Tobacco Documents Online

Smoking and Health Many scientists challenged current anti~igarette I theories at the headings before the U. S. Senate Coln- ! mlltee on Commerce in March and April 1965, This booklet contains the full testimony of the following i scientists: page Dr. Milton S. Rcaenblatt ........... 2 Associate Clinical Professor of Medicine New York Medical College New York, New York Professor K. A. Brownlee ........... 15 Associate Professor of Statistics University of Chicago Chicago, Illinois Dr. Henry Easton MeMahon 19 New York Polyclinic Medical School New York, New York Names and qualifications of other expert witnesses whose testimony challenged antidzigaret te theories are listed on Page 20.

Smoking and Disease: Etiological Perspective BY MILTON B- ROSENnLATT, M.D, Associate Cllnical Pro fes.,or of Medicine New York Medical College T11o~e ~ltll c~llnol r~m~mber 1lie pus! ~r~ ~¢t~ttn~ed to repeal it, --G. SAN [ AyAN,t. lntroductinlL The inereasc in lung cancer in the past few decades has fostered the concept that this is a modern disease produced fly exposure to recently introduced carcino- genic substances. In the intensified search for causal agents nlany fundamental facts have been disregarded. The proper differentiation between apparent and ab solute increase requires a perspective that encom- passes far more than a mere assessment of epidemio- logical statistics. The rapid rise of a disease coincident with improvement in diagnostic techniques has oc- curred many times in the course of medical history. At [he turn of the century, when the markeA increase in cancer deaths in England was atlributed variously to parasitic infestation, meat consumption, and exces sire rainfall, King and Newsholme (1) observed pro- pbetieally in 1893 that whenever a cancer became diagnostically accessible the incidence automatically increased. The reporl of the Advisory Committee has based its judgments of causality almost entirely on cpidemiological data, has invoked hypothetical con- siderations to substantiate the statistical correlations, and gas not sufficiently considered pertinent refutatory evidence. YBstofical Considerations The report of the Advisory Committee discusses the history of lung cancer only insofar as it relales to the studies on smoking and completely ignores the fact that the disease bad been prevalent for 100 gears be~ fore cigarette smoking became widespread. Nowhere in the report is it even intimated that the basic knowl- edge of the pathology of lung cancer was developed in the t9th century, that all the characteristic symptoms and physical signs were discovered in the 19th century, and that astute clinicians, both here and abroad, were emphasizing, in the 19th century, that lung cancer oc- curred far more frequently than was generally realized. In 1815, La~nncc (2) differentiated lung cancer from tuberculosis and other pulmonary diseases in an article in the Dietionnaire des sciences mgdicales. Four years later he published a treatise on physical diag- nosis, "De I'auseultadon mddiate," which was trans [ated by Forbes (3) and evoked considerable interest in the disease in England. Anthal (4), in 1821, investi- gated the possibility of diagnosis by sputum examina- tion and in 1837, edited the fourth edition of Lai:n net's classic work contributing a profusion of footnotes on lung cancer. Bouillard (5), in 1826, discussed the possible relationship between antecedent pulmonary infection and lung cancer. Heyfelder (6) won a cita- tion for his studies on the subject in 1837 and, in 1839, Bayle "Traits des maladies cancdreuses" listed "can cer du gnumon" as an established primary maligna ncy. Other French articles were published by BOrard (7) in 1821, Arnal (g) in 1844, Lebert (9) in 1845, Charcelay (10) in 1837, and Munneret and Fleury ( 11 ) in 1846,/he latter authors contributing a sizable bibliography. Lung cancer has beer1 a subject of considerable in retest in Britain since the publication of Stokes' text- books, "Dise~-ses of the Chest," in 1837, The descrip- lions of the manifestations of the disease and the prob- lenls of differential diagnosis were indicative of a wide clinical experience. Stimulated by Stokes' teachings many other British physicians made important con- tributions. A report by Hare (12) in 1838 described a neurological complication which was rediscovered ( 13 ) a hundred years later in this country Significanl observations in pathology and elinieaI medicine were made during this era by Hughes (14), MacLachlan (15), Burrows (16), and Graves (17). In April 1843 the British and Foreign Medical Review re- viewed five articles on lung cancer with an editorial comment on the increasing interest in the disease and its greater prevalence than previously suspected. In the 2d half of the 19th century, English reports showed an increasing interest in the microscopic studies of lung cancer in order to better differentiate it from metastatic tumors of the lung. Histological findings established in 1857 by Quain (18), Cockle (19), Mayne (20), Pitman (21), Budd (22), Page (23), and others are just as valid today as Ihey were a cen tury ago. The greatest advances in the pathology of lung can- cer were made in Germany in the latter decades of the 19th century. Waldeyer's (24) studies t~f the ori- gin of cancers stimulated investigative research culmi- nating in the establishment of the origin of lung cancer from bronchial epithelium. The number of case re- ports, articles, and inaugural dissertations on "luugen- krehs" contributed by German physicians in the i9th

