GENE-LEVEL GERMLINE INVESTIGATION OF BREAST CANCER SUBTYPES AND MORTALITY AMONG WOMEN OF EUROPEAN AND AFRICAN ANCESTRY

Genome-wide investigations to date have uncovered over 210 germline variants associated with BC incidence, and only a handful of mostly non-replicable variants associated with BC mortality. There remain a few key gaps in knowledge across genetic investigations of BC incidence and mortality. First, BC is a heterogeneous disease, with subtype-specific outcomes. Understanding of the germline genetic underpinnings of BC subtypes is sparse among individuals of European and even sparser for individuals of African ancestry. It is also thought that integration of BC subtype (i.e., stratification by) into germline investigations of BC mortality may a key step towards bettering understanding on this front. In this dissertation work, we address these gaps in knowledge by leveraging a statistically efficient and highly interpretable framework for genetic association testing in Transcriptome-Wide Association Study (TWAS). In Aim 1 (Chapter 4) multi-tissue (normal breast, breast tumor), multi-ancestry TWAS-based germline-regulated gene expression (GReX) analysis of 396 BC-related genes in relation to BC subtype, we find 40 GReX-prioritized genes for BC subtype, including ten shared, and 34 and six unique to Luminal-like (LL) and Basal-like (BL) subtype, respectively (after Bayesian correction for potential test statistic inflation and at global False Discovery Rate (FDR) <0.05). Among individuals of African ancestry, we see suggestion of association (global FDR = 0.06 to 0.18) for five genes. By comparison, the largest genome-wide study to date among African ancestry individuals has reported no loci at or near genome-wide significance. In Aim 2 (Chapter 5) ancestry and subtype specific GReX analysis of BC mortality, we find no associations at global FDR