cigarette smokers (6, 42, 112), but the magnitude of this relation- ship is not as great as that with squamous cell cancer in man. Reduction in Tumorigenicity The importance of reducing total particulate matter in cigarette smoke is reflected in the dose-dependent results of the Auerbach- Hammond study. A major objective of experimental tobacco car- cinogenesis must be the reduction in the tumorigenicity of cigarette smoke and other tobacco products. In a recent article (320), Wynder and Hoffmann have reviewed the various methods applied to achieve this goal. Among these methods are the modification of the tobacco itself, the modification of the conditions of tobacco pyrolysis, the use of additives, and the use of filters. The use of filters should produce a reduction of particulate matter as well as of gas phase components. Bross (44) studied 974 cases of lung cancer at Roswell Park Memorial Institute and concluded that smokers who switched to filter cigarettes showed a decreased risk of developing lung cancer. However, even after switching, heavy smokers were still found to have a mortality risk five times that of nonsmokers. More recently, Wynder, et al. (324) reported on an interview study of 350 patients with histologically confirmed lung cancer and 552 age and sex-matched controls. They found that subjects who had switched from nonfilter to filter cigarettes ten or more years prior to the study incurred a lower relative risk of lung cancer at all consumption levels than that incurred by those who continued to smoke nonfilter cigarettes, The authors suggest that this difference in relative risk may be due to the lower “tar” content in filter cigarette smoke. Prospective studies concerning the effects of filter cigarette smoking are presently being conducted. Apart from variations in “tar” exposure due to filtration, it appears that different patterns of smoking result in the inhalation of varied amounts of “tar.” Graham, et al. (103) simulated dif- ferent inhalation patterns with the use of an analytic smoking ma-~ ~ chine. He found that smoking a given number of puffs over a long period of time results in greater “tar” retrieval than smoking them over a short period. Also, he observed that taking most of the puffs at the end of the cigarette results in the highest retrieval while taking most at the beginning results in the smallest retrieval. Complementing these observations is the same author’s case/con- trol study (102) of 183 men with lung cancer and 161 men with diseases not related to tobacco smoking. He found that the lung cancer patients had significantly greater high “tar” yield cigarette smoking patterns than the controls. The risk of lung cancer was found to increase with the increase in mean number of puffs per 301 cigarette, the average length of time taken to smeke wo cigarette (except in the highest number of puffs category), wid the Taking of more puffs at the end of the cigarette. These findings, and those of the study of Averbrehy, 4 eh ELLY, add further support to the dose-response relationsh*: Between ung, cancer and total cigarette smoke condensate €Xposire.. SUMMARY AND CONCLUSIONS 1. Epidemiological evidence derived from a numer ei PrSaset- tive and retrospective studies coupled with experimminé zd! path: ological evidence confirm the conclusion that cigarette: smoking; is: the main cause of lung cancer in men. These studies seseuk titae che risk of developing lung cancer increases with the nuzd y of. cigar: ettes smoked per day, the duration of smoking, ant aeler miga- tion, and diminishes with cessation of smoking. 2. Cigarette smoking is a cause of lung cancer sen ceorpere leat accounts for a smaller proportion of cases than in rvs “S2e moc tality rates for women who smoke, although signdiwads hégier than for female nonsmokers, are lower than for mex win smoke. This difference may be at least partially attributed to differenve 12 exposure; such as, the use of fewer cigarettes per diy the use. of filtered and low “‘tar” cigarettes, and lower Jevels af inhalation. Nevertheless, even when women are compared wit men who: an- parently have similar levels of exposure to cigarette: smoke; the mortality ratios appear to be lower in women. 3 The risk of developing lung cancer among pie. andor cigar smokers is higher than for nonsmokers but significr2tayy ippsen than. for cigarette smokers. 4. The risk of developing lung cancer appears to@higher among smokers who smoke high “tar” cigarettes or smoke Gr suv a. mare ner as to produce higher levels of “tar” in the inhazd! smoke. 