Tobacco Documents Online

A STUDY OF ETIOLOGICAL FACTORS IN CANCER OF THE ESOPHAGUS ERNEST L WYNDER, DLD., AND IRWIN J. BROSS, PH.D.* D URING RECENT years, much interest has been shown in environmental factors that might influence the development of chronic diseases. The purpose of such investi- gations has been not only to comprehend more fully the pathogenesis of such diseases but also to uncover factors that, if removed, would bring about their reduction. Our group has been interested in the epidemiology of cancer front this point of view. The present report represents an investigation of cancer of the esophagus. INCIDENCE DATA • Important indications as to the role of en- vironmental factors in the development of cancer are obtained from the relative frequen- cies with which these cancers occur in differ- ent populations. With respect to cancer of the esophagus, the following observations seem of importance. Geographical. In the United States and throughout most of the Western World, can- cer of the esophagus is relatively uncommon. In 1957, $.567 deaths from cancer of the esoph- agus occurred in American male patients and 1,011 in American female patients.2s The average age-standardized death rate for cancer of the esophagus between 1952 and 1956 was 4.2 per 100,000 population for white male individuals and 1.0 for v`.tite female individ- uals. Similar death rates were reported in Western Europe (Table 1). The highest rates among male individuals are as follows: in Switzerland, 16.2; in France, 12.7; and in Fin- land, 10.9. The rate for women in Finland is 6.3, and that for women in Chile is 5.556 In From the Section of Epidemiology, Division of Pre- ventive Medicine. Sloan-Kettering Institute for Cancer Research, Ncw York. N.Y. Wegca tefully acknowledge the invaluable aid given us in the preparation of this manuscript by Mr. John Ondcrdonk, Mrs. Charlotte Zweifach, and Mrs. Ishu KriFalani, as well as that given us by Dr. Abraham Lilienfeld and Mr. Nathan Mantel in their critical appraisal of this manuscript. • Present address: Department of Statistics. Roswcll Park Memorial Institute, Buffalo, N.Y. Received for publication \tay 20, 1960. Tesus I .aGEST.9NDARDIZED DEATH RATES FOR CANCER OF THE ESOPHAGUS PER 100,000 POPULATION (1952-1956 AVERAGE)' Pt. sex Country Male Female Sex ratio $witzerland 16.2 2.0 8.L1. France 12.7 1.4 9.1:1. Finland 10.9 6.3 1,7:1. Chile 10.8 5.5 2.0:1. Japan 8.9 2.9 3.1:1. United States (nonu-hite pop.) 7.9 1.9 4.2:1. Scotland 7.1 4.4 1.6:1. Austria 6.3 1.3 -4.8:1. England & R-ales 6.1 2.9 2.1:1. Ireland 5.4 4.1 L3:1. Germanv 5.2 1.6 3.3:1. New Zcaland 5.2 2.9 1.8:1. Portugal 5.1 2.1 2.4:1. Northern Ireland 4.8 4.0 1.2:1. Denmark 4.8 2.1 2.3:1. il'etherlands 4.7 2.0 2.4:1. Italy 4.7 1.1 4.3:1. Australia 4.7 1.6 2.9:1. Belgi «m 4.7 1.6 2.9:1. United States (white pop.) 4.2 1.0 k.2:1. YonvaX. 4.1 1.2 3.4: i. Canada 3.8 1.6 2.4:1. Venezuela 3.5 2.8 1.3:1. Sweden 3.4 1.8 1.9:1. 'Derived from a report of the World Health Organi- zation.'^ comparing incidence data, one must, of course, recognize possible differences in reportability and nomenclature that may occur between countries. In the Philippines, cancer of the esophagus is less coumion than it is in the United States. It is relatively more conunon among the Chi- nese than it is among the Filipinos" Cancer of the esophagus is unusually common antong the native population of Cura4ao.ls It is also more common than expected among Bantu male individuals when they are compared to a Danish population of similar size, but the rates for women are similar.1e Burrells• 4 found esophageal cancer ammng some Bantu groups to be high, particularly in East London, South Africa. In China, esophageal cancer ha.c been reported to account for 50% of all cancers of the gastrointestinal tract and is thought to occur more commonly in the northern 0 889 _ Cancer 14

• 390 CANCER Afarch-April 1961 Vol. 14 areas.%a Khanolkar23 reported that the rate for of 36 patients with cancer of the stomach and cancer of the esophagus among Indian male 18 patients with cancer of the esophagus. individuals was 7.9% of all carcinoma cases at Craver7 found the following factors, in de- the Tata Memorial Hospital, Bombay, India, from 1941 to 1945. The rate for women dur- ing the same period was 4.1°jo?s Similar data have been reported from the Netherlands An- tilles.14 Sex Differences. Sex ratios for cancer of the esophagus vary from 1.2 to 1. to 9.1 to 1. (Table 1). In the United States for the nonwhite popu- lation, the sex ratio is 4.2 to 1. In Finland the ratio is 1.7 to 1.; in Switzerland, 8.1 to 1.; and in France, 9.1 to 1. Esophageal cancer is pri- marily a disease of inen. The one exception is in countries of northern latitude in which the disease is relatively more common in women, particularly insofar as the upper portion of the esophagus is concerned. A similar trend is also noted in the United States, where cancer of the upper third of the esophagus is rela- tively more common in women as compared to men.42 In the Philippine Islands the sex ratio is 1.5 to 1., and in CuraSao the sex rates are almost equal.ts, 34 Among the Bantus, the disease is more common among male individ- uals, except in isolated areas where it is more common among women?