Tobacco Documents Online

41 T.-h,,ku J esp: NW 154. 3.*A'+ 'ili; A Case-Control Study of Lung Cancer in Nonsmoking Women HIROYL'fil CHI?I1Zi, ML-\EHIFO MOAISHITA,* K.ATSC'1`b'KI .%llzrx-oa T.aKaO MASL'DA.$ FCiCIO OGCRA,j IIITSt•Hlko SAyTO;: MIXORC N1SHI3ECRA,4 KAZCo KC\ISHI~I-a. KAZCO KARASAII'A,- KEISUKE NISHIWAK]',„¶' MASAHIF,O YASIA)IOTO.' SHIGERL' HI'SA)IICHI and .SCKETANfI Tox1NAG:a' • Ikpart»trut of Public Nrnlth, Tolioku L'nitvrsity School of -1'lcdieiitr_ .`cadni 980: 'tbe Srcoitd' Department of Intental .1lydiciptr. \'ngoya City f-niversity: -11'edical' School'. ltagoya 467. tthr Third Drpnrttuntt of lutental Medicine, Aichi Medical l'aitYrsit y. Aatdti IS0-11.:Department of Intenral .1ledici,tr. Nntioual Nngnyn Hosptlal, Vagoya 16!i. §Deparnmrnt of Intental' .lledi'ciue. .-liciii Cancer Ctntrr Hospitnl'. Nagoya 464. "Department of Surgery; rliihi Canerr (.'rntPr Nospital, Nagoya 464. 11 Department of' Intental -1Jedici,te, Chukyo Hospital, Nagoya 457 a,.d "Die-isio/t of Epidemiology. Aichi Cancer Center Research Institute, Nagoya 464 CHtNlit'. H.. MbRISH17A. H.. MIZL'xo, K.. MASUDA. T., QoraA, Y.. SAVio: M.. `tsHtycRa. A4'.. KrviSdIS1A. K.. KaR,uaWA. K.. Vrsatww/tt, K.. YwttA>roso. SI.. llisa.ncH1, S. and Tou1.ACA. -S. .d Case-Conrrot' Study of Lung CaMen ia -VoHsmoking 1'd"omrn. Tohoku J: eSp_ )led.. 1988. 1154 (4), 389-397 - A case- contrul study of Japanese uomen in Nagoya was conducted to investigate the significance of passive smoking and other factors in relstion to the etiolotp• of fernal. lung cancer. A total of 90 nonsmoking patients with primary lun¢ cancer and their age- and: ho.cpital-mutched female controls were asl-ed to fill in a questionnaire in the hospital. Elevated relative risk (;RR), of lung cancer w•ac obsen•ed for pussive.moking from mother (RR = 4.0: p<0.05) and from hutband's father (RR = 3.2: p<0.03): No acsociution was observed between the risk of lung cancer and stnol-inE of huz:band or passive smoke exposure at work. 4ccupa- tional exposure to iron or other metals also showed high risk pRR= 4.8: p<0.051. No appreciable differences in food intakes were observed between ca<es and controlsg lung cancer : r-omen : rronsmoker : passive smoking : metal exposure Received January 7. 1988: revision accepted for publlication Harch 8. 14~M. Reprint requests: Dr. Hiroyuki Shimizu. Department of Public Health. Toboku L"nia-ersity School of Medicine, 2-1 Seiryo-machi. Send.i 980, Japan. 389 NIOTICF This rtatatial rttay bO protected by copyright law (Tide 17 U.S. Code). NOTICE: TAiS IIATERlA[ 6UIY 81 PRC1fETfD BY COPY1W LAW (T1T1F 11 U.S. CO)Lri ~:023512826

390 H'. Shimizu et al. The causes of lung cancer in Japanese women have not been clearly identified. It is widely accepted that cigarett,e smoking is causally associated with lung cancer, but the increasing trend in the urcidence of lung cancer in~ Japanese women cannot be explained by smoking alone. The proportion of smokers among Japanese women remained around 15% during the last thirty years (Tominaga 1982) and the most predominant histologic type of lung,cancer among them was adenocarcinoma, which was considered to be more weakly associated with smoking as ¢ompared' to lung cancer of other cell types (Shimizu 1983, Nakamura et al. 1986 ; Shimizu et al'. 