Caffeine Consumption and the Risk of Breast Cancer in a Large Prospective Cohort of Women FREE

Caffeine (1,3,7-trimethylxanthine), a natural purine alkaloid, is probably the most frequently consumed drug in the world.^1- 2 Common beverages (coffee, tea, and soft drinks), cocoa or chocolate-containing food products, and certain medications, including headache and pain remedies and over-the-counter stimulants, are important sources of caffeine.^1,3 In North America, coffee (60%-75%) and tea (15%-30%) are the primary sources of caffeine in the adult diet.¹

Caffeine was hypothesized to increase risk of breast cancer after a study showed that women with benign breast disease experienced relief from symptoms after the elimination of caffeine from their diet^2,4 because benign breast disease, particularly atypical hyperplasia, is a marker of increased breast cancer risk.⁵ However, clinical studies^2,6 have yielded inconsistent results regarding the effect on symptoms of benign breast disease of eliminating or reducing caffeine intake or regarding the effect of caffeine on the development of benign breast disease.

Most case-control investigations reported no association between caffeine or caffeinated beverages and foods and breast cancer risk,^7- 17 but several case-control studies found either an inverse association^18- 20 or a weak positive association.^19,21- 23 In 3 case-control studies that evaluated the association according to menopausal status, 2 studies^23- 24 observed either a positive association or an inverse association in premenopausal women only, and another study²⁵ found an inverse association for coffee in postmenopausal women only and no association for tea regardless of menopausal status. Data from prospective studies, which are less prone to methodological bias, are limited and in general do not support an overall association,^26- 33 except that 2 studies^34- 35 showed a nonsignificant positive association for black tea and 1 study³³ found a significant inverse association in postmenopausal women only. In a few studies^15,18,27 that examined the association according to history of benign breast disease, no significant association was observed in women with benign breast disease. A Norwegian cohort³¹ reported that coffee reduced risk in lean women, whereas it increased risk in relatively overweight women. However, in 3 other large cohorts^27,30,33 and a large population-based case-control study,¹⁵ no association was observed in any stratum of body mass index.

Breast cancer consists of diverse subtypes with different risk factors and varied clinical responsiveness to treatments.³⁶ However, only a few studies have evaluated the association according to hormone receptor status of breast tumors.^27,33 To our knowledge, no previous studies have evaluated the association by tumor size, lymph node metastasis, and tumor differentiation, which reflect the stage of the carcinogenic process. Using large numbers of cases and detailed information on tumor characteristics, we conducted a comprehensive analysis in the Women's Health Study, a large prospective cohort.

The Women's Health Study was established in 1992 when 39 876 female US health professionals (75.0% were registered nurses) 45 years or older and free of cancer and cardiovascular disease at baseline were enrolled in a randomized trial of low-dose aspirin and vitamin E for the primary prevention of cancer and cardiovascular disease.^37- 38 All of the participants completed a baseline questionnaire inquiring about their medical history and lifestyle factors. As of the end of the trial, March 31, 2004, the mean follow-up was 10 years, and follow-up rates for morbidity and mortality were 97.2% and 99.4%, respectively.^37- 38 The present analysis was restricted to 38 432 women after the exclusion of 1444 women who did not provide information on beverages and diet, had implausible total energy intakes (<600 or >3500 kcal/d), left more than 70 items blank of the food frequency questionnaire, or had prerandomization cancers that were reported after randomization and confirmed by medical record review.

At baseline, 39 310 women (98.6%) in the Women's Health Study also completed a 131-item food frequency questionnaire, a format that has been used and validated in the Nurses' Health Study.^39- 41 The questionnaire assessed average consumption during the past year of a specific amount of foods, including coffee, decaffeinated coffee, tea, caffeinated cola, decaffeinated cola, low-calorie caffeinated cola, low-calorie decaffeinated cola, and chocolate, and allowed 9 responses ranging from “never” to “6 or more times per day.” Intakes of nutrients and caffeine consumption were calculated using the US Department of Agriculture food composition data⁴² and supplemented by food manufacturers. In these calculations, we assumed that the content of caffeine was 137 mg per cup of coffee, 47 mg per cup of tea, 46 mg per can or bottle of cola, and 7 mg per serving of chocolate candy.⁴³ Validation studies in a similar cohort (the Nurses' Health Study) revealed high correlations between self-reported intake of coffee and other caffeinated beverages assessed by the food frequency questionnaire and by 4 weeks of diet records (r = 0.78 for coffee, r = 0.93 for tea, and r = 0.85 for caffeinated sodas).³⁹ Coffee was the primary source of caffeine intake at baseline (81.3%), with fewer contributions by tea (10.0%), low-calorie caffeinated cola (5.6%), caffeinated cola (1.2%), chocolate (0.3%), and other foods (1.7%).

