Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Int. J. Cancer: 77, 705–709 (1998) r 1998 Wiley-Liss, Inc. Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer FOOD GROUPS AND RISK OF ORAL AND PHARYNGEAL CANCER Fabio LEVI1*, Cristina PASCHE1, Carlo LA VECCHIA2, Franca LUCCHINI1, Silvia FRANCESCHI3 and Philippe MONNIER4 Vaudois des Tumeurs, Institut Universitaire de Médecine Sociale et Préventive, Lausanne, Switzerland 2Istituto di Ricerche Farmacologiche ‘‘Mario Negri’’ and Istituto di Statistica Medica e Biometria, Università degli Studi di Milano, Milan, Italy 3Servizio di Epidemiologia, Centro di Riferimento Oncologico (CRO), Aviano (PN), Italy 4Service d’oto-Rhino-Laryngologie, Centre Hospitalier Universitaire, Vaudois, Lausanne, Switzerland 1Registre The role of specific food groups and diet variety on the risk of oral and pharyngeal cancer has been considered using data from a case-control study conducted between 1992 and 1997 in the Swiss Canton of Vaud. Cases were 156 patients (126 males, 30 females) aged under 75 (median age 56) years with incident, histologically confirmed cancer of the oral cavity and pharynx, and controls were 284 subjects (246 males, 38 females, median age 57 years), admitted to the same university hospital for a wide spectrum of acute, non-neoplastic conditions unrelated to tobacco and alcohol consumption or to long-term modification of diet. After allowance for education, alcohol, tobacco and total energy intake, significant trends of increasing risk with more frequent intake emerged for eggs (OR 5 2.3 for the highest tertile), red meat (OR 5 2.1) and pork and processed meat (OR 5 3.2). Inverse trends in risk were observed for milk (OR 5 0.4 for the highest tertile), fish (OR 5 0.5), raw vegetables (OR 5 0.3), cooked vegetables (OR 5 0.1), citrus fruit (OR 5 0.4) and other fruits (OR 5 0.2). The addition of a serving per day of fruit or vegetables was associated with an about 50% reduction in oral cancer risk. The most favourable diet for oral cancer risk is therefore given by infrequent consumption of red and processed meat and eggs and, most of all, frequent vegetable and fruit intake. Diet diversity was inversely related to oral and pharyngeal cancer: ORs were 0.35 for the highest tertile of total diversity, 0.24 for vegetable and 0.34 for fruit diversity. In terms of attributable risk, high meat intake accounted for 49% of oral and pharyngeal cancers in this population, low vegetable intake for 65% and low fruit intake for 54%. Int. J. Cancer 77:705–709, 1998. r 1998 Wiley-Liss, Inc. Earlier work on dietary correlates of oral and pharyngeal cancer found little association with any specific food or nutrient (Graham et al., 1977), but subsequent studies showed decreased risks with increasing consumption of fresh fruit and green leaf and other vegetables (Marshall et al., 1982, 1992). Work from the United States (Winn et al., 1984) and Brazil (Franco et al., 1989), as well as the largest study available to date on diet and oral and pharyngeal cancer (a population-based case-control investigation conducted in 4 areas of the United States [McLaughlin et al., 1988]) indicated a protection by fresh fruit, which was not totally explained by b-carotene, vitamin C or fibre content. In case-control studies conducted in northern Italy (Franceschi et al., 1991; La Vecchia et al., 1991), the strongest protection was also related to frequent fruit consumption, and this was independent from major potential confounding factors, including tobacco, alcohol and social class correlates. As in the U.S. data (McLaughlin et al., 1988), b-carotene, but not retinol, was inversely related with risk of cancers of the oral cavity and pharynx, but the association was not stronger than with measures of fruit consumption (odds ratio [OR] for the upper level 5 0.3). It is likely, therefore, that this inverse relationship was not specific but simply reflected a