Abstract A2-33: Molecular profiling of patient-derived xenograft models across cancers

摘要: Patient-Derived Xenograft (PDX) provides important preclinical model for pharmacological testing of oncology drug candidates. Molecular profiling PDX tumors contributes to many areas discovery from target development clinical biomarker hypotheses and trial design. We established a work flow perform genomic histopathology analyses large numbers tumor models being made available Pfizer internal research. To date we have generated whole-genome sequencing (WGS), whole-exome (WES) whole transcriptome (RNA-Seq) data on spanning six cancer types including colon, pancreatic breast cancers. Bioinformatics pipelines were developed quantify gene expression detect genetic alterations mutation, copy number variations fusions. A controlled evaluation study demonstrated that in silico classification NGS reads into human/mouse origins is more effective than laboratory-based methods removing mouse tissue contamination. Comparative molecular profiles primary the same suggest patterns are retained by models. An integrative classifier was using random forest algorithm, trained data, shown identify subtypes with high accuracy. Citation Format: Zhengyan Kan, Edward Rosfjord, James Hardwick, Ying Ding, Xianxian Zheng, Julio Fernandez, Stephanie Shi, Mark Ozeck, Hui Wang, Gabriel Troche, Eric Upeslacis, Amy Jackson-Fisher, Keith Ching, Shibing Deng, Xie Tao, John Chionis, Maruja Lira, Xiaorong Li, Konstantinos Tsaparikos, Patrick Lappin, Pamela Vizcarra, David Shields, Judy Lucas, Paul Rejto. patient-derived xenograft across [abstract]. In: Proceedings AACR Special Conference Translation Cancer Genome; Feb 7-9, 2015; San Francisco, CA. Philadelphia (PA): AACR; Res 2015;75(22 Suppl 1):Abstract nr A2-33.