Tobacco Documents Online

Cancer Risk in Adulthood from Early Life Exposure to Parents' Smoking DALE P. SANDLER, MPH, PHD, RICHARD B. EvERSON, MD, ALLEN I. WILCOX,,MD, PH,D, AND JAMES P. BROWDER, MD Abstract: We obtained data on smoking by parents from 438 cancer cases and 470 controls to investigate whether cancer risk in ad'ult life is related to transplacental or childhood exposure to eitareue smoke. Gncen cases were between ages 15 and 59 at time d diagnosis. All sites but basal ccll cancer of the skin were included. Cancer risk was increased'50 ~per cent among offspring of men who amoked: Increased risk associated with father's smoking was not explained by demographic factors„social class, or individual srnok- ing habits, and was not limited to known smoking related sites. Relative risk (RR) estimates associated with father's smoking Introduction Cancer risk in adult life may be affected by wansplacen- talland'childhood exposure to cigarette smoke.t Data from studies in animals have demonstrated that many carcinogens are active when: administercd' transplacentally or during early life. In some instances, effects may be produced at lower doses than, are required for adulls.2~ The tumors resulting from these transplacental and early postnatal expo- sures may not be apparent until adulthood.t'-' Studies in humans demonstrate that the fetus of smok- ing parents is exposed:to components of cigarette smoke and is capable of bioactivating these chemicals.t01-1D For exam- ple, eotinine has been measured in the amniotic fluid of smokers and passive smokers's and thiocy'anate has been measured in fetal cord blood."-" Studies have also demon, strated increased activity of enzymes that metabolize ben- zo(a)pyrenc in placentas of women who smoke,t'-t9 and possibly even in placentas of women passively exposed to cigarette smoke.1° Similar elevations may occur in the tissues of the fetus or exposed child. Finally; increased urinary excretion of mutagcns has been found in passive smokers." Several epidemiologic studies have demonstrated in- erused risk for childhood tumors in relation to either paternal or maternal smoking?=-1• but not all studies demon- strate an increased risk.v-» Even if no increased' risk of childhood cancer were found, however, it would not rule out the possibility of increased cancer risk durini'adu)t life from fetal or childhood exposure. One recent study found elevat- eddung cancer risk for indiwiduals whose mothers smoked.r Cigarette smoke contains many known carrinogens.=' Sidestream smoke, which is the smoke released from the cigarette between active puffs. may differ qualitatively from the mainstream smoke which is inhaled by the active smok- er.- Some compounds occur in markedly, higher concentra- tions in sidestream smoke, and although this smoke is diluted by, the ambient air into which it is released„ the Address:reprtnt requests.to.DakP. Saodkr. MPH. PhD..Ep.demioloe>. Branch. Bwrnetryand Risk Assessment Program. Nationaltnstnute of Ensironmenoal HealthScicnces. Mail DropA311_:..P:O. Bo>< 1_33..Reseuch. Tnan{k Park. NC 27709- tks. Evcnon and wikox arealso.vnh NIEHS. Dr... Btouder is +iththe Departmern.of Surrerl, Sehool.of Med.cine. Univeruty d North Carolina. Chapel Hill. This paper. aubnuned to the JournalMay 1. 19b4. was revised ard acceptedfatpublucation October 16..1984. trnded to be greatest for smokers, males, and non-Whites. There was only a slight increase in overall cancer risk associated with maternal smoking. Mother's and father's smokinj were both a i-r ated with risk forr bematopoietic eancers. and a dose-res onse' relationship was seen. The RR for bematopoietic nncen incraased 5rom 1.7 when one parent smoked to 4.6 when both parents smoked. Although they should be considered tentative, study 5ndin`s sug- . gest a long-term Earard from tsansplacenul or childhood passive exposure to ciprette amoke. Ui* J Public Htafrh 1985; 75:487- 492. ). passive smoker may inhale smoke which is qualitativelyy richer in certain compounds than mainstream smoke (Hoff- man in 28). For example, the conccntration of dimethylni- trosamine in sidestream smoke is 52 times that in main- stream smoke. Such qualitative differences make it difficult to predict the biologic effect of exposure to sidestream smoke. In this study we investigate whether cancer risk in adult life is related to transp)acental or childhood exposure to cigarette smoke. Methods Our study methods have been described in greater detail elsewhere.