Physical activity and mammographic parenchymal patterns among Greek postmenopausal women

Objective: To examine whether physical activity during the last five years is related to later breast mammographic density in postmenopausal Greek women. Methods: We designed a cross-sectional study in 724 women, of ages 45-67 years. An interview-administered questionnaire was used to obtain information on duration and intensity of recreational physical activity during five years preceding study recruitment. Mammograms were evaluated according to BIRADS classification and BIRADS score was also estimated. Multivariate ordinal logistic regression analysis was used to assess associations between physical activity index and breast density according to the BIRADS classification methods. Results: We observed a statistically significant inverse association of mammographic breast density measured by the BIRADS classification method and recreational exercise (OR = -0.10:95% Cl -0.018, -0.001; p = 0.022). For one unit increase in physical activity as expressed by the MET-h/week score, the odds of lower versus higher breast density categories are 1.105 greater, given that all of the other variables in the model are held constant. A modifying effect by age at recruitment was evident among participants, with a stronger inverse association between recreational activity and mammographic breast density among older women (OR = -0.036; 95% Cl -0.063, -0.009; p = 0.009). An inverse association between physical activity and BIRADS score was evident, not reaching statistical significance (OR = 0.00; 95% Cl -0.009, 0.008; p = 0.887). Conclusions: Mammographic breast area was lower in postmenopausal women who participated in sports/recreational physical activity compared to inactive controls. Increasing physical activity levels among postmenopausal women might be a reasonable approach to reduce mammographic density. However, until more physical activity and mammographic breast density studies are conducted that confirm our findings, they have to be interpreted with caution, due to the retrospective nature of our data and the possibility of memory bias. (C) 2011 Elsevier Ireland Ltd. All rights reserved.