European Journal of Clinical Nutrition (2000) 54, 918±920 ß 2000 Macmillan Publishers Ltd All rights reserved 0954±3007/00 $15.00 www.nature.com/ejcn
Wine and other types of alcoholic beverages and the risk of esophageal cancer
C Bosetti1*, C La Vecchia1,2, E Negri1 and S Franceschi3
1Istituto di Ricerche Farmacologiche `Mario Negri', Milan, Italy; 2Istituto di Statistica Medica e Biometria, UniversitaÁ degli Studi di Milano, Milan, Italy; and 3Centro di Riferimento Oncologico, Aviano (Pordenone), Italy
Objective: To investigate the separate and combined effect of wine-drinking and other alcoholic beverages on esophageal cancer, in a high wine-consuming population. Design: Combined analysis of two hospital-based case ± control studies. Setting: Major teaching and general hospitals in the greater Milan area and in the province of Pordenone, in northern Italy. Subjects: A total of 714 incident cases of esophageal cancer, and 3137 controls admitted to hospital for acute, non-neoplastic conditions, unrelated to alcohol consumption. Intervention: Trained interviews identi®ed and questioned cases and controls using standardized structured questionnaires, including information on the average number of days per week each type of alcoholic beverages (wine, beer, spirits) was consumed, and the average number of drinks per day. Odds ratios (ORs) were calculated using unconditional multiple logistic regression equations. Results: With reference to total alcohol drinking, as compared to non- or moderate drinkers ( < 3 drinks per day), the multivariate ORs were 1.98 for drinkers of 3 ± 4 drinks per day, 4.22 for 5 ± 7, 7.60 for 8 ± 11, and 12.35 for 12 drinks per day. Higher risks were observed for wine-only drinkers and the corresponding values were 1.70, 4.21, 8.76 and 17.90. After allowance for wine intake, no association was observed between beer and spirit drinking and esophageal cancer, in a population in which 80% of alcohol came from wine. Conclusion: The amount of ethanol determines the risk of esophageal cancer, and the most commonly used alcoholic beverage appear to be most strongly associated. Sponsorship: Italian Association for Cancer Research, Milan. Descriptors: alcohol drinking; esophageal cancer; alcoholic beverages; risk factors; case±control studies European Journal of Clinical Nutrition (2000) 54, 918±920
Introduction they lacked a suf®cient number of heavy drinkers of wine or other selected beverages (Doll et al, 1993). The consumption of alcoholic beverages is consistently Wine is by far the most common type of alcoholic related to esophageal cancer, and the risk for this neoplasia beverage in Italy, and accounts by itself for over 80% of increases approximately in proportion to the amount drunk alcohol intake (La Vecchia, 1995). Thus, we were able to (Doll et al, 1993; IARC, 1988; Leclerc et al, 1987). Several investigate the separate and combined effect of wine and attempts have been made to separate the effects of different other alcoholic beverage drinking using data from two types of alcoholic drinks on esophageal carcinogenesis. companion case ± control studies conducted in northern Some authors have concluded that no meaningful separa- Italy (Barra et al, 1990). tion is possible, others that the risk is greater with spirits, others with wine, and others again with apple-based drinks Methods (Doll et al, 1993; IARC, 1988). A Danish cohort study (Grùnbñk et al, 1998) of 156 subjects with upper digestive Between 1984 and 1998 two case ± control studies were tract cancers suggested that wine drinkers may be at a conducted in the greater Milan area and in the province of lower risk than drinkers who have a similar intake of beer Pordenone, in northern Italy, the general design of which or spirits, and speculated about the potential anticarcino- has already been described (Barra et al, 1990). Brie¯y, a genic properties of wine. Most investigations, however, total of 714 incident, histologically con®rmed cases of could not analyze the association of different types of cancer of the esophagus (618 males, 96 females) were alcoholic beverages with esophageal cancer risk, since identi®ed, aged between 26 and 83 y, median age 60 y. Controls were 3137 subjects (2400 males, 737 females), aged 25 ± 83 y, median age 58 y, admitted to the same *Correspondence: C Bosetti, Istituto di Ricerche Farmacologiche `Mario network of hospitals for acute, non-neoplastic conditions, Negri', Via Eritrea 62, 20157 Milan, Italy. Guarantor: C Bosetti. unrelated to alcohol or tobacco consumption. Of these, 27% Contributors: CB conducted the statistical analyses and wrote the were admitted for non-alcohol-related traumas, 25% for manuscript. CLV organized and managed the project, suggested the