century is remarkable The authors whose works pro- vided the ground work for our current pathological concepts include Purls (25), Genrgi (26), Reinhardt (27), Birch-Hirschfeld (281, Hesse and Hartung (29), Siegel (30), Siegert (31), Wolf (32), Passler (33), Frankel (34), Schwalhe (35), Hartmann (36), and a great many others. The various cell types and architectural structures found in lung cancer were well described in these publications. 'lhe German patholo- gists noted Ihat lung cancer cases usually came to atltopsy in a far advanced stag~ of Ihe disease but that fcw had bccn diagnosed during liIe. One of the most valuable contributions was the differentiation of pri- mary lung cancer [rom secondfiry pulmonary growth and benign tumors. It is noteworthy that the patho- logical techniques and stains discovered by the Ger- man pathologists in the 19th century are stdl very much in use at present. General interest in lung cancer developed consldcr- ably later in the United States than it did in Europe but cases were described (375 at the Massachusetts Gen- eral Hospital in 1842 and I850. Articles contributed by Lehlbacb (38) in lg70 and by Loomis (39) in 1876 showed considerable familiarity with the disease. Delafield (40). the pathologist at Roosevelt Hospital, was well aequainled with primary lung cancer as early as 1868 as evidenced by his autopsy records. In the latter decades of the 19th century the number of American articles increased including contributions by Pepper (41), Van Giesen (42), Ripley (435, Kemper (445, Janeway (45), Holland (46), Hodon- py[ (47), and LeCount (485. On December 18, 1880, the Medical Record commented editorially that lung cancer would continue to interest pediologistg despite the lack of specific treatment. The most important American article on lung cancer in the 19th century appeared in the New York Medical Journal on Febru- ary 8, 1896. The author was Adler (49) whose mono- graph on the subject in 1912 has since become a med- ical classic. In the 1896 article, Adler repeatedly em- phasized that lung cancer was not a rare disease in the United States but was rarely diagnosed. He urged phy- sicians to become more familiar with its clinical mani- festations and pathologists In do more metleulous work so as to better recognize the disease at autopsy. As the 19di century came to a close, articles on lung cancer also appeared in medical journals in Italy, Po land, and Norway, and advances were being made in clinical diagnosis. It is not generally realized that the X-ray (50) and the bronehoseope (51) were begin- ning to be used to detect lung cancer before the 20th century One of the most informalive contributions ol the period was the collection of statistical data on the necropsy incidence of lung cancer. Feilcbenleld (52). in a dissertation from Leipzig in 1901, described the rise in lung cancer autopsies at the Urban Hospital in Berlin and attributed it to the larger number of autop- sies and the increased interest in the subject. Rieek (535 in 1904, noted the rise in Iung cancers at the Pathological Institutes of Munich for the period 1854 1902. Similar observations were made by Feldner (54) m 1908 at Giitlingen Kiknth (55) reported a nineIold increase in lung cancer autopsies al the Ep- gendurfer Krmlkeuhauses between Ihe periods I gg9- 99 and 1900-11 Karrenstein (56). in a review from the University of Berlin in 1908. made the significant (ihservEllion that no proper evaluation of lung cancer incidence could be made on the basis of edicial vital statistics. He also noted that refinements in histologi cal sl~ldy made il possible lot Ihe pathologist to rceog- nizc cases formerly obscured by late complications, It is very evidenl from tile many published reporl~ Ihat the rise in autopsy incidence in Germany was due to greater awareness of the disease and better pathologi eal diagnosis. Lung Cancer Mortality Proponents of the causal relationship between smoking and lung cancer have pointed to the rise in lung cancel deaths during the past three decades as corroborative evidence since this period was also char- acterized by a rise in cigarette consumption On the basis of national mortality figures it would appear that a tremendous increase in lung cancer had occurred in the United States since 1930, There is. hnwever, much evidence available to challenge the accuracy of the mortality statistics. It is not generally realized Ihat even if il had been possible to diagnose lung cancer in the early decades of this century tbere was no specific category in the vital statistics in which to record the lung cancer deaths. It was not until ]949, after revi- sion oi the International Code, that classification of lung cancer became a statistical reality In I912, Adler (57) in his famous text on lung can- ear critically questioned the accuracy of the census re- ports emphasizing that the incidence of lung cancer was far greater than realized. The total nnmber of recorded lung cancer deaths in the Unitcd States in 1914 was less than 400 reaching approximately 1,000 by 1920. On February 13, 1926, Ihe Journal of the American Medical Association commented editorially Ihat the number of cases in the preceding 15-year petted was so great as to almost defy enumeration but the current national figure for lung cancer deaths was still less than 2.000 In 1930. when the number ol lung cancer deaths in the Unded States was recorded as 2,837 Chevalier Jackson (585, the pioneer in bron- choscopie diagnosis, had records of almost 500 eases It is very apparent that the increase in lung cancer since 1930 was based on greater pathological knowl- edge and the development of techniques with which "~o diagnose lung cancer. However, the report of the Ad- visory Committee stales that a true increase in lung cancer has occurred and bases its contention on art- donee furnished by State cancer registries and necropsy records from large general hospitals. The report cites

