Molecular Landscape and Contemporary Prognostic Signatures of Gliomas

Abstract

Molecularly-driven treatments are expanding options for patients with gliomas, driving a need for molecularly-informed prognostic information. To characterize the genomic landscape and contemporary outcomes of gliomas, we analyzed 4,400 gliomas from multi-institutional datasets and The Cancer Genome Atlas (TCGA): 2,195 glioblastoma, 1,198IDH1/2-mutant astrocytoma, 531 oligodendroglioma, 271 otherIDH1/2-wildtype glioma, and 205 pediatric-type glioma. Molecular classification updated 27.4% of gliomas from their original histopathologic diagnosis. Canonical alterations helped categorize glioma subtypes, revealing mutually exclusive alterations within tumorigenic pathways. Across each glioma subtype, non-TCGA patients had longer survival compared to TCGA patients. Several novel prognostic alterations emerged, includingNF1alteration and 21q loss in glioblastoma, andEGFRamplification and 22q loss inIDH1/2-mutant astrocytoma. Certain prognostic features varied across age, with decreasing prevalence ofIDH1/2-mutation over time whileMGMT-methylation remained steady. Our findings provide a framework for further exploration and validation of glioma prognostic indicators in clinically representative cohorts and trials.