~' Cancer cases were selected from the hospital based tumor registry at the Nonh Carolina Memorial Hospi- tal of the University of North Carolina in Chapel Hill. They included all cases diagnosed between July I„ 1979 and March 31, 1981 and assumed to be alive as of March 31, 1981. Cases were between ages 1'S and 59 at time of diagnosis and included all cancer sites except basaliceU cancer of the skin. Cases were restricted to age 59 and younger, since fewer than 5 per cent of women of child bearing age in 1920 were smokers.}0j Cases were mailed a questionnaire for self completion,. followed by a second mailing and'a telephone call if rleeded.. Of 740 eligible cancer cases identified from the tumor t'egistry, 107 (14 per cent) died before we could contact them. An additional 115 (16 per cent) either refused (n - 71) to participate or could not be contacted. Cases who died or did not respond were slightly older and were more often male or non-White; cases with respiratory cancer were more likely to have been excluded, presumably due to higher case fatality. In alll completed questionnaires were obtained for 519 (70 per cent) of the eligible cases. In addition to questions on exposure to cigarette smoke, cases were asked to identify friends or acquaintances who did not have cancer and were the same race, sex, and age (_ 5 years) to serve as comparison subjects. Approximately 60 per cent of the controls were identified in this manner. For cases for whom friend controls were not successfully ob- tained, population controls were identified by systematic telephone sampling. Data were analyzed separately by con- trol selection group and the adjusted results were nearty identical to those obtained when the control groups were combined. A1PH May 1985: vol. 75, No. 5 487

SANDlEF1 ET AL Individuals were specifically requested to supply infor- mation on natural parents. Only individuals who lived with. both natural parents for all or most of the first 10 years of life are included in this report. As a result. 128 individuals were excluded (801cases and 48 controls). Transplaccntal and childhood' exposure to cigarette smoke was assessed from questionnaire reports of smoking histories of parents. Subjects were asked whether parents ever smoked, smoked before the subject's biith, smoked in, the house for most of the years before the subject was 10 years old, and whether mothers smoked while pregnant with the study subject. Subjects were also asked the usual quanti- ty of cigarettes smoked by the parents and'the frequency of smoking in the house. For ttiis report, unless otherwise specified, exposure is classified by parental smoking in the household before the subject attained 10 years of age. For this rcport, "smoking related" tumors were defined as cancer of the oral cavity and pharynx, esophagus, pancre- as, respiratory and intrathoracic organs, urinary ttact and eervix.2f Because evidence linking cervicat' cancer with eigarette smoking is not well documented, we also analyzed these data with cervical wncer, excluded from this designa- tion. The number of "smoking rclated" tumors was substan- tially reduce& by this exclusion, but the general findingss were not altered. For individual smoking status, smokers are defined to include anyone who ever smoked at least one cigarette a day for as long as six months. Nonsmokers are individuals who have never smoked. Estimates of the relative risk (RR);in stratified analyses were obtained using the Mantel1Haenszel technique'=for the summary odds ratio: The method of Gart" was use& too obtain 95 per cent confidence limits for the combine& estimates of RR. Estimates of the relative risk adjusted' simultaneously for multiple confounding variables were ob- tained using a multiple logistic model. Level of education was reported as number of years of school completed and occupation was given as usual occupa- tion. For stratified analyses, age and level of education were treated as categorical variables with4our levels of age (<30,. 30-39„40-d4; 50+) and three levels oGeducation (<12 years, 12 years, >12 years): Age was treated as a continuous variable in the multiple logistic analysis. Controls were matched one-to-one to cases to allow the selection of population controls without having an enumcrat~ ed sampling frame. The analyses presented here are un- matched to maximize the study size following losses due to missing data on exposure. In most comparisons, the factors used in control selection are taken into account by adjust- ment procedures. Analyses using matched pairs gave similar results. Results Cases and controls are distributed similarly by age. nce, and sex (Table 1): Cases and controls differ only in their distribution by years of schooling with fewer cases having completed high schooE However, cases and controls are similar in broad occupational categories. Cases and controls are similar with regard to their own smoking status with 45 per cent, of cases and 47 per cent of controls never having smoked; the similarity is largely due to the use of friends as controls. When, only cases with population con, trols are included. 