During the past two decades, a number of studies have examined the association between alcohol drinking and the risk of ovarian cancer. 1–7 The majority of these investigations have reported little or no relation of alcohol consumption to the development of ovarian cancer, but a prospective study found that drinkers with a daily intake of 10 g of alcohol or more experienced a significant 50% reduction in risk compared with nondrinkers. 1 This result is surprising considering the modest positive association of alcohol with breast cancer, 8 a hormone-dependent cancer with similar risk factors to ovarian cancer, and advances the need to reevaluate the role of alcohol in the etiology of ovarian cancer. Published reports have not included an assessment of potential histologic differences in the association of alcohol to invasive ovarian cancer or to ovarian cancer of low malignant potential or borderline tumors.
We conducted a population-based, case–control study to explore the association of a variety of dietary and nondietary factors with the estimated risk for ovarian cancer. 9 The objective of the present analysis was to examine the hypothesis that alcohol intake was associated with the odds ratio (OR) for ovarian cancer and that this risk was modified by alcohol type and histology.
MATERIALS AND METHODS
This study was reviewed and approved by the Institutional Review Board of the University of Hawaii. The methodology for this case–control study has been described elsewhere. 9 Briefly, women with epithelial ovarian cancer were identified through the rapid reporting systems of two population-based cancer registries, the Hawaii Tumor Registry and the Los Angeles County Cancer Surveillance Program, both members of the Surveillance, Epidemiology, End-Results Program of the National Cancer Institute. 10 Each of these registries is subject to annual quality control audits by the National Cancer Institute, and case ascertainment is reported to be more than 99% complete. 10 Eligible subjects included 972 women with histologically confirmed primary epithelial cancer of the ovary diagnosed between 1993 and 1999. Other eligibility criteria included age 18 or more years, residency in Hawaii or Los Angeles County for at least 1 year, and no history of ovarian cancer. Histology information was obtained from the pathology report. Interview information was obtained from 603 (62%) of the ovarian cancer subjects eligible for participation in the study. Reasons for nonparticipation included physician refusal (7%); patient was too ill, died, or refused to participate (23%); and patient moved or could not be located (8%). In this analysis, 39 additional subjects were excluded because of equivocal histologic classification, and six were excluded because their diets were considered to be unreliable. Of the 558 patients included in this analysis, 127 women were diagnosed with borderline tumors, and 431 women were diagnosed with malignant (invasive) ovarian cancer.
Population controls were frequency matched to patients based on detailed ethnicity (eg, Japanese), 5-year age group, and study site. Eligible controls had no prior history of ovarian cancer and at least one intact ovary. In Hawaii, the control pool consisted of lists of female Oahu residents who were interviewed by the Health Surveillance Program of the Hawaii Department of Health. Annually, the Department of Health identifies a 2% representative sample of all households in the state with a sampling procedure modeled after that of the National Health Survey. 11 The refusal rate for this survey is extremely low (less than 5%) because it is conducted under statutory provision. This source was supplemented with women aged 65 and older who were Health Care Financing Administration participants in Oahu, representing approximately 2% of selected controls. In Los Angeles, over 95% of the controls were selected based on a neighborhood walk procedure. 12 Lists obtained through the Health Care Financing Administration were used to supplement selection of older controls (aged 65 and older), representing about 1% of selected controls. The participation rate for the control subjects was 67% and included 607 women with complete interview information.
A structured in-person interview was conducted according to a standard protocol after obtaining written informed consent. The reference date was the year before diagnosis for the patients and the interview date for controls. The questionnaire included detailed demographic information, a lifetime history of alcohol and tobacco use, a quantitative food frequency questionnaire, a menstrual and reproductive history, exogenous hormones use, and personal and family medical history. As part of the diet survey, we obtained a lifetime history of alcohol drinking for beer, wine and sake, and spirits or hard (distilled) liquor. For each beverage type separately, we determined the age the subject began drinking on a weekly basis, the average weekly consumption of 12-oz (360-mL) cans/bottles of beer, 4-oz (120-mL) glasses of wine or sake, and 1.5-oz (45-mL) shots of spirits, the number of years of regular drinking, and periods of a year or more of abstention. Among wine drinkers, we asked about preference for red wine or white wine. In this analysis, we defined an alcohol drinker as a woman who drank any type of alcoholic beverage at least once a week for 6 months or more. A former drinker was defined as an alcohol drinker who had not imbibed for a year or more. Never drinkers of any type of alcoholic beverage were used as the reference group in all models. The total consumption of alcoholic beverages for each woman was assessed by the summation of individual drinks from each alcohol type, as each drink contained approximately 12 mL of ethanol. An alcohol drink-years variable was created by multiplying the average number of drinks imbibed per day times the number of years of regular alcohol drinking.