•'+ Change in Incidence. Unlike the marked in- crease in lung cancer in most \.Bestern coun- tries, there has been no marked change in the incidence of esophageal cancer. In fact, in the United States the incidence has been most stable. The same applies for England, where the frequency of cancer of the esophagus has not changed significantly during the last 2 decades.32 Income Group. Though data are not pre- cise, it has been a general clinical impression that cancer of the esophagus is relatively more common in the lower income groups. Data from South Carolina by Pettita" showed that esophageal cancer is particularly prevalent in the Negro. Parker and his associates29 reported that while in 1950, 5,934 Negro and 9,086 white patients were admitted to their hos- pital, during the same period 3 white and 21 Negro patients with cancer of the esophagus were admitted. Death rates in the United States demonstrate the same point (Table I). SUSPECTED ETIOLOGICAL FACTORS In 1932, Craver7 investigated a variety of factors in patients with cancer of the esoph- agus. This study consisted of a detailed history scending order, of importance in the develop- ment of cancer of the esophagus: tobacco, al- cohol, lack of water, decayed teeth, lack of teeth, and use of drugs (cathartics). In more recent literature, the following additional ref- erences are pertinent. Tobacco. In 1937, Ahlbom,' in Sweden, grouped esophageal cancer patients together with those having cancer of the pharynx and larynx and found a higher percentage of smokers in the study group than among the controls. In 1957, Wynder, Hultberg, Jacobs- son, and Bross"a showed similar findings among Swedish male patients with esophageal cancer, whereas among the female patients no relationship was found to smoking habits. Sadowsky, Gilliam, and Cornfield33 found that 96.2% of 104 esophageal cancer patients smoked, as compared to 86.8°j'o of 615 control patients-a difference significant- at the 5% level. They found no difference, however, in the type of tobacco smoked. Schwartz, Denoix, and Anguera?s in a thor- ough study in 1957 on the influence of tobacco and alcohol on different cancers, found that among 200 patients with cancer of the esoph- agus, 98°fo were smokers in contrast with 83°0 among the controls-a difference significant at the 1% level. However, they found no dif- ference with respect to the amount of tobacco consumed, which, in terms of cigarettes, was 16.1 cigarettes per day for the study group and 15.4 for the controls. They found a sig- nificant difference in the percentage of smok- ers who rolled their own cigarettes, these be- ing 63% for the study group and 37 % for the control group. Khanolkar2+ found a relationship between esopltageal cancer and betal quid chewing, particularly when the quid contained tobacco. The Bantus are frequent users of kaf6r pipes and, in some areas, women smoke pipes as often as do the men 9•' Prospective studies have shown tobacco to be associated with an increased risk for cancer of the upper alimentary tract, though the data have not listed the esophagus as an individual site. Hammond and Hornt= grouped cancer of the lip, tongue, mouth, pharynx, larynx, and esophagus. The ratio of observed to expected deaths for men who were regular cigarette smokers was 7.0, but there was no systematic tendency for this ratio to increase with the 2083038097

• • . No. 2 ETIOLOGICAL FACTORS IN ESOPHdGEAL CAN'CER - YVynfler d: B)'OSs 391 quantity of cigarettes smoked. Dorn9 who East London is high. A traditional drink of grouped deaths from mouth, pharyngeal, and all of the Bantus is kaffir beer, which is made esophageal cancer, found an increase in the from fermented millet antl qontboti and has -ntortality ratio from 1.0 to 2.5 to 4.0 as the an alcoholic content of 8%. The natives often amount of cigarettes smoked increased from fortify this kaffir beer with brandy. According 1es's than 10 per day, to 10 to 20 per day, to to Burrella.4 the Bantu women drink as much more than 20 per day respectively. qomboti in some areas as do the men. Alcohol. Schwartz, Denoix, and Anguera35 Diet. Pertinent data on diet as it relates to found that the alcohol consumption among cancer of the esophagus refer to its correlation 200 patients with cancer of the esophagus was to Plummer-Vinson disease. There is a posi- 17.1 cl. per day and that for the control group tive correlation between Pluntmer-Vinson dis- was 14.5 cl. per day-a difference significant ease, a deficiency disease particularly of iron at the 1% level. Similarly, they found a and probably also of certain vitamins, and the marked difference in the percentage of alco- relatively high incidence of cancer of the oral holics-55°jo of the study patients and 41% of cavity and upper esophagus. Ahlbomt found the control group. Several, but not all, studies that among 93 women with cancer of the have shown occupations involving alcoholic beverages, such as those of brewers, innkeep- ers, beer bottlers, and barmen, to have a higher than expected incidence of esophageal ean- cers. 13. rs, 27. 