57 per cent of cases and 47, per cent of population controls were smokers. Marrnal Smotleq There was onlyy a small': difference between cases and' controls in reported exposure to maternal smoking [estimat• ed relative risk (RRY = 1.1, 95 per cent confidence limits = 0.7, 1.6). The RR for cancer among, individuals whose mothers smoked was close to one for all measures of maternal smoking, and this lack of association persisted after adjustment for potential confounding factors including age. race, sex, education, individual smoking, and method of control selection: Site specific relative risk estimates were calculated for 13 different sites, even though for many sites the number of cases is too small for detailed analysis. For most sites, the RR in relatiomto maternallsmoking was close to one (Table 2). How•ever,lhe RR for leukemia and dymphoma was 2.7 (95 per cent confidence limits = 1.3, 5.8). The RR for hemato- poietic cancers associated with maternal smoking is greater for individuals whose fathers also smoked; (2.6 vs 1.5 for nonsmoking fathers), but the RR remained, elcvated (RR' _ 2.4, 95 per cent confidence limits = 1.0, 5.5);after adjusting for father s smoking. Adjustment for age, race, sex, educa- tion, and individual smoking did not change this finding. The numbers of specific hematopoictic cancers are small pre- cluding d'etailed', analysis. However, the crude RR fon Hodg- kins disease (RR = 4.4, 95,per cenuconfidenee limits = 1.1,. 4.6), non-Hodgkins lymphomas (RR' = 1.7. 95 per cent confidence limits = 0:5, 5.2) and acute leukemias (RR,= 8.8, 95 per cenTconfidenee limits = 2.0. 40.0)iwere greater than one. TABt.E 1-{oenp.ri.one of Ca.." and Cantrol. ' ca.as Cowva F.ccw tr (~.) N tx) Tucsl~ 436 (100) 470 (190) Mc. (y.ars) . <30 83 (19) 09 (19) 30-39 72 (1ra) D5 (20) 40-49 117(Q'7)1 110 (i231 So-R9 166 (36)1 176 (37) w.n Ape 43 43 R.c. WM. 32s (74) 340 (72) Nors.NRVte. 113 (26) 1ft0(28): &u 41aw 147 (34) 158, (34) F.maw 291 (66) 312 (66) Eouuuont <t2y.an 1a2 (42)16A . (35) 12 years 122 (26) 171, (36) >12years 133 (30): 135 (29), Om+Wflw+t Bk+e Coua, 158 (39) 1'54 (34)', wMs CoUw 154 (36). 183 (40). 'Y.lovsswrld* 97 (24) 118 (26) $nckmq Stnus Norssmoka 197 (45) 229' (47) Brtpkec 241' (SS) 247 (,S3) 1Aoo»r s smduept Plo 353 (a4) 9E9 (85) Yes 65 (76)1 66 (1s). Faew s SRCkmpr No166(u). 234 (53~. rYS 212 (s6): 2n{. (47). tra.R,o.n /.aaa c.u,.. d nr.y ....a /Jr>M Wtay 1985, vo 75. No 5 488 i

TABLE 2-Gncw RItY hcen Ilotl+.r'. SmoWnq, All Seb. OoinDw+.tl .nd S.uet.e Sns Ca.es Cn,e. 95% No (X .XVossd/t RR Corr1 r+rts A! SMS 416' (16)~ 1.1 (0.7;,1.6) -Srrn" Raatsd- 131 (13): 0.9 (0.5. 1'.6) Np -Smpump R.Naied 267 (17) 12 (06 1.6) LLlp: Orat Cavrty MO PNarynx 17 (12) 0,11 (02. 3.5)i Dgew" System 31 (10) 0.6 (02.21). R.spratory Systsm 22 (m) 0 9 (0 3. 3.2) Cunp 15 (13) 0.9 (0.2. 41) t'Ja+t. Skin and,Corwrcwe Trww 36 (6) ~ 0.5 (01:,1.6) Braast2 53 (15)', 0.9 (0.4; 2:1) f.mah li.wp.I Tractt 133 (17) 1.1' (0 7. 2 2) CMVr:S 60 (15) 0.9 rob. 1.9i Pronses§ 10 (0) 0.0 (0o. 3 7)4 Tsstisg 5 (20) 1.6 t0.2. 16 6), thwry Trfct 6 (0) 0.0 (0.0. 5.1)! Ey.. Bnm .nd oensr Nwvats System 37 (11) 07 (0.2.21) l3rain 31 , (13) 0.9 (03.27) tadOatrr Ci1.raQi 21 (19) 1.4 (0.5. 4.2) 1 t'/MnsC7polte TisM 41 (32) 2.7 (1.3. 5.6)! d!>M 6 (17) 12 (01, 10.3) 1rn.q r.M.. 1w moow.. lfoo fDnWnYa+ 56 (15*) or 455 aLreclb rws &mord C nipleyr s VMtrtp j$q pC]fit RSnWrOpn 10% a: 301,MmiM iribDl{ ww« 0tWa0O fS., pUC wrcarvon 12% d tb4nn.u oo-wow wn..pord+Ea.a~m+is,nma rya. huraa! Smdi}ng 7berc was an overall relative risk estimate of 1.5 (95 per cent confidence limits = 1.1, 2.0) for cancer among individ- uals whose fathers smoked in the household (Table 3r Adjusting for potential differences in age, sex, race, individ- ual:smoking status, smoking by spouse„ed'ucation„ maternal ' TABLE 3--Canor Risk lrom FsVi.r'i SmoL/np, All Snu comDln4d and tt.MCt.d Stt.s. Ca"s 81M Wo: (`iL ex,po.W)1 C+ud. RR 95% Con1. lwnns A1 Su" 376` (56) 1.5 (1,T, 2.0) 'Smd" RN41sd" 120 (58) 1.6 (1.0. 2.5). Nlo1'Smokmp R.)ateQ' 258 1") 1.4 (1.0. 