non-traumatic orthopedic conditions, 30% for acute surgi- hypothesis, and revised the manuscript. EN structured the data ®le and cal diseases and 18% for miscellaneous other illnesses. revised the manuscript. SF contributed to the study organization and Trained interviewer identi®ed and questioned cases and revised the manuscript. Received 19 November 1999; revised 6 September 2000; accepted controls during the hospital stay using similar structured 16 September 2000 questionnaires, including information on the average Alcohol and risk of esophogeal cancer C Bosetti et al 919 number of days per week each type of alcoholic beverage combined drinkers of wine, beer and spirits had an inter- (wine, beer and spirits) was consumed, and the average mediate risk as compared to drinkers of beer and spirits. number of drinks per day. Odds ratios (OR) and the This apparent discrepancy can be explained in terms of corresponding 95% con®dence intervals (CI) were derived different characteristics of heavy drinkers in various popu- from multiple logistic regression equations, including terms lations. Thus, where wine is a relatively selective type of for study area, sex, age, education, social class and smoking beverage, wine drinkers may have selected patterns of habits, plus type of alcoholic beverages, whenever required. drinking and=or baseline characteristics, which pose them a low risk of upper digestive tract cancers. In Italy, in contrast, where wine is by far the most common type of Results alcoholic beverage (La Vecchia, 1995), wine drinkers are at highest risk, whereas beer and spirit drinkers would share Table 1 gives the distribution of 714 cases of esophageal similar selective favorable characteristics. In contrast with cancer, and the comparison group according to consump- data from northern Europe, including Britain (Wanna- tion of different types of alcoholic beverages, and the methee & Shaper, 1999) and Denmark (Tjùnneland et al, corresponding ORs and 95% CIs. With reference to total 1999), in Italy wine drinking is not an indicator of a healthy alcohol drinking, as compared to non- or moderate drinkers diet (La Vecchia et al, 1992) and, if anything, female wine ( < 3 drinks per day), the multivariate ORs were 1.98 for drinkers less frequently reported high raw vegetable drinkers of 3 ± 4 drinks per day, 4.22 for 5 ± 7, 7.60 for 8 ± consumption in this dataset (Chatenoud et al, 1999). 11 and 12.35 for 12 drinks per day. Corresponding Our study may have some of the limitations of hospital- values for consumption of wine only were 1.70, 4.21, based case ± control studies. Since hospital controls may 8.76 and 17.90. Slightly lower risks were observed for differ from the general population with regards heavy drinkers of wine plus beer and=or spirits; the ORs for to alcohol consumption, we paid particular attention to subsequent levels of consumption were 1.83, 3.50, 6.01 and excluding from the study all subjects admitted for diseases 10.0. After allowance for wine intake, no material associa- related to alcohol drinking. However, the similar catchment tion was observed between beer and spirit drinking and area for cases and control, the use of the same setting and esophageal cancer. Compared to non-beer drinkers, the questionnaire for all study subjects, and the high participa- ORs were 0.96 for consumption of less than three glasses tion rate should have reduced any other potential source of of beer a day and 1.35 for three or more glasses a day; for bias due to this study design. spirits the corresponding ORs were 0.99 and 1.51. Our ®ndings support the conclusion that ethanol is the main component of alcoholic beverages that determines the Discussion risk of esophageal cancer, and that the most frequently consumed beverage in each area tends to be the one with This uniquely large study from a population with high wine the highest estimated relative risk (Doll et al, 1993). It is consumption indicates that wine is the alcoholic beverage clear, in any case, that wine consumption is strongly related associated with the highest risk of esophageal cancer in to esophageal cancer in individuals and populations with Italy. No consistent trend in risk was observed for drinkers heavy consumption. of beer and spirits and, for similar levels of ethanol consumption, drinkers of wine, beer and spirits showed an apparently lower risk than those of wine only. This pattern of risk is at variance from that reported Acknowledgements ÐThis work was conducted with the contribution of from a Danish cohort study (Grùnbñk et al, 1998), where the Italian Association for Cancer Research. The authors thank Mrs M only wine drinkers showed no clear excess risk, and Paola Bonifacino for editorial assistance.