generally poorer diet in the cases. Thus, foods and food patterns have a quantifiable role in oral and pharyngeal cancer risk not only in developing countries but also in developed ones (La Vecchia et al., 1997b). The precise dietary pattern that may entail a reduced oral and pharyngeal cancer risk, however, remains unclear. To provide further information on the issue, we considered data of a case-control study conducted in the French-speaking region of Switzerland, a relatively high-incidence area for oral cancer on a European scale (Levi et al., 1995). MATERIAL AND METHODS A case-control study of oral and pharyngeal cancer was conducted between January 1992 and June 1997 in the Swiss Canton of Vaud. Cases were subjects with incident (i.e., diagnosed within the year before the interview), histologically confirmed cancers of the oral cavity and pharynx (International Classification of Diseases, 9th Revision, 141.0–141.9, 143.0–146.9, 148.0–149.9, excluding lip [140.0–140.9], salivary gland [142.0–142.9] and nasopharyngeal neoplasms [147.0–147.9]), who had been admitted to the University Hospital of Lausanne, Switzerland. Cases identified and interviewed were linked with the incidence data from the Vaud Cancer Registry (Levi et al., 1997) to verify the correspondence of their socio-demographic characteristics with all incident cases. A total of 156 cases (oral cavity or oropharynx, 112; pharynx, 44; 126 males, 30 females), aged 26–72 (median age 55) years, were interviewed. Controls were patients aged #75 years residing in the same geographical area, whose admission diagnosis was of acute, non-neoplastic diseases, judged to be unrelated to tobacco, alcohol or long-term modification of diet. Controls came from similar catchment area as cases and were selected in comparable strata of age. A total of 284 controls (246 males, 38 females) aged 23–74 (median age 58) years were interviewed. They were also admitted to the same university hospital in Lausanne for a wide spectrum of acute conditions, including traumas (19%, mostly sprains and fractures), non-traumatic orthopaedic diseases (25%, mostly low back pain and disk disorders), surgical conditions (40%, mostly abdominal, such as acute appendicitis, peritonitis, urinary stones, cholecystitis, cholelithiasis or strangulated hernia) and miscellaneous other disorders (16%, including acute medical, eye/nose/ throat and skin diseases). All interviews were conducted in the hospital. Twenty-two percent of cases and 15% of controls approached for interview refused. The structured questionnaire included information on socio-demographic characteristics and life-style habits (e.g., smoking, physical exercise), anthropometric factors and a problemoriented medical history. Information on alcohol included, for each type of beverage (wine, beer or spirits), data on days per week and drinks per day, the age at starting and the duration of the habit, in years. Grant sponsor: Swiss Foundation for Research Against Cancer; Grant sponsor: Vaud League Against Cancer; Grant number: AKT 413. *Correspondence to: Registre vaudois des tumeurs, Institut universitaire de médecine sociale et préventive, Centre Hospitalier Universitaire Vaudois, Falaises 1, Casier 15, CH-1011 Lausanne, Switzerland. Fax: (41)21– 323 03 03. E-mail: [email protected] Received 4 February 1998; Revised 2 April 1998 LEVI ET AL. 706 An interviewer-administered food-frequency questionnaire (FFQ), adapted from a validated Italian one (Franceschi et al., 1993; Decarli et al., 1996), was used to assess subjects’ habitual diet, including total energy. Average weekly frequency of consumption of specific foods or food groups as well as complex recipes (including the most common ones in the diet) during the 2 years prior to cancer diagnosis or hospital admission (for controls) were elicited. Intakes lower than once a week but at least once a month were coded as 0.5 per week. The FFQ included 