 1999) are their noo-smnkmg status and their reported ETS exposure arc 2 consistently negative, lar matter which indicator of childhood important potential sotlnxs of false positive results (L.ee, 1988). An F:TS exposure was used (either parent or both parents being objective measurement of these vnriublr.m was lacking in our study, smoker,, number of smoker-years, etc.). In the Sweli,h study, andwehavenodirecrevidenccfororagainstlheprr.xaxzofabiasd subjects with at least I smoking parent had an OR of 0.5 (95% CI ln a previous validation study based on cross interviews to relnlive.s 0.1-1.9) of lung cm,cers other than squamous- and small-cell of eases and controls from 3 European cemres, including 2 centres carcinoma (Pcrshagen rraL, 1987). A decreased risk (OR 0.6.95% participating in this study, we found a very small proportion of Cl 0.2-1.7) was also reported in a previous US study (Wu et al., subjects misclassified according to either their own non-smoking 1985). status or the smoking habit of the spouse (that is, their own spousal A possible explanation of our findings is reporting bias, i.e.. ETS exposure status), without evidence of a higher proportion of cases of lung adenoctucirrunra reporting less frequenlly a history of miselassificd eases ns compared with controls (Nyberg et al., ETSexposureinchildhoodtJwancnuaols.Tirelackofasimilarbias 1998b). The higher risk from wurkplace ETS exposuro among with regard to adult sources of ETS exposure, however, reduces the younger subjects than among older subjects suggests the possibility credibility of this explanation. Similar arguments can be used to of non-differential misclassification, due to poor recall, unwng reject the hypothesis of selection bias due, for example, to the older subjects. selection of hospital controls. For S cases and 13 controls, In conclusion, our results confirm in an independent population infonnalion on childhood ETS exposure was missing (OR in this the main findings of a larger study we have carried out: a lower risk group 1.2, 95% CI 0.3-5.0), leaving open the possibility of a small of lung cancer among subjects reporting ETS exposure during "missing value" bias. Negative confounding by a protective factor childhood, a small but siuable risk after exposure from spousal or associated with childhood ETS exposure (or a risk factor with a workplace ETS with a suggestion of a dose-response relationship, negative association) is another possible explanation. When we and the suggestion of a decrease in risk after cessation of ETS repeated the analysis of childhood ETS exposure after adjustment exposure. for urban residence, education and exposure to occupational carcinogens, we obtained an OR of 1.0 (95% C10.3-3.0, based on ACKNOWLEDGEMENTS 130 subjects with complete information). Despite the lack of precision of our results, they suggest that the apparent protective SL and FN worked on this study under the tenure of Special effect of childhood ETS exposure might be due, al least in part, to Training Awards from the International Agency for Research on negative confounding from other risk factors of lung cancer in Cancer. The study was partially funded by a grant from the non-smokers. In addition, the decrease in risk was restricted to European Commission DO-XIl Contract No. EVSV-CT940555. REFERENCES AalaA, S., KATO, H. and BLOT, W.J., Passive smoking and lung cancer among Japanese women. Cnn;er Res„ 46,4804-0807 (1986). Bec:Haa, H.,'LATONSxt, W. and IocxY4 K: H., Passive smoking in Germany and Poland; comparison of exposure levels, sources m'exposure, validity, and perception. Epidemiology, 3, 509-514 (1992). TsoiYartA, P. and 26 orHrns, Multicenter case-conrrol study of exposure to environmental tobacco smoke and lung cancer in Europe. J. nat. Cancer losr, 90,1440.1450 (1998). BRowNsoN, R.C., ALAVANJA, M.C.R„ HncR, F.T. and Lnr, T.S., Passive smoking and lung cancer in nonsmoking women. Amer. J. publ. Realrh, 82. 