Document 0qBz8QoRKygLmYKjayRboOMJd

FILE NAME: Kent (KNT) DATE: 1954 Mar DOC#: KNT135 DOCUMENT DESCRIPTION: Journal Article - Etiology of Lung Cancer: Present Status Etiology of Lung Cancer: Present Status MORTON T,. LKVIN, M.D., ALBANY, NKW YORK (Asmhtnnt Com mA,shear for ,1 /alien! Serrieim, AVie York Shite Drparimt 11/ of Health) !i { vail milk evidence points to a number of lx . different possible etiologic factors for lung raneer in man. These include chemicals, radia tions, dust or fumes encountered by workers in various occupations, and cigaret smoking. Inlustries and occupations with evidence of abnornully high lung cancer risks are chromate pro.iuetioi); production of illuminating gas (studied in Kngland); mining of uranium ores in Schneelerg, (lermany, and Jaehymov, Czechoslovakia; nickel refining in South Wales; asbestos manti-1 iVttiring in Kngland; and metal grinding in! ; Kngland.1 Recent evidence points also to in: unused lung cancer among welders and other f wirkers exposed to metal fumes or dust in the ; United States.5 Suggestive evidence regarding .wral other occupations has been reported, and it teems probable th a t other as yet unknown oc! mpatinns with high lung cancer risks remain to be 1 discovered. Evidence for relationship of these occupations 1 In lung cancer consists chiefly of the increased Bwrtality rate from lung cancer observed among mirkers in these occupations. Exact, risks have i Icon estimated for two of these occupations, i rliromate production (in the United States) and I illuminating gas manufacture (in Kngland.)1 For ; 1 lew there is also experimental evidence of a mrcinogenic agent in the material to which 1 workers arc exposed, but in no instance lias lung rancor been produced in animals by such ma| Serial. In the case of the chromate industry, in | Hint'll lung cancer mortality has been shown'to | lie 14 times that of the United States population I among white workers and SO times among Negro workers, application or injection into j animals of vai ious materials to which workers are reposed has thus far failed to produce malignant .`'Idijifs rnportc:) Jicrr w e r r a i d e d b y a g r a n t f r o m th e National C a n c e r I n s t i t u t e , U.S. P u b l i c i l e a l t h Service CCS KiO). and f oin th e N e w York S t a t e D iv isi on , A m e r i c a n f'snccr Society. lung tumors. .Nevertheless, there is little or no disagreement as to the existence of a lung cancer etiologio agent in the chromate industry. A relationship between lung cancer and tobacco smoking lias been suspected for many years. Sta tistical studies, of which the first apparently was that of Midler3 in 1939, have uniformly shown a, higher percentage of cigaret smokers among lung cancer cases than among control groups. In the past few years this relationship has received increasing public notice so that it is today widely debated and discussed in the news papers, in national magazines, and, no doubt, by the public itself. Many physicians, statisticians, and others engaged in cancer research have ex pressed themselves in the public press as to the significance of the reported studies on this as sociation. Considerable, difference of opinion and of emphasis has thus been elicited. Appraisal of (he nature of the evidence regarding tobacco as a possible etiologie factor in lung cancer has thus become of increasing importance, because of the widespread public interest in and concern over the subject, the large financial and tax interests at slake, and, most important, the steadily rising; mortality from lung cancer reported each year. It is, therefore, necessary that, physicians and public health authorities appraise the situation as it now exists to determine what conclusions may be drawn and what action may be required. Most studies regarding smoking and lung can cer thus far made--18 ire listed in a recent, arti cle4---have, compared the smoking histories of lung cancer patients with those obtained from other persons, usually patients with other diseases in the same hospital. Almost without exception, these studies have found more cigaret smokers among lung cancer patients than among other patients. If these observations arc valid, it would mean that, lung cancer occurs more fre quently among cigaret smokers than among nonsmokers. March 1:", 1954 76!) MORTON L. L E V I N Several objections can be raised to this type of study, and these objections must be carefully weighed to determine what, possibilities of error east, in such studies, whether these are sufficient to invalidate the conclusions reached, and what, bearing the evidence has on the question of smok ing as a cause of lung cancer. The first question which arises with respect to these studies is w hither they prove that a real association exists between smoking and lung cancer. In