Collections > Electronic Theses and Dissertations > Comprehensive Characterization of DNA Copy Number Alterations in Mouse and Human Breast Tumors

Breast cancer is a heterogeneous disease as evident through the diversity observed between the molecularly identified “intrinsic subtypes”. These intrinsic subtypes are based upon patterns of gene expression, are predictive of relapse-free survival, overall survival, and responsiveness to treatment. Furthermore, these subtypes are in part driven by specific genomic DNA copy number alterations (CNAs), such that the identification of these intrinsic subtype-defining genetic events is of research and clinical value. To robustly identify breast cancer “driver” genes within frequently occurring DNA CNAs, we implemented multiple integrative strategies using genomic data from various human breast tumors and genetically engineered mouse (GEM) mammary models. One strategy, a cross-species conservation based method, identified “conserved genes” that are the subtype-specific DNA copy number altered genes found in both human breast tumors and GEM mammary tumors. Another strategy, incorporated gene expression signatures of oncogenic pathway activity to identify patterns of oncogenic signaling within each breast cancer subtype that correlated directly with DNA CNAs. In both strategies, additional functional data from genome-wide RNA-mediated interference screens and/or a molecular interaction network analysis were included highlighting multiple Basal-like-specific 1q21-23 amplified genes and also amplified genes unique in highly proliferative luminal breast tumors. In addition to using CNAs as a base for identifying therapeutic targets, we demonstrated that CNAs play other important roles in the advancement of “personalized medicine”. For example, when tumor DNA is used as the source DNA for genotyping, we demonstrate that CNAs should be taken into consideration as they can lead to erroneous classification of germline genotypes. We examined two separate breast cancer cohorts and observed frequent loss of heterozygosity at the CYP2D6 locus, which is a predictive marker of tamoxifen response. As result, when tumor tissue was used to determine germline CYP2D6 genotype, we observed departure from Hardy Weinberg equilibrium and misclassification of intermediate metabolizers (of tamoxifen) as either extensive or poor metabolizers. In summary, my work utilized multiple genomic data types to develop novel methods of analysis and data visualization to identify driver gene(s) within regions of DNA copy number change, which can and should be used to guide personalized treatment decisions.