the statistics in upslate New York showing an average annual rate of increase of 7 percent for males and 3 to - 3.5 percent for females during the period 1947-60. The presentation o1 the data in this manner is most misleading. Review of the total lung cancer ffeaths in New York Stale during the period under discussion shows that the rate (If increase of lung cancer had de. ciined from 42 percent between 1940 and 1945 to a low of 18 percent between 1955 and L 960. This could not have ~ccutred in tile presence of a true increase o1 the disease. The report also cites autopsy data fr~)m large hospi- tals where "diageosbe accuracy has been uniform and excellent for many years." A rise in autopsy incidence in hospitals is not an index of an absolute increase of the disease. It is well known that many factors con- tribute 1o autopsy incidence among which are ( 1 ) in- creased pathological knowledge, (2) change in au- topsy material with respect to sex and age (lung can- cer is a disease of older age groups), (3) greater per- centage of autopsies of "interesting" cases, (4)change in nature of cases hospitalized, and (5) wider selection of population (clinics in medical centers attract pa- tients from large areas). The intense interest in lung cancer primed by the first suecossful surgical treatment by Graham (61) in 1933 and snstafaed by greater awareness of the disease has resulted in the hospital referral of many more lung cancer cases with subse- quent increase in autopsies. The ~port compares the autopsy records at the Massachusetts General Hospital belween the periods 1892-1929 and 1956-61 showing an increase from 17 lung cancers to 172 it, the later period. From these figures it would appear that a true increase had recently occurred hut an earlier report from the same hospital by King (37) showed that the increase had begun as early as 1920 and that by 1938 there were 475 cases diagnosed of which 158 had been confirmed by autopsy, biopsy, or exploratory thora- cntomy. King, who had personally observed the in- crease in lung cancer hospital admissions attributed it to greater interest in diagnosis and surgical therapy. Autopsy experience in lung cancer in the United States lagged considerably behind that of the German pathologic institutes. In the 1931,t..difion of Henke and Lubarsch's Haedhnch ffer Spszinffen Patimiog- iseben Anatomic und Histologie, Fiscber's data on lung cancer included reports of almost 4,000 cases of which only a very small percentage were derived from American sources. The paucity of lung cancer au- topsy reports in the United States in the early deeedes of this century was not due to absence of the disease but to absence o1 knowledge. This is well illustrated in a report by Jeff6 (59), in 1935, showing that inng cancer comprised more than 11 percent of all the can- cers at the Cook County Hospital in Chioago. This figure is comparable to the incidence during the pres- ent era. During the same period that Jaffr, a European trained pathologist in lung cancer diagnosis, found the proportion of lung cancer to all cancers to be I I per cent, the incidence at the Vancouver General Ho~plt al was less than 2 percent and was less than I percent at the University of Oregon ( 110t There has been constant progress in die patholog- ical diagnosis of lung cancer in all areas of the world. A review by Harvey (60) of the autopsy inatedal at the Royal Prince AIIPSd Hospital in Australia revealed an error of almost 20 percent in the diagnosis 01 lung cancer. There are also marked differences in patho logical interpretation during the present era when. ac- cording to the reporl pathological diagnosis has been uniform. Spain ( 114 ) in 1959 reporled an increase in squamou~ eBll carcinoma and a decdne in ade~ocar- cinema and p~sented this as evidence of the carcino genie effect of cigarettes. In contrast, Lee ~nd Tsfa ( 115 ) in 1963 found the autopsy incidence of adeno carcinoma to he almost three limes that of squamous carcinoma. One factor responsible for a considerable portion of the lung cancer autopsy increase was the observation by Barnard (62) in England, and Karsner and Saphir (63) in the United States, that certaiu tumor iolmeriy diagilo~ed as sarcomas were. in actu- ality, cancers of the lung. The transfer of these oat cell tumors from ouc category to another automafi- cally resulted in an autopsy increase of lung cancer which was only apparent. The frequency of incidence of oat cell tumors varies with different pathofegists; Whitwell ( l 16) found that oat cell cancers comprised as high as 41 percent of all lung cancers while many others found the incidence much lower. [t is interesting to observe the disdain with which the proponents of the smoking-cancer theory relent improvement in diagnosis as a major factor in the in- el'ease of ]uRg caller. Neverthciess. every diagnostic facility used at present to detect lung cancer was either discovered or perfected within the past three decades ~e period of the great increase of inug cancer, Bronchoscopy, one of the mosl important diagnostle procedures, had been available for many years but was scarcely used prior to 1930 because of unwar- ranted anxiety as to possible hazards. It is of historical interest that at the Brompfon Hospital in London~ re nowned for specialization in pulmonary diseases, only one bronehoscopi¢ examination had been [:~rformed prior to 1925, Within three decades the nnmber of hronnhoseopies exceeded 800 per year. For more than a century physicians had been attempting to diagnose fang cancer by examination of the sputum for frag- ments containing mafi~lant tissue. Although success was achieved sporadically, the procedure was tedious and hampered by technical difficulties In the 193fi's staining techniques were developed which facilitated sputum ex~xflinalinn hut the process remained essen- finny a research procedure. In 1943, Papaidcolaou described a staining technique which within a few years became applicable to the routine examination of the sputum. The "Pap" smear for cancer cells in