1:9) LIp. Ora Cavay .na Prrryns 17 (53) 1.3 (0 a. 3.6) (aqesevs System 30 (60) 1.7 (0.8. 3.9) R..pwalury Sywm 22 (50) 1.1, (0.5. 2.9) t.w+q 1 13 (62) 1.8 (0.5, 6.6) Bar. Skin ard Cor+rrcav. Tsau. 34 (32) 0.5 (02. 12) ~ BriaStj 51. (51) 1.1. (0.6. 2.1) , F.nyM G.rrtal iTram 113 (60) 1.6 (1.0, 2.6) Cr"I - 70 (61) 1.7 (1A.3.0) Prpqate§ 9 (44) 1.0 (02:4.7) Tsna§ 5 (80) 1 52 (0.5. 125.9) , tlr+.ry Tract 5 (40), 0.8 (0.1.5.7) Ey..Br.M, an0Dew N.rvaa Syst.m 30 (63) 2.0 (0.9, 4.6) &ain 24 (67) 2.3 (0:9.60Y Endoam. Giandt 20 (55) 1.4 (0.5. 3.8); F4.mswpoas Tssw 37 (69) 2.4 (1.1, 52). qfi.. 5 (eo) 44 (0.5. 106 7) 56...q r.rm br YewaWnwp ffc. m+s+.~,a7L a a3e cc.noe ..+. smc..d rof.ew a..+o..q is...pcAC mnw+.on 4a+% oa 2>!6 r..nw CVerou .«..3Ma.w ~L. Q.o+c mr•vwvon, 43s ar rSa ~n,w oan.or ....umcra AJPH RAay 1865: Vol. 75, N1o. 5 CANCER:R(SK FROM PAREhLTS"$Mp1(IIJG smoking, or method, of control selection did not alter this finding (RR = 1.5), Estimates of the adjusted RR were obtained separately for the group with friend controls (RR = 1.6) and the goup, with population controls (RR = 1.4). The combined adjusted RR in a matched pairs analysis with a much smaller data set was also similar. The RR for tancer, associated1with father's smoking was greater for males than females (1'.7, vs 1.4); for non.Whites than for Whites (1.7 vs 1.4), and for smokers than for nonsmokers (1.7 vs 1.2). Crude estimates of relative risk for cancers at specific sites in relation to father's smoking are shown in Table 3. The RR for "smoking related" (R,R = 1.6) and for "not smoking related" sites (RR = 1.4) are similar. Specific sites with elevated RR included cervix, brain, and hematopoictic tissuc. The RR of 1.7 for eervical'l cancer among individuals ;. whose fathers smoked is unaffected by adjustmenrfor age:>' ~ rs~ e, sex, maternal'~smoking, individual smoking, or spouse sitloking. The two-fold increase in risk for brain tumors in 4( ! relation to paternaTsmoking is similarly unaffected by adjust- ment for potential confounding variables:'Although the number of lung cancer cases with data on father's smoking is smal! (n =. 13); the crude RR for lung cancer associated with father's smoking is 1.9. and isZ.5 after adjusting for age and individual smoking. The RR remains elevated when smoking by'y spouse and mother are also taken into consideration. Leukemia and lymphoma risk is also not substantially changed by adjustment for age, sex, race, spouse smoking, and individual smoking. The adjusted RR is 2.5. However, the risk is greater, for individuals whose mothers also smoked (RR - 3.1 vs 1.8 for individuals whose mothers did not smoke) and the RR is 1.9 (95 per cent confidence limits = 0.9. 4.4) after adjusting for maternal smoking. For specific hctnatopoietic cancers, the erude RR was elevated for Hodgkins disease (RR = 5,7, 95 per cent confidence limits = 1.2. 38.4), non-Hodgkins lymphomas (RR = 1.6. 95 per cent confidence limits = 0:6, 4.3), and for acute leukemias (RR'= 4.6, 95 per cent confidence limits - 0.6. 34.2). ladividual SmOking Status Overall and site specific relative risk estimates arc shown separately for individuals who smoked and those who never smoked'in Table 4. Relative risk estimates in relation to mother's smoking are similar for smokers and nonsmok- ers and are close to one for all sites but hematopoictic tissue. Increased cancer risk related to father's smoking is not 6mited to smokers or nonsmokers, although the RR for all sites combined is greater among smokers. Dcre-resaoese The elevated risks for all cancers combined and for most specific sites were related primarily to father's smoking. However„for, )eukcmia and lymphoma there is an increase in risk when both parents smoked. The RR is 1.7 when one parent smoked and 4.6 if both parents smoked (Mantel? Haenszel'chi for trend = 3.25, p < 0.001). For both mother's and father's stnoking, overall cancer risk increased only slightly with reported frequency of smoking in the house. Risk also tended to increase with reported number of ciga- rettes smoked, but a large proportion of missing values make these data unreliable. Discussion We have found overall cancer risk to be increased among the offspring of men who smoked. There was only a 489 I

