Numerous mechanisms for the effects of coffee, tea and caffeine on the risk of breast cancer have been suggested. For instance, caffeine intake has been associated with high plasma levels of female hormones(Reference Nagata, Kabuto and Shimizu1, Reference Ferrini and Barrett-Connor2). In epidemiological studies consumption of coffee, tea and caffeine has been hypothesized to either increase or decrease the risk of developing breast cancer(Reference Bhoo, Peeters and van Gils3–Reference Vatten, Solvoll and Loken9), but associations have not been clearly demonstrated. A recent meta-analysis found no association after summarizing results from cohort and case–control studies(Reference Tang, Zhou and Wang10). However, two major cohort studies showed a possible increased risk in postmenopausal women who drank more than 4 cups of coffee daily(Reference Ganmaa, Willett and Li7) and in women with tumours that were either negative for oestrogen- and progesterone-receptor status or larger than 2 cm(Reference Ishitani, Lin and Manson11). In contrast, a recent review evoked a slight beneficial effect of high consumption of coffee on premenopausal breast cancer risk(Reference Nkondjock12). This potential relationship remains to be evaluated in a large prospective cohort with detailed dietary information.
Experimental methods
The design of the Etude Epidémiologique auprès des Femmes de la Mutuelle Générale de l'Education Nationale (E3N) cohort has been described previously(Reference Tehard, Friedenreich and Oppert13). Briefly, the cohort is composed of 98 995 women living in France who were between 40 and 65 years of age at the time of recruitment in 1990 and insured by the national health insurance system primarily covering teachers. Demographic, anthropometric, reproductive, lifestyle and medical data were collected in self-administered questionnaires sent at baseline and then biennially for follow-up. Usual diet over the previous year was assessed using a 208-item dietary history questionnaire sent out in 1993, developed and validated for the French population(Reference van Liere, Lucas and Clavel14). Portion sizes were estimated using a photo album, also validated(Reference Lucas, Niravong and Kaaks15); for coffee and tea, six different cup sizes could be selected at each meal (70, 150, 200, 250, 300 or 400 ml) and were then converted into standardized 125 ml cups/d for the analysis. Food consumption was converted into nutrient intakes by means of a food composition table derived from the French national food composition table. We estimated the average daily caffeine intake by computing the consumption of all food sources of caffeine and the caffeine content found in the literature for these foods; in our study, the caffeine contents were 72·8 mg/125 ml coffee, 13·8 mg/125 ml tea, 16·1 mg/125 ml chicory beverage, 17·6 mg/330 ml cola, 2·1 mg/30 g chocolate candy and 17·8 mg/30 g plain chocolate.
Women with complete dietary data, with no history of cancer (except for skin basal cell carcinoma, breast lobular carcinoma in situ or colorectal cancer in situ) and with follow-up after the dietary questionnaire were selected for the present analysis. Women with extreme, physiologically implausible dietary values (i.e. in the bottom or top 1 % of the ratio between energy intake and required energy calculated after taking into account age, weight and height) were excluded, leaving 67 703 women for analysis.
Participants contributed person-years of follow-up starting from the date at which they had completed the dietary questionnaire up to the date of diagnosis of invasive breast cancer as first primary cancer (for cases), date of diagnosis of another cancer, date of the last completed questionnaire or July 2005, whichever came first. Because incident cases of cancer were self-reported by the women in follow-up questionnaires, pathology reports were systematically requested from patients or their doctors to confirm the diagnosis of invasive breast cancer. Histological confirmation was obtained for 96 % of cases.
Statistical analyses
Multivariable Cox proportional hazards regression models were used to estimate hazards ratios (HR) of breast cancer and their 95 % confidence intervals associated with the number of cups (defined as 125 ml) of coffee or tea consumed daily on average compared with none, and for each quartile of caffeine intake compared with the lowest. We also computed coffee, tea and caffeine intake per meal and analysed them separately. Sensitivity analyses were led by type of coffee (regular and decaffeinated). A standard method was used to adjust for energy in every model. Interactions were tested.
