Alcohol consumption and risk of bladder cancer in Los Angeles County
Version of Record online: 17 APR 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 121, Issue 4, pages 839–845, 15 August 2007
How to Cite
Jiang, X., Castelao, J. E., Groshen, S., Cortessis, V. K., Ross, R. K., Conti, D. V. and Gago-Dominguez, M. (2007), Alcohol consumption and risk of bladder cancer in Los Angeles County. Int. J. Cancer, 121: 839–845. doi: 10.1002/ijc.22743
- Issue online: 22 JUN 2007
- Version of Record online: 17 APR 2007
- Manuscript Accepted: 20 FEB 2007
- Manuscript Received: 1 DEC 2006
- National Cancer Institute, NIH. Grant Numbers: P01 CA17054, R35 CA53890, R01 CA65726
- National Institute of Environmental Health Sciences, NIH. Grant Numbers: P01 ES05622, P30 ES07048
- bladder cancer;
- cigarette smoking;
- Los Angeles
The role of alcoholic beverages in bladder carcinogenesis is still unclear, with conflicting evidence from different studies. We investigated the relationship between alcohol consumption and bladder cancer, and the potential interaction between alcohol consumption and other exposures. In a population-based case–control study conducted in Los Angeles County, 1,586 pairs of cases and their matched neighborhood controls were interviewed. Data were analyzed to determine whether bladder cancer risk differs by alcohol consumption, and whether different alcoholic beverages have different effects. The risk of bladder cancer decreased with increasing frequency (p for trend = 0.003) and duration of alcohol consumption (p for trend = 0.017). Subjects who drank more than 4 drinks per day had a 32% lower (odds ratio, 0.68; 95% confidence interval, 0.52–0.90) risk of bladder cancer than those who never drank any alcoholic beverage. Beer (p for trend = 0.002) and wine (p for trend = 0.054) consumption were associated with reduced risk of bladder cancer, while hard liquor was not. The reduction in risk was mostly seen among shorter-term smokers who urinated frequently. Alcohol consumption was strongly associated with a reduced risk of bladder cancer. The effect was modified by the type of alcoholic beverage, cigarette smoking and frequency of urination. © 2007 Wiley-Liss, Inc.
The role of alcoholic beverages in bladder carcinogenesis has been investigated in several epidemiologic studies.1 No significant association was found in most studies including the recent publication from the Framingham Heart Study.2 However, a meta-analysis of urinary tract cancers suggested a slightly increased risk of 1.3 for current alcohol-drinkers, when compared with nondrinkers.3 Results of beverage-specific analyses are also inconclusive. Some studies indicated higher risks from spirit/liquor consumption,2, 4, 5, 6 and others found an inverse association with beer consumption.2, 7 Despite the large number of published studies, questions remain about the definite role of alcohol on bladder carcinogenesis, the effect of the different alcoholic beverages, drinking pattern, and modification by factors such as cigarette smoking and frequency of urination. We collected detailed information on alcoholic beverage intake in a large population-based case–control study of bladder cancer in Los Angeles County, where we have identified risk/protective factors, such as cigarette smoking,8 use of nonsteroidal anti-inflammatory drugs (NSAIDs),9 consumption of carotenoids and vitamin C10 and use of permanent hair dyes among women.11, 12 Results from this study will help clarify the role of alcohol intake in bladder cancer.