The diet questionnaire has been described previously and included 256 items or categories that were representative of the eating patterns of the ethnic groups in the study. 9 For each food or beverage item, the respondent indicated the usual frequency consumed per day, week, or month, with yearly frequencies for particular seasonal items. The quantity of each food item consumed on a daily basis was calculated as the product of the frequency and serving size. The nutrient content of foods was determined from a customized food composition database.
The OR and 95% confidence intervals (CI) were computed by unconditional logistic regression modeling case–control status. 13 To accomplish this, binary indicator variables representing levels of alcohol drinking were entered into the model, with never drinkers as the reference group. The OR and 95% CI were created by taking the antilog of the β coefficients. All OR were adjusted for age (continuous), ethnicity (by indicator: white, Asian, other), study site (Hawaii versus Los Angeles), education (continuous), oral contraceptive use (ever versus never), parity (ever versus never), and tubal ligation (yes versus no). We also considered other potential confounders of the relation between alcohol and ovarian cancer as covariates, including body size, tobacco smoking, family history of cancer, menopausal status, as well as the consumption of calories, fat, β-carotene, and fruits and vegetables, but these did not materially alter the fit of the models. Linear dose–response was tested by entering into the model a trend variable that was assigned the median intake level for the quantile. We modeled the association of alcohol drinking with the OR for ovarian cancer in several ways, including never user versus ever user, never user versus former user, or current user, age started drinking on a regular basis, number of drinks consumed each week, and the duration (years) of use.
The distribution of subject demographics and risk factor information is shown in Table 1. Patients and controls had similar ages (mean 54.8 years), and the majority were of European or Asian ancestry. Controls were better educated than patients (OR 0.94, 95% CI 0.89, 0.98 for each year of education), had a greater number of full-term pregnancies (OR 0.59, 95% CI 0.42, 0.82 for ever compared with never-pregnant women), were more likely to have used oral contraceptives (OR 0.57, 95% CI 0.43, 0.77 for ever compared with never users), and had a higher frequency of tubal ligation (OR 0.71, 95% CI 0.50, 0.99 for tubal ligation compared with no ligation). These risk factors were used as adjustment variables in subsequent analyses.
Thirty-six percent of patients and 39% of controls were current or former alcohol drinkers (Table 2). Women who drank alcohol regularly at the time of their interview, but not former drinkers, had a significantly lower risk (OR 0.69, 95% CI 0.50, 0.96) of ovarian cancer compared with never drinkers. Neither the age started drinking, the number of years of regular drinking, the number of drinks consumed weekly, nor drink-years were associated with the OR for ovarian cancer.
Wine was the most frequent type of alcoholic beverage consumed among patients (25%) and controls (28%), followed by spirits and beer (Table 2). We found a modest reduction in risk associated with wine and beer drinking, but no association with spirits drinking. Current wine drinkers experienced a 37% reduction (95% CI 8, 56 reduction) in the OR for ovarian cancer compared with never wine drinkers. Women who drank red wine at the time of the interview had a somewhat greater reduction in the OR for ovarian cancer (OR 0.61, 95% CI 0.39, 0.94) than did white wine drinkers (OR 0.69, 95% CI 0.47, 1.01). No dose–response relations of age started regular drinking, years of alcohol drinking, the amount of alcohol consumed weekly, nor drink-years were found.
We found few differences in the association of alcohol drinking with the risk of mucinous compared with nonmucinous histologic types (Table 3). Current alcohol drinkers were at significantly reduced risk of nonmucinous, but not mucinous, ovarian cancer compared with never drinkers. Women who were current drinkers of wine and spirits were also at significantly lower risk of invasive ovarian cancer compared with never drinkers. The greatest reduction in risk associated with alcohol drinking was found among women with endometrioid carcinoma. For example, beer drinkers had a significant 60% reduction in the OR for endometrioid carcinoma compared with never drinkers.
The relation of alcohol consumption to the risk of borderline tumors was substantially different from the relation of alcohol to invasive ovarian cancer (Table 3). The OR associated with drinking spirits was significantly elevated in women with borderline serous tumors. The OR for borderline mucinous tumors was also significantly elevated for former wine drinkers.