40, ss Kwan,=8 of China, placed importance on chemical irritation such as that produced by a strong vodka-like drink, pai kan, which contains from 60% to 85% alcohol. Wu and Louckss'I found that among the Chi- nese esophageal cancer patients, 50% drank pai kan in contrast to only 12T, among the controls. According to these authors, pai kan is frequently consumed before it has under- gone an aging process that helps to remove fusel oils and other irritating substances from its tlistillates. In Japan, HirayamatT showed that 51.% of 101 male esophageal cancer patients consid- ered themselves to be heavy drinkers while 23.2% of 1,812 male controls put themselves in that category by subjective classification. Six per cent of the study group were non- drinkers as compared with 23.1% of the con- trol group. These data showed the relative risk for developing cancer of the esophagus to be 8.47 times greater for heavy drinkers than for nondrinkers. Studying the high frequency of esophageal cancer among Bantus in East London, Bur- to 8.7% of the controls. Other sutdies,t• 27,43 relt'•' noted the use of cidiviki, an alcoholic however, have shown no significant differences brew prepared in old petroleum asphalt between study and control patients with re- drums, to be quite common. He proposed that spect to syphilis. Kwan26 did not believe actual carcinogens from the drum may be a that thermal effects of hot food and drink factor in the apparent high rate of cancer of were important in China, since the women the esophagus in this particular population. eat their food at the same temperature as do This brew generally contains baker's yeast, the men. Watson,43 on the other hand, be- crude carbide, and a well known brand of lieved that thermal irritation may play a role liquid metal polish, as well as fruit unfit for in cancer of the esophagus and that hot tea sale on the European market. Apparently the drinking, particularly among Russians, may intake of cidiviki by the Bantu population in be a factor. upper esophagus, 17 had evidence of Plum- mer-Vinson syndrome. His data were con- firmed in a subsequent study by Wynder, Hultberg, Jacobsson, and Brosss3 Wu and Loucks4T believed that the Chinese diet, chiefly made up of millet, maize, and kaoliang, which is a rougher food than rice, may play a role in the high incidence of esophageal cancer in that country. Oral Hygiene. Relatively few specific data are available on the relationship of cancer of the esophagus to oral hygiene and dental con- ditions, though the general clinical impression has been that these are poorer among patients with cancer of the esophagus than in the gen- eral population. Watson and Goodner44 placed emphasis on bad teeth, intraoral sepsis, and ill fitting dentures as leading to hasty and incomplete mastication and thus to trauma to the esophagus. They proposed that trauma causes spasm and slowing of food bolus and later stagnation, esophagitis, tilceratdon, and possibly neoplasia. . Miscellaneous. Syphilis has been considered to play a role in cancer of the esophagus. Tom- lincon and Wilsona9 found that among 50 pa- tients studied in the British West Indies, 52.% had definite evidence of syphilis as compared

392 CANCER Afarch-April 1961 Vol. 14 According to Hirayama,tT the intake of hot food, especially of hot rice gruel, is associated with a high incidence of cancer of the esoph- agus in at least sonte parts of Japan. He found that in Wakayanta Prefecture in Japan, where the intake of such hot rice gruel is more com- mon than in the rest of the country, the death rate from cancer of the esophagus is 2 to 3 times higher than that for Japan as a whole. The average death rates in the former group (from 1953 to 1957) were 10.2 for men and 7.6 for women, as compared with 4.6 and 1.8 respectively for Japan as a whole (1952). The sex ratio of the esophageal cancer death rates in this Prefecture is quite low (1.3 to 1.) when compared with that for Japan (2.6 to l.), indi- cating the possibility of etiological factors common to both sexes.1T Also of interest in this regard are the observations by Burrell3• 4 that patients who have taken cidiviki, appar- ently because of the resultant anesthesia, are able to eat their food hotter than people who have not indulged in this brew. Hartzts believed that the apparent high frequency of esophageal cancer in Cura~ao might be due to pollution of the air from the huge petroleum refineries located there, since there is apparently no difference in tobacco or alcohol consumption or in dietary intake be- tween the