SANDLER ET AL TMELE 4-CancarRlak.t.om ParsrttsIlSnwklny amOny Fbm+nnkM arW Smohsxi.Iat SJt.rCombin.d anOSM.ct.d SMatt tM.Urtnal mokmyt SN tdo (% az9o..0)t RR No . M Srtes 191 (12) 12 227 -Snnr.nq RMat.d` 47 (0), 0.8 84 t4ot...SmdunqAetated144. (1'3)i 13 143 LO. Ora/ Cavrty.,anC Pt+r"y*v 0 - - 17. (?qestrvr Sysiem 13 (a) 0 7 18 Rsprnory Sywqem 4 (25) 2.9 le LWr+q 1 (0) 0 0 14 Bene. Skm, and Com.ctrve Tswe Br.ast 19 29 (11) (10) 1.0 0 9 17 24 Fvrna+. Gental Treq, 72 (11) ~ 1.0 61 (:ervis t 40 (8) 0 7 40 Eye. ljram and Other Weryo.u System 17 (6) 0 5 20 9nm, 11 (9) 0-9 20 EntlOavv GLants 11 (10) 1.9 10, hMrn.fot.,d.oc Trw» 19 (21) 23 22 Patemat mdunq Sr++okns dorw+o.es Smders (% atDC+asC)S RR Mo.. (•r..XWO.W)§~ RR No (l. aapoe.d)§: RR (19) 1.0 173 (49) 12 205 (62) 17 (151 08 41 (56) 1.7 79 (59) 1.5 (20) 1.1 132 (46) 1.1 126 (6{) 11 (12) 06 0 - - 17 (53)' 1 1 (11) 06 12 (50)! 1.3~ 1B (67): 17 (11) 0.6 4 (50)' 1:3 78 (50) 1.01 (1'4)~. 07 ~ 1~, (100)~. ~ 12 (58) 114 (6)', 03I 20 (30) 0.6 14 (36) 06 (21)' 09 28 (43) 0.9 23 (6t) 14' (25) 12 59 (51) 1.3 54 (70) 2 2 (231 1.3 34 (56) 1.7 36 (67), 2 0 (15) 0 8 15 (53): 1.5 li (73): 2 8 (15) 08 9 (561, 1.7 15 p31, 2 8 (20) 1.11'1 (55C 1~6 a (56)1.3(41)~ 3.1, 17' (65) 2.4 20 (70) 24 }S"s Rn IS o mcam nw+ tFn coneveon% i1e.oa 22a ndwoa,nq mr*n>r. eoo 10% a 235 r+n.mp ~coMtIas ..mowae narw...,n,wnq tFv cwcawon. 4T, od211 ron.mMUnq., oorboM aM 50% d227 tmo.,R9 cdNTvn ..owaa'b brnw. a.mo.mq ~ ,.S.a an.r~c cono+n.m 45% d roronolrp mr.rflr .ma..a n rrm...s rro.!^9 +^ar i% a,yc..a m ma++..w a s,rc..,p 52s d sno.n,p cwaar .mo.ad ID lam.. 6 sn.7vp. 6no 2?% .yw.ea u.mocr svmcwnp small increase in risk associated with, maternal: smoking. Increased risk associated with father's smoking did not appear to be explained by differences in such factors as age, race, sex, social class (as measured by education and occupation), or smoking habits of the case or control. The effect was not limited to known smoking related sites. Estimated relative risks associated' with father's smoking tended to be greater for smokers, males. and non-Whites. We have previously reported an increased cancer risk for individuals married to smokers,r but the apparent effect of paterna) smoking is not altered by adjustmentfor, smoking by spouse. Several findings from different' sources support the plausibility of increased cancer risk from early life exposure to cigarette smoke. in addition to the experimental studies'-' and biochemical studies in humans,ra=' limited support for the results of the present study can be found~ in other epidemiologic studies.u-j4•z'' Only' one of these studies.. however, has rcponed on cancer risk during adulthood from exposure to parent's cigarette smoke.=' Stewart, rr a!; in a large case-control study, found a very small (RR = 1.1) increased risk for cancer in children up to age 10 whose mothers smoked.`' An increased cancer risk related toSather's smoking was not seen (RR = 1.0). but 10 years may have beemtoo soon to detecran effect. Neutel and Buck== found an increased risk (RR = 1.3);in a prospec- tive study of cancer risk through age 10 among children whose motherssmoked during prcgnancy: buvdid'mot rcport on father's smoking. Questions on parental smoking during pregnancy have been includedin a number of case-control studies of particu- lar childhood tumors. Our finding of a two-fold increase in risk for brain eancer among individuals whose fathers smoked is consistent with the data of Preston•Martin, traCv In their study, which focused on exposure to nitrosamines,. an increased risk of brain cancer (RR'= 1.5)iamong children whose fathers smoked; during the mother's pregnancy was seen. Sidestream cigarette smoke which is passively inhaled is one source of exposure to ni'trosamines and other M'- nitroso compounds." Gold, er a!, did not report on father's smoking but, found a five-fold increase in risk for brain tumors among children whose mothers continued to smoke in pregnancy?'' Findings from a study, by Grufferman. rt a!}' are consistent with our finding of' a predominantly paternal effect_ In that study, an elevated relative risk for rhabdomyo- sarcoma was associated only with father's smoking, Man- ning and Carroll reported no increased risk for childhood leukemiarelated to mother's smoking.=3 ' Father's smoking was notreporaed. Although the number of cases was small and dose-response data were inconsistent, Neutel and Buck did find that the offspring of smoking women had nearly twice the leukemia risk of offspring of women who did not smoke.u Despite the small number of lung cancer cases included, we chose to look at, lung cancer risk in relation to paternal smoking because of continued interest, in passive smoking and cancer risk at this site.r',)'31 The RR for lung cancer among individuals whose fathers smoked~ was 2.5 after adjusting for age and individual smoking. The adjusted RR associated with mother's smoking was 1.8, but this was based on only two smoking mothers among 15 cases. Correa, tr a!, reported an RR of 1.7 for lung cancer associated with mother's smoking, but, no increased risk related to father's smoking.n Our finding of a possible cervicali cancer, risk related to father's smoking has not been reported elsewhere.,There is, however, growing support for a role of passive smoking (as measured by spouse smoking) in cervical cancer nsk.H 19 Data on parental smoking were obtained retrospectivelp' from offspring who, may not be in a position to, provide accurate histories. Parents or, siblings of study subjects were 490 AJPW May 1985;,Vot. 75, /,1o 5 ~