Results
Among the 67 703 women included in the present analysis, 2868 cases of invasive breast cancer (634 premenopausal and 2234 postmenopausal) were diagnosed during 707 137 person-years of follow-up between 1993 and 2005 (median duration, 11 years). Coffee was consumed by 85 % (median intake, 280 ml/d, corresponding to 2·2 cups) and tea by 59 % (median intake, 214 ml/d, corresponding to 1·7 cups) of the women. Median caffeine intake (69 % from coffee and tea consumption) was 164 mg/d (range 0·1–724 mg/d) among consumers (85 % of the population). Baseline characteristics of the study population are presented in Table 1. Overall, similar distributions were observed for all the presented characteristics in the different categories of coffee, tea and caffeine consumption.
Table 1 Baseline characteristics of the study population: women living in France, aged 40–65 years at recruitment in 1990, E3N cohort (n 67 703)
Coffee and tea consumption was not associated with risk of breast cancer, neither overall or according to menopausal status, oestrogen- or progesterone-receptor status, which was available for 2268 (79 %) cases (Table 2). Nor was caffeine intake related to risk of breast cancer. Results remained similar when considering coffee, tea and caffeine intakes in the same model or when considering different types of coffee (regular or decaffeinated).
Table 2 Multivariable hazards ratios and 95 % confidence intervalsFootnote * for invasive breast cancer according to coffee and tea consumption and caffeine intake: E3N cohort (n 67 703 women), 1993–2005
ER+, positive oestrogen receptor; ER−, negative oestrogen receptor; PR+, positive progesterone receptor; PR−, negative progesterone receptor.
* Multivariable hazards ratios and 95 % confidence intervals calculated by Cox proportional hazards regression models using age as the time scale and adjusted for baseline variables (total energy intake, ever use of oral contraceptives, age at menarche, age at menopause, number of children, age at first pregnancy, history of breast cancer in the family and years of schooling) and time-dependent variables (current use of postmenopausal hormone therapy (for postmenopausal women only), personal history of benign breast disease, menopausal status and BMI).
† 1 cup = 125 ml.
‡ Test for linear trend using median values in each quartile as an ordinal variable.
When we tested whether coffee, tea and caffeine intakes were differentially associated with breast cancer risk according to the time of day at which beverages were drunk, no differences were observed (data not shown).
Discussion
In the present prospective analysis we found no association between coffee, tea or caffeine consumption and risk of breast cancer, either overall or as defined by hormone receptor status subtype or menopausal status. No association was observed when considering the time of day at which coffee or tea was drunk (during meals) and when considering types of coffee (regular and decaffeinated).
Our results are in accordance with a recent cohort study in the Netherlands(Reference Bhoo, Peeters and van Gils3) and a meta-analysis of nine cohort and nine case–control studies, which showed no association between coffee drinking and breast cancer risk(Reference Tang, Zhou and Wang10). However, two prospective studies reported an inverse association in specific subgroups. Results from the Nurses’ Health Study(Reference Ganmaa, Willett and Li7) showed a negative relationship between consumption of caffeine-containing beverages (coffee and tea) and postmenopausal breast cancer risk in women who drank 4 cups (1 cup = 233 ml) or more per day. In the Women's Health Study, an analysis of 38 432 women showed an increased risk of breast cancer associated with caffeine intake for tumours negative for oestrogen- and progesterone-receptor status or larger than 2 cm(Reference Ishitani, Lin and Manson11). A recent review suggested a decrease in premenopausal breast cancer risk associated with drinking at least 4 cups of coffee daily(Reference Nkondjock12). In our study, we found no association pre- or postmenopause, or according to hormone receptor status. This lack of association in our study might be due to an insufficient variability in daily amount of coffee consumed by the women of the cohort. Michaud et al.(Reference Michaud, Gallo and Schlehofer16) and Bhoo et al.(Reference Bhoo, Peeters and van Gils3) have reported coffee and tea consumptions in some countries participating in the European Prospective Investigation into Cancer and Nutrition (EPIC). For coffee, Denmark had the highest coffee consumption (median 900 ml/d) and Italy and Spain had the lowest (90 and 93 ml/d, respectively), whereas the median value in our population was 280 ml/d. However, in Sweden, where coffee is largely consumed, a cohort study did not find any association regarding breast cancer risk(Reference Larsson, Bergkvist and Wolk5). Women in France consumed more coffee than women in the USA: in 1989, the mean daily consumption of coffee was 1·4 cups (same standardization) daily in US women(17) whereas in our analysis in 1993, women drank a median of 2·2 cups/d. For tea, our population had a median value of 214 ml/d and was among the highest tea consumers in EPIC; only the UK, Denmark and the Netherlands had higher median values (532, 309 and 261 ml/d, respectively)(Reference Bhoo, Peeters and van Gils3, Reference Michaud, Gallo and Schlehofer16).