population of Cura4ao and that of the adjoining islands in which cancer of the esophagus is less common. A number of benign esophageal conditions are suggested to be related to cancer of the esophagus. These include esophageal hernia, and diverticula and stricture due to lye ex- posure. Of these, oniy, the last has had an adequate statistical evaluation. Postelthwait and his associates31 found among 350 patients with cancer of the esophagus, 3 who had a history of swallowing lye that required treat- ment by dilatation. Kay and Cross20 found 2 esophageal cancers among 50 patients with lye burns. A review of the relationship of lye strictures to cancer of the esophagus was given by Kiviranta,25 and it showed that these can- cers tend to develop from 10 to 30 years after lye exposure. In 381 cases of corrosion stric- tures, nine tenths of which were caused by lye, Kiviranta=s observed 9 cancers. In about 200 patients who had lye burns of the esopha- gus, Bigger and Vinson2 found that carci- noma developed at the site of the stricture in 7 cases. Patients with cancer of the esophagus have .a higher rate of double primary cancers than do most other cancer patients. Goodner and 1Patsont0 showed that among 1,315 cases of cancer of the ezophagus reviewed at Memorial Hospital for Cancer and Allied Diseases, New York, N.Y., 9.5% were double primaries. Most of these double primaries were in the head and neck areas. All except 3 occurred in the intraoral region. Heredity has not been regarded to be a factor in cancer of the esophagus.27 One recent report, however, showed 6 cases of the disease in 2 generations of 1 family.° Clarke and his associates5 described an association of cancer of the esophagus with tylosis, a condition of keratosis of the hands and feet. In a group of tylotic patients investigated in families in which this condition was common, 30% de- veloped cancer of the esophagus. This study was made in families in which tylosis had been observed frequently. Steiner3T found a statisti- cal relationship of cancer of the esophagus to polyps of the large intestine as well as to cirrhosis of the liver. METHODOLOGY The present study is based on data ob- tained from interviews conducted by trained personnel during 1956-1959 and using pa- tients with cancer of the esophagus and con- trol patients at \femorial Hospital for Cancer and Allied Diseases, the Brooklyn and the Kingsbridge Veterans Administration ltos- pitals in New York. Each interview, con- ducted in most cases without knowledge of the diagnosis of the case, was designed to elicit information on specific factors that might have some bearing on the developmenc of cancer of the esophagus. Data obtained in- cluded family history, education, occupation, history of medical and surgical diseases, blood loss, age at menarche and menopause, number of pregnancies and lactation history, detailed dental history, history of social habits, and as detailed data on nutritional history as possi- ble. All patients with cancer of the esophagus included here had a confirmed diagnosis of epidermoid cancer. Eleven cases of adenocar- cinoma were excluded in the basic analysis. The 150 male control patients included the following: 64% with malignant tumors, of which 29% were basal cell carcinomas of the skin, 9 0 lytnphomas (including Hodgkin's disease), 5% tumors of the male genital or- gans, 5°fo tumors of the gastrointestinal tract, 5 a tumors of the rectum, 3% tumors of bone, 2083038099

• • No. 2 ETIOLOGICAL FACCORSIN ESOPHAGEAL CANCER • YVynderL" B1'OSs 393 TABLE 2 PERCENTAGE DISTRIBUTION OF ESO- PHAGEAL CANCER IN JIEN, BY RELIGION ' Religion % studp gp• (150 pt.) o control gp. (150 ptJ None I 1 Roman Catholic 56 37 Protestant 35 39 Jewish 7 21 Other 1 2 3% tumors of the thyroid, and 1% each, tu- ntors of the accessory digestive organs, the pa- rotid, the head and neck (except the thyroid), melanomas, and leukemias. Also included were 35% benign conditions of which 1207, involved the thorax, 7% the parotid, 6% the head and neck, I°/D the thyroid, and 9% other areas. The percentage distribution among the female control patients included 44% malig- nant tumors, of which 14% were of the breast, 8% of the thyroid, 5% of the rectum, 5°Jo basal cell carc,t'noma of the skin, 3% mela- noma, and 3°1' each of the gastrointestinal tract, parotid, ~nd the head and neck (except thyroid). Also included in the female control group were 56% benign conditions, of which 35% were of the thyroid, 5°fo of the parotid, 5% of the thorax, and 11°Je of other areas. In addition, we present data obtained in India by P. D. Schroff at the Tata Memorial Hospital, Bombay, India. He elicited the same data as that in our own study, with the addi- tion of information particularly pertinent to Indian habits, as, for example, detailed data on betel nut chewing. We shall also present a summary of the major factors found in our study in Sweden in which the same interview questionnaire was used. Data