Coffee contains a considerable amount of burned material, including the mutagen (Kasai et al., 1982) and carcinogen (Nagao et al., 1986a) methylglyoxal at approximately 500 to 1,000-µg/cup. It also contains the natural mutagen chlorogenic acid (Stich et al., 1981c), the highly toxic atractylosides (Nagao et al., 1986b), the glutathione transferase inducers kahweal palmitate and cafestol palmitate (Lam et al., 1982), and about 100 mg of caffeine (Ames, 1983).
Several cohort studies have been conducted to assess the risk of bladder cancer from coffee consumption. Because of the relative rarity of this disease, results are generally based on small numbers of observed cases. In a cohort study of nearly 24,000 Seventh-Day Adventists, Snowdon and Phillips (1984) found a positive association between deaths from bladder cancer and coffee consumption by people who never smoked. In two other large cohort studies, one in Norway (Jacobsen et al., 1986) and one in Hawaii (Nomura et al., 1986), investigators failed to find an association.
Case-control studies provide more meaningful data. In two early studies in the United States (Cole, 1971; Fraumeni et al., 1971), elevated risks for bladder cancer were found among coffee drinkers. In the study by Cole, these risks were restricted to females; in the study by Fraumeni and colleagues, the risks applied to black females; in neither study was there evidence of a dose-response relationship. A finding similar to that of Cole (1971) was reported by Simon et al. (1975), who conducted a case-control study among women in Boston. In two other case-control studies, one in Canada (Howe et al., 1980) and one in the United States (Mettlin and Graham, 1979), elevated risks from coffee were found among males; there was less evidence for an effect among females, but in neither study was a dose-response relationship observed. In contrast, no overall effects of coffee drinking were found in case-control studies conducted in the United States, Great Britain, and Japan (Morrison et al., 1982b; Ohno et al., 1985); in Copenhagen Jensen et al., 1986); and in Canada (Risch et al., 1988). In three case-control studies, associations were found between coffee consumption and risk of bladder cancer: in Greece (Rebelakos et al., 1985), in Connecticut (Marrett et al., 1983), and in a 10-center study in the United States (Hartge et al., 1983). In Connecticut, there was a statistically significant association in males and some evidence of a dose-response relationship. In the 10-center study, based on nearly 3,000 cases and 6,000 controls, there was a statistically significant overall relative risk of 1.4, but there was no evidence of a dose-response relationship.
Thus, the epidemiologic studies relating coffee drinking to risk of bladder cancer are inconsistent. In the positive studies, the observed relative risk is small and there is an absence of a dose-response relationship, leading to the conclusion that this association is unlikely to be causal. Residual confounding by cigarette smoking is the most likely explanation for the findings (Morrison et al., 1982b). In this context, a case-control study among nonsmokers that failed to find an association between coffee drinking and bladder cancer (Kabat et al., 1986) is particularly relevant.
Lack of an association between tea consumption and bladder cancer risk has been reported consistently in cohort and case-control studies (Hartge et al., 1983; Heilbrun et al., 1986; Howe et al., 1980; Miller et al., 1983; Morgan and Jain, 1974; Sullivan, 1982).
In the cohort study of nearly 24,000 Seventh-Day Adventists by Snowdon and Phillips (1984), no association was observed during the first 10 years, but a positive association was found between coffee consumption and fatal colon cancer during the last 11 years of follow-up. People consuming two or more cups of coffee a day had a relative risk of 1.7 (95% confidence interval, range 1.1 to 2.5) compared to those consuming less than one cup a day. There was a dose-response relationship in the last 11 years, which persisted when meat consumption was included in a multivariate analysis. A statistically significant but small excess risk was found for colon cancer but not for rectal cancer in the case-control study by Graham et al. (1978). In neither of these studies were detailed dietary data collected. It is therefore possible that the association with coffee reflects confounding with another factor, such as dietary fats, that could not be evaluated.
Two case-control studies showed no statistically significant association between tea consumption and colorectal cancer (Phillips and Snowdon, 1985; Tajima and Tominaga, 1985) as did a Canadian case-control study for consumption of