Table 1 Distribution of 714 cases of esophageal cancer and 3137 controls, and corresponding odds ratioa (OR) and 95% con®dence intervals (CI), according to daily consumption of different types of alcoholic beverages; northern Italy, 1984 ± 1998
Total alcohol Wine only Wine and other Beer Spirits
Drinks=day Ca : Co OR (95% CI) Ca : Co OR (95% CI) Ca : Co OR (95% CI) Ca : Co ORb (95% CI) Ca : Co ORb (95% CI)
0 45 : 420 45 : 420 45 : 420 441 : 2211 1c 343 : 1960 1c 1 ± 2 89 : 1005 1c 61 : 637 1c 24 : 318 1c 212 : 740 0.96 318 : 1103 0.99 (0.75 ± 1.23) (0.79 ± 1.23) 3 ± 4 114 : 701 1.98 50 : 351 1.70 62 : 346 1.83 (1.46 ± 2.67) (1.14 ± 2.54) (1.20 ± 2.79) 5 ± 7 149 : 480 4.22 51 : 147 4.21 96 : 331 3.50 (3.10 ± 5.75) (2.69 ± 6.58) (2.34 ± 5.25) 61 : 179 1.35 52 : 72 1.51 (0.91 ± 1.99) (0.95 ± 2.38) 8 ± 11 190 : 346 7.60 55 : 80 8.76 133 : 264 6.01 (5.51 ± 10.48) (5.37 ± 14.27) (3.97 ± 9.11) 12 120 : 175 12.35 10 : 12 17.90 110 : 163 10.0 (8.37 ± 18.21) (6.56 ± 48.85) (6.30 ± 15.87) aEstimates from multiple logistic regression equation including study area, age, sex, education, social class, and smoking habit. The sum of the data does not add up to the total because of missing values. bAdjusted also for other types of alcoholic beverages. cReference category.
European Journal of Clinical Nutrition Alcohol and risk of esophogeal cancer C Bosetti et al 920 References IARC (1988): Alcohol drinking. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 44, Lyon: International Agency for Research into Cancer. Barra S, Franceschi S, Negri E, Talamini R & La Vecchia C (1990): Type La Vecchia C (1995): Alcohol in the mediterranean diet. Assessing risks of alcoholic beverage and cancer of the oral cavity, pharynx and and bene®ts. Eur. J. Cancer 4,3±5. oesophagus in an Italian area with high wine consumption. Int. J. La Vecchia C, Negri E, Franceschi S, Parazzini F & Decarli A (1992): Cancer 46, 1017 ± 1020. Differences in dietary intake with smoking, alcohol, and education. Chatenoud L, Negri E, La Vecchia C & Franceschi S (2000): Wine Nutr. Cancer 17, 297 ± 304. drinking and diet in Italy. Eur. J. Clin. Nutr. 54, 177 ± 179. Leclerc A, BrugeÁre J, Luce D, Point D & Guenel P (1987): Type of Doll R, Forman D, La Vecchia C & Woutersen R (1993): Alcoholic alcoholic beverage and cancer of the upper respiratory and digestive beverages and cancers of the digestive tract and larynx. In: Health tract. Eur. J. Cancer Clin. Oncol. 23, 529 ± 534. Issues Related to Alcohol Consumption, pp 126 ± 166. Washington: Tjùnneland A, Grùnbñk M, Stripp C & Overvad K (1999): Wine intake ILSI Press. and diet in a random sample of 48,763 Danish men and women. Am. J. Grùnbñk M, Becker U, Johansen D, Tùnnesen H, Jensen G & Sorensen TI Clin. Nutr. 69, 49 ± 54. (1998): A population based cohort study of the association between Wannamethee SG & Shaper AG (1999): Type of alcoholic drink and risk alcohol intake and cancer of the upper digestive tract. Br. Med. J. 317, of major coronary heart disease events and all-cause mortality. Am. J. 844 ± 848. Publ. Health 89, 685 ± 690.
European Journal of Clinical Nutrition