on this ,tudy have been pre- sented in detail elsewhere.53 R.FSULTS AMERICAN MALE PATIENTS from those for the control group. Sixteen per cent of the patients with cancer of the esoplta- gus had no education, as compared to 5 0 of the control group; only 4°fe had a college education, as compared to 16°Je of the control group, showing a significantly lower level of education in the study group. Analysis by hos- pital status for clinic, ward, and private pa- tients showed no significant differences. A significant difference was noted when comparing the groups by religion (Table 2). The esophageal cancer group had a predomi- nance of Roman Catholics and a low per- centage of Jews, as compared to the control population. Among the patients with cancer of the esophagus, only 7°fo were Jews as com- pared to 21% among the control population. We have analyzed the patients by residence before and after 20 years of age. Before 20 1.0 r 34.IDAY I 12NIDAY T-12/0AY 0.9f- 0.31- 3-0IDAY 1-21DAY ~ i f 0 RARE-11DAY 0.3 0.1 21-34/DAY 15-2or0AY ;E „ 10-14/DAY 1-9IDAY Background Data. Among the 150 male pa- tients with squamous cell carcinoma of the esophagus, 86% were Caucasian and 14% Negro. The age distribution in percentage was as follows; 30 to 39 years, 1%; 40 to 49 years, 7%; 50 to 59 years, 29%; 60 to 69 years, 41%; 70 and more years, 22%. The study and control groups were age matched. Among the study patients 12°fc were single, 6I /a married, 20°fo widowed, and 6% divorced or separated -percentages that do not vary significantly NONE NONE 0 AMOBNr OE AMOUNI OF SMOKINC ALCOHOL Ftc. 1. Ridit analysis for smoking and drinking. Smoking is shown in equivalent number of cigarettes per day. One pipe equals 2i/ cigarettes, and I cigar equals 5 cigarettes. Drinking is shown in units of alco- hol per day. One unit equals I oz. of whiskey, 4 oz. of wine, or 8 oz. of beer. The hollow circles indicate the American male study group; the closed circles indieue the controls. ~

1 394 CANCER A1arc/i-April 1961 tb1. 14 TABLE 3 PERCENTAGE SMOKING DISTRIBUTION FOR MEN, BY EQUIVALENT NUMBER OF CIGARETTES SMOKED DAILY' No. cigarettes %(150 pt.)p %(150 pt gp p None 5 15 1-9 7 9 10-14 15 11 15-20 24 32 21-34 22 16 >34 26 17 'One cigar equals 5 cigarettes, and 1 pipe equals 2} cigarettes. years of age, 157, of the patients with esopha- geal cancer came from rural areas; after 20 years of age they lived in rural areas in 2% of the cases. The respective percentages for town dwellers are 22°fo and 5 0; for small city dwellers, 13% and 6°Jo; for large city dwellers, 45% and 77°fo; and for varied location, 6oJo and 9%. The distribution by residence did not vary significantly from that of our control population. Analysis by place of birth showed 40% of the study patients to have been born in the United States, in contrast to 507, among the controls. Other differences showed a greater percentage of the esophageal cancer patients to come from Scandinavian countries, 6% against I?lo, and Romance language countries, 21% against 9%, and fewer patients to come from the Soviet Union, I% against 8%. Smoking. The data on smoking have been analyzed by age at onset of smoking and by amount and type of tobacco consumed. Anal- ysis by age at onset and duration of smoking among smokers showed no differences between the groups. Ridit analysis demonstrated a sig- nificant difference in the amount smoked (Fig. 1). A significant difference also existed in re- spect to nonsmokers (Table 3). Whereas 15% t l k d l y o con ro s never smo e on , E h e 5 0 o f h t t e patients with cancer of the esophagus never TABLE 4 PERCENTAGE SMOKING DISTRIBUTION FOR MEN, BY TYPE OF SMOKING study gp. control gp. Type smoking (150 pt.) (150 pt.) None 5 15 Cigarette 51 55 Cigar 19 16 Pipe 9 3 Mixed with cigarette 11 9 Mixed without cigarette 4 2 smoked. There were a greater number who smoked 21 or more cigarettes per day among the esophageal cancer patients (48%) than among the controls (33%). Table 4 presents the data on smoking by type of tobacco con- sumed. There was a relatively greater per- centage of predominantly cigar and pipe smokers among patients with cancer of the esophagus, 28%, than among the control pa- tients, 19%. Data on the length to which the cigarettes and cigars were smoked showed that the pa- tients with cancer of the esophagus smoked their cigars and cigarettes to a shorter butt length than did the control patients. Forty seven per cent of those in the study group who smoked stated that they smoked their ciga- rettes or cigars down to stub length, as com- pared with 26°jo of the smokers among the controls. There is a significantly higher per- TABLE 5 PERCENTAGE TOBACCO CHEWING DISTRIBUTION FOR N]EN, BY DURATION OF CHER'ING Durat. chew. % study gp. % control gp. (150 pt.) (150 pt.) Never 79 90 <10 yr. 7 3 10-19 yr. 7 1 20-29 yr. 3 3 30-39 yr. I I 40-49 yr. I 1 50+yr. 1 1 centage of tobacco chewers among the esoph- ageal cancer group as compared to the