CANCER RISK FROM PARENTS' SMOKING also interviewcd regarding the smoking histories of the parents to validate the data obtained from subjects. We inten'iewed 649 relatives of subjects included in this report. Of these, 55 per cent were mothers and 40 per cent were siblings. For more than 3_50 suhject-mothcr pairs. agreement -on qualitative smoking questions ranged from 93 to 96;per, cent and was substantiaillt• better than chance. There was also good4greemcnt between subjects and their siblings., Our findings are not due to arn obvious recall bias. The hypothesis that parental smoking may cause cancer is not generally well known and study subjects and interviewers were told only that we were interested in smoking patterns in families. We obtained similar responses from mothers and subjects, regardless of case status, suggesting that differcn- tial recall probably did not occur. tU is difficult to distinguishl transplacental and passive childhood exposures in an epidemiologic study: women whoo smoke during pregnancy generally continue smoking after the baby is born.iO Father's smoking may produce transpla- eental as well as passive childhood exposure. An effect of father's smoking on genetic materiali in sperm~ is also a possibility.•r-`s Only 16 per cent of the smoking mothers in our study began smoking after pregnancy, and no mothers smoked only during pregnancy. This made it difficult to compare cancer risks for individuals exposed in utero with risk in individuals exposed only passiveh in childhood. Further- more. 94 per cent of the smoking fathers smoked both before and after the subject's birth. Nevertheless, if an increased risk were seen for moth. er's but not father's smoking, a transplacental effect might be a reasonable explanation. In this study and others.".=' increased~ risks were generally, related to father's smoking only. Little increased risk was associated with mother's smoking, suggesting a passive rather than a transplacentali mechanism. Our failure to find a similar effect for mother's smoking might be due to the fact that they smoked fewer cigarettes than fathers or smoked different types of ciga- rettes., Although children may spend' more time with their mothers than with their fathers, it is also eonceivable that mothers do not smoke when actively engaged in child care activities. The increasing frequency of women smoking after the 1920s should provide future studies with increasing power to detect any late effects of maternal smoking on offspring. The first sizable cohon of individuals exposed to maternal smok- ing is only, beginning to reach the age at which cancer most commonly occurs. REFERENCES 1. Evenon RB Induvdual, transplacentalty expoxd to maternaliamoking nay be at.increased cancernsk in adultlde. Lancct 1980:2:123-127;.. 2. Dnxkrey. H: Preussmann R. Ivankovic S: N-nnroso eompoundss in swFsrwxropicand tranplacenul eareirwgenesis. Ano NY AeadSci 1969: tAi.A7L~96'. 3. Rice 1M. Pennaul period ard prepsurcy: intenals of hiEh nsk for eSemiealicarcinoRcns. Ensiron Health Perspeci 1979: 29.23-:7: 4..Vksseltnovntch SD. Ran KVN. Mitiailovrch,N:.Neoplauic respoose to equse tissues during pennatal age perwds and its sgnrficance m chemical eareinofenesiv. Jn: Bailar JC. M'etsburter EK. Aaronson SA. rt a/,,ledsh Pennatal Carcino(ienesrs. Nat Cancer Inst MonoBr 1979; 31:239-250. DHEw. Puti. No. tNIHi 79-1633:.Washm6ton. DC: Govt Printirg OQice, 1979. 3. Dre. RT. BoormanGA. Hauman IK. McComseOEE. Busty. WM. Moore lA:?Ae effect of W and exposure duration on cancertnduction byy a knov.n carcinogen in rus. tnrce, aod:Aamsten..To: Appl Pbarwcol 1983. 68:120'-1 t30 6. M'echskr W. Rice.1M. VesseiinovitcbSD..TranspLxenta! and eeewsatal oductiono( neurof<en,ctumon m mtce. eomtpanson,witDreLted specxs and with buman pediatric neoplksms. 1n: Bailu1C. Wnsburyer EK.. Aanonson SA, el a1.' /edstL Pcnnatal Carcinolgenesss. Nat Cancerlost, Monogr 1979. 