Physical Activity and Risk of Male Breast Cancer

Male breast cancer is a rare disease, with a lifetime risk of 1 in 1,000. Due in part to the rarity of this cancer, relatively few studies have examined lifestyle-related etiologic factors. Leisure-time physical activity (LTPA) has been consistently associated with a lower risk of postmenopausal female breast cancer (1, 2), but associations with male breast cancer are unclear. Previous studies of physical activity and male breast cancer risk suggest inverse—but not statistically significant—associations (3–6), yet were based on small case numbers and study characteristics or designs [proxy respondents (6) and case–control (4, 5)] that may capture physical activity differently than a prospective design with self-reported data. In this analysis, with more cases and predominately prospectively collected data, we hypothesized that moderate to vigorous-intensity LTPA would be associated with a lower male breast cancer risk.

Methods for the Male Breast Cancer Pooling Project (MBCPP) have been previously published (7). In short, the MBCPP identified all case–control or cohort studies with ≥10 cases using literature searches of PubMed, citations within published manuscripts, and advertisements at the National Cancer Institute Cohort Consortium meetings (7). Of the 21 studies identified for inclusion in the MBCPP, 10 had collected baseline information on LTPA [nine prospective nested case–control studies (8–16) and one retrospective case–control study (4)]. The two cohort studies and 10 case–control studies that were not included in our analysis did not ask detailed information on LTPA and thus were not eligible for this analysis. Characteristics of included studies are presented in Supplementary Table S1. Cases were defined by the International Classification of Diseases, 10th Edition code C50 from cancer registries, medical record, or self-report. All studies contributed de-identified data following approved data sharing agreements, as well as NCI and study center Institutional Review Board clearances. Participants gave informed consent by nature of study participation.

LTPA was harmonized across studies into categories of low, medium, and high. Studies with a single LTPA question on frequency of activity (AARP, JANUS, PHS, PLCO, Kaiser Permanente) ranged from four to six response levels. For example, in the Physicians' Health Study (PHS), the frequency question was “How often do you exercise vigorously enough to work up a sweat?,” with six categorical choices ranging from daily to rarely/never. In these studies, we collapsed the categories to create a relatively even distribution for low, medium, and high. For studies with line items for more than one type of activity (Canada, CPS II-NC, EPIC, HPFS, MEC), we assigned metabolic equivalent hours per week, using intensity values previously assigned by individual studies and by referencing the updated Ainsworth Compendium of Physical Activities (17). Example activities from the Health Professionals Follow-up Study (HPFS) included separate line items with 10 categorical responses ranging from none to 11+ hour/week for walking outdoors, jogging, running, bicycling, lap swimming, tennis, squash, or calisthenics/rowing. We then divided the study-specific distributions into categories of low (<33rd percentile), medium (33–<66th percentile), and high (≥66th percentile) LTPA.

We used unconditional logistic regression with stratification by study to compute ORs and 95% confidence intervals (95% CI). We first created models adjusted for age and then additionally adjusted for race, education, marital status, diabetes history, alcohol, smoking status, and body mass index (BMI). We created missing categories for race, education, marital status, diabetes, and alcohol, as these covariates were missing for ≤5% of study subjects, and for alcohol, for which 13% were missing. Sensitivity analyses excluding observations with missing data yielded unaltered results. To assess influence by individual studies, we also performed analyses excluding one study at a time.

Because BMI was previously associated with male breast cancer risk in this population (7), and because joint effects of BMI and physical activity have been observed for female breast cancer risk, we created six categories to look at joint effects: high BMI (≥25 kg/m²) with (1) low (referent), (2) medium, or (3) high activity levels, and normal BMI (<25 kg/m²) with (4) low, (5) medium, or (6) high activity levels. We also stratified analyses by age, BMI, diabetes, smoking, and family history of breast cancer and assessed interactions using the Wald test. All analyses were performed using SAS V9.3 (SAS Institute Inc.). P values <0.05 were considered significant.

This pooled analysis of 10 studies included 449 cases and 13,855 controls. More active men were more likely to be non-Hispanic white, college graduates, have a lower BMI, and were less likely to be current smokers or report diabetes (Table 1).

Compared with men reporting low LTPA, men reporting medium and high LTPA levels did not have a lower male breast cancer risk after adjustment for other risk factors (OR, 1.01; 95% CI, 0.79–1.29; 0.90, 0.69–1.18, respectively; Table 2). Results were similar after excluding one study at a time to test undue influence by an individual study (Supplementary Table S2).

