In this study total physical activity was unrelated to MD overall. However, in heavier women we found some evidence of an inverse association between total physical activity and mammographic density, and between moderate as well as strenuous physical activity and percent mammographic density.
Our overall finding of no statistically significant association between total physical activity and mammographic density is consistent with many other studies which reported no association between physical activity and percent mammographic density (Gram et al. 1999; Suijkerbuijk et al. 2006; Siozon et al. 2006; Peters et al. 2008; Conroy et al. 2010; Samimi et al. 2008; Reeves et al. 2007; Woolcott et al. 2010).
Our results are supportive of an inverse association between physical activity and mammographic density appearing in over-weight and obese women. Our results are consistent with results of two previous studies that stratified analyses by BMI. Both studies reported inverse associations between physical activity and mammographic density in heavier women (Irwin et al. 2006; Masala et al. 2009). One of these studies was a cross-sectional analysis of 474 participants who reported their usual physical activity a year prior to their diagnosis of breast cancer from two different centers in the US (Irwin et al. 2006). No statistically significant trends were observed between total physical activity or sports/recreational physical activity and dense breast area or percent mammographic density after adjustment for BMI as a continuous variable. However, when the analyses were stratified by BMI, statistically significant inverse associations were observed between physical activity and mammographic density among obese (BMI ≥ 30 kg/m2) postmenopausal women (Irwin et al. 2006). In a study of Italian women, there was an inverse association between physical activity among postmenopausal women that was more evident in the highest BMI tertile (≥ 26.5 ) (Masala et al. 2009).
Physical activity is only inversely associated with mammographic density in overweight/obese women. In other words, one could hypothesize that the effect of physical activity would be stronger in overweight women. A year long randomized controlled trial, reported that physical activity had a favorable effect on reducing circulating sex hormone concentrations among overweight postmenopausal women (McTiernan et al. 2004; Friedenreich 2011). Thus, if this effect is stronger in overweight than in normal weight women, it could explain a possible modifying effect of BMI on the association between physical activity and mammographic density. If true, this could suggest that physical activity is particularly beneficial in reducing breast cancer risk in heavy women (Woolcott et al. 2010).
A proposed mechanism explaining the association between physical activity and mammographic density is that physical activity may alter female sex steroid hormone levels that may result in reduced mammographic density. The effect of endogenous estrogens on mammographic density is however, not clear (Tamimi et al. 2005; Tamimi et al. 2007; Boyd et al. 2002b; Greendale et al. 2005; Bremnes et al. 2007). Three studies have examined the relationship between plasma levels of endogenous sex steroid hormones and mammographic density among postmenopausal women (Tamimi et al. 2005; Boyd et al. 2002b; 2005). Boyd et al. (2002b) found no positive association between levels of circulating free estradiol and mammographic density among 189 postmenopausal women, after adjusting for age and waist measurements. Among 520 women in the Nurses’ Health Study, Tamimi et al. observed an inverse association between estradiol and mammographic density which was no longer statistically significant after adjustment for BMI (Tamimi et al. 2005). In contrast, the Postmenopausal Estrogen – Progestin Interventions study found a positive, association between endogenous estradiol levels and mammographic density (Greendale et al. 2005). An Italian study of recently postmenopausal women also reported a positive association between mammographic density and endogenous estradiol (Hofvind et al. 2011).
Thus there is some, although not overwhelming evidence for an association between endogenous levels of estrogen and mammographic density. If there is such an association, then the question is whether estrogen levels could have a stronger affect in heavier than in lighter women. In the Alberta Physical Activity and Breast Cancer Prevention (ALPHA) Trial physical activity resulted in larger reduction in estrone levels in over-weight than in normal weight women (Tretli & Haldorsen 1990).
A recent systematic review of 33 cohort studies and 40 case–control studies published until Dec 2009, found that physical activity was associated with a reduction in risk of breast cancer (Lynch et al. 2011). According to these studies physically active women had on average 25% less risk of breast cancer as compared to the least active women. Recreational activity, regular activity sustained over life time, activity of moderate to vigorous intensity and that performed after menopause had the strongest association. Physical activity provides many health benefits, including weight loss and maintenance, improved insulin sensitivity, and improved lipid profile. Alterations in the metabolism of endogenous hormones such as insulin, sex hormones and levels of insulin-like growth factor (IGF)- I, and IGF-binding proteins (IGFBPs), may form the causal pathways linking excess weight and breast cancer risk (Bianchini et al. 2002).
The cross-sectional design of our study limited our ability to draw a temporal association. Another possible limitation of our study could have been that we had measurement of mammographic density at one point in time only, and thus were precluded from assessing the effect of physical activity on mammographic density over time. As we did not have information on household and occupational activity from our questionnaire, therefore we cannot comment if there is any association with these types of activity and mammographic density. We had to rely on the information about physical activity reported by the women themselves based on the questionnaire. Thus we cannot rule out the possibility of over reporting and possible misclassification of the physical activity variables.
A fundamental problem in making a causal interpretation of associations from observational data is the possibility that such associations are due to confounding. In our study, estimates were adjusted for important confounding factors including age, BMI, education, age at menarche, use of hormone therapy. We cannot exclude the possibility that our estimates could be residually confounded. In particular BMI, as the weight and height were self reported in our study it is likely that this would have caused misclassification for the BMI variable. Moreover, the assessment of adiposity by BMI rather than a more precise measures of body fat (e.g. DXA) may appear to be a limitation of the study, however, Woolcott et al. found anthropometric measurements are likely to be sufficient for adjustment of the association between mammographic density and breast cancer risk (Conroy et al. 2012).
There were several strengths of this study. The mammographic density was assessed by trained personnel using a validated computer assisted method. Readers had no knowledge of the physical activity and risk factor data, thus minimizing the chances of systematic error due to observation bias. Another strength of our study is that we used a continuous measure of mammographic density, which more accurately represents the relation with breast cancer risk (McCormack & Dos Santos 2006). Recall bias is unlikely to have been a problem in our study, since women typically do not know their mammographic density. In addition, not knowing their mammographic density the decision to participate in the study was independent of their knowledge, hence ruling out the possibility of a potential selection bias.
In conclusion, we found some evidence of an inverse between physical activity and mammographic density among women in the over-weight BMI stratum. However, our findings are limited to one BMI stratum and to total physical activity only, thus we cannot rule out the possibility that these findings could be due to chance. Further studies are needed to confirm any associations between physical activity (including other types household, occupational activity as well) and mammographic density, and in particular to better understand the confounding or possibly modifying effects of BMI.