Abstract 2892: Refined Molecular Subtyping of GBM Discovers Both Shared and Unique Therapeutic Targets.

Abstract Background: Glioblastoma is the most common primary brain tumor in humans, which also carries a very dismal prognosis of median survival time of 14 months. TCGA consortium profiled >580 glioblastoma samples for various genomic and molecular characterizations, and has shown that the single histopathologically diagnosed cancer can be classified into four different molecular subtypes (classical, mesenchymal, proneural, and neural); these exhibit distinct characteristics, such as high expression of oligodendrocytic development genes in the proneural subtype and high necrosis and inflammation responses in the mesenchymal subtype. Our own analysis of TCGA data further sub-classified mesenchymal and proneural GBM subtypes with distinct genomic as well as survival characterization. Since treatment efficacy and outcome are highly dependent on tumor subtype, there is an unmet need to develop a computational method to tailor therapy to a particular subtype. Method: In our study, we aimed to identify genes that can serve as unique drug targets for two of our own GBM subtypes. One is enriched with mesenchymal subtype (ME) and the other with proneural subtype (PE). To this end, we developed a network-based approach that takes as input the differentially expressed genes (DEGs) in one subtype. and produces a prioritized list of candidate drug targets specific to a subtype. Network-based methods score candidates by their overall proximity in existing pathways (MetaCore™) to the DEGs, and high-scoring candidates are considered likely to cause the observed transcriptional changes. Resulting drug targets were compared in order to identify genes that are unique to each subtype and those that are common. Results: Among top 500 targets identified for each subtype, 381 were shared between two subtypes, leaving 119 high-confidence drug targets predicted as unique to each subtype. Pathway enrichment analysis revealed that drug targets unique to the ME subtype were enriched in multiple immune-related and stem-cell differentiation-related pathways. Drug targets unique to the PE subtype showed enrichment for cell adhesion, cell cycle, and development pathways. Strikingly, disease biomarker enrichments for unique ME drug targets resulted in a wide variety of neurological disorders, but no cancer in the top ranks. On the other hand, we found glioblastoma, glioma, and astrocytoma to be highly enriched for the drug targets unique to the PE subtype, indicating that these drug targets predominantly represent our current knowledge on glioblastoma vulnerability. Conclusions: The identification of subtype-specific drug targets for glioblastoma is crucial for developing novel therapeutic approaches. Our network-based approach identified a number of drug targets that are unique to the ME and the PE subtypes. These candidate drug targets can be tested in vivo and may lead to subtype-specific therapy. (Supported by NIH CA U01CA168397) Citation Format: Dorothea Emig, Seungchan Kim, Lee Lancasjore, Yuri Nikolsky, Michael E. Berens, Marina Bessarabova. Refined molecular subtyping of GBM discovers both shared and unique therapeutic targets. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2892. doi:10.1158/1538-7445.AM2013-2892 Note: This abstract was not presented at the AACR Annual Meeting 2013 because the presenter was unable to attend.