1525-1530(1992). BRowNsox, R.C., Rr:n, J.S„ Ksvti, T.L, FeRCUSaN, S.W. and PRrtzL, J.A., Risk factors for adenocarcinoma of the lung. Am J. EpidemiW., 125, 25-34 (1987). FoNTHAM, E.T.11., CORREA, P., REYNOt.os, P., Wu-WIUrALts, A., BursLER, P.A„ GREENaFRO, R.S., CHEN, VW., ALTLRMAN, T., BOYD, P., AusnN, D,F. and LJr., J., Environmenral tnhaccu smoke and luug cancer in nonsmoking womcn: a multicenter study. J Amer Med Astoc., 271,1752-1759 (1994). RJR0000001025129413 70016 5706 ` GAO, Y: T., BLOT, W,J„ ZueNG, W., Easxow, A.G., Hsu, C.W, l.EVw, L.I., ZHANO, R. end FRAOMENL J.P., lt., Lung cancer among Chinexe women. Jnt. J, Cnncer, 40, 604-609 (1987). GARFINxeL, L„ AUPARACH, 0. and IOUaERT, L., Involuntary smoking and lung cancer: a caseconlrol study. J. nus Cancer Jmrt., 75, 463~69 (1985). HACCSrtnw, A.K., lww, M.R, and WAto. N 7.. The accumulated evidence on Iung cancer and environmental tobacco smoke. Brir. rrled J., 315, 980-988 (1997). IARC, Tobacco smoking. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals lu Humans, Vol. 38, International Agency for Research on Cancer, Lyon (1986). JANERICH, D.T., TktM1P50N. W.D., VARELA, L.R., ORFRVwALD, P., CHOR05T, 8.,11rcx~L C., ZArAAN, M.B.. MELAStau, M.R., KlEt.v, M. and McKNaAU.v, M.F., Lung cancer and exposure to tobacco smoke in the household. N. Engl. J. ked„ 323,632fi36 (1990). KALANUnII, A., KATaDtIYANNI, K., VORUPOULIHI, N., BA.STAS, G., gARACtA, R. and TrucnorouLOS, D., Passive smoking and diet In the etiology of lung cancer among non-smokers. Cancer Causes Cantrol, 1, 15-21 (1990).

636 BOFFETTA ETAL TABLE 1-SrLLC7Fn CHARACTERIS11C5 OF THE STUDY POPULA'ION Coe. Colltrols N (701 N(Il8) -_.!F Age group (yrnrt) I8-54 11 15.7 48 27.0 55-(A 20 28,6 63 35.4 65-74 31 44.3 42 23.6 75+ 8 114 25 140 Gender Men 4 5.7 41 23.0 Women 66 94.3 137 77.0 Coualry Sweden Ih 22.9 25 14.0 GermMy 11 15.7 67 37.6 France 4 5.7 11 62 Italy 9 12.9 31 17.4 Romanin 9 12.9 14 7.9 Rus.ia 16 22.9 :0 11.3 Poland 5 7.1 IIl 5.6 hours of eXpOsurn multiplied by the number of years of exposure and a subjectlve index of smukrnesti of the workplace); 4. spouse and workplace combined: duration of expusure (aurn (if the indices of duration descnbed above without the weighting by workplace smokiness) mnt Ilme sina stopping exposure from both sources. For quantitative adult ETS eslurwre variables, exposed subjects ween classified in 3 groups comprising approxin)ately the bottom 75'k, the next 15% and the top 20°k of the controls, This approueh was based un Ihe results of a urinary cotinine study showing that misclassification uf yuesliounaire-based ETS exposure is greater in the 3 lowesl quaniles of the dicuibubnn than in the top quartile (Recher er al., 1992). For duration of exposure to E'1'S during childhood, only 2 categories were used, with the eatpoint at the median of the distribution among controls. We retained in the analysis only histologically confirmed cases of lung adenocarcinoma; cases classified es mixedutknocarcinoma or another histological type were excluded. We fitted muhivariate logistic regression models to estimate the odds ratios (ORs) of lung adenocarcinoma from exposure to ETS from the different sources Rnd their 95% confidence intervals. All the regression models included terms for age (10-year groups), gender and centre. In addition, we fitted models including the potontial confounders: urban residence, education and exposure to occupational carcinogens. We tested the significance of the linear trend