5. Ex-cigarette smokers have significantly lower deatir rates for lung cancer than continuing smokers. There is evidence to support the view that cessation of smoking by large numkevs: af uigarette smokers would be followed by lower lung cancer dpath: nates. 6. Increased death rates from Jung cancer have teen sbserved among urban populations when compared with pepuiatans from. rural environments. The evidence concerning the rule. ef sir peliv- tion in the etiology of lung cancer is presently incommusive. Facters such as occupational and smoking habit difference: mayr ad8G. con- tribute to the urban-rural difference observed. Detaed. epidemic- logic surveys have shown that the urban factor exexts 2 small influence compared to the overriding effect of cigavede smoking in the development of lung cancer. 302 7. Certain occupational exposures have been found to be asso- ciated with an increased risk of dying from lung cancer, Cigarette smoking interacts with these exposures in the pathogenesis of lung cancer so as to produce very much higher lung cancer death rates in those cigarette smokers who are also exposed to such substances. 8. Experimental studies on animals utilizing skin painting, tracheal instillation or implantation, and inhalation of cigarette smoke or its component compounds, have confirmed the presence of complete carcinogens as wel] as tumor initiators and promoters in tobacco smoke. Lung cancer has been found in dogs exposed to the inhalation of cigarette smoke over a period of more than two years. CANCER OF THE LARYNX Cancer of the larynx is a disease which predominantly affects males in the 55 to 70 year age group. In 1967, a total of 2,468 males and 329 females died of laryngeal cancer in the United States. With the development and application of more effective therapy during the past 30 years, the death rate for cancer of the larynx appears to be dropping slightly (282, 289); however, the incidence con- tinues to rise. Figures from the Connecticut Cancer Registry (88) show that the age-adjusted incidence per 100,000 population of cancer of the larynx for males rose from 3.0 in 1950 to 5.6 in 1961. EPIDEMIOLOGICAL STUDIES A number of epidemiological studies have investigated the rela- tionship between smoking habits and the development of cancer of the larynx. The major prospective studies, as outlined in table 20, show that smokers of cigarettes run an approximately six-to- tenfold risk of dying from this form of cancer as compared to non- smokers. Smokers of pipes and cigars incur a three-to-sevenfold risk. The retrospective studies listed in table A21 uniformly show fewer nonsmokers and more smokers among cases with cancer of the larynx than among matched controls. Table A22 summarizes— - the relative risk ratios derived from the retrospective studies. The wide variation is due to a number of factors, including type of popu- lation and interview technique. But, in general, the magnitude of most of these ratios is of the same order as in the prospective studies. Wynder, et al. (312) have distinguished between cancer of the intrinsic and extrinsic larynx. Tumors arising on the vocal cords are classified as intrinsic and constitute approximately 70 percent of the lesions, The extrinsic larynx is composed of those sections of the larynx excluding the vocal cords and may also be referred to as 303 HOE TABLE 20.—Laryngeal cancer mortality ratios (Actual number of denths shown in parentheses)?! SM = Smokers. NS = Nonsmokers. Prospective atudics Number Autbor, of year, Number and Data Follow. laryngeal Cigarcttes/day Pipes, clgars Comments country, type of collection un cancer reference population years denths Hammond 187,783 white Questionnaire 3M 24 Chyarette smokera 17/24, Cigar Data referring to mortality and males 60-69 and follow- \ SM ..24 3/24 ratio Included cancer of Norn, years of age up of death NS .. 0 Mixed esophagus and mouth, 1958, {n 9 states. certificate, 4/24 ULS.A, (120), Doll and Approximately Questionnaire 10 16 All amokera by amount Pipeand cigart + Includes data on ex- Rill, 41,000 male and follow- SM ,.16 in grama NS .... 1,00 smokers of pipes and cigars, 1964, British up of death NS .. 0 NS vivseeece nee ne SM .... 5.00 No NS died of laryngeo- Great physicians, certificate. 1-14 . « 1.00 tracheal cancer, therefore Britain 16-24 pereceeeee 1,00 1-14 gram SM act oa 1.00 (76), OG eee e seen 7.60 standard, Data combine laryngeal and tracheal carcinoma. Kahn U.S. male Questlonnalre By 54 NS oe. eee eae 1.00 (8) Pipe Refers to current cigarette (Dorn), veterana, and follow- SM ..61 | +» 827 (1) NS .... 1.00 (3) smokers only. 1966, 2,265,674 up of death NS .. 3 10-20. .. 