Material and methods
From 1987 to 1999, we conducted a population-based case–control study of bladder cancer in Los Angeles County using in-person interviews.8 Eligibility criteria included histologically confirmed bladder cancers diagnosed between January 1, 1987 and April 30, 1996 among non-Asians between the ages of 25 and 64 years. Cases were identified through the Los Angeles County Cancer Surveillance Program (SEER registry).13 In total, 2,384 cases were identified. Two hundred and ten (9%) died before we could contact them or were too ill to be interviewed. Permissions to contact 99 (4%) patients were refused by their physicians. Four hundred and four patients (17%) refused to participate in the study. For each enrolled case patient, a standard procedure was followed to recruit a control subject who was matched to the index case by age (within 5 years), sex, race (non-Hispanic white, Hispanic white or African American/others) and neighborhood of residence at the time of cancer diagnosis. We attempted to identify the sex, age, and race of all inhabitants of each housing unit; “not at home” units were systematically revisited to complete the census. The first resident along this defined route who satisfied all eligibility criteria for controls was asked to participate in this study (i.e., first eligible control). If the individual refused, the next eligible control (i.e., second eligible control) in the sequence was asked and so on until we located an eligible control who agreed to be interviewed. When we failed to find any resident who met our matching criteria after canvassing 150 housing units, we excluded race from the matching criteria. If a matched control subject based on this relaxed criterion could not be found within a maximum of 300 housing units, the case patient was dropped from the study. We interviewed 1,671 (70%) eligible patients and a matched control could not be found for 5% (85) of interviewed cases. Twenty-one control subjects were not matched to the index case by race. One thousand and ninety (69 %) control subjects were first eligible controls, 325 (20%) were second eligible controls and 111 (7%) were third eligible controls. The remaining 60 (4%) control subjects were 4th or higher order eligible controls. Therefore, a total of 1,586 pairs of cases and controls were included in the analyses. All study subjects signed informed consent forms, approved by the Human Subjects Committee at the University of Southern California, Keck School of Medicine.
Data gathering and exposure definitions
In-person, structured interviews were conducted in subjects' homes. The questionnaire requested information up to 2 years prior to the diagnosis of cancer for cases and 2 years prior to the diagnosis of cancer of the index case for matched controls (i.e., reference year). Each subject was asked to report information on demographic characteristics, height, weight, lifetime use of tobacco products and alcohol, usual adult dietary habits, lifetime occupational history, prior medical conditions and prior use of medications. All subjects were asked about consumption of water, coffee, tea, alcohol, milk and juice. Questions on consumption of hot chocolate and soda (such as Coca Cola, 7-Up, Root Beer, etc.), and specific-alcoholic beverages (beer, wine, and hard liquor) were not included in the original questionnaire, but were added to interviews conducted after January 1992. Regular drinkers were defined as those who drank the specified beverage at least once a week for 6 months or longer. For regular drinkers, information on the age at starting to drink regularly and frequency and duration of drinking was obtained. Consumption of juice was estimated based on individual questions on frequency of intakes of orange juice, grapefruit juice, and fortified fruit drinks/orange juice substitutes. Intake of milk was estimated as the summarized intake of whole milk, 2% milk and skim milk. For juice, milk and water, one glass (8 oz) was considered as one serving. For coffee, tea, alcohol, hot chocolate and soda, each subject was asked if they ever drank a specific beverage (one cup, 8 oz) on a regular basis. Subject was also asked the specific type of coffee (regular or decaffeinated), tea (black, green, herbal), soda (regular or diet, with caffeine or caffeine-free) they regularly drank. One alcoholic drink is defined as one 12-ounce can of beer (360 ml), one 4-ounce glass of wine (120 ml) or one shot of hard liquor (45 ml) whose alcoholic contents are similar (about 13 g of ethanol per drink).14, 15, 16 A total of 1,133 bladder cancer cases and 1,148 controls were asked about their frequency of beverage-specific drinking.
The association of bladder cancer with alcohol consumption was measured by odds ratios (ORs) and corresponding 95% confidence intervals (CIs) using conditional logistic regression. Total consumption of alcoholic beverages was calculated as the sum of drinks of beer, wine and hard liquor. Total fluid intake was calculated as the sum of water, coffee, tea, juice and milk intake. For subjects interviewed after 1992, intake of hot chocolate, soda, beer, wine and hard liquor was also added to total fluid intake. When the case patient or the control subject of a pair failed to answer the relevant questions, we eliminated that case–control pair from the corresponding conditional analysis. We first fitted models with no covariates, thus the analyses were only adjusted for matching factors (age, sex and race). We then repeated the analyses with further adjustment for the following identified risk or protective factors for bladder cancer: average number of cigarettes smoked per day, number of years of smoking, smoking status in reference year (smoker or nonsmoker),8 level of education (high school or below, 1–3 years of college, college graduate), lifetime use of NSAIDs (non/irregular user, <1441 pills, ≥1441 pills over lifetime),9 carotenoid intake (quintiles)10 and duration of employment as a hairdresser or barber (years).11 Results were similar for those with and without adjustment for level of education. There were no material changes in the results with further adjustment for hair dye use and thus this variable was not retained.