51:219-226. 7:. Vksselino.itch SD: Comparative uatdreson pennatal cisrcuio8enesu. ln. Tomatrs L. Mohr U teds i i Transptaetntal Carcuw*enevs..1 ARC Sct:.Pub No. 4 . Lyon: 1ARC. 1973. 14-2:: t. Napalkov NP: Some general considentiarts on the proWtm ofrranspl6- eenul carcinogenesis. ln:. Tomatis L... Mohr U'U ledsl:. Transplacental CarcinoRenesis. 1ARCSci Pub. Nb, 4. Lyoo IARC.,1973. 1-13, 9. Rrce1MAnovervrvrewof.tnnsp7accntalichcmrcalcutrcuaaeneus.Teruol-op, 1973 : 8:113-126. 10, L.ucrer GN', Lui. EMK. Lamaniniere.CA:.Meubolrc acnvataNdeactrva-tqn: reactions dunng pennatal devdopmenti Flwvoa H.ultA Pcnpect 1979i 29.7_l6. IL,Jdnes AH. Fantel AG. Kocan RA. Juchau MR: Bautivuwn of~procar, einopens tomutaQensin.liumanfetal and placental tissues. Life Set 1977, 21:1831-1836. 12. Rifkind AB: TsenQ L. Hirsch MB. Lauersen~NH: Aryl hydtncarbon bydroaylase utivuy and microsomal cytochrome contcnt of human feta1 tissues. Cancer Res1978:.38:1572-1577. 13. Smith li..Austen 1. Rollea Gi.Teniary avsokinjby the fetus /knerC. Lancet 19R2. ,1:1252: 14. Andrewsl: Thtocy'anate and smoklnt In prepnancy, Br 1 Obstet Gynaecol 1973~ 80`816-e.14. IS.. Pentgrcw• AT, Loprs RW. Willocks J::Sinotin` in presnancy-tQectson bsrthweiglit and on cyanrde and thtoeyanate kvels in mother nd baby..Br 3 Ob.stea Gy'naecol 1977; B4i31-34. 16: Bottoms SF. Kuhnen BR: Kuhnen PM. Reese AL: Matenul passive amokinlsand fetal serum thiacyanate kvels. Am I f)bstet Gynecol 1982. 1M.787-799 h 17. w'elch RM. Harrsson YE. Conney AH. Poppers P1L FinsterM: C'tgarene amoktopstimulatory effect on metaboltsm: of: 3. 4-benzpyrene by en- zymess in human placenta. Science 1968. 160..541-542. 18.. Nebert DW. Wtnker l. Gelboan, . HV:. Aryl hydrvcarbon hydiosylase acnvny in human placenta from cigarette smoking and nonsrnoktnQ women..CanceuRes. 1969. 29.:1763-1769. 19. VauQht, JB. GunooHL. Parker NB, tieBocufR.. t?octor G:. Effects of smoking on benzolatpyrxnc metabolismby" human placental mrcrosomes. Cancer Res 1979. 39.3177-3183. 20. Manchester DK. Jacoby, EH Sensitivityof human ptacental monoosy- genase acttvity.to:materssalismokmP,: Clm:Phartnacol:Ther 1961, .30:687- 692. 21. Bos RP. Ttxuws JLG, Hcnderson7T: Excretion of mutagens in human unncafter.passivesmokint. C.ancerLetten 1983t 19`.85-90.. 22:Neutel Cll Buck C: Effect of smoking during prtQuncy on,the risk.orean.cer in chi ltlre n. 1 N C I:' 1971: 47:59--63. 23. Proston-Marvn S. Yu MC. Benton.B. Henderson~BE. N-nitrosocom- pounds andchildhood'brun tumon: A casesontrol study. Cancer Res1982: 42:5240--5245. 24. Gntfferman S. Wang HH. DeLon` ER. Kimm SYS. tklzetl ES. Faiktta 1M. Envtronmentall factonins the niolm of rfubdomyosucoma' in childhood. 1NC f 1982; 68;107-I 13: 25: Manning MD. Carroll BE. Some eptdema/opeallaspecrcof kukemu metfildren. JNCI 1957;19::1087-104/ . 26. Jaffe N. Marchetto DJ: Meadov.s A, Wtnston KR. Li FP. Scher CD Clinical investigations of,the ttiolotlyof childltiood'eanccn. Proc. Am Assoc Cancer Res 1978: 19:157. 27. Cortea P. PAckle LW. Fonrham E. rral: Passive smoking and lung carcer, Lancet 1983: 2:595-597. 2!. US Depanmentof Health and Human Services: The health consequences ofamok'inp-cancer: a report of the Surgeon General. DHHS Pub, No. fPHS14'_-s0179. Washrngton. DC: Govt Printing Office, 1982. 29. SandlcrDP. Everson RB. Wilcox AJ: Passivc smokrns in adulthovd and uncer risk. Am 1 Epidemiol 185: 121:37-48. 30.. Haenuei w; Shimkin MB. Miller HPTobacco smoking patternss in the United States..PubIK Health Monop No. 45.1956.56. 31: HamsJE: Cigarette ssnokSnB among successive btrth whonsofmen.andwomen in the.UhtedSutes during 190C-0.. JNCI.1983.:7L47p-479 32. Mantel N. Haen,ucl W: Sutistrcalaspects orthe analyas of,data frvm retrospective studies. JNCI1939:.22:719-748. 33. Gan 11: Point andinterval estimaiion,of'tlxn common odd:sratio,in the eambrnation of 2 x 2 ubks .rinli fixed marpnaJs. Bametnln1970. 57:471-475. 34. Stewart A. welib J. Hcv.itt D; A surveyy of childhood malisnancKs. Br Med l 19W .1:149S-I508.. 33. GoidE..GordisL. Tonascia l:.SzkJo M: Rnk factors for~lxa.n tumors in e.hrldren. Am J Eprdcmrol1979:,109:309-319. 36. Hsrayama T: Non-smoking wives:of:heavy smoken have a 4FAer risk of l+n8 cancen a study from lapan, Br Mcd 1 1981:'3:. 183-185~ 37. Trichopoulos D. Kalandidi:A. Sparrvs L.,er o/f:ynW cancer u+dpasrve smoking lncJ Cancer 1981- 27:1.-1. NPM May 1965. Vol. 75, No. 5 491