con- trols (Table 5). Twenty one per cent of the patients with cancer of the esophagus were chewers as compared to 10°fe of the controls. The data showed no difference in the dura- tion of tobacco chewing. All of the tobacco chewers were also tobacco smokers. Table 6 presents some environmental habits of the 8 male patients with cancer of the esophagus who were nonsmokers. This group did not vary in these factors from that of the over-all group. For example, these patients did not drink more than did the ocer-all esophageal cancer group, nor was their die- tary intake different from that for the total group. Figure 2 shows the relative risk among smokers of developing cancer of the esoph- agus after st:mdardizing for alcohol consump- tion. Since in this standardization we elimi- nated the patients drinking more than 6 units 2083038101

0 No. 2 ET[OLOG[CAL FACTORS IN ESoPHdGEAL CANCER • LVyItCler & BrOss TABLE 6 HISTORY OF MALE NONSMOKERS 395 Alcohol, Tobacco Age, units/day chewing Dental Hist. Location yt. Occupation & durat. durat. condition syphilis tumor 39 . Varied with varied ex- 3/dav for Never Fair Negative Lower 3 posures; washing windmrs; hospitat work,coal porter, peddling, loading mail, dock unloading, digging for W.P.A. Carpenter, dishwasher, 24 yr. 2 }/day for ever ood ositive third iddle 63 furnace caretaker Tailor 25 yr. None Never Edent. at Negative third Upper age 52 yr. two-thirds 58 Docks, clerk for 4 mo., 12/day for Never Good Negative Middle 72 foreman for 35 yr. Textile business 32 yr. Occas. to 1/day , Never Edent. at Negative third Middle for 26 yr. age 68 yr. third 59 Salesman 1-2/day for 39 Never Edent. at Negative Unknown 59 Tool & dye maker for yr.; drank beer, 16/day for 5 yr. 3-4/day for Never age 38 yr. Fair Negative 40 yr. 38 yr. 65 Life insurance salesman None Never Fair Negative Upper third of alcohol a day, the individual cells, particu- larly in the esophageal cancer group, became rather small and thus the confidence limits are wide. The trend shows the risk of develop- ing cancer of the esophagus to increase with the amount of tobacco consumption and also ~ shows that it tends to be greater for cigar and pipe smokers than for cigarette smokers. . We also standardized for religion to deter- mine whether differences in smoking habits could influence the distribution of cancer of the esophagus among religious groups. The data showed that Jewish male patients smoke somewhat more than do the Catholic and Protestant male patients. Thus, tobacco habits cannot account for the differences in the can- cer of the esophagus frequency among the dif- ferent religious groups (Fig. 8). Alcohol Consumption. Data on alcohol con- sumption showed no differences as far as onset and duration of drinking is concerned. Pa- tients with esophageal cancer, however, drank significantly greater amounts than did the control group (Fig. 1). There was a signifi- cantly smaller number of individuals who never or rarely drank, 3% as contrasted to 15% in the control group (Table 7). Seventy- one per cent of the esophageal cancer group admitted to drinking at least 3 units of alco- hol per day as compared to 26% among the controls. Similarly, individuals admitting to 7 units of alcohol per day or going on re- peated binges are significantly more frequent among the esophageal cancer group, 34% against 13%. Table 8 shows alcohol intake by type of alcohol consumed, suggesting no con- sistent differences. Apparently, the individual who drinks his alcohol straight has about the same risk of developing cancer of the esoph- agus as does the one who drinks alcohol in diluted form. Figure 4 presents the relative risk of cancer of the esophagus among alcohol drinkers after standardizing for tobacco con- sumption. These data show a very sharp in- crease in risk for the heavy drinker, which is 6 s r= t ,L O 10 I® xon. u x cis,r su.,e t6ts pt6 tvp. »unutalelwlwlperNr fbne a H Gqtr MIrN 6tte p6r6 tPke o unns a ekoiwlper Wr Fic. 2. Relative risks among smokers of developing esophagcat cancer, after standardization for alcohu7 consumption. The base linc group was the 16 to 34 cfgareuec per day smokers, excluding the heavy drink- ers (i.e., 7 or more units of alcohol per day). 0

396 CANCER rfarch-April 1961 t'ol. 14 1.0 0.9 0.8 0.7 0.6 0.3 0.2 E C 0.1 34.IDAY u-34rDAY 15-20f0AY 10-14/DAY 6-9! DAY NONE 0 L ROMAN CATHOLIC PROTESTANT HEBREW Ftc. 3. Ridit analysis for smoking and religion. Smoking is shown in equivalent number of cigarettes per day. One pipe equals 2t p cigarettes, and I cigar equals 5 dgarettes. The hollow circles indicate the American male study group; the closed circles indicate the controls. most marked for the heavy whiskey consumer. However, the risk for the beer and wine drinker is also greater than that for the non- drinker. Because of the relatively small group of wine drinkers in this series, this group could not be subdivided by amount. Table 9 gives some details on the 17 male patients with