We also examined the possibility that the alcohol-bladder cancer association could be modified by cigarette-smoking status, sex and frequency of urination during the day. Unconditional logistic regression was used in the subset analyses by smoking status (lifetime nonsmokers, smokers of <30 years, smokers of ≥30 years) and frequency of urination during the day (<4 times, ≥4 times/day). In addition to all the covariates mentioned above, 10 age–sex strata (age groups of <46, 46–50, 51–55, 56–60 and >60 years for each sex), and race (non-Hispanic white, Hispanic white or African American/others) were included in the regression models to account for the matched sampling. Modification by sex was addressed by conditional logistic regression, since the cases and controls were designed to match on sex. For analyses of beverage-specific consumptions, all 3 types of beverage were included in the conditional analyses, which were also adjusted for other covariates mentioned above.
ORs with 2-sided p values less than 0.05 were considered statistically significant. All p values presented are 2-sided.
A total of 1,586 case–control pairs were included in the current analyses. The median age of the patients at diagnosis of bladder cancer was 58 years. Most patients were non-Hispanic whites (n = 1463), with the remaining being Hispanic whites (n = 69), African-Americans (n = 53) and Native Americans (n = 1). On average, bladder cancer patients had a lower level of education than controls. The OR for bladder cancer was 0.48 (95% CI, 0.39–0.58) for those who had at least 4 years of college education (≥16 years of schooling) and 0.69 (95% CI, 0.58–0.83) for those who had 1–3 years of college education (13–15 years of schooling) compared with those who only had a high school education or lower (≤12 years of schooling). Thus, all ORs presented below were adjusted for level of education (high school or below, 1–3 years of college, college graduate).
Table I summarizes the baseline characteristics of the study subjects. Regardless of their disease status, subjects who drank more alcohol, especially those who drank more than 4 drinks per day, were more likely to be male, less educated, current heavy smokers, regular users of NSAIDs and drinkers of other fluids. Among controls, intake of alcoholic beverages was associated with intake of coffee, soda and hot chocolate, inversely associated with intake of juice and not associated with intake of water, tea and milk. However, alcohol drinkers and nondrinkers were similar in age (mean age 55–57 years), race distributions (90–94% non-Hispanic white), intake of carotenoids and occupation.
|Category of total alcohol consumption(drinks/day)||Controls (N = 1,5771)||Cases (N = 1,5731)|
|Number of subjects||453||385||505||234||432||364||512||265|
|All other beverages (ml/day)2||2043||2057||2214||2689||2300||2114||2269||2721|
|Years of education (%)|
|≤12 (High school or below)||33||25||27||46||44||31||42||51|
|13–15 (1–3 years of college)||27||29||31||28||28||31||28||28|
|≥16 (College graduate)||40||46||43||27||29||38||31||21|
|Mean frequency (drinks/day)||0||0.47||2.03||9.33||0||0.50||2.13||9.73|
|Mean duration (years)||0||26.4||29.2||29.8||0||25.9||29.3||29.4|
|Age at 1st drinking||/||23.5||22.5||20.1||–||23.2||22.3||20.4|
|Smoking status (%)|
|NSAIDs intake over lifetime (%)|
|Carotenoid intake (%)|
|Hair dresser (%)||1||1||1||1||2||1||1||2|
Total fluid intake was not associated with bladder cancer in our study (multivariate OR for an increase of 240 ml in total daily fluid intake, 1.00; 95% CI, 0.98–1.01). The OR was 1.04 (95% CI, 0.78–1.38) for the highest quintile of total daily fluid intake (>3,229 ml/day) as compared with the lowest quintile (<1420 ml/day). There was no material difference between subjects who were interviewed with the old questionnaire before 1992 and those who were interviewed with the new questionnaire after 1992. However, there was strong evidence of an inverse association between consumption of alcoholic beverages and bladder cancer risk after adjustment for smoking (Table II). Additional adjustment for high-risk occupation, level of education, use of NSAIDs and intake of carotenoids did not change the association. The risk of bladder cancer decreased significantly with the frequency (p for trend = 0.003) and duration (p for trend = 0.017) of alcohol consumption. Subjects who drank more than 4 drinks of alcoholic beverages per day had a 32% lower risk of bladder cancer compared to subjects who never drank any alcoholic beverage (OR, 0.68; 95% CI, 0.52–0.90). This association persisted after we restricted the analyses to subjects interviewed with the new questionnaire only. Further adjustment for total fluid intake (excluding alcoholic beverages) did not materially change our results. We also examined the alcohol-bladder cancer association by volume (data are only available for subjects interviewed with new questionnaire). An increase of 240 ml in alcoholic beverage intake was associated with 4% reduction in bladder cancer risk (OR, 0.96; 95% CI, 0.93–0.99).