The primary end point of this analysis was invasive breast cancer, which was initially identified by self-report from the yearly follow-up questionnaires and then confirmed by medical record review. Deaths of participants were identified through reports from family members, postal authorities, and a search of the National Death Index. Medical records and other relevant information were sought and reviewed by an end points committee consisting of physicians for the confirmation of medical diagnoses. Medical record review confirmed 98% of self-reported breast cancer cases in the Women's Health Study.⁴⁴ During a mean follow-up of 10 years, we ascertained 1188 confirmed cases of invasive breast cancer. We also extracted detailed information on breast tumor characteristics at diagnosis from medical records, including estrogen receptor (ER) and progesterone receptor (PR) status (ER+PR+, n = 803 [67.6%]; ER+PR−, n = 125 [10.5%]; ER−PR+, n = 23 [1.9%]; ER−PR−, n = 166 [14.0%]; and unknown, n = 71 [6.0%]), tumor size (≤2 cm, n = 863 [72.6%]; >2 cm, n = 274 [23.1%]; any tumor size with direct extension to the chest wall or skin, n = 5 [0.4%]; and unknown, n = 46 [3.9%]), lymph node metastasis (absent, n = 839 [70.6%]; present, n = 284 [23.9%]; and unknown, n = 65 [5.5%]), and histologic grading and differentiation (well, n = 264 [22.2%]; moderate, n = 488 [41.1%]; poor, n = 277 [23.3%]; and unknown, n = 159 [13.4%]). Tumor ER and PR status was determined by laboratories affiliated with the hospitals where the breast cancer cases were diagnosed.

We first compared mean values or proportions of baseline risk factors for breast cancer according to the categories of coffee consumption, the primary source of caffeine, to evaluate potential confounding by these variables.

Person-years were calculated for each participant, ranging from the date of randomization to the date of confirmed cancer diagnosis, death, or March 31, 2004, whichever occurred first. Cox proportional hazards regression models were used to calculate the relative risks (RRs) and 95% confidence intervals (CIs).⁴⁵ We estimated the RRs according to quintiles of caffeine consumption and categories of caffeinated beverages and foods with adjustment for age and randomized treatment assignment (aspirin vs placebo, and vitamin E vs placebo). We also conducted a multivariate analysis that additionally adjusted for known or potential risk factors for breast cancer at baseline, including alcohol consumption (0, >0 to <10, ≥10 to <15, ≥15 to <30, or ≥30 g/d), body mass index (calculated as weight in kilograms divided by height in meters squared) (<23, ≥23 to <25, ≥25 to <27, ≥27 to <30, or ≥30), family history of breast cancer in a first-degree relative (yes or no), history of hysterectomy (yes or no), bilateral oophorectomy (yes or no), smoking status (never, past, or current), history of benign breast disease (yes or no), age at menarche (≤11, 12, 13, 14, or ≥15 years), parity (0, 1-2, 3-4, 5, or ≥6), age at first birth (≤24, 25-29, or ≥30 years), physical activity (kilocalories per week, in quartiles), total energy intake (kilocalories per day, in quintiles), multivitamin use (never, past, or current), age at menopause (<45, 45-49, 50-54, 55-59, or ≥60 years), menopausal status (premenopausal, postmenopausal, or uncertain/unknown), and postmenopausal hormone use (never, past, current <5 years, or current ≥5 years). We also conducted analyses that excluded incident cases of breast cancer diagnosed in the first 2 years of follow-up, with additional adjustment for mammography screening that was determined on the 12-month questionnaire, or that was stratified by menopausal status (premenopausal or postmenopausal), history of benign breast disease (yes or no), body mass index (<25 or ≥25), and postmenopausal hormone use (never or current). Tests for multiplicative interaction were performed by log likelihood ratio tests comparing the models with or without interaction terms.