We examined the potential for a transient effect of alcohol drinking on the OR for invasive ovarian cancer by restricting the analysis to current drinkers (Table 4). Women who started drinking later in life had a reduced OR for invasive ovarian cancer, although the gradient was not monotonic. Short-term users of alcohol (less than 10 years), but not longer-term users, were at a significantly reduced risk of ovarian cancer. A significant (P = .009) inverse gradient in the OR for invasive ovarian cancer associated with the total number of alcoholic beverages consumed on a weekly basis was found: Women consuming 14 or more drinks per week had an OR of 0.36 (95% CI 0.19, 0.70) compared with never drinkers. We found no evidence for a dose–response relation of drink-years to risk.
The results of our study are in agreement with previous investigations of the relation of alcohol drinking and ovarian cancer in showing little overall association with risk. 2,3 However, we found an inverse association of current alcohol consumption with the OR for invasive ovarian cancer, especially among women who drank wine or spirits. Although few histologic differences in the effect of alcohol on the development of ovarian cancer were evident, beer drinkers experienced a significantly reduced OR for endometrioid carcinoma compared with never drinkers.
Our results combined with those of the Iowa Women's Health Study 1 and the Cancer and Steroid Hormone Study 5 suggest that current, rather than lifetime, alcohol drinking practices are relevant to the risk of invasive ovarian carcinoma regardless of the sources of alcohol consumed. This finding, if substantiated by further study, is consistent with the existence of a chemo-preventive (antipromotional) agent in alcoholic beverages that reduces the risk of ovarian cancer. Ethanol may downregulate estrogen production by interfering with circulating levels of gonodotropins or by directly influencing intraovarian concentrations of insulin-like growth factor-1 and nitric oxide, which subsequently inhibit ovulation. 14 Alcoholic women have greater variability in menstrual cycle length than nonalcoholic women, 15 and irregular menstrual cycles may decrease the risk of ovarian cancer. 16 Ethanol may also decrease progesterone and dehydroepiandrosterone concentrations, 17 which may lower the risk of ovarian cancer. 16,18 Phytoestrogens contained in alcohol beverages, including flavonoids, isoflavonoids and resveratrol, may display both estrogenic and antiestrogenic effects depending upon the presence of estradiol. 19–21 Wine contains potent aromatase inhibitors that reduce the conversion of androgens to estrogens. 22,23
Borderline ovarian tumors do not invade the ovarian stroma and have an intermediate or transitional histologic appearance between benign epithelial tumors and carcinoma. 24 These tumors tend to occur at younger ages and at less advanced stages than do carcinomas. 24,25 Epidemiologic investigations of borderline cancers have been scant because of the rarity of these malignancies, but comparisons between borderline and invasive tumors suggest some common risk factors, particularly those related to reproduction. 24,25 The data suggest a significantly increased risk of borderline ovarian cancer among ever drinkers and former drinkers of spirits, especially among patients with serous histologic types. Former, but not current, drinkers of wine were at significantly increased risk of borderline mucinous tumors.
The positive association between alcohol consumption and borderline ovarian cancer is inconsistent with our findings for invasive ovarian cancer and must be viewed with caution because of the small number of borderline cases, especially among specific histologic groups. We adjusted for the effects of other factors that were not in the final statistical model, such as weight and caloric intake, which might confound the relation of alcohol to borderline ovarian cancer, but this had little influence on the OR. Of potential relevance to our results for borderline tumors is the weak, but generally positive, relation between alcohol drinking and the risk of breast cancer. 8 Several hypothesized mechanisms for alcohol in breast carcinogenesis may also be applicable to the risk of borderline ovarian cancer. These include inactivation of phase 1 and phase 2 enzymes involved with carcinogen metabolism, inhibiting repair of deoxyribonucleic acid damage, antagonistic effects on some dietary antioxidants, such as folate, carotenoids, and tocopherols that may be etiologically significant, and the production of reactive oxygen radicals and oxidative stress through alcohol metabolism. 26,27 Alcohol consumption has also been inconsistently associated with reduced circulating concentrations of follicle-stimulating hormone, increased endometriosis, and increased ovulatory infertility, which may increase the risk of ovarian cancer. 28,29 Clearly, these pathways are equally applicable to the risk for invasive ovarian cancer, but we have no ready explanation for the inconsistency in our alcohol findings by malignant behavior.