other words, do the studies prove that persons who smoke cigarets develop lung cancer more frequently than do those who have novel' or rarely smoked, for this is the meaning of "association" as used with respect, to this type of evidence. "Association" does not mean, as lias been implied by some writers in the press, merely that the rise in lung cancer mortality has ac companied the rise in the use of cigarets. The guilt, attributed to cigarets is not based on this latter type of association. The concomitant rise of lung cancer and cigaret consumption would be expected to follow a true association between tire two, but by itself cannot be con sidered as more than suggestive evidence of it. The evidence which is at hand is th at cigaret smoking is more frequent and heavier among per sons who develop lung cancer than among those who do not. The fact that many studies by independent in vestigators in this country arid abroad show the same result is accepted by all or most authorities as proof that some association exists, i.e., that cigaret smoker's do experience a higher incidence of lung cancer than do nonsmokers. It may lie of interest,, however, to consider why there should ho any doubt at all on this point. Two possible sources of error must, be consid ered, those due to bias in obtaining the history arid those due to bias in sampling. Whore the smoking history is taken as part of a special investigation, one must consider whether the in terviewer, by reason of his own bias or conviction, influences the patient in the answers the latter gives. In at least two studies the element of in terviewer bias would seem to have been eliminated if it existed. Doll and Hill considered that they had demonstrated this error did not exist in their study since in .TTi patients believed to have bronchial carcinoma at the time of interview, but shown later not to have the disease, "the smok ing habits of the wrongly diagnosed group were sharpl}' distinguished from the habits of the cor- --n roctly diagnosed groups and were statistic;! ' | indistinguishable from those of all the otb* patients suffering with other disease." 1 In tbi jj Roswell Park Memorial Institute studies,6,6h w i in 1938, we consider this source of error to Ion been eliminated, first, because the studies wen ; undertaken in the belief that no association rca!!) 1 existed and, second, because of the fact that tv ! smoking histories were not part of a separate i> ' terview but were obtained as part of the routin' ; history taken on admission before any diagram; had been made. A second possible source of error which to ! been pointed out is th at one could not, be, ccitv * that tile lung cancer cases studied were represent, l tivc of ail lung cancer eases or th at the conlr- patients were representative of the general popub- j; tion. However, those two conditions arc irai : necessary for a valid study of the association < ; two variables. First of all, the cases and contrd '] need not come from the general population In;' ; from the same population, i.e., the contrit j should he persons of the type most likely- to ent<" : the study as eases. Second, the cases and con- : trois need be representative only with respect li the factors studied, in this instance smokier ; history. , The theoretic, sampling questions raised aW with respect to the case studies may thus be sum- : marized in the following questions: 1. .How likely is it that the control and lumr, cancer cases in the studies reported were draw: i from two different populations which diffre; ! significantly with respect, to smoking and diffre: i always so th at the control cases came from t, | lightciVRinoking population than the lung canm 1 rases/O 2. flow likely is it that these sampling bins.* ; would be more marked for cigaret smokers thus for pipe and cigar smokers and more marked (a I heavy cigaret smokers than for light cigareb smokers? 3. How likely is it, that lung cancer cases wh smoke are more apt to go to the hospital or that ; patients with other diseases who smoke are It*.* apt to do so? j 4. How likely is it th at those sampling bia.w : should occur repeatedly in studies made in (lb feront countries and in different sections of lb i same country? Xone of these contingencies seems very likeh and their occurrence to so marked a degree rat show the differences between smokers and m ETIOLOGY OF LUNG CANCER: PR E VE NT ST AT UE VHU,F, I . -- E r t i m a t k d P r f .v a T Kn c f , R a t o s '1 o f L unvj C a n c k i i p k u 100 , 00 0 M m . kh o f R k m c o t k d A gi os a .v d V a r i o u s C i g a k k t S m o k i n g H t r t o h i k s '' -Xu tu b e r - - --Dung Cancer Control Estim ated Preva,lenco Hates ^ -Quant ity Smoked Daily'' All Qua.nti- Heavy'' ties .Gone Relative ^--P r ev al CIH`(', L - All Qua.ntiI ion vy ties Patients aged 50 