79 foods, food groups or recipes grouped into 6 sections: (i) bread and cereal dishes; (ii) meat, poultry, fish and foods used as meat substitutes; (iii) vegetables (side dishes); (iv) fruit; (v) sweets/desserts and soft drinks; and (vi) milk and dairy products. Another section dealt with alcoholic beverage consumption. At the end of some sections, 1 or 2 summary questions were included concerning all food items or dishes of a certain type (e.g., any type of meat). Several questions aimed at assessing fat intake pattern were also included. Data analysis Food items and recipes were categorised into 14 groups: milk and dairy products (10 questions), bread and cereal dishes (8 questions), soups (2 questions), eggs (2 questions), poultry (2 questions), red meat (beef, veal, lamb; 6 questions), pork and processed meat (4 questions), fish (3 questions), raw vegetables (4 questions), cooked vegetables (8 questions), potatoes (2 questions), citrus fruit (2 questions), other fruit (8 questions) and cakes and desserts (6 questions). The total weekly frequency of intake of food items and recipes included in the same group was subdivided into approximate tertiles based on the control distribution. Diversity was computed as the total number of different foods consumed at least once per week (La Vecchia et al., 1997a). Besides total diversity, vegetable, fruit and meat diversity were considered. Subjects were subdivided into tertiles of diversity based on the distribution of controls. ORs and the corresponding 95% confidence intervals (CIs) were computed using unconditional multiple logistic regression models (Breslow and Day, 1980). All regression equations included terms for age in quinquennia, years of education, tobacco (never, ex-smokers, current smokers of ,15 and $15 cigarettes per day plus pipe/cigar) and alcohol (non-drinkers, drinkers of ,15 and $15 drinks per week and ex-drinkers) consumption and total energy (other than alcohol, quintiles) intake. Intake frequency of each food group was also introduced as a continuous variable. In these models, the unit of measurement for each food group was set at 7 per week. Hence, these models give an estimate of the OR relative to an increase of 1 average serving per day. Using the distribution of the risk factors in the cases and the ORs from the logistic models, population attributable risks were computed, i.e., the proportion of cancers of the oral cavity and pharynx that would have been avoided if all subjects were in the most favourable intake tertile level. The method described by Bruzzi et al. (1985) implies knowledge of the ORs and of the joint distribution of the risk factors in the population of cases only and, thus, can be applied to hospital-based case-control studies. The attributable risk for each factor describes the theoretical percent of cases that would be prevented if all subjects were moved into the exposure level associated with the lowest risk of that factor. Variance estimates and the corresponding 95% CIs were based on the method described by Benichou and Gail (1990) (Mezzetti et al., 1996). RESULTS Table I gives the distribution of oral cancer cases and controls according to sex, age group, education and tobacco/alcohol consumption. Cases were somewhat less educated and significantly more frequently smokers (OR 5 7.1 for current and 1.6 for ex- vs. TABLE I – DISTRIBUTION OF 156 CASES OF ORAL AND PHARYNGEAL CANCER AND OF 284 CONTROLS BY SEX, AGE GROUP, YEARS OF EDUCATION, SMOKING STATUS AND ALCOHOL CONSUMPTION: VAUD, SWITZERLAND (1992–1997) Characteristic Sex Male Female Age (years) ,45 45–54 55–64 $65 Education (years) ,9 9–12 $13 Smoking status Never-smoker Ex-smoker Current smokers ,15 cigarettes/day $15 cigarettes/day pipe/cigar only Alcohol (drinks/week) Non-drinker ,15 $15 Ex-drinker Cases Controls Number % Number % 126 30 80.8 19.2 246 38 86.6 13.4 14 58 53 31 9.0 37.2 34.0 19.9 54 70 80 80 19.0 24.6 