Tobacco Documents Online

~x,~I'R RESEARCH 49, 4919-4924• September I, 1989l R'01e of Tobacco and Alcoholic Beverages in the Etiology of Cancer of the Oral ,,,,r Cavity/Oropharynx in Torino, Italy1 iwr Franco 5Ierletti," Paolo Boffetta, Giovannino Ciccone, Arthur Mashberg, and Benedetto Terracini • ~. ~ anccr Epidemiology, Unirersity of Torino, Torino. Italy IF. M.. P. El.. G. C.. t?. T.]. and l'eterans .4dministration Medical Center, East Orange..'Vew Jersey, •.. t~partment of Surgeo', University of~l,ledicine a_nd Dentistry, .Vow Jer.~ey 31edical School [4. M.] rival ~BSTIL&CI- diffused and socially accepted cultural habit in both sexes and is common in nonsmokers as well as smokers; (d) in Italy there dria A population-based case-control study of cancer of oral cavity-oro- has been a tendency to shift from cigarettes prepared with air- ~sa~ ~lmrynx was conducted in the city of Torino, Italy, between 1982 and " 1984. One hundred twenty-two cases (86 males and 36 females) and 606 cured black tobacco to cigarettes made with flue-cured blond tobacco. The Italian population includes sizeable proportions dol. ~trols (385 males and 221 females) were compared with respect to adS- lifelong alcohol and tobacco consumption. A 4- to 6-fold increase in risk that smoked exclusively or predominantly, either blond or black tmong subjects with medium or high tobacco consumption was observed, tobacco. Among these, differences in the carcinogenic risk rum ts ~'ell as a trend in increasing risk with duration and with earlier age at entailed by either type of cigarette can be investigated. A greater ~766. ~e start of smoking• Other findings included a sharp reduction in risk risk for smokers of black tobacco has been shown for cancer of • ith cessation of smoking, no clear protective effect of usage of filter, no the lung (13), larynx/hypopharynx (14), and bladder (15). To differences in risk according to color of tobacco, and a higher risk for the best of our knowledge, this issue has not been investigated ate- cigar versus pipe/cigarette smokers, in any previous epidemiological study on cancer of the oral ,i,,u An effect of alcoholic beverages was found in subjects with an average cavity/oropharynx, including the only case-control study car- Icx. ~l.v consumption of 120 or more grams of alcohol, with a higher risk in ried out in Italy of which we are aware (16). ~r drinkers. for This paper compares lifelong alcohol and tobacco consump- 7o. Among heavy consumers of alcohol and tobacco, risks of both oral and 0mphaD~geal cancer were very high. A positive association betwee.n oral tion among cases and controls. ~iae. ~mcer and low educational level, after adjustment for alcohol and tobacco, ~as found. Attributable risks for alcohol and tobacco in the population MATERIALS AND METHODS pine ~ere 23% and 72% in men and 34% and 54% in women. Leart The study included all incident eases who were residents of the city IN-L'RODUCTION of Torino and diagnosed for oral/oropharyngeal cancers between July ~asc 1. I982, and December 31, 1984. Controls were assembled with the It has long been known that tobacco smoke and alcohol abuse same inclusion criteria, multicentric population-based case-control play a role in the etiology of cancer of the oral cavity/orophar- study on cancer of the larynx/hypopharynx described in detail elsewhere .rex (1, 2) and that the two agents act synergistically. Other risk (11). They were a random sample stratified by sex and age from the nts factors shown or suggested for oral cancer include chewing of files of residents of the city of Torino and were interviewed between up- tobacco alone or in mixtures (3), poor oral hygiene and use of 1980 and 1984. ~ka. 0ral prostheses (4), low intake of fresh vegetables and fruit (5), Cancers at ~he following sites were considered (topographical code DH- Work in the textile industry (6-8), and mouthwash use (9, 10). according to ICD-9): mucosa of lip (140.3, 140.4, 140.5); tongue (141); apa- ~93. The reasons for undertaking the population-based case-con- gum (143); floor of the mouth (144); other and unspecified parts of the tr01 study on cancer of the oral cavity/oropharynx in the city mouth (145); and oropharynx (146). All cases were histologically con- wa. firmed as squamous invasive carcinomas with the exception of one ~m- 0fTorino (northwestern Italy) were: (a) the consideration that