8.45(10) SM ....10.83 (6) ULS.A, person years, certificate, 21-39 «28.62(11} Pipe and cigar (139), >, 1.18.85 (3) NS... 1.00 (8) All viv ceec ev eenes 9.95(28) SM ..,. 7.28¢11) Hammond, 440,558 males Interviews 4 57 NS ocisceeeeeeeaee 1,00 (8) Pipeand cigar Male data only. 1966, 562,671 fo by ACS SM ..B4 SM (age 45-64) ., 6.09(82) NS .... 1,00 (8) Pipe and cigar data refer to U.S.A. males 35-84 volunteers, NS .. 3 | SM (oge 65-79) .. 8.99(18) SM ..,. 8.87 (4) males 66-84 yoarp of age. (118), years of age in 25 states, SOE TABLE 20.—Laryngeal cancer mortality ratios (cont.) (Actual number of deaths shown in parentheses)? SM = Smokers, NS = Nonsmokers, Prospective studies Number Author, of year, Number and Data Follow. laryngeal Clgarettes/day Pipes, cigars Comments country, tyne of collection up eancer reference population years deaths Weir and 68,153 males Questionnaire 5-8 1k NS icceee ee ee eee OS No nonamokers died of Dunn, in various and follow- SM ..11 SLO Lis eeeee sees 1,00 lurynucal carcinumna, 1970, occupations up of death NS .. 0 20 ieee ee 6.99 therefore £10 amoher set U.S.A, in California. certificate. P90 eee eee eee B84 as 1,00 atundard, (306), NS includes plpe and clyar smokers, SA includes ex-amokers. ' Unless otherwise specified, disparities between the total number of deatha and the sum of the individual smoking categories are due to the exclusion of either occasional, miscellaneous, mixed, or ex-amokers. the hypopharynx. These authors noted that the percentage of heavy smokers among the patients with cancer of both the extrinsic and intrinsic larynx was significantly greater than that among controls. However, it 1s of interest that the excess risk of laryngéa] cancer among cigar and pipe smokers in this study could be attributed to the extrinsic laryngeal group. As in studies of oral cancer, it appears that alcohol consumption should also be taken into account in studies of Jaryngeal cancer. Wynder, et al. (312) reported a significantly increased risk of extrinsic cancer among those with alcohol intake above 7 ounces of whiskey per day. With less than this amount, no increased risk was evident. Schwartz, et al. (248), noted no effect in relation to alcohol intake. Further research into the interaction of these two variables is necessary. PATHOLOGICAL STUDY Auerbach, et al. (9) studied histological changes in the larynges of 942 men, age 21 to 95, who were autopsied at a single hospital between 1964 and 1967. Cases of primary cancer of the larynx were excluded from the study. Smoking histories for all cases were obtained from family members of the deceased by trained inter- viewers. The randomized histological sections were graded by one observer. Tables A23 and A24 summarize the findings in the true vocal cord. Of the men who never smoked, 75 percent had no cells with atypical nuclei, only 4.5 percent had sections with areas con- taining 60 to 69 percent of cells with atypical nuclei, and none had a higher percentage. The 116 ex-smokers had laryngeal histology similar to that of the nonsmokers, as far as atypical nuclei were concerned. However, disintegrating nuclei were found in 40.5 per- cent of the ex-cigarette smokers and in only 0.4 percent of the remaining cases. Only one of the 94 cigar and/or pipe smokers had no atypical cells. Three had carcinoma in situ, and one case had a section showing early Invasive primary carcinoma. The highest percentage of atypical cells was found among the cigarette smokers. The proportion of cases with a high degree of cellular change increased with increased daily smoking. None of the pack-or-more-a-day smokers was free of atypical nuclei in the laryngeal epithelium. Of those who smoked two or more packs per day, 85 percent had lesions with 60 percent or more atypical cells as compared to 4 percent of the nonsmokers. Between 10 and 18 percent of the cigarette smokers had areas of carcinoma in situ, and 4 of the 644 cases showed early microscopic invasion. The thickness of the basal level of the true vocal cord was also directly related to the amount smoked. 