28.2 28.2 13 96 47 8.3 61.5 30.1 38 159 87 13.4 56.0 30.6 11 20 125 11 109 5 7.1 12.8 80.1 7.1 69.9 3.2 109 72 103 21 75 7 38.4 25.3 36.3 7.4 26.4 2.5 12 16 122 6 7.7 10.2 78.2 3.9 56 126 89 13 19.7 44.4 31.3 4.6 never-smokers) and heavy alcohol drinkers (OR 5 10.2 for $15 drinks per week as compared to non-drinkers). The association was present for beer (OR 5 4.9), wine (OR 5 9.4) and spirit (OR 5 5.1) drinkers. Table II gives the upper limit of intake tertiles of each food group and the corresponding multivariate ORs after allowance for education, alcohol, tobacco and total energy intake. Intakes of bread, soups, poultry, cheese and cakes/desserts were not significantly related to oral and pharyngeal cancer risk. Significant trends of increasing risk with more frequent intake emerged for eggs (OR 5 2.3 for the highest tertile), red meat (OR 5 2.1) and pork and processed meat (OR 5 3.2). Inverse trends in risk were observed for milk (OR 5 0.4 for the highest tertile), fish (OR 5 0.5), raw vegetables OR 5 0.3), cooked vegetables (OR 5 0.1), citrus fruit (OR 5 0.4) and other fruits (OR 5 0.2). Table III gives the ORs for an increase of 1 serving per day of each food group significantly related to oral cancer risk. These were 0.8 for milk, 1.5 for eggs, 1.4 for meat and pork, 0.8 for fish, 0.6 for raw vegetables, 0.4 for cooked vegetables, 0.7 for citrus fruits and 0.5 for other fruits. Thus, an additional portion of fruit or vegetables per day seemed to convey a protection of 30–50% on oral cancer risk. Simultaneous inclusion of meat and vegetables or fruit led to ORs of 1.3 for meat and pork, 0.6 for raw vegetables, 0.4 for cooked vegetables, 0.7 for citrus fruits and 0.5 for other fruits. Vegetable, fruit, meat and total diversity are considered in Table IV. Compared with subjects in the lowest tertile, ORs were 0.24 for those in the highest tertile of vegetable, 0.34 of fruit and 0.35 of total diversity. The trends in risk were significant. In contrast, the ORs for subsequent tertiles of meat diversity were 1.14 and 1.82, but the corresponding trend in risk was nonsignificant. DISCUSSION Our present data confirm that dietary patterns and, in particular, food-intake patterns have a relevant role in the risk of oral and pharyngeal cancer. This holds true also even after allowance for alcohol and tobacco, by far the major risk factors for oral cancer in Western countries, as well as for social class indicators and total energy intake. As previously reported in 11 of 13 studies on the DIET AND OROPHARYNGEAL CANCER 707 (CI)1 TABLE II – ODDS RATIOS (OR) AND 95% CONFIDENCE INTERVALS OF ORAL AND PHARYNGEAL CANCER AMONG 156 CASES AND 284 CONTROLS ACCORDING TO INTAKE TERTILE OF SELECTED FOOD GROUPS: VAUD, SWITZERLAND (1992–1997) Food group Milk Upper limit (servings/week) OR (95% CI) Bread and cereals Upper limit (servings/week) OR (95% CI) Soups Upper limit (servings/week) OR (95% CI) Eggs Upper limit (servings/week) OR (95% CI) Poultry Upper limit (servings/week) OR (95% CI) Red meat Upper limit (servings/week) OR (95% CI) Pork and processed meat Upper limit (servings/week) OR (95% CI) Fish Upper limit (servings/week) OR (95% CI) Cheese Upper limit (servings/week) OR (95% CI) Raw vegetables Upper limit (servings/week) OR (95% CI) Cooked vegetables Upper limit (servings/week) OR (95% CI) Potatoes Upper limit (servings/week) OR (95% CI) Citrus fruits Upper limit (servings/week) OR (95% CI) Other fruits Upper limit (servings/week) OR (95% CI) Cakes and desserts Upper limit (servings/week) OR (95% CI) Intake tertile 12 x21 (trend) 2 3 4.0 1 11.0 0.33 (0.18–0.62) 0.38 (0.21–0.70) 11.31** 17.3 1 26.5 1.71 (0.94–3.13) 1.69 (0.82–3.49) 2.75 0.5 1 1.0 1.33 (0.75–2.34) 1.42 (0.77–2.62) 1.23 1.0 1 3.0 1.33 (0.76–2.34) 2.32 (1.28–4.22) 7.46** 0.5 1 1.0 1.05 (0.59–1.90) 0.92 (0.46–1.11) 0.08 2.0 1 3.5 1.42 (0.78–2.56) 2.14 (1.15–3.99) 5.78* 2.0 1 4.5 2.29 (1.20–4.36) 3.21 (1.66–6.24) 11.35** 1.0 1 1.5 0.76 (0.43–1.35) 0.54 (0.30–0.96) 4.40* 4.0 1 7.7 1.35 (0.78–2.32) 1.19 (0.66–2.14) 0.38 5.0 1 8.5 0.63 (0.37–1.08) 0.30 (0.16–0.58) 13.07** 5.2 1 8.6 0.42 (0.25–0.72) 0.14 (0.07–0.19) 31.69** 2.0 1 3.5 0.92 (0.51–1.65) 1.22 (0.70–2.14) 0.50 1.5 1 3.5 0.34 (0.20–0.60) 0.38 (0.20–0.73) 12.95** 5.2 1 11.0 0.42 (0.25–0.73) 0.22 (0.11–0.44) 29.66** 1.2 1 4.0 0.93 (0.54–1.61) 1.02 (0.52–2.03) 0.00 1Estimates from multiple logistic regression equations including terms for age, sex, education, smoking, alcohol and non-alcohol total energy (other than alcohol) intake.