Tobacco Documents Online

~, IMt1M1auunal cW1uenn1U,ug1cat s+ssociation t~au. Printed in Great Britai. I Association of Indoor Mr Pollution a nd Lifestyle wi'th Lu ng Ca ncer i n Osaka, Japan TOMOTAKA SOB'UE' Sobue T (Division of Epidemiology, Department of FieldiAlesearch, The Centen for Adult Diseases, Osaka 1-3-3 Nakamt ich, Higashinari+ku, Osaka 53Z;,Jepan). Association of indoor air pollution and lifestyle with lWng,cancer in Osaka, Japan. InternationalJournal oCEpidemiology 1990,19 ($upplll): S62-S66. A hospital-based case-eontrol studyamong non-smoking women was conducted to clarify risk factors in non-smok:ing females in Japan. Cases consisted of'144:non-smoking female lung cancer patients, and these were compared to 713 non-smoking female controls. The odds ratio (95% cbnfidence:interval) for use of'wood or straw as:cooking fuels when subjects were 30 years old was estimated as:1.77Y'1.08 to 2.91). For those whose household members, other than hus- bands, had smoked, the odds ratio was estimated as 1150 (1M Ito 2.32). For those whose mothers had smokedj the odds ratio was estimated as 1.28 (0:71 to 2.31). Use of heating appliances did not show:an elevated risk. Some points to be noted in the study of low-risk agents for lung cancer are discussed. In Japan, lung cancer was the second leading,cause of cancer deaths for males and females in 1987.' In males, alrthough, smoking,rates have been drrcreasing,gradu- ally since the 1970s, 61% of males smoked in 1988,, which is considerably higher than in other developed' countries. In females, however, smoking rates have beeni quite constant since the 1950s--onlv 13% of females smoked'':in 1988, which is low for a developed country: As a resudti, population attributable risks for lung cancer caused by smoking were estimated at 71% in males but only 26% in females.2 In the standard mortalirty ratio (SMR)lanalysis of the geographical distribution of lung cancer risks, ai higher. SMR' was observed in coastal urban areas than in inland rural areas for males, but' for females no such tendency was observed.'' This indicates that occupa- tional exposure and ourtdbor a~irpollutiionseemto hav:ee little influence as lung cancer risks for Japanesee women. Therefore, it is necessary to investigate riskk factors for f'ema(es which might be related to d'aily lifestyle. This study aims to clarify the risks of lung cancer caused by iindbor air pollution arnong no:nsmokiing, females by means of a, hospital-based case-control study. This: work is part ofl a joint project of'the research group for lung cancer prevention in OSaka~ The mem- bers are listed in Appendix 1. DiRisinnref Epidhmiolugt, Ucrartmrntiof Fiddlftr.rarc.h. Chr Center, for Adult Dise,:r.rs.,C)%aka I-3;3 N.uk:amich Hii:ashinari-ku. Os;nku 537 Japan MATERIAL AND METHODS. According to Osaka Cancer Registry„ 24'811 primar> lung cancer (:1977 males and 504 females) patients wer( . diagnosed in Osaka Prefecture in 1985: Of these abou one-quarter were registered from the top eight hospi tals, which have special departments for lung cancer These eight hospitals participated in a rnulti-centire. hospirtal-based case-controll study with the support, ol the Osaka Anti-Lung Cancer Association, Both cases and controls were collected from thosr newly admitted to the eight hospitals from 1 January 1'9'86 to 31 December 1988, and their ages ranged frorr 40 to 79 years:at the time of hospitalization. Of the above eight, hospitals, all wards for lunt: cancer and one or two wards for other diseases werc inv:olv:edl in this study. All newly-admittedl patients. both, males and females, in these wards were investi- gatedlby a self-administered questionnaire at the timt ofladtnissibn to the hospital. A uniform questionn:air( was used in all hospitals, designed specifically for thi~, study, which included questions about'smoking habits exposure to environmental tobacco smoke (ETS) ant. exposure to possible indoor air pollution. A total o' 1079 lung cancer patients: and 1369 patients of otlhei diseases were investigated for males,, and 295 lun, cancer patients and 1073 patients of other disease lot females. Males were not included in this analysis. Fcar females, there were 55 current srncnkers,, 64' exx smokers, 156 nonsmokers and 20 patients witll unknown smoking,startus for luinf;,cztrncer patienrts, :nlnt S62

