Association of Interactions Between Mammographic Density Phenotypes and Established Risk Factors With Breast Cancer Risk, by Tumor Subtype and Menopausal Status

Hongjie Chen; Lusine Yaghjyan; Christopher Li; Ulrike Peters; Bernard Rosner; Sara Lindström; Rulla M. Tamimi


Am J Epidemiol. 2021;190(1):44-58. 

In This Article

Abstract and Introduction


Previous studies suggest that the association between mammographic density (MD) and breast cancer risk might be modified by other breast cancer risk factors. In this study, we assessed multiplicative interactions between MD measures and established risk factors on the risk of invasive breast cancer overall and according to menopausal and estrogen receptor status. We used data on 2,137 cases and 4,346 controls from a nested case-control study within the Nurses' Health Study (1976–2004) and Nurses' Health Study II (1989–2007), whose data on percent mammographic density (PMD) and absolute area of dense tissue and nondense tissue (NDA) were available. No interaction remained statistically significant after adjusting for number of comparisons. For breast cancer overall, we observed nominally significant interactions (P < 0.05) between nulliparity and PMD/NDA, age at menarche and area of dense tissue, and body mass index and NDA. Individual nominally significant interactions across MD measures and risk factors were also observed in analyses stratified by either menopausal or estrogen receptor status. Our findings help provide further insights into potential mechanisms underlying the association between MD and breast cancer.


Breast tissue composition can be determined via radiographic imaging through mammography. Radiologically dense area (DA) on mammogram is a quantitative measure of the epithelial, stroma, and connective tissue in the breast, while the nondense area (NDA) reflects the adiposity in the breast.[1] Mammographic density (MD) has been widely accepted as one of the strongest risk factors for breast cancer.[2–4] Women with 75% or greater percent mammographic density (PMD) are estimated to have a 4- to 6-fold higher risk of developing breast cancer, compared with those with 25% or less PMD.[5–9] In addition, DA and NDA are both independent risk factors for breast cancer.[9–13] DA is positively associated with breast cancer risk, while NDA is inversely associated with the risk. DA, NDA, and PMD are all strongly related to all molecular subtypes of breast cancer, although the magnitude of association has been observed to be slightly different across subtypes.[2,14]

Although the association between MD and breast cancer risk is well-established, this association is heterogeneous. Several breast cancer risk factors—including parity, oral contraceptive use, body mass index (BMI), and menopausal hormone therapy (MHT)—have been found to interact with PMD in relation to breast cancer risk.[6,7,15–20] In contrast, there is limited data on the interactions of DA and NDA with breast cancer risk factors. We have previously studied the interactions between MD measures and breast cancer risk factors in postmenopausal women, and we found that the association between PMD/NDA and breast cancer risk differ significantly according to MHT use and parity, respectively.[20] To our knowledge, no studies have comprehensively assessed the interactions between MD measures and breast cancer risk factors, in both pre- and postmenopausal women or by tumor subtype. Therefore, we set out to study interactions between MD measures and established breast cancer risk factors in relation to the risk of invasive breast cancer overall, as well as according to menopausal and estrogen receptor (ER) status. Specifically, we evaluated potential interactions of MD measures (PMD, DA, and NDA) with alcohol consumption, history of benign breast disease, first-degree family history of breast cancer, smoking status, parity, breastfeeding history, BMI, physical activity, age at menarche, MHT use, and age at menopause, using breast cancer cases and controls nested within the Nurses' Health Study (NHS) and NHS II.