|Alcohol consumption||Number of cases/controls||OR (95% CI)1|
|Crude2||Adjusted for smoking3||Multivariate4||Adjusted for total fluid5|
|<1 drinks/day||364/385||0.99 (0.82–1.21)||0.85 (0.68–1.05)||0.85 (0.68–1.06)||0.85 (0.68–1.07)|
|1–4 drinks/day||512/505||1.07 (0.89–1.30)||0.75 (0.60–0.93)||0.77 (0.62–0.96)||0.78 (0.62–0.98)|
|>4 drinks/day||265/234||1.19 (0.95–1.50)||0.66 (0.51–0.86)||0.68 (0.52–0.90)||0.67 (0.51–0.89)|
|1–20 years||311/303||1.08 (0.87–1.33)||0.81 (0.64–1.03)||0.83 (0.65–1.05)||0.82 (0.64–1.06)|
|21–30 years||275/292||0.99 (0.79–1.23)||0.71 (0.55–0.90)||0.72 (0.56–0.92)||0.71 (0.55–0.93)|
|31–40 years||376/338||1.19 (0.97–1.47)||0.85 (0.67–1.08)||0.87 (0.68–1.10)||0.88 (0.69–1.13)|
|41+ years||188/199||1.00 (0.77–1.30)||0.64 (0.47–0.86)||0.66 (0.48–0.89)||0.67 (0.49–0.92)|
|Age at 1st use|
|25+ years||296/294||1.06 (0.86–1.31)||0.84 (0.67–1.07)||0.86 (0.68–1.09)||0.82 (0.64–1.04)|
|18–24 years||607/605||1.07 (0.89–1.29)||0.76 (0.62–0.94)||0.78 (0.63–0.96)||0.80 (0.64–1.00)|
|<18 years||247/232||1.13 (0.90–1.42)||0.69 (0.53–0.90)||0.70 (0.54–0.92)||0.74 (0.56–0.98)|
To test a possible difference in alcohol-related risk of bladder cancer by smoking status (i.e. an interaction effect), a product term of frequency of alcohol consumption (0, <1, 1–4, >4 drinks/day) and duration of cigarette smoking (0, <10, 10 to <20, 20 to <30, 30 to <40, ≥40 years) was used in a conditional logistic regression. We found the association to be significantly different by smoking status (p for interaction = 0.030; df = 15). No significant associations between consumption of alcoholic beverages and bladder cancer risk were found among nonsmokers (Table III). Most of the reduction in bladder cancer risk from alcohol consumption was seen among smokers, and this was primarily confined to shorter-term smokers, i.e. smokers of less than 30 years. The risk of bladder cancer was 0.46 (95% CI, 0.30–0.70) times lower among shorter-term smokers who drank at least 4 drinks of alcoholic beverages per day compared to shorter-term smokers who did not drink any alcoholic beverages. There was no significant association between consumption of alcoholic beverages and bladder cancer risk among subjects who had been smoking for more than 30 years (p for trend = 0.46). Consumption of alcoholic beverages was also a modifier of the smoking-bladder cancer association. Alcohol drinkers had lower risk from regular cigarette smoking (OR, 1.95; 95% CI, 1.57–2.43), compared to nondrinkers (OR, 4.34; 95% CI, 3.19–5.90). These associations were similar when the analyses were restricted to men only. The association between consumption of alcoholic beverages and bladder cancer was not significant among females (p for trend = 0.27, data not shown), and it was not modified by smoking. However, consumption of alcoholic beverages among females was not as common as among males. There were only 35 females who reported drinking more than 4 drinks of alcohol per day.