Strengths of this study include the population-based sampling procedures, the large multiethnic composition of study subjects, specific inquiry regarding the lifetime use of each alcohol type, the ability to adjust for multiple confounding factors, the availability of histology information for all cases, and the inclusion of women with borderline ovarian malignancy. Recall bias is a potential problem in this study, especially because of the well-known link between alcohol and some cancers. An analysis of the Nurses’ Health Study data suggested that recall bias had only a small effect, if any, on reported alcohol consumption between breast cancer patients and controls. 30 We have focused considerable attention on validating our dietary assessment method against food records, 31 and we have demonstrated that our dietary data are reproducible. 32 It is unlikely that ovarian cancer patients would both underestimate (invasive cases) and overestimate (borderline cases) alcohol consumption depending on their diagnosis. Furthermore, an inverse association of alcohol intake to the risk of ovarian carcinoma has been reported in the Iowa Women's Health Study in which differences in recall between patients and controls is nondifferential because of its prospective design. Subjects were asked about their average lifetime alcohol consumption, so we were unable to account for variation in intake levels, especially binge drinking. Finally, the participation rates in this study for patients (62%) and controls (67%) were not optimal and may have affected the validity of our findings.
The results of this study furnish an inconsistent picture regarding an association of alcohol with borderline and invasive ovarian cancer, so it is undoubtedly premature to speculate on the potential clinical implications of our findings. Nonetheless, the data do suggest that current drinkers have a reduced OR for invasive ovarian cancer, especially women who consume at least two drinks daily. This reduction in the OR for ovarian carcinoma risk was found for all alcohol types. A positive association of borderline ovarian cancer, especially serous tumors, with drinking spirits is a novel finding. This investigation provides preliminary evidence that borderline and invasive ovarian cancers may be related to alcohol intake, but these results will have to be confirmed in prospective studies.
1. Kushi LH, Mink PJ, Folsom AR, Anderson KE, Zheng W, Lazovich DA, et al. Prospective study of diet and ovarian cancer. Am J Epidemiol 1999;149:21–31.
2. Byers T, Marshall J, Graham S, Mettlin C, Swanson M. A case-control study of dietary and nondietary factors in ovarian cancer. J Natl Cancer Inst 1983;71:681–6.
3. Whittemore AS, Wu ML, Paffenbarger RS Jr, Sarles DL, Kampert JB, Grosser S, et al. Personal and environmental characteristics related to epithelial ovarian cancer. II. Exposures to talcum powder, tobacco, alcohol, and coffee. Am J Epidemiol 1988;128:1228–40.
4. La Vecchia C, Negri E, Franceschi S, Parazzini F, Gentile A, Fasoli M. Alcohol and epithelial ovarian cancer. J Clin Epidemiol 1992;45:1025–30.
5. Gwinn ML, Webster LA, Lee NC, Layde PM, Rubin GL. Alcohol consumption and ovarian cancer risk. Am J Epidemiol 1986;123:759–66.
6. Tzonou A, Day NE, Trichopoulos D, Walker A, Saliaraki M, Papapostolou M, et al. The epidemiology of ovarian cancer in Greece: A case-control study. Eur J Cancer Clin Oncol 1984;20:1045–52.
7. Kato I, Tominaga S, Terao C. Alcohol consumption and cancers of hormone-related organs in females. Jpn J Clin Oncol 1989;19:202–7.
8. Singletary KW, Gapstur SM. Alcohol and breast cancer: Review of epidemiologic and experimental evidence and potential mechanisms. JAMA 2001;286:2143–51.
9. Goodman MT, Wu AH, Tung KH, McDuffie K, Kolonel LN, Nomura AM, et al. Association of dairy products, lactose, and calcium with the risk of ovarian cancer. Am J Epidemiol 2002;156:148–57.
10. Ries LAG, Eisner MP, Kosary CL, Hankey BF, Miller BA, Clegg LX, et al, eds. SEER cancer statistics review, 1973–97. National Institutes of Health publication no. 00-2789. Bethesda, Maryland: National Cancer Institute, 2000.
11. Oyama N, Johnson DB. Hawaii Health Surveillance Program survey methods and procedures. Research and statistics report no. 54. Honolulu, Hawaii: Hawaii State Department of Health, Research, and Statistics Office, 1986.
12. Pike MC, Peters RK, Cozen W, Probst-Hensch NM, Felix JC, Wan PC, et al. Estrogen-progestin replacement therapy and endometrial cancer. J Natl Cancer Inst 1997;89:1110–6.