306 EXPERIMENTAL STUDY Dontenwill (76) has recently reported the development of an effective and practicable method by which small rodents (ham- sters, rats, mice) can be exposed to long-term passive inhalation of cigarette smoke in a manner which circumvents the fatal effects of acute toxicity which ruined earlier attempts but allows for a dosage of smoke great enough to induce the development of chronic patho- logical changes. The Syrian Golden hamster was found to be the most suitable species for such inhalation experiments for several reasons: its resistance to pulmonary infections, its resistance to the effects of nicotine as compared to that of rats or certain strains of mice, and, especially, its susceptibility to develop tracheobronchial cancers after treatment with carcinogens, in contrast to its almost total freedom from the spontaneous development of these tumors. Dontenwill demonstrated that the concentration of deposited cigarette smoke was greatest in the hamster’s larynx as compared to the other portions of the exposed respiratory tract (table 25), and that the laryngeal epithelium was the tissue which underwent the greatest smoke-induced histological changes. In studying the changes in the larynx, the author differentiated five stages of epithelial change, using as his reference the Atlas of Tumor Pathology of the Armed Forces Institute of Pathology (5). Table 26, quoted by Dontenwill, describes the five types of change. They range from benign, such as epithelial hyperplasia, to pre- malignant, exemplified by pseudoepitheliomatous leukoplakia, The results of the inhalation experiment are presented in figure 4 in which a dosage-related increase in the severity of the epithelial changes is represented in graphic form. The author also reported, and depicted with photomicrographs, the finding of an early inva- sive squamous cell carcinoma. This form of cancer is the predomi- nant type involving the human larynx. SUMMARY AND CONCLUSIONS 1. Epidemiological, experimental, and pathological studies sup-— port the conclusion that cigarette smoking is a significant factor in the causation of cancer of the larynx. The risk of developing laryngeal cancer among cigarette smokers as well as pipe and/or cigar smokers is significantly higher than among nonsmokers. The magnitude of the risk for pipe and cigar smokers is about the same order as that for cigarette smokers, or possibly slightly lower. 2. Experimental exposure to the passive inhalation of cigarette “. smoke has been observed to produce premalignant and malignant changes in the larynx of hamsters. 307 TABLE 25.—Deposition of C-labeled smoke particles in particular regions of the respiratory tract* Traced Traced Estimated Deposition Proportional deposition radio- radio- of areaofthe in relation Organ activity Organ activity particles respiratory to the (nCi) (nCi) (%) tract proportional area Head and palate ... 6-12 Head, palate 5.5 37.4 Tongue ..-.------- 0.41 Oralcavity 16 10.9 in total. Larynx .....------ 0.39 0.1-0.3 X561-187 Trachea .....--.-- 0.26 7.6 (traced) 5L.7 0.6 X62.3 Lungs ...-.------- 6.95 1000 Xl Total ......- 14.12 714.7 100.0 1 Cigarettes labeled with 11C_1-n-hexadecan; data represent calculated from surface distribution in the head. 2 The value of 14.7 contains 0.58 nanocuries as estimated from mean values from 10 animals, quantity of deposition in the nontraced oral cavity regions (calculated as to proportiona) area}. Sounce: Dontenwill, W. (76)- 368 60€ TABLE 26,—-Classification of the five registered stages of epithelial changes at the larynx’? Dyskeratogis (pre+ mature atypical Acanthosis (thicken- Hy perkeratosss Parakeratosis (ine cornification : Stage ing of stratum increased complete cornifica- changes in the Mitosis spinosum multi- cornification tion of nuclelin nucleus prolifera. cellular layer) (stratum corneum) the stratum comeum) tion 7 {pe basal ayar) 1, Pachydermia (epithelial hyperplasia) ........ Leeeaee + + t t t 2. Leucoplakia ........, ees sees tases at ee nes + + t t t 3. Verrucous leucoplakia .........000. Feb eeeeaee teens + + + t z 4. Papillomatous leucoplakia 0.0... ...ccsaes cece, vteee + t t ++ t 5. Pseudoenitheliomatous leucoplakia pee eey yyy Seeeaey + + + p+ + Symbols: t+ = negative; t=ominimal; += weak; +-+ = medium: +++ = atrong. 7From Atlas of Tumor Pathology of the Armed Forces Institute of Pathology. Source: Adapted from Dontenwill, W. (78), TOTAL 146 34 as as | 6 10 { 7 12 4 5 17 4 | ° . eoue ye o° ° ose ee eene je oeenle lee le jee eesion . ee en . . HISTOLOGY Pa%s%e te e é peee 2 4 6 8 10 12 14 16 18 20 22—> 23 SMOKE EXPOSURE, months e =ONE ANIMAL 6 =ANIMAL LIVING O=LARYNX CANNIBALIZED Ficure 4._Effects of chronic cigarette smoke inhalation on the hamster larynx. Review of the results of the inhalation experiments: number of smoke-ex- posed animals with and without changes in the larynx, duration of smoke exposure, and number of animals still alive. Source: Dontenwill, W. (76). ORAL CANCER The cancers included in this category are those of the lips, tongue, floor of the mouth, hard and soft palate, gingiva, alveolar mucosa, buccal mucosa, and oropharyns. It is estimated that 15,000 of these cancers will be diagnosed in the United States in 1970, accounting for about 2.5 