Does mammographic density mediate risk factor associations with breast cancer? An analysis by tumor characteristics

Background: Though mammographic density (MD) has been proposed as an intermediate marker of breast cancer risk, few studies have examined whether the associations between breast cancer risk factors and risk are mediated by MD, particularly by tumor characteristics. Methods: Our study population included 3392 cases (1105 premenopausal) and 8882 (3192 premenopausal) controls from four case–control studies. For established risk factors, we estimated the percent of the total risk factor association with breast cancer that was mediated by percent MD (secondarily, by dense area and non-dense area) for invasive breast cancer as well as for subtypes defined by the estrogen receptor (ER+/ER−), progesterone receptor (PR+/PR−), and HER2 (HER2+/HER2−). Analyses were conducted separately in pre- and postmenopausal women. Results: Positive associations between prior breast biopsy and risk of invasive breast cancer as well as all subtypes were partially mediated by percent MD in pre- and postmenopausal women (percent mediated = 11–27%, p ≤ 0.02). In postmenopausal women, nulliparity and hormone therapy use were positively associated with invasive, ER+ , PR+ , and HER2− breast cancer; percent MD partially mediated these associations (percent mediated ≥ 31%, p ≤ 0.02). Further, among postmenopausal women, percent MD partially mediated the positive association between later age at first birth and invasive as well as ER+ breast cancer (percent mediated = 16%, p ≤ 0.05). Conclusion: Percent MD partially mediated the associations between breast biopsy, nulliparity, age at first birth, and hormone therapy with risk of breast cancer, particularly among postmenopausal women, suggesting that these risk factors at least partially influence breast cancer risk through changes in breast tissue composition.