A comprehensive evaluation of interaction between genetic variants and use of menopausal hormone therapy on mammographic density

Anja Rudolph, Peter A. Fasching, Sabine Behrens, Ursula Eilber, Manjeet K. Bolla, Qin Wang, Deborah Thompson, Kamila Czene, Judith S. Brand, Jingmei Li, Christopher Scott, V. Shane Pankratz, Kathleen Brandt, Emily Hallberg, Janet E. Olson, Adam Lee, Matthias W. Beckmann, Arif B. Ekici, Lothar Haeberle, Gertraud Maskarinec, Loic Le Marchand, Fredrick Schumacher, Roger L. Milne, Julia A. Knight, Carmel Apicella, Melissa C. Southey, Miroslav K. Kapuscinski, John L. Hopper, Irene L. Andrulis, Graham G. Giles, Christopher A. Haiman, Kay-Tee Khaw, Robert Luben, Per Hall, Paul D. P. Pharoah, Fergus J. Couch, Douglas F. Easton, Isabel dos-Santos-Silva, Celine Vachon, Jenny Chang-Claude

Mammographic density is an established breast cancer risk factor with a strong genetic component and can be increased in women using menopausal hormone therapy (MHT). Here, we aimed to identify genetic variants that may modify the association between MHT use and mammographic density. The study comprised 6,298 postmenopausal women from the Mayo Mammography Health Study and nine studies included in the Breast Cancer Association Consortium. We selected for evaluation 1327 single nucleotide polymorphisms (SNPs) showing the lowest P-values for interaction (P int) in a meta-analysis of genome-wide gene-environment interaction studies with MHT use on risk of breast cancer, 2541 SNPs in candidate genes (AKR1C4, CYP1A1-CYP1A2, CYP1B1, ESR2, PPARG, PRL, SULT1A1-SULT1A2 and TNF) and ten SNPs (AREG-rs10034692, PRDM6-rs186749, ESR1-rs12665607, ZNF365-rs10995190, 8p11.23-rs7816345, LSP1-rs3817198, IGF1-rs703556, 12q24-rs1265507, TMEM184B-rs7289126, and SGSM3-rs17001868) associated with mammographic density in genome-wide studies. We used multiple linear regression models adjusted for potential confounders to evaluate interactions between SNPs and current use of MHT on mammographic density. No significant interactions were identified after adjustment for multiple testing. The strongest SNP-MHT interaction (unadjusted P int <0.0004) was observed with rs9358531 6.5kb 5' of PRL. Furthermore, three SNPs in PLCG2 that had previously been shown to modify the association of MHT use with breast cancer risk were found to modify also the association of MHT use with mammographic density (unadjusted P int <0.002), but solely among cases (unadjusted P int SNP×MHT×case-status <0.02). The study identified potential interactions on mammographic density between current use of MHT and SNPs near PRL and in PLCG2, which require confirmation. Given the moderate size of the interactions observed, larger studies are needed to identify genetic modifiers of the association of MHT use with mammographic density.