Red wine consumption and risk of prostate cancer: The California Men's Health Study
Red wine contains polyphenol antioxidants that inhibit prostate cancer development in animal studies. We investigated the effect of red wine intake on the risk of prostate cancer using data prospectively collected in the California Men's Health Study (CMHS). CMHS is a multiethnic cohort of 84,170 men aged 45–69 years who were members of the Kaiser Permanente Southern and Northern California Health Plans. Information on demographic and lifestyle factors was collected using mailed questionnaires between 2002 and 2003. We used Cox models to estimate the effect of red wine on prostate cancer risk, adjusting for potential confounders. A total of 1,340 incident prostate cancer cases identified from Surveillance, Epidemiology and End Result-affiliated cancer registries were included in the analyses. We did not find a clear association between red wine intake and risk of prostate cancer. Hazard ratio (HR) estimates for consuming <1 drink/week, ≥1 drink/week but <1 drink/day and ≥1 drink/day were 0.89, 95% confidence interval (0.74–1.07), 0.99 (0.83–1.17) and 0.88 (0.70–1.12), respectively. Further, we observed no linear dose response. The lack of association for red wine intake was consistently observed when we restricted the analyses to those with and without a history of PSA screening. In addition, we also did not observe any association with prostate cancer for beer, white wine, liquor or combined alcoholic beverage intake (HR for combined alcoholic beverage intake of ≥5 drinks/day = 1.16 (0.83–1.63). Neither red wine nor total alcohol consumption were associated with prostate cancer risk in this population of moderate drinkers.
Red wine is a rich source of polyphenol antioxidants such as resveratrol. Recently, resveratrol has received considerable attention because of its potent anti-cancer properties.1, 2 Resveratrol inhibits cancer cell growth in many cell lines, including cancers of the lung, prostate, colon, breast and ovaries.3–14 Resveratrol also reduces tumor incidence in animal models by interfering with carcinogenesis through a wide array of anti-cancer actions. Several animal studies have demonstrated the ability of resveratrol to inhibit prostate cancer development by downregulating androgen receptors, inducing apoptosis/cell cycle arrest and inhibiting disease progression.15–20
Despite the large number of epidemiologic studies that have evaluated the association between alcohol consumption and risk of prostate cancer, the association between red wine intake and prostate cancer has not been clearly established. This is because most prior studies did not separately examine the effect of red wine from white wine. Although both types of wine contains similar amount of alcohol, the concentration of polyphenols is higher in red wine than in white wine.21–23 Furthermore, because resveratrol is mostly found in grape skin, only red wine contains considerable amount of these compounds. The few studies that examined the association between red wine intake and risk of prostate cancer reported mixed results.24–27 Given the disease burden of prostate cancer in the United States,28 the potential beneficial effect of moderate red wine intake on this cancer warrants further examination.
In this study, we investigated the hypothesis that moderate red wine consumption reduces risk of prostate cancer. We used data prospectively collected from the California Men's Health Study (CMHS), a cohort study of middle aged men who were members of Kaiser Permanente (KP) Southern and Northern California Health Plans. The CMHS was specifically designed to study prostate cancer etiology and collected comprehensive information on known risk factors of this disease.
Material and Methods
CMHS is a multiethnic cohort of men who are members of the southern and northern California regions of KP, the largest managed care organization in California. Eligible participants included all male KP California members, aged 45–69 years in January 2000, who had been members of the health plan for at least 1 year at recruitment. Nearly 850,000 men met the eligibility criteria. Recruitment was carried out between January 2002 and December 2003 using a 2-step process in 3 waves of mailing. A total of 84,170 men including a nearly 40% minority participation completed the questionnaires and hence joined the cohort. Details of the cohort design and recruitment methods were described previously by Enger et al.29 Between study baseline and December 2007, 14,177 men in the cohort terminated their health plan membership (not including death), resulting in a follow-up rate of 83% by end of 2007.