INt)(1()R AIR POIiLI%11[)N ANl) LUN(i ('AN('ff:R S63 corre:spondinl;~ 122. 92. 789 annd! 7d1 for patients of herdise,lsc.r rrspcc.Uivtilv: An'alvsis was focuscul (tn, ,6 nonsmcnkiing femalc lunk canccr lnaticm+ as cases td 789 female pzttients o11(nther dlseases .rs conUr(tls. omaticli'ink procedures were c(nnducticdl between ixes and c(nntrols. There were 12 cases antil58'.ctntnurols <cluded beca'usc of missing infornnati(rnon expo`u!re. ,s a result, 144 cases amd 731 controls comprised the ttal for this studv. Adjusted odds ratios werecalcul'atcd by the Mrlniicl- lacnszel method° using four levels of agera'tegories att dmissionia'nd two levels of education. Logistic regres- on analysis was performed including the variables ~hich showed significant increase of risk in univariate nalysis.' tESULTS )LIl cases were microscopically confirmed, and had the oUlr ing distributiion-adenocarcinoma (7&%)„ qu'amouscell carcinoma (8%), small cell carcinomai 5%), la'rge cell carcinoma (5%), and other histologi- alltypes (4'%). Controlswere diagnosed as having the ollowing; diseases; breast cancer (46%);, stomach ~ancer (13%); other cancers (16%), benign neoplasms 8%'~)4 circulatory diseases (4%), respiratory diseases 3%), infectious disease (2%) and! digestive diseases, 2%). Table 1 shows the distribution of' age and edu- .ational llevel for cases and controls. The mean age at idmission to hospitals was 601for cases and 56.for con- orols. Higher education lovels were observed for con- [rols as compared to those for cases. Table 2 shows adjusted odds ratios fon: lung,cancer associatedl with use of wood'~ ar straw as cooking fuels accordingto the age at exposure. Significantly elevated ris' were observed for, subjects 30 years of age who had used wood or straw as cooking fuels. Use of these fuels at age 15, showed a~ sligHt~ increase or risk although it was nor: statistically significant. When the exposed were defined as those who used these fuels TABLE I Disvibiuiorr of age at'adnrissiorr artd " vrars of education for cases and corurols Charactcristics Case Control N % N % Age at adtnissioni 40-49 20 13.9 238! 32.6 5() 59 49 34.0 229 31.3 6t1--b9 41 2Ft':5' 186 25.4 7(4-79 34 23.6 78' 10.7 Years of education Icssthan 9 69 47.9 229 31.3 1'0 or over, 75 52.1 502 6R.7 eitlner at age 15'(tr agr 30. the odds ratio was esti!nratcd a. 1.28 with an (I.8ti-1.87'conifii'_lencc iintervall. In the ctilkuilatit,n uf thc odds ratio. thr u:a• of heat- iing appli,tnccs-krrose,ne. grt`, ctnal, charcoal andd wood stoves without chimneys were regarded as pKtss- iiblcstnurcesof exp(nsuncwhichcould pttlluteindtnorair with combustion producUs. Electric air conditiitners, stoves with chimnevs and electric stoves werc not regarded as sources of exposure. There were no risk elevations observed for exposure at any age (Table 3). The charcoal foot warmer was popuihrly uscdluntil the 1960s. but i's now rarelv used in Japan. Again, risk ele- vation was not observed fo'r exposure at any age (Table. 4),.. Odds ratios for lung cancer associated with ETS dlaring childhood were shown by source of exposure (Table 5). A slight increase of risk was suggestedl for those with smoking mothers, although statistical sig- nificance was not observed. As regards ETS in adu'lthood~ an elevated risk was observed for those whose household members, other than husbands, had smoked'(Tab'le 6). Smokers among other household members consisted chiefly of the h'us- bandPs father and sons. Table 7'shows the results of logistic regression analy- sis, including the three variablr=s in the model, which were, suggested to raise the risk of lung cancer in