the sputum and b[onchial aspirate has added signifi eantly to the diagnostic approach during the two dec- ades in which it has been utilized. Surgical exploration of the chest is another pro- eedure which has added to the number of cases diag- nosed. Prior to the 194fi's, exploratory diotacotomy was often considered a procedure of last resort and the indications were very limited. Since then ~finements in surgical technique and anesthesia have made this a safe pr~)cedure with wide diagnostic indications in 1949, a report from the Mayt] Clinic by Johnson (64) ~fiowed that 50 percent Of the lung cance~ had been diagnosed only after surgical exploralion~ During the past two decades interest in thoracic surgery has in- creased tremendously and thoracotomy is a routine procedure in praedeally all hospitals, disclosing cases of lung cancer which would otherwise remain undiag- nosed, Surgery has also been of diagnostic value in the increased number of biopsy examinations utilizing new techniques. The grea~st contribution to lung cancer diagnosis has been the vast increase in X-ray examinations whioh are now routine procedures in hospitals, clinics, doctors' offices, industry, and mili- tary organizations, Many of the lung cancers detected by routine X-ray examination have been entirely un- suspected¸ E seems hardly necessary to emphasize that the tremendous improvement in diagnost~ facili- ties has resulted in a far greater recognition of lung cancer than was previously possible. Clinical experi- ence has shown how easily lung carleer may be mis- taken for pneumonia, tuberculosis, lung abscess, brain tumor or even abdominal disease in the absenoe of ap- propriate diagnostic inve*tigation. How much reli- ability can be placed on lung cancer mortality statistics during an era when the disease was not clinically diagnosabfe? Accuracy fa Diagnosis The van improvements in diagnostic techniques have made it possible to dett~t greater numbers of lung cancers bul have also revealed that there are many pitfalls in diagnosis not snspec~d previously. Autopsy studies have shown that cases diagnosed din- ical[y as lung cancer, even with biopsy confirmation, may actually be secondary cancers, ft is now well es- tablished that the lung has a great affinity for meta- static tumors and that 30 to 40 percent ( 112, 113 ) of all cancers produce secondary pulmonary growths, some of which may simulate beonchogenle carcinoma. When non pulmonary cancers spread to the bronchial tubes all the lnadifcstations Of primary lung cancer nmy be reproduced resulting in erroneous diagnoses. In spite of all the progress made in diagnostic tech- niques, a positive differentiation between primary and secondary lung cancer may not be possible without autopsy exclusion of other primary sites. The present classification of lung cancer deaths con- tains two categories: ( 1 ) lung cancer, specified as prl- mary; and (2) lung cancer, unspecibed as to primary or secondary. Examination of lung cancer mortality statistics for [96[ and 1962, the latest years for which dnta have been completed, reveals some interesting facts For the gear 1961, out of a total of 38.929 long cancer deaths, only 19,462 (Sfi percent) were speci fled as primary lung cancer. For the year 1962, out of a total of 41.376 lung cancer deaths only 18,866 (46 percent} were specified as primary. The re mainder were unspecified which means thai the hmg cancers may have been primary le the lung, or. may have spread to the lung [rgm tumors originating in officr organs. The statistician and epidemiofoglst, lacking experi- ence in clinical diagnostic problems, are often totally unaware of the difficulties in differentiating between primary and secondary lung cancer Even when cases are specified as primary lung cancer this does not necessarily mean that the diagnosis was correct in all eases, it has become increasingly apparent to clini- cians and pathologists working in the field of lung cancer that sputum examinatinns for cancer cells and even biopsy examinations may yield misleading in- formation. Great reliance is placed on the histological findings obtained by examination of a lymph node or of bronchial tissue but positive differentiation be tween primary and secondary cancers is not always possible, fn a recent autopsy study (65) of cancer of file pancreas, it was found that 12 percent of the cases had been diagnosed during life as primary lung cancer, Other reports of erroneously diagnosed lung cancers on the basis of biopsy examination have included can- cers of the kidney, adrenal, ovary, thyroid, stomach, and rectum. Inasmuch as the cases specified as pri- mary lung cancer were not all autopsied, there is a significant potential source of error to be coiisidered, The Advisory Committee to the Surgeon General placed the greatest reliance on statistical computations in arriving at its conclusion of a causal relationship between smoking and lung cancer. The report states, "In recent years, about two-thirds of the certification of lung cancer deaths have been based on microscopic examination of tissue from the primary site and Ihe percentage is even higher for deaths under 75 years" The two references cited to validate the above state- ment appear as very tenlallus endot'sem~nts. One re port (66) is an evaluation of accuracy in a series of 1.837 death certificates in Pennsylvania. The con clusion with respect to primary ltmg cancer was based nn a total of 21 cases with no mention made of the nmnber of diagnoses based on autopsy The second reference (67) discusses the general aspects of cancer diagnosis but presents no particular data on primary lung cancer, A study based on 21 king cancer cases without autopsy confirmation is inadequate authority for a broad statement implying that over two-thirds of all the lung cancer deaths were confirmed patho- logically.