to 5b y e a rs Doll arid HUH SYvmlcr a n d G r a h a m 2 fiiulowskv, G illiam, a n d C o r n f ie ld 2 Hreslow, Ho aglin, R a s m u s s e n , a n d Roswell Bark M e m o r i a l I n s t i t u t e Patients a g e d 0.) a n d 69 y e a rs Doll an d H i l l 2 Wynder and 'Graham2 Sadowskv, Gilliam, and Cornfield2 Breslow, Ilo ag lin , R a s m u s s e n , a n d Roswell P a r k M e m o r i a l I n s t i t u t e Abrams2 Abrams2 276 275 91 42 7 12 (3 2.58 21 0 117 47 4 27 1 1 198 208 87 4 1 12 7 4 212 212 102 4 6 6 17 7 172 657 141 109 15 9 7 197 199 127 62 0 187 160 128 70 5 29 15 76 71 4 t 7 l .5 9 4 168 169 422 78 17 25 5 197 675 479 280 50 9 5 * Estimated according to th e m ethod of C ornfields *> All d at a up pou ring in th i s ta b le , ex c e p t for Ros we ll P a r k M e m o r i a l I n s t i t u t e figures, a r e t a k e n fr o m th e p a p e r b y Bres- `mw<1nl? <Quantity am o v ed dai ly, ex c e p t fo r t h e p r e s e n t s t u d y , is all f o r m s of s m o k i n g ex p res se d as r i g a r c t s with pipe s a n d cig ars inverted to theii e q u i v a l e n t s in cigavets as follows: Dull and Hill--- 1 ou n ce t o b a c c o = 4 e i g a r e ts a d a y \Yynder a n d Gj a h a m -1 ci g ar -- 5 eiga r e t s ; ! p ip e fu l --- 2 1/? ei g ar et s Endowsky, Gilbr.m, a n d Co rn fie ld I r i g a r -- 10 ei g ar et s ; 1 o u n c e p ip e to b a c c o = 20 ei g ar et s Da the Bicslow c nl. s t u d y q u a n t i t y s m o k e d da ily in cl ude s o n ly eigaret. sm o k in g . d Relative pr ev al en ce ar e the. following ratio s: h e a v y -- n i tc ai m in g h e a v y s m o k e r s / r a t e a m o n g n o n s m o k e rs ; all q u a n t i t i e s rate among sm :k e rs/ ra te am o n g nonsniokers. ' Heavy s m o k in g is defi ned as t h e f olio win g n u m b e r of e i g a r e t s or eq u i v a l e n t s of cigaro.t.s po.r d a y : Doll a n d Hill, 50 or m o r e ; Wymler a n d G r a h a m , 25 or mo re ; S a d o w s k y , G ill ia m , a n d Cor nf ie ld, 41 or m o r e ; B r es lo w et nl., 40 o r m o r e ; p r es en t s t u d y , At or more. `mokers elicit xl hv the various studies is oven >sa likely. Their likelihood is probably least in tmlies made .n hospitals, such as those at Rosvcll Park Memorial Institute,5 in which all pa tents arc admitted without- charge, and in the British studies of Poll and Ilill, made in patients in hospitals coming within the National Health Insurance scheme. Wyndcr and Cornfield's `twly7 made of smoking histories of physicians who died of lung cancer also is not subject to the possible bias of hospitalized eases. Tire fact tin t 13 separate studies have shown the same result is strong presumptive evidence that an association exists between cigarct smok ingand lung cancer, i.e., that lung cancer occurs more frequonily among cigarct smokers than among nonsmokers. Indeed, one may predict that the "pros- >eetive" studies now under way will fully confirm 1he fact of a substantially increased risk of lung cancer among cigarct smokers. How ever, they should provide more precise estimates of the magnitude of this risk. M agnitude o f In c re a se d R isk o f R u ng Cancer A m o n g C igarel: S m o k e rs Estimates of the extent of increased risk to lung cancer suffered by cigarct smokers vary markedly. 'Fable I, taken in part from the paper by Breslow el a l.2indicates that in the age groups fifty to fifty-nine and sixty to sixty-nine years, male smokers have lung cancer prevalence rates from four to 15 times as great as among nonsmokers. Among "heavy" smokers, according to varying definitions, the relative prevalence is estimated at seven to 29 times that of nonsmokers. Tabic IIA presents estimated rates based on a tabulation of smoking histories 'of patients ad mitted to Roswel! Bark Memorial Institute from 1938 to the present. These data, indicate that cigarct smokers have at least six times the risk of lung ca.ncer incurred by nonsmokers. 'Flic ex cess risk among pipe and cigar smokers is much less. In 'liable III A these data- have been recal culated, after adjusting the place of residence, classified into nine population density categories, of the noneancer cases to th at of the lung cancer cases. The indicated relative prevalence among smokers is not significantly altered by this adjust ment. The relative, incidence ratios between smokers and nonsmokers arc based on the as sumption that the differences in smoking history between lung cancer patients and noncancer March 15, 195 771 MORTON L. L E V I S TART,K If A . Ksti m.v i t i ) A \ m : m . I N<:n>i;.\ci'; ok i j ; \ d ( '.s\r:i;u i>i-;u 100,001) I'o i t i .a i hiv A m o m i P u t ;, C i c a n, ano Ck a hut R\fi>Mr AND N' o NH Md K I'.IIN ( M a l k s ) Age ( roup ( Years) X onsmokery A n n u a l I n c i d e n c e ............ -% rigarci. Pipe ( 'gar Smokers Smokers Smokers R atio of Inridotiro to N onsm okris Among ( 'igaret, Pipe Cimr .i Smokers S m o k e rs Smokre ? 35 l o d i 45 to 54 55 to 64 (5 to 74 75 ami over 1 .2 0.3 28.5 55.1 If. 9 10.2 64.8 106.2 343.5 302.1 15 . 0 26.4 76.0 71.8 11.3 50.6 40.8 01 .4 8.5 7.0 0.0 6 2 23.2 ;i l .6 1.2 1 0.0 l.H 1 1 .4 o fi r 4 .2 5.4 " T h e r e worn no 1m m c a n c e r case s a m o n g p ip e a n d ci g a r s m o k e r s in th is ag e groii] l. XOlMHKit o r ( !ASIPS 6 A(io (r o u p ( Years) '- - .X o n s m o k e rs --> Lung Non- ( 'ivnce.v cancer -Pipe" Smokers- lain:g Xon- Cane "v r a n e e i' .-C g a r" Smokers-- Lung Xon- Cancer cancer --- ( 'igaret/' Smokoi*1- [i Lung Win - Cancer riuicci 35 to 44 45 to 54 55 to li t r>5 to 74 75 a n d over 1 5! r> 122 Hi 1Sii 13 168 1 80 37 (it 3 1 1 2, 37 6 75 10 05 2 41 21 250 16 2 34 8 52 5 71 2 32 17 205 30 Hi i ! 00 3|R ; 182 301 j. 772 151 j 2fl p 400 m) i " Inr hides " p u r e '' p ip e a n d r i g a r sm o k ers only. i fi I n cl ud es cigavet s m o k e r s w h o also s m o k e d p ip e s a n d / o r ci gars. ;j TABLF, iH A. L stimatku L unin C anukr A n n u m , I'N rin h N r r : m u 1 0 0 , 0 0 0 P o p u l a t i o n A m o s '<; M a l l C i g a r k t S m o k k u .h a n d XONSMOKKKS Age. C r o u p (Years) Xcw York State"- -- Xon- AH smokers Smokers Males P a t i o of Inoi- dnni'e Smokers Xonsmokors 30 to 30 40 to 40 50 to 50 60 to 6!) 70 to 70 0.8 4.2 14.8 50.3 33.6 2.0 25.4 108.0 280.4 435.8 2.3 18.2 70.7 12 5 . 7 124.5 3.6 6.0 7.3 5.6 13.0 " R ates h \scd on incidence rates for lung 'ancor in Xcw Y o r k S l a t e , exc lus ive of N e w Y o k C i t y , for 1040 -1051. proximate true differences in the relative ridO developing lung cancer. These data I.VO) indicate that among males aged thirty yrvf and over who are light smokers (less than ty| oigarets a day) the lung cancer risk is aim,:- doubled, in smokers of five to 19 cigarets a A r the risk is about six times its great, and in thy) smoking 20 or more cigarets, approximately g) times as great a,s in nonsmokers. t In Table V is presented a- further analysis ofp smoking history of male lung cancer patientd .11oswell Park Memorial Institute, compared*!! patients admitted with a presumptive diagim.j Age (Yea rs) 30 to 30 40 to 40 50 to 50 60 to 00 70 to 70 R. X o n s t n o k e r s - - Lung Xon- Cancer canear 1 38 3 86 8 1 56 23 206 4 150 30 636 ---( igaret Smokers---- Lung Yon- Cancer cancer 16 70 57 244 152 366 140 225 25 72 300 1,077 ^..........O th e r ........ i lamp; Xon- i C a n c e r cancer i| ! 30 if tA ,J 13 iw ) 25 244 r 10 180 ! 52 OX if patients reflect similar differences between lung cancer patients and the population from which they come. Since there is no reason to believe that the smoking habit per so influences admis sion to the hospital, these ratios probably ap- of lung cancer but whose symptoms upon fe:|! diagnosis were not due to the presence of!-; cancer or any lung tumor. This is a severe ton) the difference as to smoking history betweenfy cancer and other patients since smoking per*): ETIOLOGY OF JAJXG CAXCFll: P U F F ES T E TA T US T A K l,P IV A, In t i m a rt:i ' A n n u a i, L ung O an ci -.u I nt idk nck i'i'.u ] 00,00 0 A mong M auks, C ica hkt S mokfuh and X onhmok kus, hv Avkuagk A mo unt S mokkd D aidv ( N kw Yoinc St a t i / ) Age ( Years) - I >;t iIy C i g a r e t s S m o k ed Less than 5 :> to 19 20 or more Nnns lookers All S m o k e ::o t<> 59 -10 to U) 50 to : 9 fit) to r.9 70 to :o 80 to 7 9* -1 ,S 8.0 88.2 fd). 1 82.2 20.3 1 . 7 18.8 SO. 8 212. 5 475.0 92.7 ' 8.1 8 1.7 111.1 479.0 000.-1 10 2 .0 0.8 l 2 IPS 50.8 88.0 15 . 0 2.9 25. 1 10S 9 280.4 485.8 180.0 " l'or the m e diod of eaPula.ting incidence rat.es, see Cornfield.8 h Based on re p o r te d h i n g c a n c e r infici ene e fo r N ew Y o r k Stri, te (exelnsi ver of Y ew Y o r k ( li t y ) for HI 10 -195 1. f A dj us te d (n p o p u l a t i o n , male's ag e d M0 t o 70 y e a rs , N e w Y or k S ta t u , ex c lu siv e of N e w Y o r k C i t y , 1000 ce nsu s. P . Pi-:n C u n t 1 ) i s t k i n f i i o n " o f C i e M t i r r S m o k f u s i n D at um . Vh m i i k h o f C i g a u k t s S m o k o d '' Age (Years) - - Daily Xu mbor of Ciga rets <--- -Hess t h a n 5- - --- > t.o H) ......... - Non- hung Non- bung cancer ( 'a licer cancer ( 'ancer 50 to 89 40 to 49 50 to 59 00 to 09 70 to 79 2.71 3.59 0.83 5.89 4.21 5.50 1.59 2.99 2.50 2.78 28. HI 23.00 20.05 ] 5.81. 0.54 10.07 57 . 14 53.89 3.8.97 30.50 n A d ju s te d t< pla ce of res id e n ce of t h e lu ng c a n c e r g ro u p . &Pipe a n d ei ?ar sm okers n o t shown in this table. - - 20 or More Non- bung cancer ( 'a ncer 40.01 37.34 29.55 12.10 0.84 05.07 71 .43 59.28 40 . 15 33.31 Xonsmokera - Non- Tmng cancer Cancer !5 . 