uni- variate analvsis, Use of wood or straw as cooking fuels at age 30 showed!a risk 1.7 times higher, with~statistical siignificance. The other twolvariables showed slightly increased risks, but were not statistically significant:. The results from the same a!nalvsis,,when breast cancer patients (:controls) were excliuded, showed~ si~milar results.. DISCUSSION From the results of this st!udv: the use of wood or straw as cooking fuels was suggested as aipossible risk factor for current female lungcancer cases in Japan4 despite Twet_e 2 Odd.oratios fctrlnnQ cancer associated with the use of wood or straM• as cooking flmis according to ag4 at exposure Case/Control OR'~ (95% CI"') ' Age 15 No 59/361 1.(Nl Yes 85/37(1 l .'-4! ((1:8C>-1.81) Age 3(t No 1'12766(J 1.4ND Yes 3J 71 1.89 (;I.U6-3.(Xi) Present' No t4U731 1!.(Kf Yes (1/ (D - 'Confldt:ncc imernal

TAbuE3 Odds ratios jor Iung cancer associated with the use of hearing equiptnent+ polluting room air with combustion products, according to - age at,exposure Case/Control OR (95% Ct') Age 15 '. No 40/201 l .OGI Yes 1041530 1'A1 (11.68-1,.52) Age 30 No 51/29+a• i 10) Yes 93/437 1.18 (0.81-1.72) Present No 77/404 1.00 Yes 671327 1.11 (0.77-1.60) •Con6dence interval its being an oldlpractice. These types of coo'kin!gfuels were widespread until the 1960s, even in suburban areas, but now very few people use them even in rural areas. Of those who used wood or straw at 30 years of age, 90°/'p had also used these coo'king fuels at 15'years of age. This indicates thar those exposed at age 30 must have been, exposed for a longer diura'tion. It is reported that use of coo'king'oil, especially rape- seed oil, increased the risk of lung cancer among Chinese women in Shanghai.5 In the same report, how- ever, the use of cooking fuel including coal, gas and wood' did not show an elevated risk of lung cancer. In Hong Kong,, the use of kerosene oil as cooking fuel appeared to increase the risk of lung cancer among Chinese women although the effects of these factors seemed to be l'imitied.b It is also reported fromi Sin} gapore that there was no difference of risk for lung cancer between those who used!wood or charcoal andd those who used petroleum or gas.' However, all these reports provided'i information concerning Chinese women, who practice different, methods of cooking from Japanese women. Also,, in these studies, the exposure from cooking,fuels were defined a's ever ver~- sus never or were based on only recent startus, and! the _ABLE 4 Odds ratios for lung cancer associated' wlih the use of exposures variable may not, correc:uty reuccttne suatus of pa st exposure. In fact, when ever versus never anal- ysis was used, the use of wood!orstravv as cooking fuels did not show a significant elevation of risk. Iln the present studiy', no one was found! who uses wood or straw as cooking fuels at present, so this does not constitute a factor.for primary prevention in this country. However„this showed that the environmental exposures occurring 20 years ago coul'd'aff'ect the inci- dence of lung cancer, which in turn means that some lifiestyles widospread at present can be risk factors for lung cancer in the future altllough conventionall epi- demiologica'1 studies cannot reveal these factors at present. It has beenireported that some compounds found in wood' smoke-benzo(a)pyrene and formaldehyde- are possible human carcinogens." It, has been shown that the aromatic fraction of wood smoke, which con- tains various polycyclic aromatic hydrocarbons has murtagenicactivity." Also, the polar fraction of organic extracts from emissions of wood combustion h'asbeen shown to have direct mutagenic activity.'t' It is reported that natural inhalation exposure to wood smoke increased the incidence of' liung,cancer in mice.'t' The use of'heatang equipmenr for room air,,including kerosene, gas, coal, charcoal and wood stoves without chimneys, didlnotshow an elevated risk of'lung cancer. Of these, charcoal and kerosene was most