The questionnaire collected information on demographics, education, income, anthropometrics, family history of cancer, existing health conditions, physical activity, tobacco smoking, diet and vitamin use. A section devoted to men's health collected detailed information on history of prostate-specific antigen (PSA) testing and genital and urologic conditions. Information on alcohol, dietary and multivitamin intake was assessed with a detailed semi-quantitative food-frequency questionnaire adapted from the Women's Health Initiative and other studies, and specifically modified for men.30–33 Frequency of consumption of red wine, white or rosé wine, beer and liquor, as well as the size of the drink, small, medium or large relative to the standard medium serving size (i.e., 12-ounce can or bottle for beer, 6 ounce glass for wine and 1.5 ounce shot for liquor) were elicited.
Ascertainment of prostate cancer
Using unique medical record numbers, we conducted data linkages to identify incident prostate cancer cases through the health plans' cancer registries. Cancer case ascertainment is expected to be highly valid as reporting of cancers is mandated under state law and the KP Cancer Registries fulfill the reporting requirements for the State of California Cancer Registry and the National Cancer Institute (NCI) Surveillance, Epidemiology and End Result (SEER) program. We extracted data on date of diagnosis and stage at diagnosis. A total of 2,028 incident prostate cancer among the CMHS members were identified between study baseline (2002–2003) and December 2007.
Men who had a previous cancer diagnosis (except nonmelanoma skin cancer) based on KP cancer registry records or self-report were excluded from the analyses (N = 7,506). Prostate cancer cases diagnosed within 12 months after study baseline (465 cases) were also excluded to avoid temporal ambiguity of red wine use and other reported behaviors.
We first calculated the distribution of demographics and other potential prostate cancer risk factors by case and noncase status. Crude associations between these variables and prostate cancer were then obtained using univariate Cox model and likelihood ratio test. To understand the associations between consumption of red wine and potential prostate cancer risk factors, distributions of these risk factors were calculated by levels of red wine consumption and evaluated by chi-square test or Kruskal–Wallis test.
Multivariable Cox's regression was used to investigate the association between red wine consumption and time to prostate cancer. Follow up started at the time of the baseline survey and ended at the time of prostate cancer diagnosis, diagnosis of other cancers, termination of health plan membership, death or end of year 2007, which ever came first. Factors that were associated with both prostate cancer status and red wine consumption in these crude analyses (at α = 0.10 significance level, 2 sided) were selected as potential confounders for adjustment in the multivariable model. These include race/ethnicity, household income (annual income of <$40,000, $40,000–$80,000, <$80,000), body mass index (BMI) calculated from reported weight in kilograms divided by the square of height in meters (BMI categories of <18.5, 18.5–24.9, 25–29.9, 30–34.9, ≥35), meat consumption (modeled as categorical variable), family history of prostate cancer, utilization of PSA testing (number of PSA tests in the 5 years before study baseline), sexually transmitted infections (STIs), benign prostate hyperplasia (BPH), BPH surgery, prostatitis/prostate infection and diabetes mellitus. Age was used as the time axis in the proportional hazards model. Although beer, white wine and liquor intake were not associated with prostate cancer status in the crude analyses, they were included in the multivariable model to study the effect of red wine independent of other alcoholic beverage intake. Red wine consumption was modeled by 3 frequency categories: <1 drink/week, ≥1 drink per week but <1 drink per day, ≥1 drink per day, in reference to nondrinkers. The “nondrinkers” group was defined specific to red wine, regardless of the intake of other types of alcoholic beverages. The nondrinkers could include those who drank red wine occasionally (i.e., less than once per month). Red wine consumption was also modeled as a continuous variable in the unit of number of drinks per month. Intake of the other types of alcoholic beverage was modeled similarly. As there was no obvious heterogeneity in the effects of different types of alcoholic beverages, the effect of combined alcohol consumption from these types of alcoholic beverages was modeled. To examine the impact of higher alcohol intake, we modeled 5 frequency categories: <1 drink per day, ≥1 drink but <3 drinks/day, ≥3 drinks but <5 drinks/day and ≥5 drinks/day, in reference to nondrinkers of any alcoholic beverage. A total of 65,972 men (1,340 prostate cancer cases) with complete data for these covariates were included in the multivariable analyses.