We also performed an analysis according to combined ER and PR status (ER+PR+, ER+PR−, or ER−PR−), tumor size (≤2 or >2 cm), lymph node metastasis (present or absent), and histologic grading and differentiation (well, moderately, or poorly differentiated). All statistical tests were 2-sided.

In this population, median and 90th percentile values of caffeine intake at baseline were 283.4 and 658.2 mg/d, respectively. At baseline, 9262 women (24.1%) never drank coffee, 4996 (13.0%) drank less than 1 cup per day, 5448 (14.2%) drank 1 cup per day, 12 623 (32.8%) drank 2 to 3 cups per day, 5900 (15.4%) drank at least 4 cups per day, and 203 (0.5%) had missing information on coffee intake (Table 1).

Table 1 presents the distribution of baseline risk factors for breast cancer according to the frequency of coffee consumption. Women who drank more cups of coffee were more likely to be leaner, less physically active, postmenopausal, and current smokers and to have had more births. However, they were less likely to have experienced late age at menarche; to have experienced late age at first birth; to take postmenopausal hormones; to have a history of hysterectomy, bilateral oophorectomy, or benign breast disease; and to undergo mammography screening. They also tended to consume more caffeine, alcohol, and total energy but were less likely to consume tea, decaffeinated coffee, decaffeinated cola, or low-calorie decaffeinated cola. Age, age at menopause, family history of breast cancer, and consumption of caffeinated cola, low-calorie caffeinated cola, and chocolate did not seem to differ substantially according to coffee consumption.

Intakes of caffeine, coffee, tea, caffeinated cola, low-calorie caffeinated cola, chocolate, decaffeinated coffee, decaffeinated cola, and low-calorie decaffeinated cola were not statistically significantly associated with overall risk of breast cancer in the models adjusted for age and randomized treatment assignment (Table 2). Additional adjustment for risk factors for breast cancer did not materially change the results; the multivariate RR comparing the highest to the lowest quintile of caffeine consumption was 1.02 (95% CI, 0.84-1.22). Compared with those who specified their consumption as almost never, the multivariate RRs were 1.08 (95% CI, 0.89-1.30) for coffee (≥4 cups daily), 1.03 (0.85-1.25) for tea (≥2 cups daily), 1.17 (0.87-1.57) for caffeinated cola (≥1 can or bottle daily), 0.88 (0.68-1.13) for low-calorie caffeinated cola (≥2 cans or bottles daily), and 0.97 (0.78-1.20) for chocolate (>1 bar or packet per week). The results did not appreciably change after excluding breast cancer cases diagnosed in the first 2 years of follow-up, with additional adjustment for mammography screening determined on the 12-month questionnaire; the multivariate RRs were 1.02 (95% CI, 0.83-1.25) for caffeine (top vs bottom quintile), 1.04 (0.85-1.28) for coffee (≥4 cups daily vs almost never), 1.12 (0.92-1.37) for tea (≥2 cups daily vs almost never), 1.16 (0.84-1.60) for caffeinated cola (≥1 can or bottle daily vs almost never), 0.92 (0.70-1.20) for low-calorie caffeinated cola (≥2 cans or bottles daily vs almost never), and 1.01 (0.81-1.27) for chocolate (>1 bar or packet per week vs almost never).

In women with a history of benign breast disease, a borderline significantly increased risk of breast cancer was seen for the highest quintile of caffeine (multivariate RR, 1.32; 95% CI, 0.99-1.76) and for daily consumption of 4 or more cups of coffee (1.35; 1.01-1.80) (Table 3); tests for interaction were marginally significant (P = .05 for caffeine and for coffee). The associations between consumption of caffeine, coffee, decaffeinated coffee, and tea and risk of breast cancer did not seem to differ by body mass index (P for interaction = .23 for caffeine) (Table 3), menopausal status (P for interaction = .53 for caffeine) and postmenopausal hormone use (P for interaction = .08 for caffeine) (Table 4). Although decaffeinated coffee consumption was not associated with risk of breast cancer in all postmenopausal women, a significant inverse association for decaffeinated coffee was observed in never users of postmenopausal hormones (multivariate RR, 0.58; 95% CI, 0.36-0.93; P = .02 for trend) (Table 4).