acinar cell carcinoma• th:. ~ong Italians (as well as in other countries of southwestern One hundred thirteen cases were identified in 14 stomatology, otor- ani- Europe) intake of alcoholic beverages is closely associated with hinolaryngology, radio- or chemiotherapy units operating in the city of !64: raeals: thus reliable estimates can be obtained through inter- Torino or its outskirts. They were interviewed in the hospital within a Ior. !iews focusing on diet including food and drink intake (11); (b) few days or weeks after diagnosis. In addition, early in 1983, 1984, and :tile 111 northern Italy, mortality rates for cancer of the oral cavity/ 1985, the files of all public and private pathology services operating in °r01~harynx are higher than rates in the central or southern the city of Torino and its outskirts were surveyed. A further 27 histo- ~ak. tro- regions, with a clear trend according to latitude (12); (c) tobacco logically confirmed incident cases were thus identified, of which 20 sra0ke exerts its effect on the oral mucosa of low alcohol were alive and could be traced. Twelve agreed to be interviewed at of ~0nsurners or teetotallers (2); in many countries the assessment home. Clinical information recorded for each case at the time of diagnosis 0fthe effect of alcohol per se alone encounters a major limita- included, among other items, a detailed topographical description of ~a. ti0n rim due to the low proportion of moderate or heavy drinkers the lesion and was reviewed by one of us (A. M.) in order to assess the iria. a~°ng nonsmokers. In northern Italy drinking is a widely site of origin of the cancer in each case. ~ Cases and controls were p~rsonally interviewed by 8 trained inter- Received 9/22/88: revised 2/24/89: accepted 5/3/89. viewers using the same questionnaire• There were no changes in the the The costs of publication of this article were defrayed in part by the payment ~01. ~age charges. This article must therefore be hereby marked advertisement in interview process, interviewers, or questionnaire over the period of the .Ordance with 18 U.S.C. Section 1734 solely to indicate this fact. studv. Lifelong smoking and drinking histories and a detailed descrip- I~ m investigation was supported by the Consiglio Nazionale delle Ricerche, tion of lifelong occupational history, of current diet, and of any major -ing ~_e.(Progetto Finalizzato Oncologia, Contracts 85.02391.44 and 86.00595.44); ~m,~azione Italiana per la Ricerca sul Cancro: and Ministry of Public Education. change in the past were collected by a standard questionnaire (11). :cry l~puter facilities Were made available by a grant from Consorzio per il Sistema Brands of cigarettes (including tobacco used for hand-rolled cigarettes), LSo7O ~rmativo, CSl-Piemonte, Torino. The International Agency for Research on tobacco smoked in pipes, and cigars were classified as black or blond ~'icer, Lyons (France), contributed to the training of the interviewers. Partici- • i ~n in the study took place within the framework of the United States-italy on a priori information from the Italian State Monopoly of Tobacco3 tug i a l~ration for Cancer Research• , . . and, for a few brands, on interviews of tobacconists. Two % of cigarettes "i'a To whom requests for reprints should be addressed, at Epiuemiolog~a dee could not be classified with regard to tobacco type. Cigarettes of mixed re j,i..~0ri, DipartimentO di Scienze BiOmediche e OncOIOgia umana, Universith di d wind, via Santena 7 10126 Torino, Italy• ~ Dr. G. D. Noia, personal communication. 4919

type (which represented a small proportion of all cigarettes) were classified according to the prevalent color. For the conversion of alcoholic drinks into ethanol dose, 1 liter of wine, beer, aperitif, and liquors was considered to correspond, respectively, to 94. 40, 145. and 3t7 g of ethanot (17). Further information on place of birth, civil status, and educational title was collected for all eligible eases and controls from the files of residents in the city of Torino. Educational level was considered as an indicator of socioeconomic status under 3 categories: primary (up to 5 .x r of school): secondary (6 to 8 yr): and high sehool-universit.