frequentilyy used at age 15'and 30; respectively. Wood was used!for heating fuel only for less than 5% of' the population, therefore the risk due to wood stoves could not be eval- uated. It is reported from Hong, IKong, that the use off kerosene stoves increased the risk in wornen.'' Ini Japan,,no increase of risk was observed for the use of kerosene stoves." ETS from the mother during childhood seemed to raise the risk but did not show statistical significance. It, has been established' that ETS for children increases the occurrence of lower respiratory illnesses, particu- TABLE 5 Odds ratios for lung eancerassociated with environmental charcoal foot warmers for sleeping, orcording to age at exposure tobacco smoke dtering,childhood'by source of exposure CaselCon t rcrl O R, (95'%q Cl') Case/Control, O'R' (95% Cl') Age 15 No y1J470 I.IKD Father smoked No 35/143 1.(KT Yes 531261 1llIP (().t/9-I148) Yes 109/588 (1;79 (l):52-11.21) Age 3(l No 1J2/61iN I,.IN) Mother smukedi No 1271668 10) a ~. Yes 31ZC,1 I5 l (R5 (0.6ti--I 6d4) Yes 17163 1.33 (11.74-2.37) ~ Present , _ . Other household 'members. ra No 143m5 I.IKI No 113687 1. tK) Yes W In 0 67 (l/ tW1~5 12) Yes 31/7441 11li 1 (11 7(>•-L.84) . . . . . lV 'Cimtiilrnce interval 'Cunlidencc interval N GLf lU

INDOOR A~IIR POLLlIlI1ICJfit AND LUMCICANCF.R, i i 6 Odds ratins for lrrnt; crurrt'r rtoan ia4•rl w'i!h enrirunntrntal tr+lrrtrror crnrrkr°in urlidthr,rnd hp,.crrurct, uf t'cfrr+surt•r caw;Cr,ntrol () R cl•1 .haod I.a»ut,c.d 64/336 I.ctt1 c. ier househtnid,membern }4f1[39i1 1.13 (11.7P+-1.(nz) Jrn 911/5511 ILIN1, res 53rq8'1I 1!.57 (III.I)7-2.31). .infiiknce intervat -ly, early in life, andl incrcases the frequency of ronic respiratory symptoms. ""K Its relartion, to lung nrer, however, has been less clear. Iitis reportied that e odds ratio for lung cancer associated with exposure a smoking,mother for nonsmoking,females was 1.7 the U.S." and 4:0 in Japan." Concerning ETS in adulthood, ETS fromi the hus- md did not show an elevated, risk in this study. In pa. . a 5(D-1i0C1%o increased risk foriung cancerassoci- ed with ETS from the husbands was reported's'" though some studies found no increase." It is esti- ated from the meta-analysis dealing with two cohort udies and tien case-control studies that the increased ;k of lung cancer by ETS from the husband would be. IQ/o."' In the present study,, ETS from household embers other than the husband showed an increased sk of lung cancer. This is consistent with a report from ipan'that ETS fromithe husband's father elevated the A o ' lung cancer 3.2'tirnes." Some methodologicall problems should be con- diered in this stvdy: First, a substantiallproportilon of )ntrols consisted of cancer patients, especialQybreast. Incer. Although use of! cancer controls has various terits and demerirts,"' irt is obviously'not appropriate to ;e cnntrols from a single disease. When breast cancer as i_...:Nuded from controls„the odds ratios for use of' ood or stranv as cooking f'ulels,,ETS from the smoking tother, and ETS from household members other than ie husband became 1.65,, 162 and 1.47„respectively; hichidid not show substantial change. Second, smoking status of tlhe study subjects was ivestigatedi by se'lf-admimistered questionnaires and o validation was conducted by otherobjective means, lchlas testing for cotinine in urine or eanbon monox- je'in expired breath. However, these methods cannot e applied to determi'nesmlokingstartus in the past only j recent smloking',status. Further studies are neededliln 'lts area. Third, the exact duration of intensity of exposure ouldlnot be investigated for use of cooking fuels and :TS from various sources of exposure. However, S65 dctailed information obtained from individuall memory may not be reliable enough to conduct dose- ~ respontic analy5is.