|Alcohol (drinks/day)||Nonsmokers (N = 856)2||Smokers (<30 years) (N = 1131)3||Smokers (≥30 years) (N = 1163)3|
|Case/control||OR (95% CI)1||Case/control||OR (95% CI)1||Case/control||OR (95% CI)1|
|<1||89/141||1.67 (1.17–2.40)||152/157||0.75 (0.52–1.08)||123/87||0.69 (0.47–1.03)|
|1–4||68/133||1.32 (0.89–1.95)||181/227||0.58 (0.41–0.82)||263/145||0.92 (0.65–1.31)|
|>4||17/32||1.22 (0.62–2.40)||76/98||0.46 (0.30–0.70)||172/104||0.77 (0.52–1.15)|
|<1||22/36||1.63 (0.83–3.22)||32/33||1.29 (0.62–2.71)||32/19||0.52 (0.24–1.15)|
|≥1||8/18||1.14 (0.44–2.99)||31/43||0.89 (0.43–1.85)||53/21||0.75 (0.36–1.57)|
|<1||67/105||1.74 (1.13–2.68)||120/124||0.64 (0.41–0.99)||91/68||0.73 (0.45–1.17)|
|1–4||60/116||1.35 (0.87–2.09)||155/191||0.51 (0.34–0.77)||228/128||0.96 (0.63–1.44)|
|>4||17/31||1.28 (0.64–2.57)||71/91||0.38 (0.24–0.63)||154/100||0.74 (0.48–1.16)|
Because alcohol is a strong diuretic, we next examined the alcohol-bladder cancer association by frequency of urination, which was requested only in the new questionnaire used in interviews conducted after January 1992. The effect of alcohol consumption was marginally modified by frequency of urination during the daytime (p for interaction = 0.057; Table IV). Among shorter-term smokers who urinated less frequently (<4 times per day, median urination frequency among controls), no association was found between consumption of alcoholic beverages and bladder cancer risk (p for trend = 0.91). However, among shorter-term smokers who urinated more frequently (≥4 times per day), consumption of alcoholic beverages was inversely associated with bladder cancer risk (p for trend <0.0001); drinking at least 4 drinks per day significantly reduced the risk of bladder cancer by 76% (OR, 0.24; 95% CI, 0.12–0.47).
|Alcohol (drinks/day)||Daytime urination <4 times/day||Daytime urination ≥4 times/day|
|Case/control||OR (95% CI)1||Case/control||OR (95% CI)1|
|<1||31/29||1.46 (0.61–3.47)||64/60||0.58 (0.32–1.06)|
|1–4||31/35||1.04 (0.45–2.44)||71/94||0.38 (0.21–0.67)|
|>4||16/17||1.24 (0.44–3.50)||34/60||0.24 (0.12–0.47)|
Beverage-specific data was requested only by the new questionnaire used in interviews conducted after January 1992. Each beverage had a different impact on the risk of bladder cancer (Table V). There was a significantly decreased risk associated with beer drinking (p for trend = 0.002), after adjustment for other types of alcoholic beverages, i.e. wine and hard liquor. Subjects who drank more than 4 cans of beer per day had a 0.54 (OR, 0.54; 95% CI, 0.35–0.83) times lower risk of bladder cancer compared to subjects who did not drink any beer. Wine consumption was also associated with a slightly lower risk of bladder cancer (p for trend = 0.054), although the numbers were very small (only 15 cases and 16 controls drank more than 4 glasses of wine per day). Drinking at least 1 glass of wine per day was associated with 0.59 (OR, 0.59; 95% CI, 0.44–0.81) times lower risk of bladder cancer. No association was found between hard liquor consumption and bladder cancer (p for trend = 0.85). We also examined the alcoholic beverage-specific associations by volume (in 240 ml). Results were materially unchanged, and beer was the only alcoholic beverage statistically significantly associated with a lower risk of bladder cancer.
|Alcohol (drinks/day)||Number of cases/controls1||OR (95% CI)2||OR (95% CI)3|
|<1||243/266||0.83 (0.65–1.06)||0.81 (0.61–1.07)|
|1–4||141/152||0.81 (0.61–1.08)||0.68 (0.49–0.95)|
|>4||82/78||0.88 (0.61–1.28)||0.54 (0.35–0.83)|
|Each 240 ml/day||1.00 (0.98–1.03)||0.96 (0.93–0.99)|
|<1||231/266||0.72 (0.57–0.91)||0.84 (0.64–1.09)|
|1–4||86/115||0.57 (0.41–0.79)||0.65 (0.44–0.95)|
|>4||15/16||0.77 (0.37–1.60)||0.91 (0.41–2.02)|
|Each 240 ml/day||0.88 (0.76–1.02)||0.92 (0.80–1.05)|
|<1||251/245||1.33 (1.06–1.69)||1.18 (0.90–1.55)|
|1–4||156/135||1.57 (1.18–2.10)||1.01 (0.72–1.41)|
|>4||71/54||1.74 (1.16–2.63)||1.01 (0.63–1.62)|
|Each 240 ml/day||1.36 (1.09–1.70)||1.02 (0.81–1.28)|
More than 80% of our bladder cancer cases are Ta and T1 (i.e., superficial or noninvasive bladder cancers). Only less than 20% are T2+ (i.e., invasive bladder cancers). Nonetheless, we examined the association between alcohol and bladder cancer separately for superficial and invasive bladder cancers and the alcohol-related bladder cancer protection was mostly confined to superficial bladder cancers (p for trend = 0.003 and 0.88, for superficial and invasive bladder cancers, respectively).