in risk across increasing categories of quantitative ETS exposure variables by fitting an ordinal vnriable. We conducted additional analyscs, restricting the study population to women and separating young and old subjects with a cutpoint at the age of 65. RESULTS A total of 70 non-smoking adenocarcinoma cases and 178 non-smoking cantrol s were included in the study. Their distribution according to age, gender, and centre is shown in Table 1. Cases (mean age 64.4 years) were older than controls (60.9 years); 5.7% of cases and 23.090 of controls were men. All countries contributed at least 15 subjects to the study. One-hundred and seven controls were healthy individuals, and 71 were hospital patients treated for a variety of diseases, including orthopaedic conditions (N - 17), non-tobacco-related malignant neoplasrns (N = 17), acute respim- tory conditions (N 6 14), arthrosis (N = 5), digestive diseases (N = 4) and benign neoplasms (N - 4), Exposure to F.TS during childhood was reported by 33 cases and 110 controls, resulting in an OR of 0.6 (95'k C10.3-1,2) (Table II). More subjects reported exposure to father's smoke than to mother's smoke, and the decrease in risk was more pronounced, although not statistically significant because of small numbers for the latter type of exposure. Three cases and 9 controls reported exposure from RJR0000001025129413 70016 5704 1 TABLE It-ODDS RATIOS DFLUNO ADENOCARCINOMAFROM E%Pp51JRCT0 ENVIRONMENTAL TOBACCO SMOKG OP PARENTS DURtNO CHILDHOOD Nmnbtr ulcnms Number urcrnyab "~~-" OR YS'xCl Unexposed 32 55 1.0 Ref. Ever exposedl 33 I l0 0.6 0.3-1.2 Exposed to father's 33 101 0.7 0.3-1.3 smoke Exposed to rnother's 3 18 0.4 0.(19-I.8 smoke Duratian of exposure' (weightRdyear.s) I-10 8 2 .0 .4-2.4 10.1+ 10 50 0.3 0.1-0.9 Trend 0.06 ~ OR: odds ratio adjusted for age, gender and centre; CI: confidence interval; ReL: reference category. 1.5 cti.srs and 8 controls had missing information. ?10 exposed cases and 21 exposed controls had missing infunoation. TAatP. nl-ODDS RATIOS OF LUNOADENOCARCINUMA 1'ROM axP(1SIrRE TO 6NVIRONMEh7AL'NUAt,CGSMOKEFROMTIBSPOU$E Number Numeer ur<ontroU OR 94K(1.I Unexposed 37 103 1.0 Ret. Ever exposed 33 75 1/) 0.5-1.8 Duration of exposurer (hoursl day X yeara) 1-120 9 4 .9 .4-2.0 121-258 6 1l 0.9 0.3-3.0 259+ 4 7 1.9 0.3-12 Trend 0.8 Cumufarrve exporurel (pack-years) 0.1-14.0 20 48 0.9 0.4-1.8 14.1-26.0 5 12 0.8 0.2-2.7 26.1+ 6 8 1.5 0.4-5.9 Trend 0.9 OR: odds ratio adjusted for age, gender and cemre; CI: confidence interval; Ref,: reference category. 14 exposed cases and 13 exposed controls had missing infnmra, tion. ?2 exposed ceses and 7 exposed controls had missing rnforrna- tioni both parents (OR 0.8, 95% C10.2-3.6). There was a decrease in the risk of lung adenucarcinoma according to duration of exposure to ETS during childhood after weighting this variable according to type of smoker. Results were similar for the correaponding unweighted variable (not shown). The reduced risk of lung udenucarcinoma was present among women and men, while the trend in decreasing risk was clearly present only among subjects below age 65. Tablelll presents the results according to ETS exposure from the spouse. There was no overall increase in the risk of lung adenocar- cinoma among the subjects exposed to spousal ETS (OR 1.0, 95% Cl 0.5-L8). A non-slgnifrcant increased risk, however, was present in the highest category of duration and cumulative exposure, these two variables being correlated (Pearson correlation coefficient among controls 0.81, p< 0.01). The results of the analysis restricted to women were similar to those based on the whole study populatitm, sinee there were no male cases and only 6 male controls exposed (OR of ever exposure among women 1.0, 95% Ci 0.5-1.9). No clear pattern emerged from the analysis based on different age groups. The main results of the analysis of ETS exposure at the workplace are presented in Table IV, Exposure was reported by 38 cases and 97 controls, yielding an OR of 1.5 (95% Cl 0.8-3.0) without a clear indication of a dose-response relationship for duration of exposure. The risk of lung adouncarcinoma from