percent of the estimated 600,000 malignant neo- plasms reported (289). A variety of histological types of malig- nant neoplasms can affect these tissues, but squamous cell car- cinoma is by far the predominant type, accounting for about 90 percent of the cancers. The incidence of and mortality from oral cancers has remained steady over the past 20 to 30 years. The Connecticut Cancer Reg- istry (88), which is a fairly reliable index of incidence, noted that the incidence among males remained between 15.8 and 16.3 per 100,000 population during the years from 1950-1961, Examination of mortality rates over the past 20 to 30 years (282, 289) reveals a similar constancy. The apparent lack of change in mortality from oral cancer in 310 contrast to the sharp increase that took place in lung cancer rates in those years is probably due to several of the following factors. First, pipe and cigar smohing are both significantly related to can- cer of the oral cavity, and the increase in cigarette smoking among men, noted between 1920 and 1955, has been, to a large degree, accompanied by corresponding reductions in the use of pipes and cigars. Second, aside from the various changes which the Interna- tional Classification of Diseases (ICD) had undergone during that period, the diseases discussed above are recorded in ICD Codes 140-148 which include some neoplasms not found to be related to the use of tobacco. The various sites of cancer themselves do not contribute equally to the overall rate and are subject to widely dif- ferent cure rates, so that their contributions to the total incidence rate is different from their contribution to the overall mortality rate from oral cancer. Although more than 20,000 cancers of the oral cavity were estimated as newly diagnosed in 1967, the total number of individuals recorded as dying from oral cancer during that year was only 6,718 (289). Ora] cancer occurs predominantly in people of the middle and older age groups. More than 90 percent of all oral cancers occur in persons over age 45, with the average age at time of diagnosis approximating 60. Although the majority of oral cancers occur in men, there is recent evidence that the ratio of males affected to females affected is decreasing (257). EPIDEMIOLOGICAL STUDIES The use of tobacco in various forms has been associated with the development of cancer of the ora] cavity and pharynx. The studies in this area of concern are truly international, many having been carried out in Asian nutions as well as in the West. The major prospective epidemiological studies have found in- creased rates of these cancers for cigarette smokers as well as for pipe and cigar smokers (see table 27). Pipe smoking, per se, has_ _ long been recognized as a cause of lip cancer (291). The methodol” ~ ogy and results of the numerous retrospective studies are sum- marized in tables A28 and A28a. These studies almost uniformly show significant relationships between the various forms of tobacco use and crnecrs of the oral cavity and pharynx. Studies in Asian notions have examined the prevalence or inci- dence of premaliynant change, such as oral leukoplakia, as well as that of cancer of the oral cavity. In many of these studies, forms of tobacco use not prevalent in Western countries have been investi- gated, including reverse smoking (in which the lighted end of the cigarette is kept in the mouth close to the palate) and the chewing 311 ZTE TABLE 27.—Oral cancer mortality ratios—prospective studies (Actual number of deaths shown in parentheses) NS = Nonsmokers. SM< Smokers, Author year, Number and Data Follow- Number country, type of collection up years of Cigarettes Pipes, clgars Comments reference population deaths Hammond = 187,783 white Questionnaire M4 56 20/56 Pipe Mixed Data referring to mortality and males in 9 and follow-up 1SM. GL 5/56 21/66 ratlo do not include cancer of Norn, States 60-69 of death NS. ..3 Cigar larynx and esophagua, 1968, yeors of age, certificate, 6/66 ft Excludes two oceanlonal U.S.A, only emoubess, (180), Dol} and Approximately Questlonnatre 10 19 Allamokera by amount Pipe and cigar No NS dhed of oral cancer, Hall, 41,000 male and follow-up SM ..19 in graeme NS ovisseee 1.00 therefore 1-14 gram 1964, British of death NS 6.0 NG cicccerceeveee —_ SM... seee 1.00 smoker act as 1.00 Great physiclana, certificate. 1-4 veeaee 1,00 standard. Britain 16-24 saves 0.25 (74). >25 . 5.25 Kahn U.S. male Questionnaire 84 61 NS wee ee eee 2,00(11) Pipe Data do not Include pharynx, (Dorn), veterans, and follow-up SM ..50 tCigs/day 1-9 .... 0.86 (1) NS ....... 2.00(41) f Refers to current clgarctte 1966, 2,265,674 of denth NS wi 10~20 . 