1986). Several studies have been conducted with emphasis laid on passive smoking and lung cancer, since the first positive results were presented by Hirayama (1981)) and Trichopouios et al. (1'981). Some of these studies showed a clear association of passive smoking with lung cancer (Correa et al. 1983 ; Qarfinkell et al. 1983 : Akiba et al. 1986; Inoue et al. 1986). However, the results of other studies were equivocal or negative (Garfinkel 1981 ; Kabat and Wynder 1984 ; Koo et al. 1981: Wu et al. 198.5. Lee et al. 1986). This paper reports a case-control'study of lung cancer in Japanese nonsmok- ing .t•omen. in a•hich passive smoking and other factors such as occupational histon•. domestic heating system and dietary habits were investigated. XAT$RIAIS AVD MSTHODS Our case: consisted of female patients with primary lung cancer who were treated in 4 hospitals in \agoya from August 1982 to July 1983. One oC the hospitals (Aichi Cancer Center Hospital) was a cancer hospital and the remaining three were general hospitalk. Nagoya is the fourth largest city in Japan with a population of 2.L million and located in the middlr of the main island. Honshu. Durhig the above period 118 female lung cancer patients were pathologically identified. The physiciars or nurses asked all of them to fill in a questionnaire for this study on the first or second day of admieaion to the hospitals. Out of 118 lung cancer patients 4 refused'to fill in the questionnaire and 24 reported that they were current or ezsmokas. The remain, ing 90 nonsmoking patients were selected as the caaes for the following analyses.. The questionnaire mainly consisted of the questions about smol-ing, occupational history. dietarc haHit,c. personal disease history and about the kinds of fuel for cooking. As regards passive smoking. we asl•ed' them about the smoking habits or the number of cigarettes smoked per day by parents, siblings, children or husband't parents in the home. We alto asked them about the length of time which the woman spent witb her husband in the same room. the period of married life and the number of cigarettes smol•ed'by her husband. The passive stuoke exposure at working places was assessed only in terms of the presence or absence of smokers. As regards dietary history, we acl-ed the frequency in recent five years of intake of food items and divided into four categories (no intake, I or 2 days; w•eeh. 3 or 4 davs week. and almost every day). We asked directly the number of glasses of milk and the number of orattges taken per week. The 90 lung cancers included 69 adenocarcinomaa (T7°/a). 13 spuamous cell caseinomas (14°.o), 4 Irrae cell carcinomas (4%). 3 amallioell carcinoma (3°0) and 1 adenoid cystic c,trcinom+ The number of eases in the age group of 30-39. 10-49. 50-39. 60-69. 70- ?9 and 89) -Verrs were 3(3°0), 16 (17%a). 38 (31%), 2-1 (30%), 14 (16°,%) and 2(2°:0) respectix•eh- The minimum and maximum ages of the ca+es were 35 and 81 years and those

Risk Fjctr*r~ for Female Lung Cancer 391 nf controls uere alsta.i and'R1 yea". respect it'elr. The mean age at admissinn '.~.L, - •i`.iyears for ca%es and 9A cears for controls. As a control. up :k.,ked female in-patientF other than those with lung canee- tn the same or adjacent wards ui! :i:cho.cpital to fill in the questionnaire as we did for lunc cancer patients (i.e.. povnttrllcontrul:!, We selected two controls matched' in txrms of h.--pital'. (the same hospitul). age (± I}•ear), and dat. of admission Eoreach case from these }xirentiall rnntrols. For 17 ca,~e, we could find only one control which satisfied the cntrria. The controls finally used' for this,anall•sis compricedl 163 patients with the following dixases : breast cancer 67. (ll^„) : diabetes mellitus, 11 (7°,p) : stomach cancer. 11! (%"; t: hepatitis and other Iii-rr dtsew-es. 8(5°0) : malignant I~ mphoma, 7(a%) : heart diseiyr> 5. (3%) : hcpertensiou 5. (3°0) : ga(l stone. 