Coffee Consumption and Invasive Breast Cancer Incidence among Postmenopausal Women in the Cancer Prevention Study-II Nutrition Cohort

A 2017 World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) Continuous Update Project (CUP) systematic literature review showed a weak inverse association between coffee consumption and postmenopausal [hazard ratio (HR), 0.98; 95% confidence interval (CI), 0.95–1.00 per 1 cup/day], but not premenopausal (HR, 1.00; 95% CI, 0.97–1.03 per 1 cup/day) breast cancer risk (1). In the largest prospective study included in that analysis, the highest versus lowest category of caffeinated, but not decaffeinated, coffee consumption was associated with lower postmenopausal breast cancer risk (HR, 0.90; 95% CI, 0.82–0.98; ref. 2). Subsequently, at least two other studies found that coffee consumption was associated with lower risk of breast cancer–related mortality (3, 4). Despite accumulating evidence, the 2018 WCRF/AICR CUP comprehensive review classified the association as limited-no conclusion (5). To provide further evidence, we examined total, caffeinated, and decaffeinated coffee consumption in relation to postmenopausal invasive breast cancer incidence overall, and by tumor estrogen receptor (ER) and progesterone receptor (PR) status using data from the Cancer Prevention Study-II Nutrition Cohort (CPS-II NC).

The CPS-II NC is a prospective study of cancer incidence initiated in 1992 (6). Beginning in 1997, follow-up questionnaires were sent to participants every 2 years to ascertain updated and new exposure information, and self-reported new cancer diagnoses. The CPS-II NC has been approved by the Emory University Institutional Review Board (Atlanta, GA).

After excluding women who did not complete the 1999 survey (n = 14,297), were lost to follow-up (n = 3,125), reported a personal history of cancer other than nonmelanoma skin cancer prior to 1999 (n = 15,182), were not postmenopausal or aged 55+ years in 1999 (n = 286), and were missing both caffeinated and decaffeinated coffee consumption information (n = 7,817), 57,075 women were included in this analysis.

Coffee consumption was self-reported on the 1999 survey using a 152-item modified Harvard food frequency questionnaire (7). Usual intake of one cup of caffeinated and decaffeinated coffee (assessed separately) over the previous year was reported in one of 10 possible frequencies ranging from never to ≥6 per day. Participants who left either type of coffee blank (but not both), or whose reported consumption was never or <once per month were assigned 0 cups per day. Average cups of total coffee were calculated by summing the contribution from each type; categories were nondrinker (<1 cup/month), <1 cup/day, 1–<2 cups/day, and ≥2 cups/day.

We identified and verified (through medical record abstraction or linkage with state cancer registries or the National Death Index) 2,980 women with incident invasive breast cancer (International Classification of Disease for Oncology Third Editions site code C50) diagnosed between the date the 1999 questionnaire was returned and June 30, 2015. Hormone receptor status was available from medical records or state cancer registries.

Person-years of follow-up was calculated from the completion date of the 1999 questionnaire to breast cancer diagnosis (including censoring on in situ disease), death, last completed questionnaire, or June 30, 2015. Cox proportional hazards regression was used to calculate HRs and 95% CIs for total, caffeinated, and decaffeinated coffee consumption modeled as categorical and continuous variables. All models were stratified on single year of age and adjusted for confounding (see Table 2 footnote). The proportional hazards assumption was assessed using likelihood ratio tests comparing multivariable-adjusted models with and without cross-product terms for follow-up time and coffee consumption; no violations were observed.

Associations of coffee consumption with risk of ER⁺ or PR⁺, or with ER⁻ and PR⁻ breast tumors were assessed using joint Cox proportional hazards regression analysis. All analyses were conducted using SAS (version 9.4).

Among women in this analysis, 17.2% were noncoffee drinkers and 39.6% reported consuming ≥2 cups per day. Those in the highest category of total coffee consumption more frequently reported ever cigarette smoking and current alcohol drinking, whereas other potential confounding factors were similar across categories of coffee drinking (Table 1).

Neither total nor caffeinated coffee consumption was associated with postmenopausal invasive breast cancer risk in age- or in multivariable-adjusted models (Table 2). In age-adjusted analysis, consumption of ≥2 cups per day versus nondrinking of decaffeinated coffee was associated with a marginally statistically significant higher risk (HR, 1.12; 95% CI, 1.00–1.25). However, this association was attenuated in multivariable analysis (HR, 1.06; 95% CI, 0.95–1.19). There were no associations of coffee consumption with risk of ER/PR subtypes.

Some epidemiologic (1) and mechanistic evidence (8) supports a potential inverse association between coffee consumption and breast cancer risk. However, in this large prospective study of postmenopausal women, among whom nearly 3,000 were diagnosed with invasive breast cancer over approximately 16 years of follow-up, we found no evidence that total, caffeinated, or decaffeinated coffee consumption was associated with a lower risk of invasive breast cancer overall or by ER/PR subtypes after adjusting for age, alcohol consumption, smoking history, and other breast cancer risk factors. These results may be useful in future meta-analyses of coffee and breast cancer risk.

No potential conflicts of interest were disclosed.

The views expressed here are those of the authors and do not necessarily represent the American Cancer Society, Inc. or the American Cancer Society–Cancer Action Network, Inc.

S.M. Gapstur: Conceptualization, supervision, methodology, writing–original draft. M.M. Gaudet: Methodology, writing–review and editing. Y. Wang: Methodology, writing–review and editing. R.A. Hodge: Formal analysis, methodology, writing–review and editing. C.Y. Um: Writing–review and editing. T.J. Hartman: Writing–review and editing. M.L. McCullough: Data curation, methodology, writing–review and editing.

The authors express sincere appreciation to all Cancer Prevention Study-II participants, and to each member of the study and biospecimen management group. The authors acknowledge the contribution to this study from central cancer registries supported through the Centers for Disease Control and Prevention's National Program of Cancer Registries and cancer registries supported by the NCI's Surveillance, Epidemiology, and End Results Program. The American Cancer Society, Inc. supports the maintenance and follow-up of the Cancer Prevention Studies.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.