Abstract
Glioblastoma (GBM) is the most common brain tumor, with rapid proliferation and fatal invasiveness. Large-scale genetic and epigenetic profiling studies have identified targets among molecular subgroups, yet agents developed against these targets have failed in late clinical development. We obtained the genomic and clinical data of GBM patients from the Chinese Glioma Genome Atlas (CGGA) and performed the least absolute shrinkage and selection operator (LASSO) Cox analysis to establish a risk model incorporating 17 genes in the CGGA693 RNA-seq cohort. This risk model was successfully validated using the CGGA325 validation set. Based on Cox regression analysis, this risk model may be an independent indicator of clinical efficacy. We also developed a survival nomogram prediction model that combines the clinical features of OS. To determine the novel classification based on the risk model, we classified the patients into two clusters using ConsensusClusterPlus, and evaluated the tumor immune environment with ESTIMATE and CIBERSORT. We also constructed clinical traits-related and co-expression modules through WGCNA analysis. We identified eight genes (ANKRD20A4, CLOCK, CNTRL, ICA1, LARP4B, RASA2, RPS6, and SET) in the blue module and three genes (MSH2, ZBTB34, and DDX31) in the turquoise module. Based on the public website TCGA, two biomarkers were significantly associated with poorer OS. Finally, through GSCALite, we re-evaluated the prognostic value of the essential biomarkers and verified MSH2 as a hub biomarker.