__.. ,M1 Fourrth, no systematic review for histopatholoe;ieal' diagnosis was carried out, hut routine pathology reports were used. Hiowever, since all pathologists involved in the eilght hospitials were specialists in lung cancer andlhad worked at least five years in this area, validity of'these reports were thought to be qµirte higMi as far as the determination as to whether it was malilg.- nantor benign. The analysis in this study, was not con* ducted by dividing' lung cancer into histological types, and it is believed the effects of'this on the results wouldd be minimal. There are some epidemliolbgical points to be dis- cussed& ini the study of low-risk agents. First, subjectss were limited! to low-risk individuals for lung cancer,, which in' this study were Japanese females who had never smoked. It is generally thought that focusing on low-risk individuals can strengthen the association between the disease and exposure,2" making,it easier to find possible associations, except when positive inter- actions exist. Secondly, when we categorize the study subjects into exposed and non-exposedi, it is important to pay atten- tion to the timing between exposure and disease. According to the mechanisms of carcinogenlesis,, this appropriate time difference willl vary. For example, if the agent in question acts mainly in t~heearly st~agesofs carcinogenesis, there should be a longer latency time between exposure and disease, but if the agent acts mainly in the larter, stages,, the lag time' between expo- sure and disease will be short. In this stiudy, exposures were defined according to the patient's age, and were able to reveal the association betweencookingfiuel and lung, cancer. However, if we use ordinary classifi- cations; such as never-user versus ever-user,,or, present use, the association would not be seen. Thirdly, even if we can use the appropriate classifi- cation of exposure, considering its timing in the occur- rence of the disease, it is important that the popu!lationi has the appropriate diversity in terms of exposure classifiicatiion, In other words, there should be some proportion, of people who, willi be classified as non- TABLE 7 Odds ratios estimated' by logistic regression analysis. Adjusted jor age at hospitalization. Variable OR (95% Cl') Use oflwood or straw anage 30 1.77 (1.0$-2.91). Other household members smoked in adulthood 1.50 (1.01!-2.22) Mother smoked in childhood 1.28 (0.71-2.31) ~ 0 N ~ N+. N ~

S66 r IN7 ER1ATIfJNAL JIJUfcNAL Wr- trtvrallWLVt, r' (>UI't•LLtMlt\4' l)' exposed together with people who will be classified as exposedi.This is not always the case in the situation of cooking or heating practices, for which most peoplt;: share a comrnon, t'raditionL In Japan, there have been drastic changes in lifestyle since World War II. Sanitary conditions in most houses were not very good in the :: 1950s, but have drarnatiically improved in the 1980s, and this can be regarded as an appropriate non- exposed situation. Mixed practices in cooking andd hearting,were prevalent during this transitional periodd betweeni the 1950s and 19,80s, which provides a good opportunity to identify a low-risk agent for lung,ca!ncer associatedl withi daily lifestyle. ACKNOWLEDGEMENT We would like to thank the late Dr Shizuo Okada for his contribution to this study. We also thank hrts N. Nakayama, Ms T Imubusi and Y Ivliarai for their tech- nical! assistant:e. REFERENCES "Statisticsand IinformationDepartment, Minister'sSecretariat; Min+ istry of'Health and'Welfare. Wiral Sratistics 1987Japan- Tokyo. 1989: Sobue T. Suzuki T. Horai T. Matsuda M. Fujimoto l,. Relationship between cigarette'smoking and histologic type of lung cancer~ with special reference to sex difference. Jpn J'Clin Oncol'1988; 18: 3-13. ''The Research Committee on Geographical Distribution of Disease. National Atlas of major disease mortalities for cities, towns and villages in Japan: All causes of death, cancer, cardiovascular diseases, diabetes mellitus:, liver cirrhosis and tuberculosis:. 1969-1978. Japan Health, Promotioni Foundation. Tokyo. 