~ degree. ORs~ stratified on age (3 classes) were calculated according to Mantel and Haenszel (18) with test-based confidence intervals (19). Uncondi- tional logistic regression was used to control confounding. Age (6 classes), education (3 classes), and area of birth (2 classes) were present in all models. Lifelong nonsmokers and teetotallers were excluded from models estimating risks for different patterns of tobacco and alcohol exposure, respectively. ORs for smoking and drinking hal~its were estimated from a basic model including smoking variables (consump- tion: duration; time since quitting; type of product, cigarettes, pipe, and cigars), average consumption of alcohol, and type of alcoholic beverage. When age at start of smoking was entered in the model, duration of smoking was excluded. Roles of filter and type of tobacco were studied in cigarette smokers only. The combined effect of both exposures was analyzed through a model including--in addition to age. education, and area of birth--a term for each combinatiOn of alcohol and tobacco exposure, the reference cute- goD" being subjects with lowest consumptions (see Table 5 for details). The average risks for alcohol and tobacco were estimated from a single model and. therefore, they were adjusted for each other. The ARs in the population, based on marginal ORs of Table 5, were computed according to Miettinen (20); approximate 95% confidence intervals of AR were obtained from a 4-fold table as suggested by Walter (21). SAS programs were used for both stratified and multivar- iate analyses (22, 23). Throughout the period covered by the study, eligible cases were 103 men and 40 women, corresponding to annual incidence rates (age standardized on the world population) of 5.76 and 1.72 × 10-s. Eight2,.'- six and 36. respectively, were interviewed. Table 1 reports the site of the cancer among interviewed cases according to the classification of one of us (A. M.). Among noninterviewed cases of either sex, 9 were aged 65+ (18.7% in this age class), and 12 (12.6%) were in younger age groups. Six cases refused the interview, 2 were not traced, and 13 deceased or too sick to undertake a satisfactory interview. Participation Of controls was less satisfactory: only 385 of 679 men (56.7%) and 221 of 425 women (52.0%) were interviewed. Most failures were due to refusals (46%) or the fact that the person to be contacted was untraceable (41%). Table 2 shows the response rate of eligible subjects by age, education, and area ofbirth. Controls who were younger and more educated were more likely to respond, while no difference was suggested according to area of birth. These variables were intro- duced in the logistic re~gression models. The distribution of cases and controls interviewed by each interviewer was not homogeneous; however, no major differences were found among interviewers with respect to numbers of changes in smoking habits, jobs in occupational histories, and numbers of items recorded in the dietary questionnaire. TOBACCO. ALCOHOL. AND ORAL CANCER Table 1 Site of cancer among interviewed cases Men Women Site (no,) (no.) RESULTS In men, 5 of 86 cases reported to be lifelong nonsmokers versus 13 of 36 among women. Nonsmokers accounted for only 1 of 26 cancers of the floor of the mouth versus 7 of 27 cancers of the oral tongue and 4 of cancers of the alveolar ridge. Age-adjusted ORs for tobacco smoking and alcohol drinking are reported in Table 3. In both men and women, a 4- to 5-fold increase in risk was observed among subjects with medium or high average lifetime tobacco consumption. A trend in increas- * The abbreviations used are: OR. odds ratio; AR. attributable risk expressed as percentage: CI. confidence interval. Floor of mouth Oral tongue Soft palate complex Soft palate 2 Anterior pillar 10 Lingual aspect of rctromt)lar trigone 2 Alx eolar ridge 2 Labial mucosa 1 Buccal mucosa 3 Hard palate 0 Posterior pillar 5 Oropharynx Base of tongue 6 Vatleculae 6 NOS° 10 • Totals 86 ~ NOS, not otherwise specified. 