Results in the literature on tea consumption and breast cancer risk are inconsistent. While most cohort studies reported no significant association between overall tea consumption and breast cancer risk(Reference Ganmaa, Willett and Li7, Reference Goldbohm, Hertog and Brants18, Reference Michels, Holmberg and Bergkvist19), one study suggested that black tea may be positively associated with risk of tumours having positive oestrogen- and progesterone-receptor status(Reference Larsson, Bergkvist and Wolk5). Other case–control studies suggested a negative association for green tea in China(Reference Sartippour, Pietras and Marquez-Garban20, Reference Zhang, Holman and Huang21). In our study in a French population, we found no association between tea consumption and breast cancer risk. Although the type of tea drunk was not available in our data, it is reasonable to assume that it was mostly black tea, traditionally consumed in France.
To explain a potential association between coffee and tea consumption and risk of breast cancer, considered an oestrogen-dependent disease, hypotheses regarding the effects of compounds of the methylxanthine family, including caffeine, have been proposed. Caffeine has been associated with increasing blood levels of oestrone, sex-hormone-binding globulin and decreasing plasma free oestradiol(Reference Kotsopoulos, Eliassen and Missmer22), and all of these factors are associated with reduced oestrogenic exposure of breast cells(Reference Ishitani, Lin and Manson11). Moreover, in experimental studies, caffeine was suggested to both stimulate and suppress the development of mammary tumours(Reference Wolfrom and Welsch23). Other components of coffee and tea have antioxidant properties (e.g. polyphenols).
Caffeine may differentially affect glucose metabolism as a negative relationship between caffeine intake and glucose absorption, glucose hepatic output and glucose storage has been shown(Reference Scalbert, Manach and Morand24, Reference Greer, Hudson and Ross25). Therefore it might be hypothesized that its effects differ according to the time of day at which the beverage is drunk, thereby potentially affecting breast cancer risk(Reference Lajous, Boutron-Ruault and Fabre26). However, our data did not confirm this hypothesis.
In conclusion, our prospective study showed no association between coffee, tea or caffeine consumption and breast cancer risk.
Acknowledgements
The E3N study receives financial support from the ‘Mutuelle Générale de l'Education Nationale’, the European Community, the French League against Cancer, the Institut Gustave Roussy and the French Institute of Health and Medical Research (INSERM). G.F. was supported by the French Ministry of Research. The present study was also supported by the ‘Association pour la Recherche sur le Cancer’. The study sponsors had no role in the design of the study, analysis or interpretation of the data, writing of the manuscript, or the decision to submit the manuscript for publication. The authors declare that they have no conflict of interest. G.F., M.S.T., I.R. and F.C.-C. designed the analysis plan. G.F. performed the statistical analysis and wrote the manuscript. M.S.T., I.R., M.-C.B.-R. and F.C.-C. revised the manuscript critically. The authors are indebted to all participants for providing data used in the E3N study and to practitioners for providing pathology reports. They are grateful to R. Chaït, M. Fangon, L. Hoang and M. Niravong for technical assistance, to Jerri Bram for her assistance with the English, and to the E3N group.