0 8 20.09 22.41 29.09 37.58 5.50 4.70 4.02 11 .08 10.20 C. O o M l 'AUl S GN o f E s t i m a t k o P f l a t i vk I x c t d k x c f o f b u x o C w u n u A mon' o C i g a u k t S m o k k r s a n d X o n h m o k k u s Age ;Yea rs) ........................ D ai ly C'igarcts S m o k e d - IjOss t b a n 5 5 to 19 20 or More All Q u a n t i t i e s 30 In 39 40 to 4 9 50 to 59 90 to 09 70 to 79 80 to 79" 0.0 2 . i 4.4 1 .9 3.3 8.5 2.2 5.8 8.5 1 .2 4 .2 8.9 2.4 IP 1 8.0 1 .8 0.2 10.8 3.0 0.0 7.3 5.0 13.0 7.2 n .Relative, in cidence (sm o k ers /tio n s m o k ers ) o b ta in ed a l te r a d j u s t in g all rate's to th e nude popu la tio n , New York S ta te , ex clusive of N e w Y or k C it y , 19.10 census. T A B L E V.-- ( hoAKF.T S mok i;hs A mono P ati unta w it h lot no C ano fu and w i t h O tii fk D i si-;asks or' t h k R ung ( Mo n t i ; mo usi in M akkh ( R o s w k u . P auk. M kmouial n s t i t u t f ) A (Yean ------- - h u n g ( 'a n c e r ...... - - ('gai'et Smokers ( 'uses" Number P e r ( 'cut. b u n g X o n tu m oi M ' -- - - Cigaret Smokers ( discs Number Per Cent 1udic uted Relative Ineidenee Smokers to All Others'' 85 to 44 45 to 5 1 55 to 0 1 05 to 74 75 a n d o -er 81 30 90.8 120 109 90.8 218 182 83.5 no 72 05.5 14 7 50.0 17 11 01.7 37 28 75.7 40 20 50.5 17 4 23.5 1 0 0 10.5 3.5 3.9 0.2 493 4 00 118 09 I nc l ud es 18 cases will) h i s t o r y of c'o m b i n a li on p ip e a n d ci gar s m o k in g , n o t t a b u l a t e d in 'Table I I B . '' Incl ude s ch ro n ic b r o n c h i ti s , b r o n c h i e c ta s i s , e m p h y s e m a , a n d o t h e r d ia gn ose s. r Comparison here is between eigaret smokers and " all o th e rs" including non-smokers a n d pipe and cigar smokers. known to cruise in some individuals chronic in- The differences in per cent, eigaret smokers bcflammation of the trachea and bronchial tree, tween the two groups is statistically significant, MORTON L. LU VJ N V7Y- ;uul the indicated relative prevalence, from these data,, of lull}? cancer among smokers compared to nonsmokers is still considerable (approximately live times as great). Because the routine method of obtaining smok ing histories is known to understate the preva lence of smokers, there is reason to believe that these estimates of relative risk are too low rather than too high. These data confirm those of previous reports from Roswell Park Memorial Institute5'6and by many other investigators that the lung cancer risk among moderate and heavy eigaret smokers is several times th at of nonsmokers. E liologic Significance The fact that eigaret smokers suffer an appreci ably increased risk of lung cancer does not of course amount to proof that this increased risk is caused by eigaret smoking. I t could also be due lo the fact that eigaret smoking was in turn as sociated with some other factor or factors which were carcinogenic or that eigaret smoking en hanced the effect of other carcinogenic factors. Possible factors of this kind which have been suggested are atmospheric pollution by in dustrial wastes, engine exhaust fumes, dust from tarred roads, and occupational exposure to carcinogenic, dusts or fumes or radiations. The air of some industrial cities is known to contain benzpyrene and arsenic, and mortality rates from lung cancer are higher in cities than in rural areas in this country, in Kngland, and in Denmark. However, this would fail to explain the marked excess incidence of lung cancer among males in cities as well as in rural areas. The extensive British studies of Doll and Ilill. also showed the same excess risk among smokers when attention was conlined to residents of London. Doll10 has recently re ported that the lung cancel' incidence among nonsmokers in Dug land was about the same in London as in rural areas and in females as compared to males. These findings suggest that the difference in lung cancer incidence between the sexes and between rural and urban dwellers may be due to differences in the prevalence of eigaret smokers in these groups, either alone or in combination with other factors. It has not been established that exposure to air pollution, to engine ex haust fumes, or to dust from tarred roads is significantly higher among lung cancer cases than among the population from which they come. Lung cancer cases known to arise in occupations which involve increased risk to lung cancer at present account, for only a relatively small per centage of the total cases. Admittedly, there may be many more such cases than are recorded and more such occupations than arc now known. Two studies correlating smoking history and or- | cupation in lung cancel' cases have been reported. In 1,357 male lung cancer cases and 1,357 i matched controls, Doll1 found an important ex cess in the cancer group in only one occupation, the production of illuminating gas: "The cancer : group containined 23 such men, the control group 14.'' Although this difference was not