–2Reference category: *p , 0.05, **p , 0.01. issue, the strongest factor was the protection conveyed by fruit and vegetables (Potter and Steinmetz, 1996). The addition of 1 serving per day of fruit or vegetable appeared to reduce oral cancer risk by about 50%. This protection by fruit and vegetables has been related to selected micronutrients, including b-carotene and other carot- enoids (La Vecchia et al., 1991; Zheng et al., 1993). The data, however, were also consistent with the hypothesis that a more varied diet (and hence probably richer in several nutrients) may be protective against oral carcinogenesis. A simple explanation for this protective effect is that individuals eating more varied foods are more likely to consume a number of beneficial compounds (La LEVI ET AL. 708 TABLE III – ODDS RATIOS (OR) AND CORRESPONDING 95% CONFIDENCE INTERVALS (CI)1 AMONG 156 CASES OF ORAL AND PHARYNGEAL CANCER AND 284 CONTROLS FOR AN INTAKE INCREASE OF 1 SERVING PER DAY OF SELECTED FOOD GROUPS: VAUD, SWITZERLAND (1992–1997) Food group OR (95% CI) Milk Eggs Red meat Pork and processed meat Fish Raw vegetables Cooked vegetables Citrus fruit Other fruits 0.80 (0.66–0.97) 1.46 (1.14–1.86) 1.35 (1.07–1.69) 1.37 (1.13–1.67) 0.85 (0.74–0.98) 0.64 (0.52–0.80) 0.44 (0.33–0.59) 0.73 (0.61–0.88) 0.51 (0.38–0.68) 1Adjusted for age, sex, education smoking, alcohol and total energy (other than alcohol) intake. TABLE IV – ODDS RATIOS (OR) AND CORRESPONDING 95% CONFIDENCE INTERVALS (CI) AMONG 156 CASES OF ORAL AND PHARYNGEAL CANCER AND 284 CONTROLS ACCORDING TO DIVERSITY OF INTAKE OF SELECTED FOOD GROUPS: VAUD, SWITZERLAND (1992–1997) Food group Vegetable Fruit Meat Total diversity Diversity1 OR (95% CI)2 OR adjusted for number of servings (95% CI)3 ,34 3–5 .5 x21 (trend) ,34 3–4 .4 x21 (trend) ,34 3–5 .5 x21 (trend) ,164 16–19 .19 x21 (trend) 1 0.38 (0.22–0.64) 0.22 (0.11–0.43) 22.306 1 0.53 (0.30–0.95) 0.25 (0.12–0.54) 14.726 1 1.14 (0.65–1.99) 1.94 (1.08–3.50) 4.665 1 0.52 (0.30–0.91) 0.31 (0.16–0.60) 12.666 1 0.41 (0.24–0.7) 0.24 (0.12–0.48) 18.546 1 0.58 (0.32–1.04) 0.34 (0.15–0.76) 8.486 1 1.14 (0.64–2.01) 1.82 (0.91–3.34) 3.51 1 0.54 (0.30–0.96) 0.35 (0.17–0.70) 9.376 1Number of items consumed at least weekly.–2Estimates from multiple logistic regression equations including terms for age, sex, education, smoking, alcohol and total energy (other than alcohol) intake.–3As in footnote 2, plus total weekly servings.–4Reference category.–5p , 0.05.–6p , 0.01. Vecchia et al., 1997a) since oral carcinogenesis seems to be favoured by a generally poorer diet (Franceschi et al., 1990). It is also possible, however, that pre-cancerous lesions of the oral cavity induce modifications in dietary practices (e.g., decreased citrus and fruit consumption), which will therefore be the consequence, rather than the cause, of the disease. In this study, the most favourable diet appears, however, to be a combination of frequent fruit and vegetable intake and infrequent intake of certain types of meat. A large, multicentric, U.S. study also found some direct association between meat, saturated fats and cholesterol intake and oral cancer risk, and in another study from northern Italy (La Vecchia et al., 1991), an association was found with retinol, an indicator of meat intake. In this study, direct relationships were observed with eggs, pork and