36 Age <65 62 (86.1)a 254 (63~2) 21 (93.8) 162 (62.1) 65+ 24 (77.4) 131 (47.3) 15 (91.7) 59 (36.0) Education (yr of school) -<5 56 (83.6) 185 (49.1) 26 (92.9) 114 (44~4) 6-8 17 (77.3) 101 (64.7) 5 (83.3) 62 (58.5) 8+ 13 (92.9) 99 (67.8) 5 (83.3) 45 (72.6) Area of birth Northern Italy 71 (84.5) 251 (58.0) 28 (90.6~ 145 (52.0) Elsewhere 15 (78.9) 134 (54.5) 8 (88.9) 76 (52.1) Total 86 (83.5) 385 (56.7) 36 (90.0) 221 (52.01 Table 2 Distribution of respondent cases and controls by sex, education, and area of birth Men Women Cases Controls Cases Controls (no.) (no.) (no.) (no.) °N " umbers ~n parentheses, percentage of respondence among eligible subjects. ing risk with duration of smoking is evident in men (OR over 5.0 for duration over 30 yr), but not in women. Young age at start of smoking is associated with higher ORs in both sexes; exsmokers show a reduction of risk when com- pared with current smokers. Cigarette and pipe smokers have similar risks (OR = 3.8), while cigar smokers, with or without the combination of other tobacco products, seem to be at very high risk (OR = 14.6, lower 95% confidence limit = 4.7). Women smoked only cigarettes. In men a higher risk is also suggested for subjects who smoked black cigarettes. No clear difference in risk is observed accord- ' ing to the proportion of filter cigarettes smoked. Among men, an effect of alcoholic beverages is obvious only in subjects with an average daily consumption of 120 or more g of alcohol. Among women, a dose-effect relationship is sug- gested. An elevated risk in both sexes is found for subjects who ~. arank beer with or without other beverages, eul Temporal aspects of drinking habits, such as age at start, duration, and time since quitting, did not show any trend, tw~ When both age and education were considered in this strati- . an fled analysis, ORs showed no major changes, ruling out a~ he, important confounding effect of education. The multivariate exl analysis confirmed the results of the stratified analysis (Table r-,) is t 4). However, in men, no difference in risk related to tobacCO color of cigarettes was identified at this stage. ~o~ att An increased risk was confirmed for cigar smokers in me~ r~ i34" and beer drinkers in both sexes. Table 5 shows the distribution of cases and controls for each combination of alcohol and tobacco consumption with the ~ be 4920 24 2 15 12

TOBACCO, ALCOHOL. AND ORAL CANCER Table 3 Number of cases and controls, ORs adjusted for age, and 95% confidence intervals for smoking and drinking habits, by sex " ~ Men wom~n Cases Controls Cases Controls (n = 86) (n = 385) ORs 95% CI (n = 36) (n = 221) ORs 95~ CI Tobacco Nonsmokers 5 1-7 3 8-15 27 10-25 37 Duration of smoking (yr) 1-20 4 21-30 5 31--40 29 41-50 26 >50 17 Age at beginning (yr) <15 21 15-17 27 18-20 24 >20 9 Yr since quitting: 0-I 68 >5 2 Type of smoker Only cigarettes 68 Pipec 7 Cigarc ' I 1 Color of tobaccoa >66% Blond 13 Mixed (both colors <66%) 7 >66% black .. 48 Usage of filtera >66% with filter 20 Mixed (both types <66%) 13 >66% without filter 35 Alcohol (g/day)a Nondrinkers" 4 1-20" 8 21--40 9 41-80 29 8t-120 14 > 120 22 Alcoholic beveragesr Wine only 21 Beeff 31 Aperitifs~ 30 Liquor~ 45 85 58 91 106 45 54 52 79 77 38 48 91 119 42 195 42 63 263 30 12 84 37 142 97 42 124 13t 142 62 32 132 86 95 196 1.0 0.9 4.6 5.2 5.2 1.0 1.7 - 5.0 5.0 7.1 7.0 4.6 3.1 3.4 5.4 4.4 0.4 3.9 3.8 14.6 2.4 2.9 4.8 3.2 4.7 4.2 1.0 0.5 1.0 1.1 3.3 0.5 1.2 1.0 0.8 13 137 1.0 0.2-319 10 32 5.6 2.0-i5.3 1.8-11.9") 2.1-13.3~ 13 52 5.9 2.3-15.0 1.9-14.6J 0.2-5.1 ~ 9 0.4-6.9 J 60 5.5 1.8-16.8 1.6-16.2 8 15 6.1 2.1-18.1 1.8-13.8 ~. 1.9-26.0 j 6 9 5.7 1.9-16.9 2.6-18.41 1.8-12.2 ~" 13 53 6.7 2.4-18.4 1.2-8.4 .J 1.1-10.6 10 31 4.7 1.8-12.4 2.3-16.8 18 68 7.4 3.0-18.3 1.6-12.4~ 0.1-2.7 J 5 16 3.7 1.3-10.8 1.6-9.4 1.1-12.6 4.7-45.6 23 84 5.4 2.4-12.5 0.8-7.2 12 63 6.0 2.2-16.1 0.9-10.1 4 9 . 4.7 1.3-16.3 1.9-12.1 7 12 6.9 2.5-19.1 1.2-9.0 15 69 6.3 2.4-16.6 1.5-14.9 3 6 5.4 1.3-22.2 1.6-11.0 5 9 5.2 1.8-15.1 5 32 ). 1.0 6 80 J 0.2-1.2 13 73 1.6 0.7-3.8 0.5-2.0 "] 0.4-2.8~ 12 36 3.1 1.3-7.8 1.5-7.4J 0.2-1.7 14 100 0.9 0.3-2.6 0.4-3.8 9 27 3.7 0.9-15.6 0.3-3.1 4 26 0.9 0.2-3.8 0.2-2.4 10 60 1.2 0.4-3.8 Averaged over the period of consumption. Reference category for all tobacco ORs. ~ Only or with other tobacco products , Analysis restricted to subjects who smoked cigarettes only. ~.Reference category for alcohol consumption. Reference category, nondrinkers. Only or wRh other beverages. COrresponding ORs. The small number of cases makes it diffi- trt. COlt to analyze the combined effect of alcohol and tobacco, but t% feature~ are coherent between men and women: (a) lack of tti- az effect of alcohol in light smoker males and in females who an ~eVer smoked; (b) sharp increase of the risk for subjects heavily ate !XPOsed to both alcohol and tobacco. In men, 72.4% of cases hie ~ estimated to be attributable to a consumption of more than :co g of tobacco/day; a corresponding percentage of 23.5 is l;heehe~ ~!