4 (2°,o) : colorectal cancer 3. (2°,0) : cancer of the uterine cen•ic 3. C_°;o) : and others 39: (24°0). The logistic rrgression method was applied to this individually matched caee-control study andisa* ratio was computed s estimated relative risk for each variable (1Breslow- eti al. 1978 : llresiotr and' Day 1980). The statistical significance was determined bo' using tw-oasided p values. R Es cLTs Table 1 shows the risk of'female lung cancer for several ty}x< of passive smoking. When the mother of a case was a smoker, the relative risk of lung (;-cancer was 4.0 (p <0.05): However, the risk was not elevated when her father was a smoker (RR=1.1). High relative risk was observed when the husband's, father living K ith the case smoked in the home (RR-3.2 ; p<o:05). When mother or husband's father was a smoker, the relative risk was 3.3 (p <0.01). There was no association between the risk of lung cancer and smoking by husband, siblings or children~ in the home. Passive smoke exposure at a•ork was not clearliy associated with female lung cancer, although the relative risk was slightly elevated (RR= 1.2). Table 2 shows the combine& effect of household' smoking by mother and Tsnta 1. Rs/otitr ritks (RRJ' of luny aaacer in no.rrrnok• tng taornen for senrral' typu of tobocrn tmoke etpoeurrf Smoker Frequency in controls (%) RR In the home: Husband 56 1.1 Father 41 1.1 Mother 3 4.0' Husb.nd's fat6er, 8 3.2• Husband's taother 4 0.8 Son(s) or daughter(s) 40 0.8 Brother(s) or siiter(s), 32 0.8 Someone at working place 35 1.2 N 'p<0.06. 0 , N CJ CA

392 H. Shimizu, et al. TnsLE 2. Relatitt risks of lung cnnoer in norts+noding wo+nen for srwkzng by wrothe+ and kusb'and i father in the hoiwe Smoking by husband's father ('-) (+) Saaling by mother (-) CA '"S 9'': ( + ) 6-3 2.8 'p<0.05 husband's father. Both of these two rariables showed a relatively high risk independ'entlk•. Particularly, the risk for smoking by husband's father in the absence of smoking by mother was significantly elevated (RR=3.9, p<0.05). However. no synergistic effect of the above two variables w as observed. About 60°° of the respondents had occupations. No difference was foun&in the distribution of the occupational categories between cases and' controls. How•erer, histories of occupational exposure to specific substances showed high risks of lung cancer. The relative risk for exposure to iron or other metals was 4.8 (p <O.Oa), although the frequency of such exposure was very low in controls. The relative risk for exposure to coal, stone, cement, asbestos or ceramics was 3.3. but it was not statistically significartt. For the analysis of dietary habits. cut points dividing into lower two an& higher two categories are arbitrarily chosen in general. We selected the 8/week or more as cutpoint for mandarine oranges in winter and odds ratio of milk was computed for the daily intake. Table 3 shows that there is neither positive nor negative association with food items investigated here. Only chicken showed the l'uw risk of 0.7. We observed no dose-response relationship for these variables. The per_sonai medical: history of silicosis showed the relataive risk of 2.0, but T.tetc 3. Relative risks (RR) of (epiy mnnr in wonswioLing w°wen in /r1017oM to Ae fleputncy of food illW-e Food item Frequency of intaL•e Creen-.•rllow• .•egeubles z 3d .r Fruit 23d R Oc.ng-. ( msndarine ) z 8 w Milk 2 I glass/d Fish ~ 23d-x' Pork z 3 d.r l3erf z 3 d w chick-,o 23 d w d. dj.-s : w. w•eeks. Frequency of intal•e in controlt (%) RR 86 0.9 86 1.2 T7 1.0 76 1.0 55 1.0 22 1.0 20 1'.0 40 0.7 lti

Risl: Factors for Female Lung Cancer TABLE 4. Rrlatiir niks (RRI of lung canur w Monsmoking women for typr of household' hratireg system usrd in recent yean Type of household' heating system Frequency in controls (%). RR G as 32 1.t1 Kerosene 86 1.6 Coal or charcoal 8 l.T TwBtaS: Relative risks (RRJ of lung cancer in wnsmod•ing romrn in relation to the atlaud factors (w = 66i) . 393 RR Factor Crude Adjustedt Smoking b} mother in the home 3.0 2.1 Smoking by husband's father, in the home 3.5• 3.°' Uccupationaliexposure to iron or other metals •2:8 2.4 tRR of each factor adjusted for other two factors after excluding the pairs in which one of the factors had unknown values. 