To further eliminate the possibility of residual confounding by race/ethnicity, we performed stratified analyses restricted to non-Hispanic whites and African-Americans. Since detection bias is a concern for prostate cancer in observational studies, we also examined the association between red wine intake and prostate cancer in stratified analyses restricted to various subgroups: (i) men who had at least 1 PSA test (diagnostic or screening); (ii) men who did not have any PSA test; (iii) men who had a screening PSA in the 5 years before study baseline; and (iv) men who had a family history of prostate cancer. Screening PSA was defined by excluding PSA tests ordered by urologist, a PSA test due to symptom, or as a follow up to a previously self-reported abnormal test. We also conducted subanalyses restricted to (i) cases diagnosed at Stage 2 or higher; (ii) cases with intermediate grade cancer (moderately differentiated); and (iii) cases with high-grade cancer (poorly differentiated or undifferentiated). All analyses were conducted using SAS statistical software version 9.1.3 (Statistical Analyses System Inc, Cary, NC, 2002–2003).
The 65,972 men in these analyses contributed 304,484 person-years between 2002 and 2007. The distribution of baseline characteristics by prostate cancer status is shown in Table 1. In the crude analysis, red wine intake was not associated with prostate cancer status. Table 2 presents baseline characteristics by level of red wine consumption. Age, race/ethnicity, income, education, body mass index, tobacco smoking, physical activity, dietary intake, use of other types of alcoholic beverages, family history of prostate cancer, history of PSA testing, STIs, BPH, prostatitis/prostate infection and diabetes were all significantly associated with level of red wine consumption. In general, use of red wine was associated with a healthier life style including increased fruit and vegetable intake and physical activity, and less current smoking. Men who drank red wine were also more likely to have a PSA test in the 5 years before study baseline. However, red wine drinkers were also more likely to have certain potential risk factors for prostate cancer, such as history of STIs, BPH and prostatitis. Table 2 also shows the mean red wine consumption level within each frequency category. Among men who consumed at least 1 drink or more of red wine per day, the average consumption was ∼2 drinks per day in that category.
Table 1. Characteristics of CMHS men by prostate cancer status
Table 2. Selected baseline characteristics by level of red wine consumption
In the multivariable analysis, we did not find a clear association between red wine consumption and risk of prostate cancer. Hazard ratio (HR) estimates for consuming <1 drink/week, ≥1 drink/week but <1 drink/day and ≥1 drink/day were 0.89, 95% confidence interval (0.74–1.07), 0.99 (0.83–1.17) and 0.88 (0.70–1.12), respectively (Table 3). Further, we observed no linear dose response pattern (HR = 1.00 (1.00–1.00) for each 1 drink increase per month). In analyses restricted to non-Hispanic whites, African-Americans, men with a history of any PSA testing, men without any PSA testing in the 5 years before study baseline, men with family history of prostate cancer, cases diagnosed at Stage 2–4, cases with intermediate and high-grade cancer, similar HR estimates were observed for red wine intake (Table 3), again suggesting a lack of association between moderate red wine consumption and risk of prostate cancer. Among those who had a screening PSA in the 5 years before study baseline, red wine consumption of at least 1 drink per day was associated with reduced prostate cancer risk (HR = 0.70 (0.49–1.00)). Again, we observed no linear dose-response trend.
Table 3. Adjusted hazard ratio (HR) and 95% CI for prostate cancer by level of red wine consumption
In addition, we did not observe any clear association with prostate cancer for beer, white wine, liquor or combined alcohol consumption with prostate cancer (Table 4) even for up to 5 or more alcoholic beverage drinks per day (for all cases, HR for 5 or more drinks of alcoholic beverage per day was 1.16, 95% CI (0.83–1.63)) (Table 4), with or without restricting by PSA screening/testing history.
Table 4. Adjusted hazard ratio (HR) and 95% CI for prostate cancer by level of beer, white wine, liquor, and combined alcohol consumption, n = 65,972
Our results do not support the hypothesis that red wine intake reduces risk of prostate cancer, at least not at the level of moderate consumption (i.e., 1–2 drinks per day). These results did not confirm the inverse association previously seen by one population based case-control study.24 However, our findings are consistent with three other prospective cohort studies that reported lack of association between moderate red wine intake and risk of prostate cancer.25–27 Of note, previous studies have focused on white men. African-Americans, however, have a higher disease burden of prostate cancer than other races. Our study further determined that there was no association between red wine intake and prostate cancer risk in African-American men.