~aatttiv'butable tO the habit Of drinking more than 40 g Ot~alcOhO1/ ~4~' In women the AR to smoking is 53.9; a proportion of ~.4% is attributable to drinking more than 20 g/day of alcohol. ~ e AR to the interaction between tobacco and alcohol cannot ~, clearly estimated from our data due to the absence of effect ~talcohol in nonsmokers of both sexes. 4921 Other characteristics of the subjects, considered in the logistic analysis as confounding variables, were found to be of interest. In men, place of birth other than northern Italy (15 cases and 134 controls) was negatively associated with oral cancer (OR = 0.4; 95% CI = 0.2 to 0.7). Moreover, in men a strong interactior~ between birth in northern Italy and alcohol consumption was observed. Restricting the analysis to men born in northern Italy, the risk for alcohol increased on average by a factor of 2.0; for a daily consumption of 120+ g, the OR was 4.3 (95% CI = 1.1 to 17.0, based on 21 cases and 22 controls) versus an OR of 2.1 in the whole series. Finally, people with less than 5 yr of education showed increased risks compared to more educated subjects (9 or more yr of education), with ORs of 2.1 (1.0 to 4.4) in men and 1.8 (0.7 to 4.7) in women. All these risks were

TOBACCO. ALCOHOL, AND ORAL CANCER Table 4 3[ultiple logistic regression analysis: ORs and 95% confidence intervals for smoking and drinking habits in exposed subjects by sex Tobacco (g/day)~ 1-7 8-15 16-25 >25 Duration of smoking (yr) 1-20 21-30 31-40 41-50 >50 Age at beginning (yr)c <15 15-17 18-20' >20 Yr since quitting 0-I 2-5 >5 Type of smoker Only cigarettes Pipea Cigard Color of tobaccoe >66% blond Mixed (both colors <66%) >66% black Usage of filtere >66% with filter Mixed (both types <66%) >66% without filter Alcohol (g/day)n 1-20 21-40 41-80 81-120 >120 Alcoholic beverages Wine only Beery Aperitifs"r Liquor/. Men Women ORs# 95% CI ORs _ 95~ CI 1.0 1.0 4.4 1.0-18.3~ 5.1 1.2-21.07 0.6 6.2 1.4-28.3J 1.0 } 0.7 0.1-4.4 2.5 0.3-18.4 3,9 0.4-34.6 } 34.0 2.6-436.4 1.0 3.3 2.3 0.6 0.3-1.5 1.0 0,4 0.2-0.9 J 0,4 0.1-1.1 0.5 1.0 1.5 1.0 0.7 0.3-1.8~ 0.3 0.1-1.8J 0.1-2.4 0.1-2.1 0.3-8.9 0.1-25.9 0.3-15.2 1.0 0.4 0.1-1.3 8.3 2.1-31.9 1.0 1.0 0.7 0.2-2.7 2.0 1.0 0.4-2.6 2.3 1.0 1.0 1.2 0.4-3.5 1.4 0.1-21.3 1.2 0.5-2.8 0.6 0.1-4.6 1.0 1.0 0.7 0.2-2.6 3.0 0.9,10.5 1.3 0.4-3.8] 0.6 0.2-2.17 3.4 0.9-12.9 2.1 0.6-6.8J Table 5 Distribution of cases and controls by sex and combined exposure to tobacco attd alcohol: ORs and 95% confidenddTntervals Alcohol (g/day) .Men Tobacco (g/da.~) 0-40 41-120 >120 Total 3 72 1 10 8 14~ 1__ 0,6 ] 1.0 0.2-2.0 15 50 5 /-0 27 91 3.6 8.6 4.7 1.1-12.0 1.9-3--~.0 2.0-1 l~i 25 82 16 12 51 151 3.6 21.4 5.1 1.2-11.3 5.9-77.7 2.3-11.4 43 204 22 32 86 385 1.2 3.0 0.6-2.1 1.4-6.5 0-7 Ca" Co 4 61 OR 1.0 95% CI 8-15 Ca Co 7 31 OR 3.3 95% CI 0.9-12.4 >16 Ca Co 10 57 OR 2.5 95% CI 0.7-8.5 Total Ca Co 21 149 OR 1.0 95% CI Alcohol (g/day) 21-40 >40 Total Women - Tobacco (g/day) 0-20 5 46 2 25 13 137 I.I 0.8 1.0 0.3-4.1 0.1-4.2 8 27 10 11 23 84 6.5 21.3 6.3 1.7-24.5 5.1-88.6 2.6-15.5 13 73 12 36 1.6 2.~ 0.6-4.0 1.0±7,6 36 221 0 Ca Co 6 66 . OR 1.0 95% CI 1+ Ca Co 5 46 OR 2.8 95% CI 0.7-11.1 Total CaCo I1 112 OR 1.0 95% CI a Ca, cases: Co, controls. 1.0 1.0 2.1 1.1-4.0 6.1 1.4-26.5 1.4 0.7-2.6 0.4 0.1-1.7 0.7 0.4-1.4 0.8 0.3-2.3 ~ ORs are adjusted for age, educational level, area of birth, tobacco smoking habits, alcohol consumption, and type of alcoholic beverage. n Averaged over the period of consumption. ~ Estimated in a model not including duration of smoking. a Only or with other tobacco products. "Color of tobacco and use of filtered cigarettes analyzed in the 68 cases and 263 controls who s~noked only cigarettes. /Only or with other beverages. estimated from a logistic model including age, educational level, area of birth, and the average daily tobacco and alcohol con- sumption. was related to both average consumption and duration of smok- ing, the OR for light smokers being close to unity. The associ- ation with alcohol was limited to an average lifelong consump- tion of 120 or more g per day. On the contrary, no dose- response for tobacco was found in women (perhaps due to small absolute numbers), and the OR for light smokers was around 5. Similarly, an OR around 3 was found in both light and hea~3" drinkers in women. The