Abstract 3611: Identifying gene expression markers of anticancer drug response using large scale genomic and drug response databases established from patient derived tumors

Abstract

Abstract Chemotherapy has a well-established history of checkered success for treatment in a number of solid and hematological malignancies alike. Numerous reports have demonstrated heightened levels of response to some of these agents within subpopulations of cancer patients, leading to the notion that we have effective pharmaceuticals; however, we lack reliable means to identify these subpopulations prior to administering therapy. We report here an integrated series of preclinical and translational research studies aimed at identifying those populations by capturing predictive markers of response to specific anticancer agents. Our approach to identifying molecular markers of response relies on a biobank of more than 140,000 patient derived tumor samples, and an accompanying database of ex-vivo drug resistance and sensitivity profiles for a broad range of chemotherapeutic agents. Briefly, the drug response database was created by dissociating fresh tumor into single cell suspension, and treating each cell preparation in an anchorage independent growth matrix with antineoplastic agents relevant to tumor type. Drug resistance was scored as Extreme (E), Intermediate (I), or Low (L), with the Low score providing a correlative measure of sensitivity. We identified subsets of tumors featuring differential drug response profiles, and performed Affymetrix gene expression microarray profiling to create data sets that could be interrogated to identify gene expression markers predictive of response. In all, more than 300 tumors were profiled covering 7 indications: prostate, lung, colon, breast, melanoma, ovarian, and lymphoma. Statistical analysis and biological interpretation of genomic data was performed using Ingenuity iReport. iReport, powered by the Ingenuity® Knowledge Base, identified biological processes, pathways, and cellular phenotypes most perturbed in the tumor samples, providing insight into the molecular basis of drug response in these tumor cells. This focused the analysis down to a subset of genes anchored on pathways representative of the drug response and cellular properties of the tumor samples and prioritized a set of candidate markers for further study. We were able to identify gene expression markers predictive of drug response and evaluate those markers based on functional roles consistent with pathways and processes. Additional studies are on-going to validate these potentially novel markers in a series of primary tumor cell models established directly from patient tumor biopsies registered in the Molecular Response tumor bank. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3611. doi:1538-7445.AM2012-3611