'p<0.05. it was not statistically signiScant. The risk for histories of both chronic bronchi- tis and asthma was 0.8, and the risk for history of tuberculosis R-as 1.1. No appreciablb difference was observed between cases and controls in the typee of household heating in childhood and' in the kinds of fuel for cooking in adulthood. However, a recent use of a kerosene or coall (charcoal) stove for household heating showed a somewhat higher risk (RR = 1.6 and 1. 7. respectively). However, neither of them was statistically significant (Table 4). The frequency of using cooking oil was almost the same in cases and controls. To confirm the risk associated with each variable described above. we computd ed the relative risk by using the multiple logistic regression analysis for the main 3 variables. Table 5 shows that the results are almost the same as those in univariate analysis. Discvssro N The presence of a smoking family member does not necessarily indicate that exposure to a sideatnam of cigarettes has actually occurred. To know the level of passive smoking, measurement of concentration of cotinine in the urine is useful (Matsukura et al. 1964 ; Wald et a1. 1984). However, it is very hard' to .ssess the passive smoking level over a period of several decades because the hal'f-life of serum cotinine is 72 hr. In this analysis we used only the information on smoking history of the respondents, their family members and their colleagues at working

394 H. Shimizu et al. places. tm this study we found a positive aseacist'ion between lung cancer in nonsmoking women and the smokcng history of family members, especially that of mother and husband's father. As Japanese children usually spend much longer time with their mother than other family members do, mother's smoking may be a representative index of passive smoking before leaving home at around 20 years of age. Recently we found that the saliva cotinine levell of nonsmoking school- children is not high when their fathers were smokers but high when their mothers were smokers in tliyagi, a district of northeastern Japan (unpublished data). After marriage, 33°-0 of women in controls lived with their husband's parents. The final' proportiom of control women whose husband's father smoked cigarettes in the home was as small as 8%, but that (18%) of caxs was somewhat larger. The husband's father may have retired already and may have stayed home much longer than the husbands. There is a possibility that Japanese women may be more frequently ecposed' to the smoke of cigarettes by their husband's father than that by their husband. We assessed the total length of period which a woman spent with her husband from the length of the period of marriage and the hours during which she lived in the same room, but no difference was found between cases and controls. No dose-response relationship was observed between the risk of lung cancer and' the history of smoking of mother or husband's father. Usually the respon- dents remember whether their mother or their husband's father were smokers, but they may be unable to recall the exact number of cigarettes smoked by their mother (especially in childhood) or husband's father in the home. It has been suggested that beta-carotene and preformed vitamin A decrease the risk of lung cancer (Smith 1982 ; Hinds et al. 198d): We asked a very simple question concerning the frequency of green-yellow vegetable intake, which has been referred to as a protective factor against lung cancer in a large cohort study of Japan (Hirayama 1982). No association was observed between this variable and female lung cancer risk in our study. Most of the respondents had green- yellow vegetables very frequently and we found no difference between cases and controls. There was no