Separate analyses according to hormone receptor status showed a significant positive association between caffeine consumption and risk of ER−PR− breast cancer; the multivariate RR was 1.68 (95% CI, 1.01-2.81, P = .02 for trend) (top vs bottom quintile) (Table 5). In addition, a significant positive association for caffeine consumption was found for developing tumors larger than 2 cm; the multivariate RR was 1.79 (95% CI, 1.18-2.72, P = .02 for trend) (top vs bottom quintile) (Table 5). There were no significant associations between caffeine consumption and breast cancer risk according to tumor lymph node metastasis or tumor histologic grading and differentiation (Table 5).

In this large cohort of women, we found that consumption of caffeine and caffeinated beverages and foods was not significantly associated with overall risk of breast cancer. There were also no significant associations according to menopausal status, postmenopausal hormone use, body mass index, tumor lymph node metastasis, and tumor histologic grading and differentiation. However, in women with a history of benign breast disease, we observed a borderline significant positive association between consumption of greater than 486.3 mg/d of caffeine or 4 or more cups of coffee daily (the primary source of caffeine) and breast cancer risk. We also found a significant positive association between caffeine consumption and risk of breast tumors that were ER−PR− or larger than 2 cm.

Previous findings on the association between caffeine or coffee consumption and breast cancer risk have been inconclusive. An ecologic analysis showed a strong inverse association between coffee and tea consumption and breast cancer mortality.⁴⁶ However, higher caffeine consumption has not been associated with risk of breast cancer in most case-control studies.^7- 17 Several case-control studies have found a weak positive association, but there were no clear trends of increased risk with increasing consumption,^21- 23,47 and a few others have observed an inverse association.^{18- 20,25,48} A recent meta-analysis⁴⁹ of 13 case-control and cohort studies indicates a lower risk of breast cancer associated with higher green tea consumption (the main tea consumed in Asia) but conflicting results for black tea (the main tea consumed in the United States and Europe); black tea consumption was associated with a reduced risk of breast cancer in case-control studies but with a slightly increased risk in cohort studies.

Lack of overall association between consumption of caffeine, coffee, tea (black tea), decaffeinated coffee, soft drinks, and chocolate and risk of breast cancer observed in the Women's Health Study is generally consistent with the findings from previous prospective cohort studies in North America and Europe, including the Seventh-day Adventists cohort,²⁶ the Iowa Women's Health Study,^27,32 the New York State Cohort,²⁸ a Norwegian cohort,⁵⁰ the Swedish Mammography Screening Cohort,³⁰ and the Nurses' Health Study.³³ In a French cohort study,⁵¹ consumption of coffee and tea was not associated with risk of breast cancer; however, consumption of herbal tea was significantly associated with a reduced risk of breast cancer. A nonsignificant positive association for black tea was also observed in the Netherlands Cohort Study.³⁴ In a Japanese cohort,²⁹ although coffee and black tea consumption was not associated with breast cancer risk, green tea consumption was nonsignificantly inversely associated with risk. In 2 other Japanese cohorts,³⁵ coffee and green tea consumption was nonsignificantly inversely associated with risk of breast cancer, but black tea was nonsignificantly positively associated with risk of breast cancer.

The mechanisms by which caffeine may affect breast carcinogenesis are complex and remain unclear. Caffeine has been reported to suppress cell cycle and proliferation and to induce apoptosis.⁵² Caffeine also has been positively associated with blood levels of estrone⁵³ and sex hormone–binding globulin^53- 55 but negatively associated with plasma free estradiol.⁵⁵ Caffeine and coffee can stimulate or suppress the development of mammary tumors in rodents, depending on the phases of tumorigenesis (initiation/promotion) when caffeine and coffee are administered.^2,56 Caffeine is a known antagonist of the adenosine receptor.^2,57 Adenosine, an endogenous bioactive substance, exerts its diverse biological effects through the activation of specific cell surface adenosine receptor.⁵⁷ In breast cancer cell lines, high concentrations of adenosine inhibited cell growth and induced cell cycle arrest at the G₂/M phase but had no effect on ERα levels,⁵⁸ suggesting that, through antagonism of the adenosine receptor, caffeine might stimulate breast cell proliferation independent of ERα pathway.