In joint-effects analyses of BMI and LTPA, we found that among high BMI men (≥25 kg/m²), neither medium nor high LTPA was associated with risk (OR, 1.01; 95% CI, 0.76–1.34; 0.90, 0.66–1.24, respectively) compared with the referent group of low LTPA/high BMI. However, nonsignificant ∼16% lower risks were observed among normal BMI men (<25 kg/m²) who reported either low or medium LTPA (OR, 0.83; 95% CI, 0.58–1.19; 0.84, 0.59–1.18, respectively), and a nonsignificant 27% lower risk was observed among men who had a normal BMI/high LTPA (OR, 0.73; 95% CI, 0.50–1.07; Table 2).

Stratified analyses did not suggest statistically significant interactions by median age, BMI, diabetes, or family history of breast cancer (all P_interaction values ≥0.05; Supplementary Table S3). There was a stronger inverse association between LTPA and male breast cancer among current smokers (OR, 0.25; 95% CI, 0.07–0.85) than among never or former smokers, with the P interaction showing borderline significance (P = 0.05).

Previous studies of LTPA and female breast cancer risk have suggested stronger risk reductions of about 25% among postmenopausal normal weight women compared with no association for obese women (2), which is similar to the magnitude of association we found among active men with a normal BMI.

Published male breast cancer studies have shown nonsignificant inverse associations with LTPA, but were based on limited numbers (range, 81–178; refs. 3–6) and faced challenges of collecting PA data from proxy respondents of deceased men (6) and retrospective (case–control designs), which may introduce recall bias (4, 5). Reasons for a possible interaction of LTPA with BMI remain unknown, but may reflect the cumulative effect of reduced estrogen exposure for lean active men, supported by other findings from the MBCPP of elevated risks being associated with both high BMI (7) and higher estradiol (18).

Strengths of this study include the relatively large sample size, prospective data collection in nine of the 10 studies, and use of studies with sufficiently detailed information on LTPA. Limitations of this analysis include the inability to stratify by genetic male breast cancer risk factors such as Klinefelter syndrome or BRCA status, as this information was not available. We also had LTPA and BMI data only from baseline, which does not account for changes in these factors over time. We captured only leisure-time activity in this study and were unable to examine occupational activity or sedentary time, each of which may affect risk. In summary, our findings suggest that physical activity alone may not be protective for male breast cancer risk.

No potential conflicts of interest were disclosed.

Conception and design: H. Arem, L.A. Brinton, K. Johnson, L.N. Kolonel, K.B. Michels, S.K. Van Den Eeden, E. Weiderpass, C.E. Matthews

Development of methodology: H. Arem, L.A. Brinton, S.K. Van Den Eeden, E. Weiderpass

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.A. Brinton, S.M. Gapstur, L.A. Habel, K. Johnson, L.N. Kolonel, V.A. McCormack, K.B. Michels, H.D. Sesso, G. Ursin, S.K. Van Den Eeden, E. Weiderpass, M.B. Cook

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): H. Arem, L.A. Brinton, S.C. Moore, H.D. Sesso, E. Weiderpass, M.B. Cook, C.E. Matthews

Writing, review, and/or revision of the manuscript: H. Arem, L.A. Brinton, S.C. Moore, S.M. Gapstur, L.A. Habel, K. Johnson, L.N. Kolonel, V.A. McCormack, K.B. Michels, H.D. Sesso, G. Ursin, S.K. Van Den Eeden, E. Weiderpass, M.B. Cook, C.E. Matthews

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.A. Brinton, S.K. Van Den Eeden, E. Weiderpass, M.B. Cook, C.E. Matthews

Study supervision: L.A. Brinton, E. Weiderpass, M.B. Cook, C.E. Matthews

H. Arem, L.A. Brinton, S.C. Moore, M.B. Cook, and C.E. Matthews are supported by the Intramural Research Program at the NIH, Bethesda, Maryland. The Physicians' Health Study was supported by grants CA 097193, CA 34944, CA 40360, HL 26490, and HL 34595 from the NIH (to H.D. Sesso). The Health Professionals' Follow-Up Study was supported by research grant CA167552 from the NCI, NIH (to K.B. Michels). Support for MEC was provided by NIH/NCI under grant number R37CA54281 (to L.N. Kolonel). The American Cancer Society funds the creation, maintenance, and updating of the Cancer Prevention Study II cohort (to S.M. Gapstur).