RaIe of fnerl~ase The so-called epidemic rise in certified lung cancer deaths in the United States has followed a pattern trial when scrutinized objectively, is more indicative of progress in diagnosis than of a true increase of the disease. This patter[[ shows a progressive deelthe in the rate of increase of lung cancer and was fully elab- orated upon by Gifiiam (fig) in his studies at the National Cancer Institute. it is regreltable that the Advisory Committee disregarded this significant con- trihulinn in hs cpideminlogical considerations. The earliest statishcs available (table 1) show that the number of certified lung cancer deaths in the United States was 371 in 1914, and 956 in 1920 Although these figures are ridiculously low when com- pared to the large numbers of cases reported in Euro- pean institutions during this era they represented an increase of more than 150 percent. A comparable trend was noted in the records o1 the Metropolitan Life Insurance C/o. which showed an increase of I00 percent in the standardized annual death rate from lung cancer among its policyholders between 1917 and 1926 It is to be emphasized that this tremendous in- crease occurred before the era of widespread cigarette smoking. In the succeeding years the total number of lung cancer ca~es increased but the rate el increase declined. r [~14 7, ,'1 logical explanation for the progressive decline in tile rate of increase is th at it i~ a reflection of the increasing ability to diagnose lung cancer during the past three decades. In the early 193ti's the mortality attributed to lung cancer was increasing among the'white male popeintion at the rate of approxinlately 10 gercent pet year. Thereafter, this percentage declined reg- ularly and by 1958 the lung cancer deaths were in creasing only at tbe rate of 5 peroent annually. Among Ibe white female popeladon, the rate of increase d~ dined from approximately 6 percent per year in Ihe earlier period to about I percent in 195~q. The re- corded lane cancer deaths among the nonwhite population, although much smaller in number, showed the same downward trend in the rate of increase. The higher rate of increase in the nonwhite population was consistent with the accelerated improvemenl in medical faeilgies in recent years for this group. The progressive decline in the rate of increase is a very significant factor in evaluafing the carcinogenic effect of tobacco If smoking produced lung cancer, directly or indirectly, the increased consnmption of cigarettes would have r~ulled in a sustained or greater rate of increase. The fact that the opposite has oc eurred implies that there is a relatively fixed preva- lence o1 lung cancer in the popu[atinn and that with each succeeding decade the improvement in diagnostic techniques resulted in the recognition of a greater number of eases. Eventually. the incidence of lung cancer will be stabilized as it has been for many other cancers in which the investigative procedures have been slandardized for long periods of time. The experience in the declining rates of increase of the certified cases of lung cancer is very similar to that which occurred in the German patiu~logie insti- tutes in the late 19th and early 20th centuries. This was the era when the diagnosis of lung cancer could only be definitely established at autopsy. In almost ....... ~" ..--~ all of the major hospitals the rate of increase followed ,-~ • , ,~ a definite trend consisting of a sudden rise, corre- spondinfi to fbe period in which pathologic diagnostic m Between 1930 and 1935 the increase was 79 par- cent; between 1940 and 1945 it was 53 percent; and between 1955 and 1960 it was only 39 percent. A comparative study (table I1) of the lung cancer mor- tality statistics of New York Stale (no figures avail- able before ] 93 ] ) showed Ibe greatest rate of increase, 64 percent, between 1931 and 1935. Since then the rate progressively declined and between 1955 and 1960 it was only 18 percent. How is it possible to reconcile the declining rate of increase of lung cancer with the 200-fold increase in cigarette consumption during the same period? If tobacco is a cancer-producing agent there should have occurred a progressive rise in the rate of increase. A criteria were established, followed by a decline in the rate of increase as the disease became better known. At the Moabit Hospital in Berlin, Wahl (69) found a 300-percent increase in the percentage of lung can- eers to all cancer~ between 1917 and 1923. The felinwing year the rate of increase had declfiled to 17 percent. Schonberr (70) rept~rted similar findings at the Pathologic Hygienic Institute of Chem nitz; com- paring the periods 1898-]916 and 1919-29 there was a 110-percent increase in lung cancer autopsies which fell to 24 percent in the next 4 years, Comparable observations (s~e table 11I) were made by Materna (71) in Troppau, by Schlesinger (72) in Lcipzig, by Engenolf (73) in Gfittingen, and by Holzer (74) in Prague. At St. Bartholomew's Hos~fial in London, Maxwell and Nicholson (75) found that the greatest increase in the autopsy incidence of lung cancer had

occurred between the periods 1884-88 and 1894-98. "lbe rapid increase in the incidence of a disease when new ggineal or pathological diageo~tio tech- niques are introduced has been observed repeatedly, An excellent example is endobronehinl tuberculosis. a very common ¢omg.linatlon of l~rogressiYe pul- monary lubereulosls. The condition has been Un- detected by pathologists in thousands of autopsle~ per- formedin this country and ggroad. In theearly 1930~s clinical and bronehoscoplc ~tedle~ established the ex- isbence of endobeonchial tuberculosis and this was reflected immediately in a tremendous ris~ in necropsy incidence. The report of the Advisory Committee stares (p 135 ) that "Lung cancer tonality among males has risen at a fairly constant rate since 1930" and cites Gordon, et at. (76) of the National Cancer Institute as reference. However~ there s~n~s to be som¢ dis- parity between the report's intcrpretatinn and Gor- don's article which states, "For white male~ and females and for nonwhite males the incnease in mor- tality for cancer of the lung and beonggus was great~st between 1930 and 1935 and ha~ since tended to de- celerate." The ari[efe also noted