2.93(13) SM oo... 3.12 (4) smokers only. ULS.A. person years, certificate, ZN-39 cece av eevee TOA (20) Cigar (139). >49 -. 673 (3) NS w..eaee 1.00(41) AN teens » 4.00(397) SM wi.ccaee GUE (9) Hammond, 440,468 males Interviews by 4 05 NS ova. eeee receaes 1.00 (7) ¢ Plpe and/or 1 Male data only. Pipe and 1966, 562,671 fomalea ACS volunteers SM ..BB OSM (nage 46-64) ..,. 9.00(63) chuar. clyor data refer to nmintes USA, SG-84 yours of NS ..7 SM (age 65-10) .. 2.93(26) NS ....e. 21,00 (7) 6b-B4 yours of axe. (41s), age in 26 States. ‘ SM oo... e, 4.9416) Welrand 68,163 males Questionnaire 6-8 19 1.00 SM Includes ex-amokers, Dunn, in various and follow-up .. 3.69 NS includes plpe and 1070, occupations of death 2117 elgar smokers, ULS.A, in California. certificate. 5.52 (906), . 2,96 of “pan” or “Nass,” which are mixtures of tobacco with either betel nut or lime ash, and other ingredients (241, 255, 256). Snuff dipping, a habit in which snuff is placed in the gum and retained there for prolonged periods, has also been associated with the development of oral cancer (198, 210), as has the chewing of tobacco (124, 193, 241, 298). ‘The risk of developing a second primary mouth or throat cancer, aftér the recognition of the first primary cancer, has been found to be greater in continuing smokers than in those who quit smok- ing. All of the patients studied by Moore (190) were asymptomatic for at least three years following the treatment of the first cancer. Of the 117 patients with adequate smoking histories, only 4 of 43 (9 percent) who quit smoking developed a new primary cancer. On the other hand, 27 of 74 (36 percent) who continued to smoke developed a second primary cancer. However, a study by Castigliano (53) of patients treated for oral cancer did not show a greater risk of a second primary among continuing smokers. In this study, 5 of 26 (19 percent) of those patients who did not quit smoking developed a second primary cancer as compared to 9 of 51 (18 percent) of those who did quit. The rate of quitting smoking in the two studies is markedly dif- ferent (36 percent in the Moore study and 62 percent in the Casti- gliano study). From the data presented in the two papers, it is not possible to evaluate the other significant ways in which the pop- ulations may have differed. Keller (140) studied 408 males with histologically confirmed Squamous cell cancer of the mouth or pharynx. This author dealt with the question of recurrent tumors in a somewhat different manner. The patients were observed for the development of a sec- ond or third primary cancer at an anatomically discrete site of the mouth and pharynx within a median period of three years after the first cancer. He found that a second or third cancer (termed a coexisting cancer) developed in 28 of the 408 cases. Among these 28 cases with 33 coexisting neoplasms, 21.7 percent were heavy — - smokers, but among their matched controls, there were no heavy smokers. Coexisting cancers were most commonly found on the soft palate, an anatomical site that is in dire¢t contact with the main- stream of tobacco smoke. More recently, Wynder, et al. (315) studied 63 male and 23 female patients with multiple primary cancers of the mouth and pharynx. They observed that heavy smoking prior to the develop- ment of the oral cancer was associated with a greater likelihood of developing a second primary. Also, continued smoking after the first primary was found to have a significant association with the occurrence of a sé€cond primary. 313 With or without smoking, use of alcohol appears to contribute to the development of oral cancer (124, 140, 183, 297, 322). In a study of male vetérans, Keller (150) found that heavy smoking and heavy drinking were associated with cancer of the mouth and pharynx. No studies are presently available which determine the relative contributions and possible interactions of heavy smoking, heavy drinking, and concurrent nutritional deficiencies in the etiol- ogy of these cancers. EXPERIMENTAL STUDIES In 1964, the Advisory Committee to the Surgeon General on Smoking and Health (291) reported that cigarette smoke and ciga- rette smoke condensates had failed to produce cancer when applied to the oral cavity of mice and rabbits or to the palate of hamsters and that the oral mucosa appears to be resistant in general to can- cer induction even when highly active carcinogens such as benzo- [aJpyrene are applied. Some of the difficulties in experimental de- sign were attributed to the fact that mechanical factors, such a3 secretion of saliva, interfere with the retention of applied carcino- genic agents