In the present investigation, caffeine consumption was associated with increased risk of breast cancers negative for both ER and PR or larger than 2 cm, which have less favorable prognoses. These findings indicate that caffeine consumption may affect breast cancer progression, and such an effect may be independent of the estrogen pathway. These findings, however, are not in line with the results of the Iowa Women's Health Study and the Nurses' Health Study, in which there were no associations between caffeine consumption and risk of breast cancer according to ER and PR status, although the caffeine intake levels were generally similar between the Women's Health Study and the Iowa Women's Health Study cohorts.

Consistent with the hypothesis that caffeine may increase the risk of breast cancer in women with benign breast disease,^2,4 we found significantly increased risk associated with the highest quintile of caffeine and with daily consumption of 4 or more cups of coffee in women with a history of benign breast disease. These findings suggest that high caffeine consumption may promote progression from premalignant breast lesions to breast cancers because most types of invasive breast cancer are thought to arise from certain premalignant lesions, such as atypical hyperplasia.⁵ Of note, the increased risk was apparent only in those with the highest intake, and there was no association in those consuming fewer than 4 cups of coffee daily. However, such findings are inconsistent with the results of the Iowa Women's Health Study cohort²⁷ and 2 large case-control studies,^15,18 in which no positive association was found between caffeine or coffee intake and breast cancer risk in those with benign breast disease.

A Norwegian cohort³¹ reported that coffee consumption was associated with a lower risk of breast cancer in lean women but with an increased risk in overweight women. However, we, along with the Iowa Women's Health Study,²⁷ the Swedish Mammography cohort,³⁰ the Nurses' Health Study,³³ and a large case-control study,¹⁵ found no significant association between consumption of caffeine and coffee and breast cancer risk according to categories of body mass index.

The strengths of this study include the prospective design and the high follow-up rates, which minimize the possibility that these findings are a result of methodological biases. We also minimized confounding by other risk factors by controlling for established risk factors for breast cancer comprehensively. These results are also unlikely to be explained by the potential bias that breast cancer itself (before it was diagnosed) may have affected caffeine consumption because the RRs, after excluding case patients who were diagnosed as having breast cancer in the first 2 years after randomization, were similar to those using all case patients. In addition, this study had more than 1000 incident breast cancer cases in 38 432 women followed up for at least 10 years and detailed information on tumor characteristics, which enabled us to evaluate comprehensively the caffeine–breast cancer association according to tumor characteristics.

This study also has limitations. Because we used the information on consumption of caffeine and caffeinated beverages and foods at baseline only, which did not account for changes in caffeine consumption across time, measurement error due to random within-person variation is inevitable. Such misclassification in prospective studies tends to weaken any true associations. In addition, because the number of case patients in some exposure categories and categories of tumor characteristics was relatively modest, we had limited statistical power in some subgroup analyses. Finally, we cannot exclude the possibility that the findings in some subgroups may be the result of chance because a large number of subgroups were evaluated. More studies are needed to refute or confirm the associations observed in some subgroups.

In conclusion, the findings from this prospective study suggest that caffeine consumption is not related to overall risk of breast cancer. However, these data suggest that high caffeine consumption may increase the risk of breast cancer in women with a history of benign breast disease or in breast tumors that are ER−PR− or larger than 2 cm, but these findings may be due to chance and warrant further study.

Correspondence: Shumin M. Zhang, MD, ScD, Division of Preventive Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA 02215 (shumin.zhang@channing.harvard.edu).

Accepted for Publication: April 8, 2008.

Author Contributions: Dr Zhang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Manson and Zhang. Acquisition of data: Buring. Analysis and interpretation of data: Ishitani, Lin, Manson, and Zhang. Drafting of the manuscript: Ishitani and Zhang. Critical revision of the manuscript for important intellectual content: Lin, Manson, Buring, and Zhang. Obtained funding: Zhang. Administrative, technical, and material support: Manson and Zhang. Study supervision: Manson and Zhang.

Financial Disclosure: None reported.

Funding/Support: This study was supported by research grants CA096619, CA47988, and HL43851 from the National Institutes of Health.

Additional Contributions: Eduardo Pereira, MST, provided statistical analytic support and Natalya Gomelskaya, MD, provided assistance with the manuscript. We also thank the entire staff of the Women's Health Study and the 39 876 dedicated participants in the Women's Health Study.