absence of an effect of alcohol in nonsmokers has been previously reported (5). A comparison of doses of alcohol with those reported in other studies is impaired by differences in both data collection and detail of adjustment for confounders. Nevertheless, for similar daily intakes, ORs in the present series were lower than in most studies carried out in populations of Anglosaxon origin (26-29), but quite similar to the risks for wine drinking reported by a recent case-control study (30). Although there was no correlation of alcohol and tobacco consumption among male controls, the very low pro- portion of nonsmokers among male cases (5 of 86) did not permit any meaningful analysis on the effect of alcohol per se. Among women, the proportion of nonsmoker cases was about one-third, and no effect of alcohol among nonsmokers is sug- gested (Table 5). The combined effect of alcohol and tobacco on the risk of oral cancer seems to confirm previous observations (4, 30, 31). The inverse association between oral cancerand high educa- tional level, after adjustment for alcohol and tobacco, confirms previous findings from analytical studies (5, 28, 29). This association could partly be due to selection bias in our studY, since success in interviewing subjects was related to both edu- cational level and case-control status (Table 2). In additional analysis, the effect of education was investigated using derno" graphical data available for all eligible cases and controls. ORs of 1.9 and 1.5 were found, respectively, for men and women with up to 5 yr of education compared with more educated DISCUSSION More than 30 yr have elapsed since a formal case-control study first related tobacco and alcohol consumption in a clear way to oral cancer in humans (24). Several other epidemiolog- ical studies, both cohort and case-cont~;ol, confirmed such a relationship, and most of the results of the present study are consistent with previous knowledge. The proportion of nonsmoker cases was negligible in men, whereas it was around one-third in women. A similar pattern was previously described (24, 25). In men, the risk for smokers 4922 2063629350

~seo hall und :avy I in n of ired tent ~ in out lar rol ~nd ro- not se. out C - ms is I~dy, TOBACCO. ALCOHOL. ~ubiccts (9 or more yr of education). Among respondents only, :~c corresponding ORs, after controlling for alcohol and to- ~cco consumptions and area of birth, were remarkably similar. nattacl.v. 2.1 and 1.8. These findings suggest an independent :.:~.ct of educational level on oral cancer risk. 1,,ues raised by the present study relate to type of tobacco a~.i alcoholic.beverages and to temporal variables related to ~0bacco consumption. Among tobacco products, cigar con- sumption exhibited the highest risk. This is consistent with pre~ious findings (24, 29, 32). On the contrary, no previous stud) on oral cancer attempted to discriminate between the effects of cigarettes made with tobacco of different colors. In the present series, an excess risk associated with black (air- cured) tobacco, suggested during the first stage of the analyses, ~.hen age was the only confounder accounted for, was not confirmed after adjustment for drinking, other relevant aspects of smoking, and educational level. This result contrasts with the finding that the smoke from black is more carcinogenic than the smoke from blond (flue-cured) tobacco for the lung 113), larynx/hypopharynx (14), and bladder (15). This discrep- ancy. which requires confirmation, suggests that different car- cinogens of tobacco smoke are involved in the carcinogenic process in different organs. The reduction of risk after quitting confirms previous results {25, 30, 32). To the best of our knowledge, the effect Of age ~'hen first starting to smoke has only been considered in one other study (30): both the cited and the present study suggest a ~end in reduction of risk as age at beginning increases. This phenomenon has been reported for other tobacco-related dis- ~es, including cancer of the lung (1, 33), larynx/hypopharynx {14), and bladder (15). As for alcoholic beverages, beer drinking was found to be related to a statistically significant higher risk than wine drink- ing. This finding is of interest in view of its occurrence in the t~,o sexes and its consistency with some previous observations i26, 30). Information on time of drinking during the day