Most previous studies reported either null associations2, 6, 17, 18, 19, 20, 21, 22, 23, 24 or slightly positive associations4, 5, 25, 26, 27, 28, 29 between consumption of alcoholic beverages and risk of bladder cancer. Our results regarding beverage-specific consumption are in agreement with some, but not all, previous studies. Beer was found to be significantly associated with lower risk of bladder cancer in both the Framingham Heart Study (>4 drinks/day vs. nondrinkers, OR, 0.50; 95% CI, 0.2–0.8; p for trend = 0.03)2 and an Italian study (drinkers vs. nondrinkers, OR, 0.69; 95% CI, 0.52–0.92)7 with a similar magnitude of reduction in risk. Other studies found that liquor or spirit was associated with higher risk of bladder cancer,2, 4, 5, 6 which is consistent with our findings among female subjects (OR associated with ≥1 shot of hard liquor, 2.39; 95% CI, 0.97–5.89), although the number of female subjects in our study is too small to be conclusive.
Alcohol consumption is known to be associated with an increase in risk of cancers of the mouth, oropharynx, esophagus, liver and breast, with the risk of these diseases increasing with higher volume of consumption.30, 31, 32, 33 The formation of acetaldehyde-DNA adducts, reactive oxygen and nitrogen species, and the impairment of methionine-folate metabolism and DNA methylation by alcohol, have been proposed as possible mechanisms for alcohol-related carcinogenesis.32, 33 On the other hand, moderate consumption of alcoholic beverages have been found to protect against coronary heart disease, diabetes mellitus,30, 34 renal cell carcinoma,35, 36, 37, 38, 39, 40 prostate cancer41 and non-Hodgkin's lymphoma,42 perhaps through mechanisms such as improved immune response and increased insulin sensitivity. One possible mechanism that may be responsible for the alcohol-mediated bladder cancer protection is the urogenous- contact hypothesis,17, 43, 44, 45 which we will discuss below. Other mechanisms, perhaps not as well supported by the data, are the anti-inflammatory properties of alcohol,46, 47 and the anticarcinogenic effects of polyphenols in red wine48 and beer.49, 50, 51
The following lines of evidence suggest that the urogenous-contact hypothesis, which associates bladder cancer risk with prolonged exposure to carcinogens in the urine,17, 43, 44, 45 may play a role in the alcohol-mediated bladder cancer protection. First, the fact that beer was the beverage associated with the highest reduction in risk suggests that the amount of fluid may play a role in this association. Beer is consumed in greater volume than the other types of alcoholic beverages (1 can of beer 360 ml; 1 glass of wine 120 ml; 1 shot of hard liquor 45 ml). Second, the analyses stratified by frequency of urination, in which alcohol appeared to be most protective among individuals who had a high frequency of urination, also point to the urogenous-contact hypothesis playing a role in this association. Third, previous reports document the diuretic properties of alcohol in both experimental animals and human, with acute alcohol consumption increasing urine flow,52, 53 possibly by inhibiting antidiuretic hormone.54 Studies have shown that even in chronic alcohol abusers, an acute alcohol load can induce a diuretic response similar to the observed in nonalcoholic controls.55, 56 These 3 lines of evidence suggest that increased fluid intake and frequency of urination may play a role in the alcohol-mediated bladder cancer protection by decreasing the time the bladder is exposed to carcinogens in the urine.