thai "the increase in long cancer mortality began simultaneously for all adnlt age groups suggesting it was caused by fecrea~ed diagrlo~tic awareness. Sex Dialribullon Cancer of the lung occurs predominantly among males. In the United States in 1930 ther~ were 1.ggg lung cancer deaths in males and 1~019 in females (ratio = 1,8 to l). In 1940 there were 6.057 male and 2,029 female lung cancer deaths (ratio = 3 to ] ) A decade later there were 14,922 runic'and 3,391 female deaths (ratio 4.4 to 1). For 1960 the figures were 30.800 males and 5.000 females (ratio 6.2 to I ). The progressive widening of the sex ratio during recent d~cades has been interpreted by many as an indication of greater diagnostic accuracy inas- much as lung cancer has been found to be ptedomi- nandy a disease of males. In 1947. Clemmenses and Busk (77) observed that the lung cancer sex ratio in Denmark between 1936 and 1945 was 2 4 to I where- as the ratio in a special diagnostic clinic in Copen- hagen was 72 to I (This well known sludy was omitted in the report). Low sex ratios were found in many of the lung cancer studies in the 19th century and reflected the difbeultles in differentiating between primary and metastatic involvement The disparity in sex distribution has led to assumptions by those unfamiliar with the biological characteristics of the disease that predominance in the male is the result of heavier smoking habits. The validity of this in terpretation is readily challenged by ( I ) sex distrlbu- tiofl of lung cancer before the era of eigaretic smoking, and (2) the lack of any effect o[ the increased con sumption of cigarettes by females on the sex ratio. It is apparent from table IV that lung cancer was predomleandy a disease of males during an era in which cigarette smoking was virtually nonexistent, Most of the slodies show a ~ex ratio of at least 3 In 1, and some are comparable to ratios reported within the past decade 11 is of considerable interest that as early as 1895, Wolf's cases of "prim~tre lungerkrebs" from the Dresden municipal hospitals showed a dis- tribution of 6 to 1 favoring the male. and Passler, in 1896, found a r~tio of 3 to I in ~ collected series of 68 cases. In Hare's a "Pathology. Clinical History and Diagnosis of Affaedons of the Mediastinurn," pub- lished in 1889, the ratio was also 3 to I in a series of 122 cases. Harris (78) in 1892, made a combined study of St Bartholomew's Hospital and at the City of London Hospital for Diseases of the Chest and found a ratio exceeding 3 to 1. Lenhartz (79), in Leipzig, reported a ratio of 13 to I in 1899. and Feilchenfeld (52), in 190I, a rabo of 10 to I in Berlin. n~,,~, ~~ ~:.~ ~ L~.,~ ~ .... iill ....................... Predominance of lung cancer in the male in the 19th century was evident in both the individual nec- ropsy sludie$ and in the eolleeted series of cases. Adler's (57) collected series spanned more than seven decades and included cases from London, Prague, Berlin, Jena, Rome, Paris, Glasgow, St. Petersburg,

Zurich, and even Brooklyn. Despite the variety of sources the overall picture of male predilection was "similar to Raicbelman's (80) report in 1902 from the Fricdrichsbain Hospital in Betiin, Karrenstain's (56) in 1908, from the University of Berlin, and Briese's ( 8 I ) report, in 19 [6, from the Pathologic-Hydienie Institute of Chemnitz Weder (82), who was among the first American pathologists to become interested in bronchogenic carcinoma, reported, in 1913, a series of 87 collected cases with a sex rabo of 4 tn I th some instances it was pessiffie to follow the sex ratio of lung cancer thruugh several periods in the same institutions. In 1925, Bcrbliager (83) corn pleted a necropsy study previously originated by Laesehka at the Pathologic Institute of Jena. Lung cancer predominated in the male for each lustrum beginning with 1910~ and the ratio for the 15-year period was 4 to 1, Peters (8d) found a ratio of 3 to ] between 1905 08 at the Moahit and Urban Hospitals of Berlin and a ratio of 5 to I between 1917-22. At the Dresden and Friedrlehstadt hospitals and nursing bomes, Junghanns (85) found a ratio of 4 to 1 in the period 1893-97 and again in the period 1898- 1902. Dormanns (86) found a sex ratio of almost 7 to I at the University of Munich in 1901-11 and again in 1922-31. In Cologne, Eichengrum, and Esser (87) found a ratio of approximately 6 to I in the period 1902-14 and a 94o-1 ratio in 1914-19. One of the most significant studies was that of Passey and Holmes (88) from a group of teaching hospitals in England and Scotland covering a p~riod of more than 30 years and including the eras before and after the onset of heavy cigarette smoking. They found a sex ratio of 3 to I for each of the periods 1894 98, 1899-1903, and 1909-13. Between 1914- 18 the ratio was 4 to 1, and between 1919~23 it was again 3 to I. The highest ratio was 45 to ] which occurred in the period 1924 28. The affinity of lung cancer for the male was present before and after the onset of the cigarette smoking era. Some of the sex ratios in the 19th century re- ports were actually higher than those obtained in re- cent studies. The alleged correlation between smok- ing and male predominance of ]t~rlg cancer is based on a statistical association which ignores the fact that the disease favored the male when cigarette smoking was scarcely known. This point was emghac, ized in a report by lbrahim (89) from the Dacca Medical College Hospital in East Pakistan where he found a sex ratio of 9 to I in a group of 20 lung cancer eases which inehlded 14 nonsmokers, 4 occasional smokers and 2 heavy smokers. Cigarette smoking is a rarity in East Pakistan; the popular method of smoking is the hukga pipe with the water filler. Lung Cancel" hi Females There has been a progressive increase in cigarette consumption by women for more than four decades. A survey of smoking patterns by tile U S Bureau of the Census in 1955 stated thai 18 million women, 18 yeats of age and over, had repotted use of tobacco, and that 15 million had at one thne or another been regular cigarette smokers The corresponding figures fl~r men were 38 million and 31 millgm respeefvely The survey acknowledged that it had underestimated cigarette consumption by 15 percent on the basis of comparable tax data, and