dose-response relationship between the frequency of intake of green-yellow vegetables and lung cancer risk. We also assessed the efficacy of vitamin supplements over a period of more than one year in this analysis, and found the risk of 0.5. However it was not stuisticall,r significant. Other dietay factor such as vitamin C and cholesterol may be related to the development of lung cancer (Hinds et' al! 1983, 1984; Byers and Graham 198#), but no appreciable association was observed between the risk of lung cancer and the intake of food items listed in this study. To evaluate the effect of dietary habits, more precise measurement of food intake is neededi A slightly elevated risk for disease history of silicosis is consistent with the

Hi;k F.tctnn fi7r Femnla Lung C:ni Pr 39.i data in recent repon- (Finkelstein et al. 1982 : Lvnge et aL 19.16). despite the fact rhat our resu)ts tt-erc tiaae& on the infonnatiom reported by the respondents andd that the number vf rises with silicosis was rerv small. An excess risk of aelenocarcinoma of the lung observed preriously for those with occupational exlw4ure to iron or other metals in \agoc,i area (Shimizu 1983'). Even if the risk for these occupational ecposure is confirmed. contribution of these factors is small because the frequencr of such exposure is very low in Japan. Possibly tli-rv :- some bias in our studk•. Lung cancer cases tcere not derived from general population but from the patients of a limited number of' hospitals;. The proportion of adenocarcinoma patient• in our series was ten percent larger as coatp<tred, with that :n total lung cancer patients of this area. The proportion of squamous cell carcinoma showed ut opposite tendency (Karasawa 1983). We :,electPd the controi- from the same hospitals considering that both cases and' controls in the same hospital may have similar backgrounds. Hoa•ecere one of the hospital w:Ls a can, •r hn<pital' and wp had to include many breaut cancer patients in the controls. For this reason we compared the status of passive smoking among the breast cancer patients with that among other controls, but we found no difference. Furthermore, the risk of lung cancer for the survivors of cancer of thee brew;t was not high when assessed by the data of a populattion-based cancer registrr (Takano and Oltuno; personal, communication). Our study showed that the exposure to tobacco smoke from household members (i.e., mother or husband's father) could be associated with female lung cancer. As the precise situation of passive smoking in the home or other places is stilli unclear, further studies are needed to clarify the significance of passire smoking in relation to the etiology of lung cancer in Japanese tt•omen. Acknowledgments We aie grateful to Ms. K. Nirose of Aichi Cancer Center Research Institute and Ms. Y. Takah,onhi of Tohoku [?nirenity School of Medicine for their techmcul aimi=tance. This aud}• w•as supported b~~ a(:ran~in-Aid for Cancer Re;earch from the llinistn• of Health and l4elfare (l:rant Number 57S): References 1) .t,kiba, S.. Kato. H. k Blot. W:J. (1986) Passive smoking and lung cancer among Japanese women. G'ssar ltes., 46. 48W-1807. 2) Breslovr, N.E. & Day, N.E. (1980) The analy:u of caa•control studies. In : Starisricn! .1Tuhods ix Caroe. Rarcarc6. VoL 1. IARC Scientific Publications No. 32. International Agency for R+uearoh on Cancer. Lyon. 3) Brrrslo.r. N:E.. Dsy; N.E:. Halvorsen. K.T.. Prentice, RL: dt Sabai. C. (1978) Estimation of multipk relative risl• functions in matcbed caaeronuol studies: Ame+. J. Epide+wiof.. 11116, 299-307. 4) Syers. T: & Graham, 8. (1984) TThe epidemiology of die4and'cancer. In : Adf+aROrs ie ['awerr RtxaneJi, Vol. 41, edited by (.. Klein & S. Weinhouse, Academic Press, Orlando-Fbrida, pp. 1-69. 5) Correa, P.. Pickle, L.W., Fontbam. E.. Lin. Y. & Haenszel. W. (1993) PPassive