sausages, which again may reflect an unfavourable effect of animal fats and cholesterol (Franceschi et al., 1991; Marshall and Boyle, 1996) and possibly of carcinogens in broiled meat (de Meester and Gerber, 1995) but also may be a general indicator of a poorer diet. As in previous studies (Marshall et al., 1992; Marshall and Boyle, 1996), little indication emerged for an association between bread and cereals and oral cancer. Our study was hospital-based and is, therefore, open to some of the related criticisms (Breslow and Day, 1980). However, the link of its case series with a cancer registration system (Levi et al., 1997), the satisfactory participation of cases and controls and the inclusion of only acute conditions unrelated to alcohol, tobacco and long-term diet modification in the comparison group are reassuring against selection bias. With reference to information bias and confounding, we administered in the same structured way a validated FFQ, allowing for a measure of total energy intake (Willett and Stampfer, 1986) besides alcohol, tobacco and sociodemographic correlates of oral cancer. Most multivariate ORs, however, were not materially different from age-adjusted ones, suggesting that even more detailed allowance for covariates is unlikely to explain the associations observed. The results obtained for alcohol drinking and tobacco smoking, moreover, were consistent with available knowledge on the issue (Blot et al., 1988; Franceschi et al., 1990; La Vecchia et al., 1997b). Since cases were representative of all incident cases in this population, we were able to derive population attributable risks (Bruzzi et al., 1985). Tobacco accounted for 69% of oral and pharyngeal cancers in this population (95% CI 51–88%) and alcohol for 72% (95% CI 62–82%). Although these 2 factors accounted for the largest proportion of cases, the population attributable risks for dietary factors were far from negligible, being 49% (95% CI 27–71%) for high meat or pork intake, 65% (95% CI 46–84%) for low vegetable consumption and 54% (95% CI 33–76%) for low fruit consumption. There is, therefore, ample scope for prevention of oral and pharyngeal cancer (Negri et al., 1993), a neoplasm whose prognosis has shown little sign of improvement over the last few years (Franceschi et al., 1992). ACKNOWLEDGEMENTS The contributions of Mrs. G. Bidois, Dr. E. Brossard (of the Vaud Cancer Registry’s staff) and Dr. E. Negri with computing assistance are gratefully acknowledged. REFERENCES BENICHOU, J. and GAIL, MH., Variance calculation and confidence intervals for estimates of the attributable risk based on logistic models. Biometrics, 46, 991–1003 (1990). BLOT, W.J., MCLAUGHLIN, J.K., WINN, D.M., AUSTIN, D.F., GREENBERG, R.S., PRESTON-MARTIN, S., BERNSTEIN, L., SCHOENBERG, J.B., STEMHAGEN, A. and FRAUMENI, J.F., JR., Smoking and drinking in relation to oral and pharyngeal cancer. Cancer Res., 48, 3282–3287 (1988). BRESLOW, N.E. and DAY, N.E., Statistical methods in cancer research. Vol. I, The analysis of case-control studies, IARC Scientific Publication 32, IARC, Lyon (1980). BRUZZI, P., GREEN, S.B., BYAR, D.P., BRINTON, L.A. and SCHAIRER, C., Estimating the population attributable risk for multiple risk factors using case-control data. Amer. J. Epidemiol., 122, 904–914 (1985). DECARLI, A., FRANCESCHI, S., FERRARONI, M., GNAGNARELLA, P., PARPINEL, M.T., LA VECCHIA, C., NEGRI, E., SALVINI, S., FALCINI, F. and GIACOSA, A., Validation of a food-frequency questionnaire to assess dietary intakes in cancer studies in Italy. Results for specific nutrients. Ann. Epidemiol., 6, 110–118 (1996). DE MEESTER, C. and GERBER, G.B., The role of cooked food mutagens as possible etiological agents in human cancers. A critical appraisal of recent epidemiological investigations. Rev. Epidémiol. Santé Publ., 43, 147–161 (1995). FRANCESCHI, S., BIDOLI, E., BARON, A.E., TALAMINI, R., SERRAINO, D. and LA VECCHIA, C., Nutrition