common observation sug- gests even a greater margin ol err~r According to the various prospective and retrll spective slodies cigarette smoking is alleged to be a potent carcinogenic agent. It is difficult In reconede this conclusion with the continued predominance o1 the disease in males despite the prodigious rise in cigarette consumption by females During all the years of increased female smoking there has not been the slightest narrowing of the sex ratio indicative of a relative increase of lung cancer among women. It has already been emphasized that the sex ratio has actually widened. Inasmuch as diagnostic facilities are equally available to both sexes, it is more logical to assume that the predominance in males is the result of sex predilection of the disease rather than of snmk- inc. Data from the U.S. National Office of Vital Stars tics show that lung cancer death rate between 1930 and 1958 increased 12 times for men and 3 dines for women, The greatest increase among women occurred between 1930 and 1950 during which time the rate rose from 1.5 per 100,000 to 4 per 100,000. Since 1950 the rate showed little increase The male female ratio with respect to mortality rates was 1.7 to 1 in 1930 and 6.6 to 1 in 1958 emphasizing the relative decrease among women d urine an era of heavy cigarette smoking. Mortality statistics in succeeding years show no indication of changing trends. The total increase of male lung cancer deaths between 1959 and 1962 was 4,665? The increase of female lung cancer deaths for the d-year period was 333 cases, averaging less than 85 eases per year. It is evident from the forgoing that the tremendous increase in cigarette snloking by women did not result in any comparable rise in lung cancer deaths or in reduction o{ the sex ratio of the disease. During the past few years the female lung cancer death rate has remained virtually stationary. It is scientifically in congruous to attribute the etiology of lung cancer to cigarettes when more than 15 million regular women smokers show an annual increase of ]ess than 85 ease~ The rapid adoption of smoking habits by women about 40 years ago would have resulted in a ctmtinaing epidemic nf lung cancer if tobacco were the causative agent. ' Mortagly statistics for 1962 ¢~timated by American Can- cer Soelet y

Age Distribution of Lung Cancer Observations for considerably mary than a century have shown that lung csncer is primarily a disease of older age groups with most cases occurring in the fifth, sixth, and seventh decades. Simons ( I I I ) in a eohec~ed sed:cs of more fhan 5,000 c~es found ~h~l 80 percent had occurred between 40 and 70 years of age. Lung cancer is uncommon below the age of 40 and rare during the first three decades o1 life¸ II is seen, in table V, that the peedilecti0n for older age gron~ W.~s noted co:~slslnndy in all the statistical re- ports before and after the onset of heavy cigarette smoking¸ ~ ~ + + • ~ + ..... ÷ • "lhe age distribution of lung cancer has a difinite relevancy to the etiological considerations, it has been alleged thai cigarette smoking produces a cumulative carcinogenic effect over a period of approximately two de, cadet. If this concept is v~fid, the se~nent of the cigarette smoking population which began smok- ing in the early teen ages should develop lung cancer before 40 years of age. No indication of this has OCCUrred in over 40 years of heavy cigarette smoking. Data from the World Heuldi Orgeulz~fions show a sin~ilar curve of age distribution in the United States, the United Kingdom, Germany, and Italy with the greatest number of eases between 55 and b4 years of age and only a small proportion before 44 years of age. The age ~t which the lung cancer pebe~t started to smoke has no relation ~ the age in which the disease was acquired. Patients who began smoking in the early teen ages and those who began 20 or more years later developed lung cancer during the same later period of life, namely, in the fifth, sixth, nr seventh decade. Furthermore, there is no evidence that the lung cancer patient who was a heavy smoker dcvelngcd the disclose at an earlier period of liic than the paffent who was a light smoker• If tobacco is a potent carcinogen the amount consumed and the duration of snloking should have some relationship to the onset of the disease Passeg (90) studied the snluking histories of 495 naen with lung cancer and found that they had acquired tile disease at approxiraately die same age regard]ess of whether smoking had been started at 6 years or at 41 years of age. It was also found that the nunther of cigarettes consumed daily did not influence the age of onset of disease; the light smoker was affected with lung cancer at the same age as the heavy smoker• Relation of Lung Cancer In Total Cancers During recent years lung cancers have comprised roughly 10 to 15 percent of all cancers diagnosed at necropsy and national mortality statistics have fol- lowed a similar pattern of distribution. It is of con- siderable interest, therefore, to observe that the same relationship was present before the era of cigarette smoking in certain European institutions with a long history of experience in lung cancer. Maxwell and Nicholson (75) found that between 1894 and 1898, the percentage of lung cancers to total cancers at the St. Bartholomew's Hospital was 7.83 Bonser (9i) found the percentgge of lung cancers at the Leeds General infirmary to be 13.6 in 1896, 8.1 in 1899, and 10.9 in 1910. FIampeln (92) found a 12 5 per- centage between 1894 and ] 899 at the Riga Hospital• In Dresden, Junghanns (85) found the percentage to be l 1.6 between 1893 and 1897 and 9.3 between 1 ggg and 1902. Seyfarth (93) found apercentage of I 1.23 in Leipzig between 191a and 1918 Jaffe and Stern- berg (94) found a 10.73 percet~tage in Vienna be tween the years 1915 and 1918, Lipschitz (95) found 11.3 percent between 1919 and ]923 at die National Research Institute in Zwickau. The large proportion of lung cancers among total cancers in these reporls is comparable to current relationships In institutions unfamiliar with lung cancer the ratios were Inwer in the preclgarette era a~d rose steedffy with increasing knowledge of the disease. Tite Pathological Evidence The Report of the Advisory Commdtee clted cer- tain plldioJogicul investigations (96, 98, ]25) as cor- roborative evidence of a relationship betweert smok- ing arid lung cancer. The results of these investiga- tions tlhowed that pathological changes occurring in the epithelial lining of the trachea and large bronchial tubes were much more common in s4nohe~s and in lung cancer paticntt than in nonsmokers. There were three types of changes observed and each was found