396 H. Shimizu et al. smoking and lung caneer, Lancrt. 2, 393-597: 6): Finkektrin„M.. Kusiak, R. & Suranyi, G. (1982) i[ortality among miners receiving workmen~s compettsation for silicosiz in Ontario : 1940-1975. J. aacup. Med.. 24„ 663-667. 7) CTarfinktl. L. (1981) Time trends in lung cancer mortality among nonsmokers and a note on passive smoking. J. nat. Cancer Iest., i6: 1061-1066. 8) Garfinkel. L, AueTbach: O. & Joubert, L. (198b) Involuntary smoking and lung cancpr: A case-contr+ol study. J. nw. Cancer last., 75, 463-469;. 9) Hinds. 11.W.. Kolonel. L.N., Hankin, J.H. k Lee, J. (1983) Dietary, cholesterol and lung cancer risk in a multiethnic populationin Hawaiil, htt. J. Cancer. 32. 727-732. 10) Hinds. M.W., Kolonel, L.N., Hankin, J.H. & Lee„ J. (1984) Dietary vitamin A, caroten-. vitamin C and risk' of lung cancer in Ha.raii; Ame+. J. 6pidemioi., 119, _r!7-237. 11) , Hirs}•ama. T: (1931) Non-smoking wives of heavy smokers have a higher risk of lung cancor: A studY fr+om Japan_ B.it. twed. J.. 262, 183-188. 12), Hiraranma. T. (1932)' Epidemiological aspects of lung cancer in the Orient. In: Lung Cancer 194.'_ edited' by S. Ishikacra, Y. Hayataa & K. Suemasu, Excerpu \ledica. Atnsterdam-Oxford-Princeton,, pp. 1-13: 13) Inoue. R.. Ohtsuk+t. T:, Shimura. K. & Hirayama. T. (1986), A case-control study of lung cancer. Lung Canaer, 26. 763-767: (Japanese) 1i) Ksb&t. G_C. & 1i•.-nder, E.L. (198{ ') Lung cancer in tansmokers. Cancer, 53, 1214- 1-221. 15) Karasawa. K. (1985) Distribution of histological types of lung cancer in Aichi Prefecture. Jap. J. Chest Dis.. 44, 809-813: (Jrapanese) 16) Koo. L..E'.. Ho. J.H. & Savr, D. (1984) Iis passive smoking an added risk factor for lung cancer in Chinese women. J. ezp: clin. Cancer f2u., 3,,277-283. 17) Lee. P..;.. Chamberlain, J. & Alderson, M.R. (1986) Relationship of passive smoking to ri<k ofllung cancer and other.smoking-aasociated diseases. Bru. J. Canarr. 54. 97- 103. 18) L}•nge. E.. Kurppa.,K., Kristoferson. L., Malker, H. & Sauli. H. (1986) Silica dust and lung eanoer : Results from the Nordic occupational mortality and'caneer incidence register>: J. nat. Cancer Inst., 77. 883-889: 19) Matsukura. S.. Tominato, T., Kitano„ N., Seino. Y., Hamada, H.. Uchihashi. M.. \al-ajima. H. & Hirata, Y. (1!984) Egecta of environmental tobacco smoke on urindrn•, cotinine excretion in nonsmokers. Evidence for passive smoking. Ye+c E,tgf: J. Mrd:, 311. 828-832., 20)', `+tkamura. M., Hanai, A., Fujimoto• I., M'atsuda, M. & Tateishi, R(1986) Relktion- ship between smoking and the four major histolbgic types of lung eancer. Lung F`n ne.r. 26. 13 7-1 48. (Japanese) 21) Shimizu: H. (1983) A ca.e-oontrol study of lung cancer by histologic type. Lrng Canner. 23. 127 -137, (Japanese) 22) Shituii:u: H., Hisamichi„S., Motomiva, M., Ouumi, K., Konno, K.. Hashimot,o: K. dt Nal•ada. T. (19861 Risk of lung cancer by histologic type among smokers in lliyagi Prefecture. Jap. J. edin. Oaacl., 16, 117-121. 23) Smith. A.H. (1982) Relationship between vitamin A and lung cancer. Vat. Cannr Inst. Monoqr., 62, 165-166. 24) Tominaga. S. (1982) 8Smoking in Japan. In: Tbe C'ICC Sneaiing Cowttnl tlorE- shop.,edited: by S. Tominaga & K. Aoki, Univeraity of Nagoya Ptess.,Nagoya. pp. 2i- 35: 25) Trichupoulos. D. Kalandidi, A.. Spartos. L & llacMahon: B. (1981)' Lung cancer and'paa<ire smoking. lat. J. Cancer, 27: 1-4. , 26) 11'aldl N ,f.. Boreham, J.. Bailey. A.. Ritchie, C.. Haddoa-:,J.E. & Knight, G. (1984) Urinrn- cotinine as marker of breathing other people•s tobacco smoke. Lancet, 1.

Rt<k F.tcwr< for h'-•ni. b LunR C'auc,•r 39'+ 330~231. :"s l 13'uI AA.. HeudPrwn. B E.. Pike. KC & 1'u. M ('. (;19Ra1 Srnukine -:ad othrr risk factors Lor l6ug c:utcer in w-otuvn- J. rnt. Fn."r 1Ma1.. 74'. 747-7:rl