13. Breslow NE, Day NE. Statistical methods in cancer research. Vol. I. The analysis of case-control studies. IARC scientific publication no. 32. Lyon, France: IARC, 1980.
14. Dees WL, Srivastava VK, Hiney JK. Alcohol and female puberty: The role of intraovarian systems. Alcohol Res Health 2001;25:271–5.
15. Becker U, Tonnesen H, Kaas-Claesson N, Gluud C. Menstrual disturbances and fertility in chronic alcoholic women. Drug Alcohol Depend 1989;24:75–82.
16. Risch HA. Hormonal etiology of epithelial ovarian cancer, with a hypothesis concerning the role of androgens and progesterone. J Natl Cancer Inst 1998;90:1774–86.
17. Valimaki MJ, Laitinen K, Tiitinen A, Steman UH, Ylostalo P. Gonadal function and morphology in non-cirrhotic female alcoholics: A controlled study with hormone measurements and ultrasonography. Acta Obstet Gynecol Scand 1995;74:462–6.
18. Helzlsouer KJ, Alberg AJ, Gordon GB, Longcope C, Bush TL, Hoffman SC, et al. Serum gonadotropins and steroid hormones and the development of ovarian cancer. JAMA 1995;274:1926–30.
19. Miodini P, Fioravanti L, Di Fronzo G, Cappelletti V. The two phyto-oestrogens genistein and quercetin exert different effects on oestrogen receptor function. Br J Cancer 1999;80:1150–5.
20. Shen F, Weber G. Synergistic action of quercetin and genistein in human ovarian carcinoma cells. Oncol Res 1997;9:597–602.
21. Bhat KP, Pezzuto JM. Cancer chemopreventive activity of resveratrol. Ann N Y Acad Sci 2002;957:210–29.
22. Eng ET, Williams D, Mandava U, Kirma N, Tekmal RR, Chen S. Anti-aromatase chemicals in red wine. Ann N Y Acad Sci 2002;963:239–46.
23. Jang M, Cai L, Udeani GO, Slowing KV, Thomas CF, Beecher CW, et al. Cancer chemopreventive activity of resveratrol, a natural product derived from grapes. Science 1997;275:218–20.
24. Harris R, Whittemore AS, Itnyre J. Characteristics relating to ovarian cancer risk: Collaborative analysis of 12 US case-control studies. III. Epithelial tumors of low malignant potential in white women. Collaborative Ovarian Cancer Group. Am J Epidemiol 1992;136:1204–11.
25. Modugno F, Ness RB, Wheeler JE. Reproductive risk factors for epithelial ovarian cancer according to histologic type and invasiveness. Ann Epidemiol 2001;11:568–74.
26. Forman MR, Beecher GR, Lanza E, Reichman ME, Graubard BI, Campbell WS, et al. Effect of alcohol consumption on plasma carotenoid concentrations in premenopausal women: A controlled dietary study. Am J Clin Nutr 1995;62:131–5.
27. Navasumrit P, Ward TH, O'Connor PJ, Nair J, Frank N, Bartsch H. Ethanol enhances the formation of endogenously and exogenously derived adducts in rat hepatic DNA. Mutat Res 2001;479:81–94.
28. Helzlsouer KJ, Alberg AJ, Gordon GB, Longcope C, Bush TL, Hoffman SC, et al. Serum gonadotropins and steroid hormones and the development of ovarian cancer. JAMA 1995;274:1926–30.
29. Jensen TK, Hjollund NH, Henriksen TB, Scheike T, Kolstad H, Giwercman A, et al. Does moderate alcohol consumption affect fertility? Follow up study among couples planning first pregnancy. BMJ 1998;317:505–10.
30. Giovannucci E, Stampfer MJ, Colditz GA, Manson JE, Rosner BA, Longnecker MP, et al. Recall and selection bias in reporting past alcohol consumption among breast cancer cases. Cancer Causes Control 1993;4:441–8.
31. Hankin JH, Wilkens LR, Kolonel LN, Yoshizawa CN. Validation of a quantitative diet history method in Hawaii. Am J Epidemiol 1991;133:616–28.
© 2003 The American College of Obstetricians and Gynecologists
32. Hankin JH, Nomura AM, Lee J, Hirohata T, Kolonel LN. Reproducibility of a diet history questionnaire in a case-control study of breast cancer. Am J Clin Nutr 1983;37:981–5.