PASSIVE SJIOKING ANO LUNG APENOCARCINOMA 637 P - ~ Ofr ~ TABLPIV ~SU~TO workplace ETS exposure was 1.2 (95% CI U.fr2.5); that of the THEw RKPLAC ,A'f S E ONMER ' ALTOBACCO MOK m D • l k l NvMSer arc>res Nemxr ofcvawie OR vsve cl Unexposed 31 81 1.0 Ref. Everexpnsed, 38 97 1.5 0.8-3.0 Durariort of e.rposure' (hoursl day X years) I-61.2 6 4 .9 .9-4.1 61.3-157.0 4 13 0.7 0.2-3.1 157.1+ 6 8 1.7 o.4-6b Trend 0.4 OR: odds ratio adjusted for age, gender end centre: CI: confidence intervah Ref.: reference category. 'I case had missing infnrmatinn.-t2 expused cases and t2 exposed cnntmlx had missing information. TABLE Y-ODD9 RA]rO9OF LUNGADENUCARC'IFUMA7 RUM P.xI7)SUBE"rt) F.NYIRONMEMALTOaACCn SMOKE fAUM THB SPOUSE OR AT T89 WORKPLAC'E Numl!r nferes Nurnmr ofanmrak OR 9J4>CI Une.xposed 211 52 1.0 Ref. fiver exposed 50 126 1.2 0.6-2.5 Duration of e.ryosure' (hoursl day x ytara) 1-111 0 3 .2 .6-.26 112-200 R 20 1.0 0.3-3.1 201+ 9 II 1.8 0.5-6.2 Trend 0.5 Years since lasr expo. surer 15+ 9 40 0.5 0.2-1.5 3-14 19 36 1.4 0.6-3.3 0-2 22 49 1.9 0.".5 Trend 0.08 c OR: odds ratio adjusted for age, gender and centre; CI: confidence interval; Ref.: reference category. '3 exposed cases and 12 exposed controls had missing informa• tion: 'l exposed control had missing information. workplace ETS exposure was higher in men (OR 2.6, 95% CI 0.05-140) than in women (OR 1.2, 95% CI 0.6-2.5), but this difference was not statistically significant and the risk estimate in men was highly unstable. The increased risk was present among subjects who were younger than 65 at the time of the study (OR 4.5, 95% C1 1.6-13) but not among older subjects (OR 0.6, 95% Cl 0.2-1.5; p-value of difference 0.004). ETS exposure either from the spouse or at the workplace was associated with an OR of 1.2 (95% CI 0.6-2.5) (Table V). Increasing duration of exposure showed a non-significant increas- ing trend in risk ip-valuc 0.4) with subjects In the category of longest duration most clearly at risk. The results, after atratification of the study population on gender and age, suggested a higher risk in men than in women, as well as among younger subjects B.s compared with older subjects (results not shown). These results are explained by the higher risk from ETS exposure at the workplace among men and among subjects below age 65 (see above). An effect of stopping ETS exposure is suggested by the results obtained according to time since cessation of EPS exposure from the spouse or at Ihe workplace: the OR of lung adenocarcinoma was 1.9 (95% CI 0.8-4.5) among subjects currently exposed but only 0.5 (95% C10.2-1.5) among those unexposed during thc last 15 years (Table V). This effect was prcscnt among wnmen and men as well as among subjects above and below age 65. Adjustment for exposure to occupational carcinogens or educa- tion did not modify the results substantially. After adjusting for occupational exposure, the OR of ever exposure to spouse or w or r p o exposure to spousa highest category o durat on o f f i E"fS was 1.8 (95%CT 0.5-d.2). Adjustment for education resulted in ORs of 1.5 (95% C10.7-3.3) and 1.6 (95% CI 0.4-6.1) for the same ETS exposure variables. Information on urban residence was missing for 24 cases and 39 controls; the adjustment suggestrd a decrease in the risk estimates and in their precision (OR of ever exposure to spousal or workplace ET'S 1. 