was available for current consumption only. The overwhelming majority of wine was drunk during meals, while over one-half 0fbeer was drunk between meals, the difference being stronger among controls. Time of alcohol consumption might explain the different carcinogenic effect of beer and wine, but more detailed information on lifetime consumption is required to i~vestigate this problem. However, alternative explanations, such as the different concentration of nitrosamines among beverages (34), are also plausible. In men, risks attributable to alcohol and tobacco in the ~01~ulation of the ciff of Torino were, respectively, 23% and 72%. The corresponding figures in women were 34% and 54%. F0r. public health purposes, these estimates further focus the ~rnl~ortance of the two habits in the etiology of oral/oropharyn- geal cancer. Nevertheless, other agents should not be disre- ;,,tiea. In the Iirst place, the present study did not consider ~ther known risk factors, such as dentition status. In addition, the effect of place of birth and educational level, which persisted AND ORAL C-XNCER has been adjusted for, at least partly, by the inclusion of age and education (both related with success of interview) into the logistic regression models. Identification of eligible cases was satisfactory: incidence rates based on the present study are quite similar to those estimated in residents in the city of Torino after the inauguration of the local cancer registry in 1.985.s ACKNOWLEDGMENTS The authors acknowledge the Group for the Study of Laryngeal Cancer in Latin Countries for the preparation of the questionnaire, Dr. A. Tuyns, Dr. J. Est~ve (IARC, Lyons), Dr. L. Garfinkel, and Dr. S. Stellman (ACS, New York) for helpful discussions, and Dr. Clara Latino and Dr. Rita Giacometti for technical help. !fter adjustment for alcohol and tobacco, indicates the need for ~Vestigating other risk factors, such as diet and occupation. The present finding of a higher proportion of nonsmokers ~ra0ng patients with cancer of the oral tongue is consistent with e recently reported observation of a weaker effect of smoking REFERENCES I. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans: Tobacco Smoking, Vol. 38. Lyons: International Agency for Research on Cancer, 1986. 2. IARC Monographs on the Evaluation of the Carcinogenic Risk of chemicals to Humans: Alcohol and Alcoholic Beverages, VoL 44. Lyons: International Agency for Research on Cancer, 1988. 3. IARC Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans: Tobacco Habits Other than Smoking: Betelquid and Areca-Nut Chewing: and Some Related Nitrosamines. Vol. 37. Lyons: International Agency for Research on Cancer, 1985. 4. Graham. S., Dayal, H., Roher, T., Swanson, M., Sultz, H., Shedd, D., and Fischmann, S. Dentition. diet. tobacco, and alcohol in the epidemiology of oral cancer. J. Natl. Cancer Inst., 59:1611-1618, 1977. 5. Winn. D. M.. Ziegler, R. G., Pickle. L. W., Gridley, G., Blot, W. J., and Hoover, R. N. Diet in the etiology of oral and pharyngeal cancer among women from the southern United States. Cancer Res., 44: 1216-1222, 1984. 6. Moss. E.. and Lee, W. R. Occurrence of oral and pharyngeal cancer in textile workers. Br. J. Ind. Med.. 31: 224-232, 1974. 7. Whitaker, C. J., Moss. E., Lee, W. R., and Cunlife, S. Oral and pharyngeal cancer in the North-west and West Yorkshire regions of England and occupation. Br. J. Ind. Med., 36: 292-298. 1979. 8. Winn. D. M., Blot, W. J.. Shy, C. M.. and Fraumeni. J. F. Occupation and oral cancer among women in the South. Am. J. Ind. Med., 3:I"61-167, 1982. 9. Blot. V¢. J.,'Winn. D. M.. and Fraumeni, J. F. Oral cancer and mouthwash. J. Natl. Cancer Inst., 70: 251-253, 1983. 10. Wynder, E: L., Kabat, G.. Rosenberg. S.. and Levenstein, M_Oral cancer and mouthwash use. J. Natl, Cancer Inst., 70: 255-260, 1983. 11. Riboli. E., P6quignot, G., Repetto, F.. Axerio. M.. Raymond, L, Boffetta, P.. Zubiri, A.. Del Moral, A.. Est~ve. J., and Tuyns, A. J. A comparative study of smoking, drinking, and dietary habits in population samples in France. Italy, Spain. and Switzerland. L Study design and dietary habits. Rev. Epidemiol. Sante Publique, 36: 151-165, 1988. 12. Cislaghi. C.. Decarli. A., La Vecchia. C., Laverda, N., Mezzanotte, G., and Smans, M. Data, Statistics, and Maps on Cancer Mortality. Italy 1975/1977. Bologna: Pitagora Editore, 1986. 