cancer of the esophagus who were light or nondrinkers. Analysis for alcohol consumption by re- ligion showed that the difference in the alco- hol consumption in the total group is not chiefly the result of differences in religious distribution, though the Jewish control group contained fewer heavy drinkers than did the Catholic and Protestant ones (Fig. 5). Dental Pactors. Analysis for the age at which the patient became edentttlous showed a tendency for esophageal cancer patients to become edentulous somewhat earlier than did control patients. Seven per cent of the study patients were edentulous prior to the age of 80 as compared to 1% of the controls. Thir- teen per cent of the cancer patients were edentulous prior to the age of 50 as compared to 5% of the controls. In all, there were 29% edentulous study patients as compared to 26% in the control group. Analyzing various factors among the esophageal cancer paticnts who lost their teeth prior to the age of 40 showed no outstanding differences in this group as compared to the over-all esophageal cancer group, though the majority of the patfents be- longed to the heavy drinking group. We found no differences with respect to the presence of sharp teeth. The esophageal cancer patients had a greater percentage of decayed teeth (42ofo) thar did the controls (21%). The study patients also brushed their teeth less often (69°fo once a day) than did the controls (87% once a day). Condition of teeth, however, was seen to be related to other variables such as education, religion, tobacco and alcohol con- sumption, oral hygiene, and dietary factors, (particularly consumption of milk). Dietary Factors. We studied in as much detail as possible the dietary habits of the pa- tients, though admittedly this is one of the most difficult aspects of the interview tech- nique. In this study, we attempted to obtain data on major differences rather than minute dietary information. As for special diets such as salt-free or reducing diets, the groups showed no differences. Tables 10 and 11 sum- marize the average intake of mi]k, butter, and eggs and green and yellow vegetables. The group with esophageal cancer consumed these items, particttlarly milk, to a lesser extent than did the conn-ol group (Fig. 6). Whereas only 12% of the control group stated that TABLE 7 PERCENTAGE DISTRIBUTION FOR ;\tEN BY AMOUNT OF ALCOHOL CONSUMPTION, IN NUMBER OF DRINKS PER DAY Amt, alcohol consump[., units o % study gp. control gp, (150 pt.) (150 pt.) !Vever 2 5 Rarely (<1/moJ 1 10 1/mo. to <1/day 9 33 1-2/day 17 26 3-6/day 37 13 7-12/day 19 7 >12/day 10 5 Sporadic heavy binges 5 1 2083038103

No.'2 ETIOLOGICAL FACCORS IN ESOPHAGEAL CANCER - I'VyLd6'P SY BYOSS 397 3.00, of the esophageal neoplasm to be independ- i • 6a ent of these factors As reviousl indicated y p .. for cancer of the oral cavity, more double pri- iw maries occurred among patients with cancer of the esophagus (8.o-/0) than among other can- cer groups (3.7°,/n).i1 Nine of 12 of these cases 2W occurred in the oral cavity and larynx, sug- osa tion of sore throats and colds, of which there cases. te o o er tseases s owe no stg- e nifrcant differences-with the possible excep- r ar r v g o p e tous tseases,Jo o e - study patients had a history of syphilis (either ~ a clinical history or one proved by serology), IBM 1.0o with a similar percentage among the control v' w f th d' h d R ° gesting common environmental tactors ( i aote ~ 12). e ~ 150- With d t ' d' 8 f th 0L_ <1 1-6 )6 Q 1-6 >5 Wine dents. There were no differences with respect wniste geei to bronchitis, rheumatoid arthritis, nephritis, y peptic ulcer, pneumonia, and allergies. There Fsa. 4. Relathe risks among drinkers of developing was a ne ~tive association to diabetes. Seven esuphageal cancer, after standardization for tobacco u' <nnsunspaon. The base line group was the other than cases were found in the control group, and more ntan 6 units of alcohol per day drinkers. The none in the study group. The esophageal can- risks shown are only for those patients smoking be- tween 16 and 34 cigarettes per day. cer group included no clear-cut case of Plum- mer-Vinson tlisease. One of the patients gave a they never or but rarely drank milk, the re- history of oral fissures. One had a history of spcctive percentage for those with cancer of glossitis, 3 a history of koilonychia, and 5 a the esophagus was 29%. The drinking of milk history of brittle nails. has been cross analyzed against alcohol con- Data on occupation showed a somewhat sumption, since the data suggest that the con- higher proportion of those exposing workers sumption of milk is less, though not signifi- to higher alcohol intake. There were 12 bar- c.tntly so, among patients with cancer of the tenders, brewery workers, and waiters in the esophagus upon standardizing for alcohol study group, and 7 among the controls. Among consumption. The difference is seen to be in- this group of 19 "alcoho6exposecP" workers, 9 