on the tissues of the oral cavity and pharynx. Positive results with certain carcinogens have, however, been obtained in the hamster cheek pouch, but it has also been pointed out that the cheek pouch lacks salivary glands and that its structure and func- tion differ from those of the oral mucosa. The majority of these studies are outlined in table A293. Although cigarette smoke condensate acts as a complete carcino- gen on mouse skin, the work of several authors (319) supports the concept that cigarette smoke contains cancer promoters that may be of special importance, particularly in oral carcinogenesis. Elzay (90) has reported that whole cigarette smoke is a promoting agent for the hamster cheek pouch. More importantly, regarding the chewing of tobacco, Bock, et al. (27,30), Van Duuren, et al. (294), and Wynder and Hoffmann (321) have shown that unburned to- bacco products contain tumor promoters that might contribute to the promoting activity of the smoke. . Roth, et al. (226, 227) have shown that the dye-binding capacity of the DNA of oral epithelial cells is significantly enhanced in cigarette smokers in contrast to nonsmokers, probably reflecting an increase in the DNA content of oral epithelial cells in smokers. Smokers had values of dye-binding capacity intermediate between nonsmokers and 21 patients with proven oral cancer. Those smok- ers who refrained from smoking for up to six months showed a significant decrease toward more normal! values. 314 SUMMARY AND CONCLUSIONS 1. Epidemiological and experimental studies contribute to the conclusion that smoking is a significant factor in the development of cancer of the oral cavity and that pipe smoking, alone or in conjunction with other forms of tobacco use, is causally related to cancer of the lip. 2. Experimental studies suggest that tobacco extracts and tobacco smoke contain initiators and promoters of cancerous changes in the oral cavity. CANCER OF THE ESOPHAGUS Esophageal cancer accounted for 4,306 deaths among American males in 1967 and 1,321 deaths among females. The death rate from esophageal cancer has remained relatively constant since 1949. EPIDEMIOLOGICAL STUDIES The major prospective epidemiological] studies (table 30) have indicated a significant relationship between smoking and esopha- geal cancer. Overall mortality ratios for male cigarette smokers range from 1.74 to 6.17. There are insufficient data concerning females for establishing firm conclusions. A number of retrospective studies concerning the relationship of smoking and esophageal cancer are outlined in table A351 and A31a. Smokers incur risk ratios ranging from 1.3 to 6.6 when compared with nonsmokers. As in studies of oral cancer, the effect of alcohol consumption must be taken into account in studies of esophageal cancer. Because a relationship between alcohol consumption and tobacco use is known to exist, Wynder and Bross (310) analyzed the association between tobacco consumption and esophageal cancer after adjust- ing for alcohol intake. They found that in the absence of alcohol - consumption, there was no association between the use of tobacco and esophageal cancer but that in the presence of alcohol consump- tion, an increasing relative risk with increasing number of ciga- rettes smoked was apparent, as well as an association between cigar and pipe smoking and esophageal cancer. More recently, Takano, et al. (272), ina retrospective study of 200 patients with esophageal carcinoma, found an increased risk with smoking which was magnified by increased alcoho] consump- tion. Martinez (183) analyzed the association of tobacco usage and esophageal cancer after controlling for age, sex, and alcohol consumption. Increasing relative risks with increasing tobacco use 315 gtt TABLE 30.—E'sophageal cancer mortality rattos—prospective studies {Actual number of deaths shown in parentheses}? SM = Smokers. NS = Nonamokers. Author, Number of year, Number and Data Follow- esophageal country, typeof collection up years cancer Cigarettes/day Pipes, cigars Comments reference population deaths Hammond 187,783 white Questionnaire 3Y, 34 Cigarette smokers Pipe Mixed Data referring to and males ing and follow-up NS ... 1 16/33. 2/33 elgarette mortality ratios Horn, States 60-69 of death SM... 433 Cigar smokers included cancer 1958, years of age. certificate. 2/38 13/33 of mouth U.S.A. and larynx. (220). Doll and Approximately Questionnaire 10 29 All emokers by amount t+Pipe and cigar tIncludes ex- Hill, 41,000 male and follow-up in grams NS ..., 1,00 amokers of pipe 1964, British of death NS w...se 1,00 SMo.... 2.00 and cigars. Great physicians. certificate,