Abstract 4974: Pharmacogenomic interactions in glioblastoma cell line models

Abstract

Abstract Glioblastoma (GBM) is the most common and malignant brain tumor. These tumors display a uniform and very short survival time even with treatment, but are highly heterogeneous at the histological and genomic level. To identify effective treatments and dependencies, we profiled the sensitivity of a panel of cancer cell lines to a small molecules and integrated this with systematic analysis of genetic and non-genetic determinants associated with chemical response. Methods. We profiled 381 drugs described in the Cancer Therapeutics Response Portal (CTRP) at 16 different duplicated concentrations across 78 GBM cell lines belonging to two different models: Patient-Derived GBM Cell Lines (PDGCL) and Long-Term GBM Cell Lines (LTGCL) previously included in the Cancer Cell Line Encyclopedia (CCLE). Cell lines were deeply characterized as to genotype and phenotype. As non-genomic determinants we considered: model, growth rate, behavior, stem cell and differentiation markers. Genomic determinants included mutations and somatic copy number alterations (SCNA) computed from whole exome sequencing. Each of these were integrated to determine oncogene and tumor suppressor gene pathway disruption (p53, RB and RTK signaling). At transcriptomic level we considered expression of 20.647 genes and patters described in GBMs (proneural, neural, classical, mesenchymal). Overall we correlated 10,859,643 pharmacogenomic features to discover associations with drug sensitivity. Summary. We developed a brain tumor living tissue bank as platform for preclinical pharmacogenomics analysis. Large-scale phenotypic characterization of GBM models showed increased cellular and molecular heterogeneity among PDGCLs compared to LTGCLs. PDGCLs better recapitulated patient GBM copy-number profiles. GBM cell lines exhibited all major driver mutations in human GBMs, except IDH1. PDGCLs and LTGCLs enriched for proneural and mesenchymal phenotypes, respectively. We identified NAMPT inhibitors as among the compounds with highest activity across cell lines. Integrative pharmacogenomic analyses showed MDM2/4 inhibitors were able to effectively suppress TP53 wild type GBM models, being CDKN1A (p21) expression a robust predictor of drug response in vitro and in vivo. Overall drug resistance across the screen in lines was highly associated with TP53 mutation, however a specific subset of TP53 mutant cell lines bearing simultaneous CDKN2A/B deletions were sensitive to CHK1/2 inhibitors, revealing a potential synthetic lethal interaction of clinical significance in these highly refractory cells. Analysis identified genetic alterations associated with vulnerabilities targeted by small molecules. About 85% of GBM patients display p53 pathway disruption, our results suggest independent pharmacological strategies for two genetic subtypes of GBM determined by TP53 and CDKN1A status. Our analyses provide molecular insights to drive targeted therapies in the new era of precision medicine. Citation Format: Ruben Ferrer-Luna, Shakti H. Ramkissoon, Karl H. Olausson, Lori A. Ramkissoon, Steven Schumacher, Rebecca Lamothe, Jaime H. Cheah, Kristine Pellton, Sam Haidar, Yun J. Kang, Brenton R. Paolella, Cecile Maire, Wenyu Song, Alice Meng, Ahmed Idbaih, Mikael L. Rinne, David A. Reardon, Patrick Y. Wen, Paul A. Clemons, Stuart L. Schreiber, Alykhan F. Shamji, Rameen Beroukhim, Keith L. Ligon. Pharmacogenomic interactions in glioblastoma cell line models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4974. doi:10.1158/1538-7445.AM2017-4974