It is well-documented that the diagnosis of prostate cancer is strongly linked to the use of PSA screening. In our study, we found that red wine drinkers were significantly more likely to have a PSA test than those who did not reported red wine drinking (Table 2). This suggests a potentially strong detection bias in studies where PSA screening behavior is not properly adjusted. Our study helps to fill the gap of current literature by comprehensively adjusting for PSA testing and screening behavior as most previous studies were unable to adjust for PSA screening in satisfactory detail and were unable to separate diagnostic PSA from screening PSA. Although an inverse association was seen between daily red wine drinking and prostate cancer among those who had a screening PSA before study baseline (HR = 0.70 (0.49–1.00)), no association was observed among men who did not have any PSA testing in the same period. Given these results, consumption of up to a mean level of 2 drinks of red wine per day may not confer any important benefit to prostate cancer prevention.
We also did not find any clear association with prostate cancer for other types of alcoholic beverage intake, i.e., beer, white wine and liquors. For total alcoholic beverage intake, we did not find an association for up to 5 or more drinks per day (HR = 1.16 (0.83–1.63)), mean drinking level in this category was 7.5 drinks/day). However, given the majority of the alcohol users in our cohort were moderate drinkers, we were not well powered to study the effect of heavy alcohol intake (there was only 37 cases at consumption level of 5 drinks or more per day). Despite the significant number of studies that have examined the association between total alcohol intake and risk of prostate cancer,26, 34–52 few studies have examined the effect of heavy drinking. Most studies, however, find moderate drinking not linked to excess prostate cancer risk.53, 54
There are several potential limitations to be considered when interpreting our results. First of all, we do not have long-term data on alcohol use, which limited our ability to study the effect of former drinking, duration of drinking and the effect of red wine consumption on prostate cancer initiation. These factors need to be further investigated in future studies. Use of food frequency questionnaire can also result in imprecise measurements for alcoholic beverage intake compared with instruments specifically designed to collect this information. In addition, we used “current nondrinkers” as the reference group, which could include former drinkers in addition to life time abstainers. Because people tend to quit smoking and drinking when clinical symptoms appear (sick quitter effect), using such a mixed reference group could result in biased estimation of the effect of interest.55–57 To mitigate this problem, we excluded prostate cancer cases diagnosed within 12 months of the baseline assessments. Second, members of the CMHS cohort on average reported moderate alcohol consumption. This limited our ability to examine the effect of heavy consumption of red wine and other types of alcoholic beverages. We also did not have sufficient cases diagnosed at advanced stage (Stage 3–4) to examine the effect of red wine consumption on advanced cancers. As we did not capture cause of death information, men who died because of previously undiagnosed prostate cancer were not included in the analysis. However, given the relatively low-overall death rate in this cohort of middle-aged men, it is unlikely that such an exclusion affected our results substantially. Despite these limitations, our study has several important strengths, including the availability of detailed information on alcoholic beverage consumption, PSA screening behavior and other lifestyle/medical history factors, and reliable and sensitive prostate cancer ascertainment. Furthermore, the characteristics of the CMHS cohort was similar to the general male population who participated in a random digit dial survey in California.29 Therefore, our results are likely generalizable to the broader male population in similar age range.
Our study was motivated by the findings from preclinical studies of the anti-cancer effect of polyphenols found in red wine, such as resveratrol.58 Although red wine is a relatively rich source of resveratrol and other antioxidants, the effect of these compounds from red wine in humans remains largely unknown. Resveratrol is quickly metabolized into conjugated forms after ingestion both in human and laboratory animals. A human study demonstrated that it is unlikely for resveratrol to reach sufficient concentration in its original form at prostate gland.59 However, this does not preclude the potential beneficial effect of resveratrol metabolites on prostate. The biological effect of conjugated resveratrol remains unknown and is a subject of current investigation. Should the cancer-preventive bioactivity of resveratrol metabolites be confirmed, further studies to determine the effect of resveratrol intake (e.g., in the form of supplements) on prostate cancer prevention and treatment in humans are warranted.