, sANDLER ET AL.. )8. BucV'k> JD. Harris R W C. Doll R. el al' Cauianerot 'utdy d the Aus6.nds of wornen with dys}+lasia or carcinoms of tAe cervis uterf. Lancet 1981 :,2:1010-1015. 39. BroWn DC.,Pereira L. Garner JB: Caocer ot the cervuc and the smokiae Mub.nd Can F.m PA>sicun 1982. 2d:494-50_: 40. McMalion B. Alpen M: SaJbcr EJ: Maternal .cieM and pren.tal smotsog 6abits. Am 1~ Epdemal 1466: 82:247_261. , 41. Tovuus L. Cabral JRP, Likluchcv Al. Ponoaartov V: lncreased aecer ocidence in the pm8cny d mak nts exposed to ethyYnitrosouro beforx wuung In0 1 Cancer 1981: 2t.:473.•t78. 42. Nomura T: Ptisrsetal e:posure to X-rays and chemicals itduces hentabk tuawurs .nd.nomahes in mice. Nature 1982: 2%:375-577. U:, Evans Hl: Parental swujenesis and famitul eaacer. Natwe 1962: 29e:488-489:, 44. Evans Hl. FlncAer 1. Torrance 1. HarQtavc TB: Sperm aboormalntes and c4arette sewkrr>j Lancet;.1981; 1:(t:7-62'9: 45. CmsQerman S. Delxell ES. Maule MC. Mrclueloqoulos G: Parents' c3prette Smolcios and ehildliood cancer. Med Hypoth 1983; 12:17.20. AGKNOWLEDGAfENTS The authon would Ue to thank the staC of the Caoeer D.u Base. Nonli Carolina Memonal Hospital. for.Ilorin8 the use of the tuma rcystry,for case identificauon:, David L. Sbore., Karen L. Milne. and Sur W. Ward tor assistin8 in data mana8emem and analysis: and Dr. Roben, S. Sandler for etiunl rtview o( the manuscnpt; A pvrhon of this material was pesemed at the Annual Meeuat of the Society foe Epidlmialofic Researcli. June 1984. Houston. Tesas. Mortality Rates In Boston (and Other Large Cities), 1911 Boston's death ratr for 1911 was 17.1, which is high compared with rates of most other large American citits. The 1911 rrates for the other cities having over SOt7,000 inhabitants wQrr-Cltvtland 13.53;. Pittsburgh 14.94; Chicago 14:55; New York' 13.22; St. Louis 13J6; Philadelphia 16.31, and Baltimore 18.43. A brief analysis of these rates is desirable. Typhoid Fever-.-From typhoid Boston had the lowest rate of all the cities. The rates per 100.000 were as follows:-Boston 9:14; Chicago 10.78; Ne>,• York 10.99; Philadrlphia 1+1.11: Cl'rveland74.46; St. Louis 15.36;,Pittsburgh 23.81'; and'Baltimort 27.28, Scarlet Fever:-Thr rates were Baltimore 7.79; Pittsburgh 9.95: Boston 10.74: Philadelphia 11.33; New York 13.25; Chicago 2120; St. Louis 27.26: and Cleveland 31.12. Diphtheria:-Tht rates were Baltimore 12.05: St. Louis 16.84: Boston 18.00: Cleveland 21.94; Pittsburgh 23.60;,Nrw York 25.84,;:Philadtlph'ia 31.31; and Chicago 39.1}. Ivleasles:-Thr rates were Chicago 5.75; Cltvtland 6.63: Pittsburgh 939; Boston 10.74; New York 1313; Baltimore 13.63; St. Louis 15.84; and Philadtlphia 1930. Whooping Cough:-The rates were Chicago 2;43: St. Louis 4:57: Philadelphia 734; New York 7.75; Baltimore 830; Cleveland 14.80r Boston 15.68: and'Pittsburgh 19.54. Taking all thrst diseases togethrr. Boston had the lowest rale with 64.30 per 100,000; the others in order were Baltimore 69.13; New York 70:88; Chicago 7.29; St. Louis 80.07; Philadelphia 83.59; Pittsburgh 88.49; and Cleveland 88.93. Tuberculosis of the Lungs:-Thr rates per 100,000 were Cleveland 121.77; Pittsburgh 130:88: St. Louis 13519; Boston 154.88; Chicago 165.98; New York 17739; Philadelphia 187J1: and Baltimore 205.12: This analysis of the communicable diseases should be extremely gratifying to the people of Boston. HowevEr, it fails to reveal the reason of Boston's higher rate. But a study of the figures for 1910;, for which year more complete figures are available, will show much nwra.... Cancer, cerebral hemorrhage, organic diseases of the heart, pneumonia and violent deaths stand out as the principal causes which have comparatively high rates in Boston.... Thrrt are two reasons why Boston has such a large number of deaths of non-residents. First it has a population in its immediate suburbs greater than the population of the eiry itse/f. These people labor in Boston and when ilPcomt to Boston hospitals. The other reason is that Boston is the rtcogniZed' medical centre for dl New England and attracts those afflicted with cancer, for esamplt, a disease more prevalent in the New England States than in any of the other registration states. Moreover in 1910 Boston had a larger percentage of its deaths over 6 years of age than any of the cities under considcration. In a word, Boston is an old'city and'has many old people. ... But in a word it may be sqfily said that Boston's high rate is largely due to its geographical position in the centre of populous suburbs, and to its famr as a medical crntrt, and not, as has bttn so often intimatrd, to the unhraltl4ful'conditions in the city. -Davis WH: Boston's death rate. Am ] Public Health 1912; 2:638-640. 492 /.1PM May tp85. Vol. Z5,,NO: 5