Reply to: Alcohol consumption and ovarian cancer risk in a population‐based case–control study by Peterson <i>et al.</i>

We read with interest the paper by Peterson et al.,1 providing information on the role of total alcohol, and different types of alcoholic beverages, on ovarian carcinogenesis. Authors considered data from a population-based case–control study conducted in the USA on 762 ovarian cancer cases and 6,217 controls. A significant increased risk of serous invasive ovarian cancer was found for women drinking at least 1 beer per day at early ages (20–30 years). However, no association with total alcohol, wine, beer, and spirit drinking was found for all ovarian cancers and for drinking in the recent past, confirming findings from a pooled analysis of 10 cohort studies.2 However, a recent cohort from the USA based on 90,371 women and 253 cases with ovarian cancer found no association with total alcohol, beer or spirit intake, but a direct association with wine.3 To provide further information on the issue, we pooled data from 2 large case–control studies of ovarian cancer,4, 5, 6 conducted in Italy, i.e. in a population with frequent and regular consumption of wine by women and relatively limited consumption of beer and spirits.7, 8 The first study, conducted between 1983 and 1991 in the greater Milan area,4, 5 included 971 cases and 2,503 controls; the second one, conducted between 1992 and 1999 in 4 areas of the North, Center and South of Italy,6 included 1,031 cases and 2,411 controls. Overall, cases were 2,002 women with incident, histologically confirmed epithelial ovarian cancer, admitted to major teaching and general hospitals in the areas under surveillance. Controls were 4,914 women admitted to the same hospitals as cases for acute, nonneoplastic, nongynaecological, nonhormone-related conditions, unrelated to alcohol drinking. Among controls, 29% were admitted for nonalcohol-related traumas, 30% for other orthopedic disorders, 15% for surgical conditions and 25% for other miscellaneous illnesses. Less than 5% of both cases and controls refused to participate. Trained interviewers administered questionnaires, including a detailed and validated section on weekly alcohol consumption for different types of alcoholic beverages.9 The section on alcohol drinking of the first study included the number of days per week each alcoholic beverage (wine, beer, and spirits) was consumed, the number of drinks consumed per day and the duration of the habit. The section on alcohol drinking of the second study included questions on the weekly number of drinks of wine, beer, grappa (a typical Italian spirit), amari and digestives (other types of Italian liquors drunk after meals) and spirits (whisky, cognac, brandy). In both studies, 1 drink corresponded approximately to 125 mL of wine, 330 mL of beer and 30 mL of liquors and spirits, i.e. about 13–15 g of ethanol. We estimated the odds ratios (OR) and corresponding 95% confidence intervals (CI) for various measures of alcohol drinking using unconditional multiple logistic regression models, including terms for study, centre, age, education, body mass index, parity, menopausal status, oral contraceptive and hormone replacement therapy use, and family history of breast/ovarian cancer. When considering different types of alcohol, ORs were derived after further allowance for the continuous term of other types of alcoholic beverages. Table I shows the distribution of ovarian cancer cases and controls according to consumption of different types of alcoholic beverages. Compared with abstainers, no association was found with total alcohol, the continuous OR for an increment of 1 drink per day being 1.02 (95% CI: 0.98–1.06). There was no evidence for the risk to increase with dose. The OR for ≥21 vs. <7 drinks per week was 0.86 (95% CI: 0.67–1.10). After allowance for other types of alcoholic beverages, the continuous ORs were 1.02 (95% CI: 0.97–1.07) for wine, 1.01 (95% CI: 0.82–1.23) for beer and 1.01 (95% CI: 0.85–1.21) for spirits. The OR for high vs. moderate consumption of wine was 0.82 (95% CI: 0.63–1.08). Again, we did not find any trend with dose in risk when the 3 separate alcoholic beverages were considered. There was no modification effect of education as an indicator of socioeconomic status, or any of the other covariates considered. When we limited the analysis to the second study, where we could exclude former drinkers from the reference category, the estimates for current drinkers were substantially unchanged. Alcohol drinking is socially accepted in this population. Consequently, the reproducibility (r = 0.81)10 and validity (r = 0.70)9 of alcohol consumption, particularly wine, were satisfactory. Moreover, findings were derived from a uniquely large dataset of southern Europe, participation was almost complete, catchment areas of cases and controls were comparable and we had the possibility to adjust the models for a number of potential confounding factors, including mutual adjustment for different types of alcohol. In conclusion, the present findings, based on uniquely large datasets from a population characterized by regular and frequent wine consumption in women,7, 8 are in broad agreement with previous large cohort2 and case–control studies.1 They provide therefore definite evidence that wine is not materially related to ovarian cancer risk, the continuous estimate for wine being 1.02, based on over 1,300 wine drinking cases and 630 abstainers. It also confirms that none of the other types of alcoholic beverages is associated to ovarian cancer risk.2, 11 Yours sincerely, Silvano Gallus, Lorenza Scotti, Renato Talamini, Silvia Franceschi, Luigino Dal Maso, Eva Negri and Carlo La Vecchia This work was conducted with the support of the Italian Association for Cancer Research, the Italian League Against Cancer and the Italian Ministry of Education (PRIN 2005). The work in this paper was undertaken while CLV was a senior fellow at the International Agency for Research on Cancer. Silvano Gallus*, Lorenza Scotti*, Renato Talamini , Silvia Franceschi , Luigino Dal Maso , Eva Negri*, Carlo La Vecchia* ?, * Istituto di Ricerche Farmacologiche “Mario Negri,” Milano, Italy, Unità di Epidemiologia e Biostatistica, Centro di Riferimento Oncologico, Aviano (PN), Italy, International Agency for Research on Cancer (IARC), Lyon, France, ? Istituto di Statistica Medica e Biometria “Giulio A. Maccacaro,” Università degli Studi di Milano, Milano, Italy.