signifi- cant, it corroborated other evidence (gathered also by Doll) tha t gas workers have an excess risk to lung cancer. Breslow2in a study of the smok ing and occupational histories of 518 lung cancer cases and 518 controls found 77 lung canper eases (about 15 per cent) "had an exposure of more 1 than five years to suspect occupations." In one : of these, welders and sheet metal workers doing welding, part of the excess risk was attributable to the high proportion of smokers, and a signifi cant portion could not be so explained. Unless occupational hazards causing lung cancer arc ; many times more prevalent than any existing $ evidence indicates, they could not possibly ex- ? plain more than a minor fraction of the observed e increase in lung cancer or the excess risk among i smokers. 1 This conclusion has been further tested by a e study of the occupational histories and smoking ! histories of male lung cancer patients and other t male cancer patients at Jtoswcll Bark Memorial | Institute. In this study (to be reported more ( fully in a subsequent publication) 137 male lung [ cancer cases and 390 control cases were matched 1 for age and place of residence and adjusted for of- i cupation. The excess of eigaret smokers among ; the lung eaneer eases remained after the factors of j age, residence, and occupation were equalized ; (Table VI). : Interpretation of the ctiologic significance of : the smoking-lung cancer association rests on several types of evidence. First, it conforms ' with what is known about other carcinogenic agents. Tobacco smoke is known to be an irritant and to come into contact with the bronchia! mucosa. All or most known carcinogens maybe classed as irritants, although the reverse is not true, and direct contact is one of the most fre quent modes of effective application of ca.rcina ETIOLOGY OF LIJNG CANCER: P R E S E N T S T A T U S T A U L F , V f . ---- K h T I M A T K . I ) H K E A T I V K U l H K O K T , U \ 0 ( , ' A N C K H R V S m O K I N O C ' A T K C i O K I K S A N D O C C U P A T I O N ( M m . K S ) Pnaclj listed for o c c u p a ti o n fuing cancer Control (othcn cancer) Ding cancer incidence among cig.vet smok ers co m p ar e d to nonsmokers Control cases a d j u s t e d to occupational distribu tion of lung cn. cor cases Lung ca ncer inc idence among cigarc' smokers compared to nonsmokers, ad ju sted for occupation Nonsmokers Other than Ciyarcts r>. i i 24. 24 8.03 28.28 1.00 1 .33 23.02 27.81 1.00 1 .32 S m o k i n g ( 'a te g or ie s ........................ - ( Per Cent) Cigarets by Daily Units aiul Duration ^-- Less than .---20 or M o r e - - ' 20 per Dav per Dav 'Total Under 40 Years U nder 40 Years Cigaret 10 Y ea r s or M o r e 40 Ye;,1* or M o r e S m o k e r s 13.87 10 . 0 1 10.07 0.87 27.01 13.80 35.03 13.13 80.80 47.48 0.24 1J.87 5.20 0.38 0.21 12.71 14.18 13.74 8.71 40 . 17 5.32 5.32 8.04 I 1.50 8.05 'Total Cases Num- Per her Cent. 337 100.0 300 100.0 100.0 I'Piis. Second, it rests on the failure to demon strate other mown causes of lung cancer among smokers, as compared with nonsmokers, to explain the excess risk among the former. Third, and more indirect 1y , it rests on the conformance of the etlogic hypothesis with the known epidemio logic facts. These require that the suspected cliologic age it be one known to have increased markedly in prevalence within the past thirty pears, as has cigaret smoking, and to have af fected males more extensively than females. The product .on of skin cancer in mice by tar ob tained from cigaret smoke, reported by Wynder, Graham, and Croningcr,4adds further support to the belief that cigaret smoke may be a truly ctiologic factor i s lung cancer. All of the available evidence thus indicates that cigaret smoking is one of the contributing causa tive factors in the production of lung cancer. It does not indicate that it is the sole causative fac tor or even chat it is operative in the absence of other factors, such as susceptibility and, per haps, other as yet. undiscovered etiologic factors. Perhaps llic fairest statement of the case at present is that there is a strong presumption that cigaret smoking is an important etiologic factor in lung cancer. I t has been estimated previously6 that in males cigaret smoking may account for from 50 to 75 per cent of lung cancer on the as sumption that other etiologic factors are equal in smokers and nonsmokers. Recently, Doll" has estimated that in England, "about one in live of the lung cancer deaths in persons aged twentyfive to seventy-four in 1950 were attributable to causes other than smoking,'' or about 80 per cent to smoking. A n ticipated R esults o f F u rth er R esearch Total population studies, whether "retrospec tive" or "prospective," should serve to furnish more, precise estimates of the magnitude of the excess risk associated