13. Benhamou. S,. Benhamou. E,, Tirmarche, M.. and Flamant, R. Lung cancer and use of cigarettes: a French case-control study. J. Natl. Cancer Inst., 74: 1169-1175, 1985. 14. Tuyns, A, J.. Esteve. J., Raymond. L.. Berrino, F., Benhamou, E., Blanchet, F,, Boffetta, P., CrosignanL P., Del Moral, A., Lehman, W.~ Merletti, F., Pequignot. G., Riboli, E., Sancho-Garnier, H., Terracini, B., Zubiri, A., and Zubiri, L. Cancer of the larynx/hypopharynx, tobacco, and alcohol. Int. J. Cancer. 41: 483-491, 1988. 15. Vineis. P., Esteve. J., and Terracini. B. Bladder cancer and smoking in males: types of cigarettes, age at start, effect of stopping, and ihteraction with occupation. Int. J. Cancer, 34: 165-170, 1984. 16. Bruzzi. P,, Margarino. G.. Tonetti. R.. Bonelli. L.. Catturich. A., and Scala. Minerva Med., 74: 19-24, 1983. - 17. Pequignot, G., Crosignani, P.. Terracinl, B.. Ascunce. N., Zubiri. A., Ray- mond, L., Est~ve, J.. and Tuyns, A. J. A comparative study of smoking, drinking, and dietary habits in population samples in France, Italy, Spain, and Switzerland. III. Consumption of alcohol. Rev. Epidemiol. Sante Pub- lique, 36: 177-185, 1988. 18. Mantel, N., and Haenszel, W, Statisticqal aspects of the analysis of data from retrospective studies of disease. J. Natl. Cancer Inst., 22: 719-748, 1959. 19. Miettineh, O. S. Estimability and estimation in ease-referent studies. Am. J. Epidemiol., 103: 226-235, 1976. ~a tongue cancer compared to other oropharyngeal sites (30). 20. Miettinen. O~ S. Proportion of disease caused or prevented by a given thThe major limitations of the present study were its small size, exposure, trait, or inter,'ention. Am. J. Epidemiol., 99: 325-332, 1974. • e lack of consideration of some known risk factors such as 21. Walter. S. D. Calculation of attributable risks from epidemiologic data. Int. ~ntition status, and the potential biases brought about by the J. Epidemiol., 7: 175-182, 1987. ~fferent response rate between cases and controls. The latter ~ Registro Tumori Piemonte, unpublished data. 4923

TOBACCCI~-. ~LCOH~3L~ AND OR,-kL CANCER 22. SAS Institute, Inc. SAS User's Guide: Statistics, Version 5. Car2,.', NC: SAS Institute, Inc., 1985. 23. Harrell, F. E., Jr. The logistic procedure. In: Supplemental User's Guide. Cary. NC: SAS Institute, Inc., 1983. 24. Wynder, E. L., Bross, I. J., and Feldman, R. M. A study of the etiological factors in cancer of the mouth. Cancer (Phita.), 10: 1300-1323. 1957. 25. Wynder, E. L., and Stellmann, S. D. Comparative epidemiology 0ftobacco- related cancers. Cancer Res., 37: 4608-4622. 1977. 26. Mashberg, :A., Garfinkel, L.. and Harris, S. Alcohol as a primary risk factor in oral squamous carcinoma. CA, 31: 146-156, 1981. 27. Herity, B., Moriarty, M., Bourke, G. J., and Daly, L. A case-control study 6f head and neck cancer in the Republic of Ireland. Br. J. Cancer, 43:177- 182, 1981. 28. Elwood, J. M., Pearson, J. C. G.. Skippen, D. H., and Jackson. S. M. Alcohol, smoking, social and occupational factors in the etiology of cancer of the oral cavity, pharynx, and larynx. Int. J. Cancer, 34:603-612, 1984. 29. Williams, R. R., and Horm, J. W. Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: interview study from the Third National Cancer Survey. J. Natl. Cancer Inst., 58: 525- 547, 1977. 30. Blot, W. J., McLaughlin, J. K.. Winn, D. M, Austin, D. F., Greenberg. R. S.. Preston-Martin, S.. Bcrnslein, L., Schoenberg. J. B., Stemhagen, A., and Fraumeni. J. F., Jr. Smoking and drinking in relation to oral and pha~'ngeal cancer. Cancer Rcs., 48: 3282-3287, 1988. 31. Rothman. K,, and Keller, A. The effect of joint exposure to alcohol and tobacco on risk of c~mccr of the mouth and pharynx. J. Chronic Dis,. 25: 711-716, 1972. 32. Martinez. I. Factors a',.',ock~ted with cancer of the esophagus, mouth, and pharynx in Puerto Rico..I. Natl. Cancer Inst., 42: 1069-1094, 1969. 33. Brown, C. C., and Chu. K. C. Use of multistage models to infer stage affected by carcinogenic exposure: example of lung cancer and ciga_rette smoking. ,I. Chronic Dis. 40 (Suppl. 2): 1715-1795, 1987. 34. Preussman, R. Occurrence and exposure to N-nitroso compounds and pre- cursors, ln: I. K. O'Neill, R. C. Von Borstel, C. T. Miller, J. Long, and H. Bartsch (eds.), N-Nitroso Compounds: Occurrence, Biological Effects. and Relevance to Human Cancer, pp. 3-15. Lyons: International Agency for Research on Cancer, 1984. 4924 h h t~ ol