Improving pre-clinical screening of drug candidates for pancreatic cancer by applying a systems genomics approach to pinpoint ideal cell line models.

Abstract

e15268 Background: Despite substantial research efforts in pancreatic cancer the estimated 5-year survival rate remains 5-6%. The lack of early detection methods compounded by a startlingly low clinical trial success rate precipitates the need for innovative approaches to pancreatic cancer drug development. Choice of cell lines used for pre-clinical compound screening is usually based on ease of culture, popularity, and availability. However, we have observed that the constellation of pathway derangements in commonly used pancreatic cancer cell lines may not optimally model pancreatic cancer tumors. Methods: With this motivation, we leveraged gene expression (GE), copy number variation (CNV) and targeted sequencing data from The Cancer Genome Atlas (TCGA) and the Cancer Cell Line Encyclopedia (CCLE) to predict cell lines that best model pancreatic tumor genomic architecture. Inspired by the hypothesis that a disease relevant CNV should impact GE, we used an eQTL-like approach independently in pancreatic c...