Coffee drinking, which has also been postulated to have a diuretic effect, was not associated with bladder cancer risk in our study. The available literature has shown that doses of more than about 250–300 mg caffeine (equivalent to 2–3 cups of coffee) have a mild diuretic effect, while doses of less than 250 mg do not show any effect.57, 58 Thus, doses of caffeine equivalent to the amount normally found in standard servings of tea, coffee and carbonated soft drinks appear to have no diuretic action.57 In contrast, alcohol has a more potent diuretic effect and 1 gram is sufficient to increase urine output by ∼10 ml.52
Cigarette smoking was a modifier of the alcohol-bladder cancer association. Consumption of alcoholic beverages appeared most protective among shorter-term smokers, less protective among longer-term smokers and had no effect among nonsmokers. One could hypothesize that smokers would have benefited more from alcohol consumption than nonsmokers, if the mechanism of protection involves reducing exposure to cigarette-derived carcinogens in urine. Accordingly, we would expect the much lower level of carcinogens to preclude our observing the beneficial effects of alcoholic beverages among nonsmokers. On the other hand, the protective effect of alcohol by increasing urine flow and reducing exposure in urine to carcinogens may be overwhelmed in longer-term smokers, in which the levels of carcinogens are too high to be eliminated by these anticarcinogenic properties of alcohol.
The present study is a population-based case–control study, therefore designed to minimize selection bias. However, nonresponse rate was 31% among first eligible controls and only 17% among cases, leading to a potential selection bias problem. It is possible that nonresponse is related to a less altruistic, less healthy and thus drinking behavior, which would mean that alcohol consumption could be underestimated among our participating cases and controls, or more so among our controls (as one may expect because the cases have disease which may make them all more cooperative), but that underestimation would upwardly bias our estimates, and thus become a force moving them toward the null, given that the associations we observed are inverse. We also believe that recall bias has probably not played an important role in our findings. Even though consumption of alcoholic beverages has been related to increased illnesses, it is not an established risk/protective factor for bladder cancer. One limitation of our study is the lack of data on drinking pattern, such as consuming the same total weekly amount of alcohol over a period of 1 or 2 days as compared with several days, which is correlated with other lifestyle factors and could have impacted on the effect of alcoholic beverages, for example, by modifying the diuretic effect of alcohol. To our knowledge, no studies that looked at the possible association between alcohol consumption and bladder cancer have examined the effect of drinking pattern. In addition, because we did not collect information about the type of wine consumed, we were unable to test the hypothesis that a protective effect of red wine may have been due to its high polyphenol content. Even though residual confounding from cigarette smoking has been a serious concern for most of the studies of alcohol that reported positive associations, it is not likely to explain the inverse associations we found between alcohol consumption and bladder cancer risk. However, confounding may have resulted from unmeasured socioeconomic and other lifestyle factors that differentiate drinkers and nondrinkers of beer, wine and hard liquor. As in any other epidemiological study, our findings may have been due to chance, although we believe this is not likely considering the consistency of our findings. For our large well-established population-based case–control study, we were able to match our cases and controls on age, sex, race and neighborhood of residence. Because we had also available data describing consumption of fruit and vegetable, use of NSAIDs and use of hair dye, we were previously able to identify these factors as risk or protective factors for bladder cancer and subsequently control for them in our current analyses.
In conclusion, our data showed that consumption of alcoholic beverages was associated with a decreased risk of bladder cancer. Beer was the beverage offering the highest protection. There was also some suggestive evidence of a moderate association between wine consumption and reduced risk of bladder cancer. The urogenous-contact hypothesis and the anti-inflammatory properties of alcohol are proposed as possible mechanisms for the alcohol-mediated bladder cancer protection. More well-designed epidemiologic studies are needed to confirm these findings.
- 1Alcohol Drinking. Biological data relevant to the evaluation of carcinogenic risk to humans. IARC Monogr Eval Carcinog Risks Hum 1988; 44: 101–52.
- 13Cancer in Los Angeles County. Los Angeles, CA: University of Southern California, 1991.,
- 15Agriculture USDo. Nutrient database for standard reference, release 10th edn. Bethesda, MD: Human Nutrition Information Service, USDA, 1992.
- 16Agriculture USDo. USDA nutrient database for individual food intake surveys and data sets, release 7th edn. Springfield, VA: National Technical Information Service, 1994.
- 32ChoCH,PurohitV, eds. Alcohol, tobacco and cancer. Basel, Switzerland: Karger, 2006.