with smoking of various types and durations. Such studies cannot be ex pected per so to furnish more information than we now have rewarding the etiologic significance of smoking unless accompanied by a concomitant stud.y of all other known or suspected etiologic factors for lung cancer among smokers and nonsmokers. Experimental studies now aimed at production of cancer in lung tissue by application of the components of tobacco smoke, if successful, may add to our confidence in the etiologic nature of the smoking-lung cancer relation. Further study is also indicated of the reasons for the much smaller increased lung cancer risk found among cigar and pipe smokers. Possibly, sonic! sensitivity test may be devised to distin guish between the majority of smokers who may smoke with relative impunity and the minority who will develop lung cancel1. The identification of and removal from tobacco of the presumed car cinogenic agent, without removal of the tobacco, is a hopeful possibility. P u h lic H ealth Im p licatio n s bung cancer is a rapidly increasing cause of death in this country. 1n New York State it now exceeds tuberculosis in number of deaths among males. Retween 1931-1933 and 1948-1950, ageadjusted lung .cancer mortality increased among males in New York Slate by 385 per cent, among March 15, "lo t 775 uiiim an m s mm s m MOttTOK L. L E V IX TA BLK ( 'a n c k k VI I. A<r,-Ai).ir:sTKi)" a v i A lk O tiiku S itkh S t a t i ,'' ( I ) k a t i i m p k k M ohtamtv from o k ( ' a n c k k i n N kav 100,000 L i v i n c ) L k n < Y ork T A L L F , Y J I ] . - P f.r ( ' p a t r n o n A M r . r n o f D kyf, b o t a r I , K N<; ( 'ANCKK ANI ( ' AX OK It OK O t I[ KK S l T K S FROM P ] ITITI n \ ( M 2 I 9 M ANI 1 9 4 9 I M I ) AJales 1931 -1933 19 18-1.950 Females 1931 -1933 194,8 1950 Lei- c e n t c h a n g e Males Females I,unti Oiinci'r 4.7 22.8 2.5 4.2 + 385 1-08 ( -;i tif'r r O ther Sitos 110,3 112,7 151,2 128.8 --2 -- 15 -Site Male All sites Lung Female AH sites Lung 1912 1944 1f t. -l:' 0.91 19.37 0.22 1949-1951 2 20 ..0117 23.27 0.34 1949 Mb) I (Adj usted'1) 18. 5f> 1 .87 21 .40 0.31 <' O b t a i n e d b y h o ld in g 1042 1044 m o r t a l i t y ra t e s from all ca u se s c o n s t a n t a n d a p p l y i n g 1040 1951 r e p o r t e d eancer inci dence rates. " Adjusted to 1940 population. New York Stale, t Kxclusivo of N'rw Y or k ( li ly . females h r OS per cent (Table VIJ). Since can cer is a reportable disea.se in New York State, it is possible to calculate the probability of developing cancer of various sites throughout the, life span. For lung cancer, within the short period between 1942-1944 and 1949-1951, this probability has more than doubled for males (Table VIII). At present rates of incidence, 2 per cent of males may be expected to develop lung cancer. If the present rate of increase continues, this figure may double again within the next fifteen years. Lung cancer is a public health problem rapidly mounting in importance. Public health authori ties as well as private medical practitioners must take cognizance of the available facts regarding its occurrence and etiology. The following ap proaches to the problem are indicated: 1. Case-finding: Since early diagnosis and treatm ent offer the only hope of cure in lung cancer, early case-Inuling should be intensified. Special attention should be given to the following groups: (a) males aged forty and over, (b) cig arct smokers, (c) persons exposed to known oc cupational hazards. The application of the community chest x-ray program, so extensively developed in this State, to the finding of lung cancer has already begun. Further refinement and adaptation of this program to the diagnostic problems of lung cancer are needed. 2. Public education: The public looks to the medical profession and to public health authori ties for information regarding health matters and for an appraisal of existing evidence regarding health hazards. W hat shall be our reply to questions regarding the relation between rig,a,ret smoking and lung cancer? The existing evidence justifies the following statements: (a) Lung can cer occurs more frequently among cigarct smok ers, and particularly among heavy smokers, than among nonsmokers. We do not know exactly how much more frequently it occurs, but the evi dence we have indicates th at the exces.s risk is ap preciable. (b) In the absence of evidence to the contrary, the most reasonable assumption is that this increased risk is due to the effect of cigarct smoking, either alone or in conjunction with other factors, (c) Persons who arc concerned about lung cancer cannot be assured th at cigarct smoking will not increase their chances of developing the disease, (d) Cigarct smokers, particularly those past forty .years of age, would be well advised to have chest x-rays taken at