1982. ' Breslow N E, Day N E.,Statistical'methods:in cancer research. The anatysis of case-control studies. IARC scientific:puMications:c vol I. 1980, Lyon France. `Gao Y T. Blot W'J,,,Zheng W; etal: Lung cancer among Chinese women, 1nt J Cancer 1987; 40: 6f}I3-9! Koo L C„Lee N, Ho J H-C. Do cooking fuels pose a risk for lung cancer? A case-control study of women i Hong Kong: Ecology of Disease 1983,: 2:255-b5. ' MacLennan R, DaCosta J. Day N E; Law'C'Hl Ng YK. Shanmug- aratnam K. Risk:factarrs for lungcanceriin Singapore Chinese, a population, with high'female incidence rates. !nt J'Cancer 1977; 20: 854-860. "Pierson W E. Koening:l Q. Bardana E'J. Potential adverse hcalth, effects of wood smoke.,Wesr! Med IN8Pd; 151: 339-42. ° Mumford J L. He X Z. Chapman R S, et,al: Lung cancer and'indrnor airipollution in Xuan Wci„China. Science 19t+7: 235: 2117'-2^_f1;. "" AlIfheim Il Ramdahl T.,Contrihution of wxxxdcombustium to indoor air pollmion as measured by mutagenicity in salmonella and polvcyclic' aromatic hydrocarhon, concentr:rtion. Ena•irnn Muuugen 1984; 6; 121-30. " Liang C K. Quan N Y;,Cao S R„He X Z„Ma F. Natural inhalation exposure to coal'smoke and wood smokrinduees lung cancer in mice andlrats. Blomed'Environ Sci 1988; 1: 42-50. '= Leung J S M. Cigarette smoking. the kerosene stove and Ilung cancer in:Hong Kong. Brl Dis Chest 1977; 71: 273-6. "Shimizu H, A case-control study ofllung cancer by histologic:type: J Jap Lung Cancer Assoc 1983: 23: 127-37: "U,.S. Department of'Health and Human Services„ Public Hbalth' Service Office: Srrroking and Healtli: The health conseqf4encess of smoking; chronic nbstructive lung disease: A reporrof'them surgeon general. Washington. DC: US Government Printing Office. 1984: DHHS(PHS) 84- 50205. "U.S: Department of Health and Human Services, Public Hl:alth! Service Office: Smoking and Health: The health conseqyerrcess of involuntan• svnoking: A report of the surgeon general. Washington. DC: US'Govermment Printing'office, 1986. '" Wu A H, Henders'on' B' E. Pike M C; Yu fvl C. Smoking and other risk factors for lung cancerimwomen: l Nall'Cancerlnst 1'985t 74: 747-511 L "Shimizu H. Tominaga S. Nishimura M! Urata A. Comparison of, clinico-epidemiological features of lung cancer patients with and without a histtnry of smoking. Jpn J Clin Onco!' 19,ti3; 14: 59540). "'Hirayama T. Non-smoking wives of heavy smokers have a higherr riskoflungcancer: a'study from Japan. BrMi•dJ'19811:282'. 183=-85'. '" Akiha S, Kato H„Bilot'W J. Passive smoking and lung canccr among Japanese women. Cancer Res 1986: 46: 480:1'-7. '' Blot W J. Fraumeni J F. Passive smoking and lung cancer: J Nail Cancer lnst:1986: 77: 993=1 Olttl. ="Smith'F:,HL Pearce N E'. Callas P W.' Cancercase-control Istudies with other cancers' as contrvls. 1nt J'Epidentiol' 1988`, 17: ^^89-3t46. -- Kolonel L N. Hirohatu T. Nomura A M Y: Adequacy of survey data collected,from substitute respondents. Arn J Epidentiol'1977`, 106; 476-84: =' Lerchen M',L.Samet J'M. A,n assessment oflthe validity ofquestionr naires responsesprovidtd by a surviving spouse. Am,J Epi.- demiolog 1986; 123: 481-9. =' RothmanKJ.,Fonle C. Astrengthcningproeramme:forweak associ- ations: lnr J Epidenriol 1988: 17: 955-959. APPENDIX I Research group for lung cancerprcvcntie?niin Osaka.. Tomotaka Sohue, MiD'', Takaichiro Suzuki. MD'. Minoru Matsuda. MD'. Osamu:Doi, MD'. Takashi Mori. MD. Kiyovuki Furuse. titD'. Masahiro Fukuoka. MD'. Tsutomu Yasumitsu. MD', Osamu Kuwah- ara,,MDa, Michio lchitanii MID`: Masahikto Kurata. MD'. Ma.avoshi Kuwahara~ MD'. Kazuva Nakahara, MD".,Shozo Endra.,MD'. Kcnji Sawamura:. MID". Shoji Hattori. MD'. ' ThrCenter for Adhlt Disc.asc„ Osaka. 537 Japan ' National Kinki Central Hospital forEhust,Disu;tses, 591 Jal?an ' Osaka Prefectural Huhikino Hospital„iJ8 JaP:m ' National Tonevama Hospital. 5ND Japan ' Osaka Rcd Crtiass Huspital. 543 Japan "TazukaKofukai Medical Rcsearchilnstitutc, 53tDJapan Kansai Dcnrv0ky " Hrospital'. 553 Japan Fir:st Departmcnt,of Surgun:. Osak Univvr.itc Mcdic;tl'Schi ul, 553' Japan "Osaka Ant Lung Citnccn Assoiiatiun, 541 Jatpani