1. 95% CI 0.5-2.8). DISCUSSION Our present analysis was performed to assess whether the results of n bage ca.econtrol study of ETS exposure and lung cancer we had conducted (Boffeihr er al., 1998) would he repl n:ated in another study population and to provide addidonnl evidence un the association between ET5 exposure and lung adcnocarcinoma in r:umpean populahnns. We restricted this study to lung adenocarci- noma, the most common histnlogwul type of lung cancer among non-smokers, mainly due to the small number (f cnses in our study with other histological types of lung cancer. The etrrcinogenic effect of tobacco smoke on adenocarcinomas has been considered weaker than un ruher histological types of lung cancer, namely squamous-cell and small-cell carr:inonm (IARC., 198G). Other evidence, however, suggests that the difference in risk miglN he smaller than previously thought (WU-Williams and Samet, 1994). A nnmtxr of epidmuiological studies have reported results on risk of adenocarcinoma of the lung following exposure to ETS, mainly from the spouse. Significant (or borderline significant) increases in risk, with relative risks in the order of 1.3-2.1, have heen reported in studies from the USA (Garfinkel er ai., 1985; Fonthttm er nl., 1994), Oreece (Kalandidi et al., 1990). Russia (Zaridze er al., 1999) and Hong Kong (Lam er at, 19g7). A similar increase in risk, albeit not significant, has been reported by Stockwell et al. (1992) and Brownson st al. (1987) in the USA and by Koo er a7. (1987) in Hong Kong. Only 1 study, from the USA, failed to detect on increased risk (Brownson e( at, 1992). In Europe. a study of Swedish non-smoking women found an increase in the risk of lung cancers other than squamous- and small-cell carcinomas after cumulative ETS exposure from the spouse of 22.5 or more pack-years but not for lower exposure (Pershagen er al., 19B7). In a multicentre study from Western Europe, the OR of adenocarcinoma following ever exposure to spousal ETS was low, 1.08 (95% CI 0,82-1.42); the OR following workplace exposure was 1.06 (95% CI O.B 1-1.40), and the OR for prolonged duration of exposure from the spouse or at the workplace was 1.58 (95% CI 0.98-2.54) (Boffetta et al„ 1998). In most of the studies reporting results on different histological types of lung cancer and ETS exposure, the increase in risk of ademx:arcinuma was smaller than that of squamous- or small-cell carcinomas (Akiba er al., 1986; Brownson et al., 1987; Gao er a/.. 1987; Pershagen er a7., 1987; Stockwell er aJ., 1992; Fontham er af., 1994; Boffetta er aj., 1998; Zaridae er al., 1998). In 2 studies from Hong Kong (Koo er al., 1987; Lam er al., 1987). 1 from the USA (Garfinkel er al., 1985) and I from Greece (Kalandidi et al.. 1990), however, the risk of adenocarcinoma was equal to or higher than that of other histological types. Despite the lack of statistical significance, the results on adult exposure confirm the presence of a small increased risk in lung adenocarcinoma following adult exposure to ETS. In addition, the finding of a concentration of the risk among the most heavily exposed subjects is consistent with a earcinogenic effect. The replication of our previous finding of a protective effect of stopping ETS exposure adds to the evidence that this dimension of ETS exposure is an Important one that should be carefully examined in future studies conducted in populations in which important reductions in the prevalence of smokers has taken place (Nyberg et ae., 19980). We observed a decrease in adenocarcinoma risk following childhood ETS exposure, a small increase In risk after spousal or RJR0000001 025129413 70016 5705