In this large prospective cohort study of post-menopausal women, we found some evidence of an overall inverse association between decaffeinated coffee and endometrial cancer risk, and no such overall associations for total coffee, or caffeinated coffee. However, within the subgroup of obese women (baseline BMI ≥ 30 kg/m2), we observed a lower risk of endometrial cancer among obese women who drank ≥2 cups/day of coffee. Though these observations were based on a small number of cases, the results suggest that coffee consumption may lower endometrial cancer risk among obese women, who also have the highest absolute risk in comparison to thin women.
Three other studies [17
] have prospectively evaluated coffee consumption and endometrial cancer risk. The Swedish Mammography Cohort, the large study of predominantly post-menopausal women who primarily drank caffeinated coffee, reported an overall 25% lower risk in the highest category of coffee drinkers [17
]. Additionally, they showed that the reduction in risk was strongest among heavy women and not present in normal weight women. Our results are most consistent with this Swedish study as we too found an inverse association between caffeinated coffee consumption and endometrial cancer risk among obese women [17
]. A second prospective study conducted in Japan reported a 62% overall lower risk among women with the highest coffee consumption [18
]. The smallest prospective study in Norway did not show an inverse association between coffee intake and endometrial cancer risk [19
]. However, the referent category in this study included women who drank ≤2 cups of coffee per day and over 60% of the women were pre-menopausal at baseline. Furthermore, neither of the two studies from Japan or Norway evaluated risk by BMI strata. To our knowledge our study is the first to prospectively evaluate decaffeinated coffee and endometrial cancer risk. Our results were slightly suggestive of an inverse association, with heterogeneous results across BMI strata. Only two case-control studies have evaluated decaffeinated coffee and both showed null associations [22
The unopposed estrogen hypothesis provides a compelling explanation for the etiology of most endometrial cancer [5
]. It posits that estrogen exposure stimulates uncontrolled cell division in the endometrium, increasing the likelihood of deleterious mutations that may result in endometrial cancer [24
]. Consistent with this hypothesis, studies indicate postmenopausal women with higher serum estrogens have an increased risk of endometrial cancer [7
]. In addition, lower blood levels of SHBG are also associated with increased endometrial cancer risk [7
], presumably because SHBG binds to estradiol, thereby reducing the levels of bio-available estrogen. Diabetes and hyperinsulinemia are also hypothesized to play a role in endometrial carcinogenesis [1
]. Specifically, insulin may promote endometrial cancer development directly as a mitogen [25
] or indirectly through its effects on estrogen availability [26
]. Several biological mechanisms related to these two pathways could account for a possible inverse association between caffeinated coffee intake and endometrial cancer risk. In line with the well-established estrogen-mediated etiology of endometrial cancer, higher intake of caffeinated coffee or caffeine has been associated with increased blood levels of sex hormone binding globulin in postmenopausal women [10
]. In postmenopausal women, caffeinated coffee intake has also been associated with blood markers of decreased insulin secretion and improved insulin sensitivity, including lower C-peptide levels [29
], and higher adiponectin levels [30
]. In a study by Wu and colleagues, the inverse association between caffeinated coffee consumption and plasma C-peptide levels was stronger in obese and overweight women than in normal-weight women [30
]. This observation is consistent with our findings and those of Friberg and colleagues who showed that the reduction in risk through caffeinated coffee consumption was greatest among heavier women [17
]. Coffee also has a number of different phytochemicals other than caffeine that could also play a role in endometrial cancer carcinogenesis. For example, chlorogenic acid has been suggested to also delay glucose absorption and improve insulin sensitivity [31
]. Other constituents of coffee may act as antioxidants [34
] or have anti-estrogenic properties [35
] that could also have relevance to endometrial carcinogenesis, particularly in heavier women.
We observed an overall association with decaffeinated coffee, although the pattern was not consistently observed across all body mass index categories. Decaffeinated coffee intake has been linked to decreased C-peptide levels in postmenopausal women [29
]. Contrary to observations for caffeinated coffee intake, however, decaffeinated coffee intake has not been linked to elevated blood levels of adiponectin [30
], elevated levels of SHBG [28
], or lower levels of total estrogens [28
]. A possible explanation for inverse association for decaffeinated coffee we observed is chance, as only a small percentage of our population drank decaffeinated coffee daily. Alternatively, our findings for decaffeinated coffee may reflect confounding by unknown factors, as decaffeinated coffee drinkers may be a unique population. The results for decaffeinated coffee are unclear and additional large studies that are able to separate caffeinated and de-caffeinated coffee may shed further light on this issue.
Major strengths of our study include its prospective design, detailed information on coffee consumption and potential covariates, and adjudication of endometrial cancer cases. Due to the prospective nature of the study and limited loss to follow up, the likelihood of recall bias and selection bias are minimal. Additionally, the large study population allowed us to examine the independent effects of caffeinated and decaffeinated coffee consumption. Our study is limited by its reliance on self-reported values for coffee consumption without specification of serving size per cup, which may have attenuated our results to towards the null. We additionally did not have information on other sources of caffeine, such as soft drinks and energy drinks, nor were data available on use of additives such as sugar, honey, and cream. Furthermore, we did not have access to DNA samples and were not able to evaluate potential gene-diet interactions. For example, polymorphisms of CYP1A2
gene are associated with differential metabolism of caffeine [37
] and estrogen [38
]. Examination of such gene-diet interactions in future studies may help to elucidate the underlying mechanisms associated with coffee consumption and endometrial cancer risk.