least once a year. 8. Research: .Further research into causes, improved treatment, and methods of control ol cancer is a public health responsibility. In New York State this has long been recognized. The expanded Roswell Park Memorial Institute at Buffalo12and the Bureau of Cancer Control of the New York State Department of Health are de voting considerable attention to investigation of the relation of tobacco to lung, larynx, lip, oral, and other types of cancer. Investigation is pro ceeding along the following two main lines: (nj investigation of the carcinogenic properties of tobacco smoke components and (b) epidemiologic investigation of the characteristics of persons who develop lung cancer and of smokers compared to nonsmokers in an effort to uncover all pertinent associated factors. Sum m ary 1. The existing evidence regarding the eti ology of lung enneer points to the existence ol multiple etiologic factors, including occupational haza rds and ciga rct smoking. (2. Known occupational hazards do not ac count for more than a relatively small proportion of lung cancel' cases. 3. The evidence regarding excess risk of lung cancer attributable to cigarct smoking indicates that it is several times that among nonsmokers 770 New York State J. Med, L T IO M O Y OF LU KO C A S C E ll: PU KH K KT S T A T U S snd may acenuui for a. major part of the incidence <iflunp, cancel', ]articukirly among males. 1. Tim evidence for excess risk of lung cancer aiming smokers persists when comparison is made, between lung cancer patients and patients with uthcr conditions affecting the lung and bronchi ml when the :influence of residence and of oc cupation is equalized in lung cancer and control roups. 5. The available evidence is sufficient to justify public health action in the direction of i',ise-finding and public health education to the effect that eigaret smoking must be presumed to play an important role in the production of lung cancer. 6. Intensive further research in all aspects of the problem s urgently indicated. Further studies may be expected to provide more pre cise data on the magnitude of the excess risk to lung cancer among smokers. Additional evi dence regarding etiology may be forthcoming from the results of experimental studies but may rest in the fined analysis on the study of other' possible etiologic factors found in smokers but not in nonsmoxers. At present no such differ ences are known to explain the excess risk among (lie former. Tim weight of the evidence there fore indicates that the relationship between eigaret smoking and lung cancer is causal and not merely an '`association." 7. In less than a decade the chance of de veloping lung ( anccr at some time (luring life has more than doubled among males in New York Slate. Two per-cent of all males may be expected to develop lung cancer at present rales of inci dence. Since a majority of adult males are cig aret smokers and most, of those over forty have smoked for many years, a good deal of the smok ing effect probably has already been exerted on these. Hence, no appreciable decrease in rates may be expected in the next, decade, even if everyone forthwith gave up smoking. Public health efforts hence must, concentrate on early case-finding and on research. G r a t e f u l a c k n o w l e d g m e n t ip m a d e t o M r . A r t h u r K m us an d Miss R ita C u sh m an , of the B u reau of C an cer Control, N'ew York S ta te D e p a r t m e n t of Health, for assistance in the statistical analysis. Noth : T w o c h a r t s a n d a n ad d it ioual I ah In will ho i n c lu d ed in t h e a u t h o r 's r ep r i n ts . R eferences 1. Doll, R .: B r it . M. J . 2: a 21, 585 (Sept., 5, 12) 1953. 2. Bres low , L., I lo a glin , I-., R a s m u s s e n , (J., a n d A b r a m s , II.: Presen ted at th e A nnual M e e tin g of the A m erican P ublic H e a l t h A sso ci at io n , Oct,., 1953. I n press. 3. M ll e r, V. II,: Z t s d i r . f. K re b s fo rsc h , 4 9 : 57 (1959), 4. W y n d e r , K. L., G r a h a m , E. A., find C r o n i n g e r , A. B.: Cancer Research 13: 855 (1038). 5. bovin, M. I,,. Goldstein, H., a n d G erh ard t, P. IT: J.A.M .A. 143: 330 (1950). 0. Levin, M. L.: Acta, Union I n te rn atio n a le ce n tre le C ancer, 9: 531 (1953). 7. W y n d e r , E. 7,., a n d Co rn fie ld , J . : N e w E n g l a n d J. M ed. 248: 441 (1033). 8. Co rn fie ld , J . : N a t . C a n c e r J n s t . 11 : 1.200 (10.31). 0. Dol l, R . t a n d Hill, A. B . : B r i t . M . J . 2 : 730 (1950). 10. hid. 2 : 1271 (1952). 31. Doll, It.: Brit. Can car 7 : 303 (1053). 12. M oore, G. E., a n d W eh r, W. II.: N u w Yomc S tatk ,1. M kd. 5 3 : 2 7 08 (1953). M e d ic a l S o c ie ty o f t.lie S ta te o f N e w Y o rk llo tel S 'a lle r A N jN . U A L M K II T I N U M ay 10 to 4, 19:14 N ew Y ork C ity March 15, 1054 777