OPTC-3. Differential Synapse-Related Gene Expression Identifies Glioma Subtypes and Predicts Prognosis

Synaptic connectivity between gliomas and adjacent brain was recently implicated in tumorigenesis. However, the interaction of synapse-related genes (SRGs) with patient survival and with established clinical and molecular glioma subtypes merit further study in a large patient cohort. We characterized differential expression of SRGs in gliomas and investigated SRG expression as putative clinical biomarkers in a large glioma cohort. Expression of 1189 SRGs was interrogated via RNA sequencing analysis in 603 gliomas (LGG n=451, GBM n=152) of The Cancer Genome Atlas. SRG expression patterns partitioned gliomas into two clusters that were more distinctive than priorly identified glioma subtypes. The two glioma clusters showed significantly low (14.9 months) or high median survival (105.1 months; Hazard Ratio = 7.1, 95% CI: [5.32, 9.78], p = 1.29e-46 using a log-rank test). The high survival cluster showed overrepresentation of known pro-neural and neural subtypes and IDH-mutated gliomas (p<0.00001). The mesenchymal and classic GBMs and IDH-wild type gliomas were overrepresented in the low survival cluster (p < 0.00001). In addition, an Elastic Net Cox Regression model identified 34 SRGs whose expression significantly predicted differential survival and a prognostic Synapse Gene Score (SGS) was created. Using an accelerated failure time model, SGS predicted survival in the overall gliomas after adjusting for IDH-1 mutation (HR = 2.51, 95% CI: [2.27, 2.7496], p<0.00001), and in the IDH-1 mutant cohort after adjusting for 1p/19q co-deletion (HR = 2.05, 95% CI: [1.73, 2.37], p<0.00001). Our analysis shows that gliomas can be distinctively clustered by differential SRGs expression, suggesting that synapse-related proteins may contribute to tumorigenesis via multiple mechanisms. Furthermore, the potential utility of SRGs as clinical glioma biomarkers is supported by our creation of a prognostic SGS. Future studies elucidating interactions between tumorigenesis and synaptic mechanisms may reveal additional insights for glioma biology and therapeutic targeting.