to increase with the numher of cigarettes smoked; anions the ex-cigar~t te smokel~, the ehange~ were far less notlceablc The specific pathological changes were: ( I ) loss el cilia of the surface cells, (2) hyper- plasia of the basal cell~, and (3) the presence of atypical cells. Because the changes were found ill smokers and in lung cancer patients the report de- duced that tbese changes, especially the advanced stages, probably represented precancerous condilions. In the judgment of the rcporl, the missing link between smoking and lung cancer had bce~ established. There is, however, a major fallacy in the assuulpdofl that these padlologieal changes fend to lung cancer¸ The changes that were found in ~mokers were ~ocated in those parts of the bronchial tubes in which lung cancer seldom develops¸ There were other significant inconsistencfes. The advanced pathological changes (carcinnma in situ) were found equally distributed among the large and small bronchial tubes whereas it is now well established (126, 127, 128) that lung cancer predomingntly originates in the smaller bronchi) or, more peripherally. In a recent patho- logical study of lung cancer cases operated at the Doctors Hospital during the past 15 years it was found that almost 90 percent of the cases had originated in the smaller bronchi or in the distal bronehioLes. "[he tendency for the agcged precancerous changes to occur in parts of the bronchial tubes seldom in- volved by lung cancer was also noted in a study (129) similar in scope to those cited in the report. It was found that most advanced changes occurred at the carina (the site of the largest bronchi) andthat the changes were least evident in the smaller bronchial tubes where lung cancer develops most frequently. It was also noted that "bronchial epithelium may be per- fectly normal in heavy smokers, even in men who have smoked from adolescence and reached a consumption of 40 clgarcttes daily." An interesting inadvertent contribution to this sub- ject is a study (130) of the distribution of cigarette smoke in the bronchial tubes¸ 11 was demonstatnd that the greatest con~ntration of smoke occurred in the large bronchi just below the trachea (where lung cancer is seldom found) alld th~,io~t concentration of smoke in the peripheral parts of the lung (where ldng cancer is cornmeal. This inverse r¢latlonship between the concentration of the inhaled ~mobe and the site of origin of lung cancer d~s not suppOrl a causal association. The o0nccntration of smnk~ wag greatest in the parts of the bronchi associated with the faust "precancerous" changes and the fewest lung cancers. There were other paradoxical results in the path~ logical studies cited in the report. The moderate and fees advanced "precancerous" conditions were found with greatest frequency in the large bronchial tubes where lung cancer seldom occurs, and wcn~ found with least frequency in the small tubes where lung cancer commonly ouculs. Tile finding of so many in stances of carcinoma in situ among the smokers was a mosl unusual observatlon. This lesion is gerlcragy conside~d a rarity in noflcancer patients and a sum mary (Jl)0) of eleven other studies comprising almost a Ihousafld noncanucr patients showed that carcinoma in situ had occurred in only one instance. Ibe re port saw fit to include in its evidence it study of certain cells in brorfchial epithe]fem which were pre- sumed to typify the ca-smoker "lfie allegro] ~geciflcity of these cells is extremely doubtrld. They ;ire degen crated cells o1 the epithelial lin feg in which the nucleus has contracted to assume an eccentric position becanse of accumulation of mucus These cells have been oh- served in many inflammatory conditions of the respir a tory tract. Thay may also be Iound in the nasal secrc tions of smokers, as nonsmokers, alike, who happen to have common colds. It must be emphasized that the alleged precancerous changes obeerved in the bronchial lining of smokers are the same changes that have been observed in patients with inflammatory diseases of the lung for many decades before cigarette smoking became wide- spread. As early as 1876, it was noted (97) that one of the complications of pneumonia was an atypical growth of the cells o1 the bronchial ]feing. These changes were also found in eases of chronic pulmonary tuberculosis and were well described in 1895 (32). Similar changes were noted (98) in bronchopneu- inertia and in influenza. The pathological alterations described in the report as identifiable with cigarette smoking occur commonly in a variety of pulmonary conditions including bronchfeetases, bronchopneu monia, tubereul~is, and ldngabscess They represent the natural protective response of the bronchial epi- thelial cells to injury and occur, alike, in smokers and in nonsmokers. The proliferation of cells from the germinal layer of the epithelium to form a new type of tissue (squamous metaplasia) is a natural process which attempts to recreatc the type of epithelium which existed in the embryonic state. These changes are very common and if they led to lung cancer, the incidence o1 the disease would be many times greater. There have been studies (117, I I g ) which showed a relationship between postinflam- matory changes and lung cancer but the latter changes were not observed in the studies associating smoking and lung cancer. In spite of all the evidence that negates the etiologi- cal significance o1 Ihe pathological changes found in smokers, the report concluded that these changes help to establish a causal relationship between smoking and lung cancer. The precise wording of the conclusion is most interesting: "It may be concluded * * * that some of the advanced epithelial hypcrplasdc lesions with many atypical ceils, seen in the bronchi of sortie cigarette smokers, are probably premaggeant." One may speculate as to why a conclusion couched in such 10