creased in the heavier drinking group (Fig. 7). drank 7 or more drinks per day. There were Analysis for vegetable consumption also indi- 6 longshoremen in the study group, and none catetl a lesser intake among patients with can- in the control group. A review of the drinking cer of the esophagus, whereas there was no habits of the longshoremen found them to be difference between this gr^up and the con- particularly heavy drinkers-4 of them con- trol group in the consumption of ineat. sumed 7 or more drinks per day. We also anal- ilfiscellaneotts Data. Data for the location yzed for additional exposures that patients of the esophageal lesion showed that the upper third of the esophagus was slightly more com- monly involved in women than in men. The TABLE 8 percentage distributions for the upper third, PERCENTAGE ALCOHOL DISTRIBUTION FOR middle third, and lower thirds of the esopha- MEN, CON UE[P" [ONCOHOL gus respectively were 25, 45, and 21°fe for men and 29, 49, and 14% for women. One per cent of the men had the site diagnosed as benign in the upper two-thirds and 7% in the lower two-thirds. Nine percent of the women had the site diagnosed as the lower two-thirds of the esopiragtts. Analysis for site of tumor by quantity of alcohol consumed and by type and amount of tobacco used showed the location 11 ® were 15 cases among the esophageal cancer patients and only 6 among the control pa- mnsunipthel %(150 p[.)p %(1control 50 pt.) p Never 2 5 Whiskey straight 13 lt Whiskey mixed 8 9 W&smeyesdtratght 3 li Wine 12 10 Beer 32 32 Varfed 30 21

0 398 CANCER Alnrch-April 19G1 Yol. 14 may have, such as sprays, paint, dust, dirt, smoke, and sawdust, and found no consistent differences between the study and control groups. Specifically, there was no patient who was exposed to lye. We also studied the possible effect of food temperature on the development of cancer of the esophagus. The patients were asked whether they consumed their drinks or food at the same, hotter, or colder tctnperature than the average person. The data showed that 78ofo claimed that they eat at the same temperature, 23% hotter, and 2% colder than others. The respective percentages for the controls were 83%, 15%, and 3%, indicating no significant difference in the 2 groups. Even if there were a difference in the consumption of hot foods, the "recall factor" would have to be consid- ered. It may reasonably be supposed that a patient who knows lie has a disease of the esophagus would be more likely to recall any traumatic experience or habit that might affect that organ than would a patient free of such disease. Similarly, we analyzed the data for spicy food eaten. There was a trend for the esophageal cancer group (35°Jo) to eat more spicy foods than did the controls (21%). How- ever, this difference, though suggestive, dis- appeared when standardized for religion, since it was shown that Catholics tend to eat more spicy food than do Jews. Upon analysis for consumption of fried foods, it was found that 25% of the control patients rarely or never ate fried foods as compared with 19% of the esophageal cancer patients. Analyzing for body build, it was not found TABLE 9 HiSTORY OF NONDRINKERS IN THE AMERICAN MALE STUDY GROUP i Age, yr. Occupation Smoking 51 Construction 5-10 cigarettes/day engineer for 20 yr. 52 Civi1 engineer 20 cigarettes/day for 32 yr. 56 Mechanic 12 cigarettes/day for 35 yr. 59 Salesman, outside 2-3 cigarettes/day work for 25 yr. 61 Garage attendant, 20 cigarettes/day for railroad worker 4 yr., 5-6 cigars/day for 39 yr., 7-8 pipes/day for 46 yr. 62 Bartender 5 pipes/day for >40 yr. 62 Laborer, plasterer, 30 cigarettes/day hot metal worker for 35 yr., 5-6 p~pes/day for 42 yr. 63 Tailor Never 64 Construction, 10-15 cigarettes/day heavy labor for >40 yr. 65 Grocer 40 cigarettes/day for 30 yr.; 2 cigars/day for 30 yr. 65 Life insurance Never salesman 65 Pninter 10 cigarettes/day for 52 yr.; 2-4 pipes/day for 50 yr. 71 \Yaiter 4-6 pipes/day for >40 yr.; 2-4 cigars/day for >40 yr. 72 Salesman Rarely 72 Factory worker 16-20 cigarettes/day for >40 yr.; 1-6 pipes/wk. for <10 yr 2-4 cigars/day ~ for <10 yr. 75 Leather worker y cigar/day for >40 yr. 84 Master mariner 10 pipes/day for 66 yr. Tobacco Dental Hist. Lfxation chewing condition syphilis tumor Never Good Negative Unknown Never Good Negative Middle third Never Good Negative Upper third Never Fair Negative --Jpper third 5-6 chews/day Poor Negative Lower for 39 yr. third 6 yr. Edent. at Negative Middle age 53 yr. third 15 chews/day Edent. at Negative 'tfiddle for 1-1 yr. age 50 yr. third Never Edent. at Negative Upper age 52 yr. two-thirds Never Good Negative Middle third Never Poor Negative Middle third Never Fair Negative Lrppcr third Never Poor Negative Upper third Never Edent. at Unknown Upper approx. third age 65 yr. Never Edent. at Negative Middle age 68 yr. third Never Good Negative Lower third Never Poor Negative Lower two-thirds 4 yr. Poor Negative Upper third 2083038105