and cancer of the oral cavity and pharynx in north-east Italy. Int. J. Cancer, 47, 20–25 (1991). FRANCESCHI, S., LEVI, F. and LA VECCHIA, C., Decline in 5-year survival rates for cancer of head and neck. Lancet, 340, 47 (1992). FRANCESCHI, S., NEGRI, E., SALVINI, S., DECARLI, A., FERRARONI, M., FILIBERTI, R., GIACOSA, A., TALAMINI, R., NANNI, O., PANARELLO, G. and LA VECCHIA, C., Reproducibility of an Italian food frequency questionnaire for DIET AND OROPHARYNGEAL CANCER cancer studies: results for specific food items. Europ. J. Cancer, 29A, 1–10 (1993). FRANCESCHI, S., TALAMINI, R., BARRA, S., BARON, A.E., NEGRI, E., BIDOLI, E., SERRAINO, D. and LA VECCHIA, C., Smoking and drinking in relation to cancers of the oral cavity, pharynx, larynx, and esophagus in northern Italy. Cancer Res., 50, 6502–6507 (1990). FRANCO, E.L., KOWALSKI, L.P., OLIVEIRA, B.V., CURADO, M.P., PEREIRA, R.N., SILVA, M.E., FAVA, A.S. and TORLONI, H., Risk factors for oral cancer in Brazil. A case-control study. Int. J. Cancer, 43, 992–1000 (1989). GRAHAM, S., DAYAL, H., ROHRER, T., SWANSON, M., SULTZ, H., SHEDD, D. and FISCHMAN, S., Dentition, diet, tobacco and alcohol in the epidemiology of oral cancer. J. nat. Cancer Inst., 59, 1611–1618 (1977). LA VECCHIA, C, MUÑOZ, S.E., BRAGA, C., FERNANDEZ, E. and DECARLI, A., Diet diversity and gastric cancer. Int. J. Cancer, 72, 255–257 (1997a). LA VECCHIA, C., NEGRI, E., D’AVANZO, B., BOYLE, P. and FRANCESCHI, S., Dietary indicators of oral and pharyngeal cancer. Int. J. Epidemiol., 20, 39–44 (1991). LA VECCHIA, C., TAVANI, A., FRANCESCHI, S., LEVI, F., CORRAO, G. and NEGRI, E., Epidemiology and prevention of oral cancer. Oral Oncol. Europ. J. Cancer, 33, 302–312 (1997b). LEVI, F., LA VECCHIA, C., LUCCHINI, F. and NEGRI, E., Cancer mortality in Europe, 1990–92. Europ. J. Cancer Prev., 4, 389–417 (1995). LEVI, F., TE, V.C., RANDIMBISON, L. and LA VECCHIA, C., Statistics from the Registry of the Canton of Vaud, Switzerland, 1988–1992. In D.M. Parkin, S.L. Whelan, J. Ferlay, L. Raymond and J. Young (eds.), Cancer incidence in five continents, Vol. VII, pp. 686–689, IARC Scientific Publication 143, IARC, Lyon (1997). MARSHALL, J.R. and BOYLE, P., Nutrition and oral cancer. Cancer Causes Control, 7, 101–111 (1996). 709 MARSHALL, J.R., GRAHAM, S., HAUGHEY, B.P., SHEDD, D., O’SHEA, R., BRASURE, J., WILKINSON, G.S. and WEST, D., Smoking, alcohol, dentition and diet in the epidemiology of oral cancer. Oral Oncol. Europ. J. Cancer, 28B, 9–15 (1992). MARSHALL, J., GRAHAM, S., METTLIN, C., SHEDD, D. and SWANSON, M., Diet in the epidemiology of oral cancer. Nutr. Cancer, 3, 145–149 (1982). MCLAUGHLIN, J.K., GRIDLEY, G., BLOCK, G., WINN, D.M., PRESTONMARTIN, S., SCHOENBERG, J.B., GREENBERG, R.S., STEMHAGEN, A., AUSTIN, D.F., ERSHOW, A.G., BLOT, W.J. and FRAUMENI, JR., J.F., Dietary factors in oral and pharyngeal cancer. J. nat. Cancer Inst., 80, 1237–1243 (1988). MEZZETTI, M., FERRARONI, M., DECARLI, A., LA VECCHIA, C. and BENICHOU, J., Software for attributable risk and confidence interval estimation in case-control studies. Comput. Biomed. Res., 29, 63–75 (1996). NEGRI, E., LA VECCHIA, C., FRANCESCHI, S. and TAVANI, A., Attributable risk for oral cancer in northern Italy. Cancer Epidemiol. Biomarkers Prev., 2, 189–193 (1993). POTTER, J.D. and STEINMETZ, K., Vegetables, fruit and phytoestrogens as preventive agents. In B.W. Stewart, D. McGregor and P. Kleihues (eds.), Principles of chemoprevention, pp. 61–90, IARC Scientific Publication 130, IARC, Lyon (1996). WILLETT, W.C. and STAMPFER, M.J., Total energy intake: implications for epidemiologic analyses. Amer. J. Epidemiol., 124, 17–27 (1986). WINN, D.M., ZEIGLER, 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). ZHENG, W., BLOT, W.J., DIAMOND, E.L., NORKUS, E.P., SPATE, V., MORRIS, J. and COMSTOCK, G.W., Serum micronutrients and the subsequent risk of oral and pharyngeal cancer. Cancer Res., 53, 795–798 (1993).