Construction of Novel DNA Methylation-Based Prognostic Model to Predict Survival in Glioblastoma

Glioblastoma (GBM) is a most aggressive primary cancer in brain with poor prognosis. This study aimed to identify novel tumor biomarkers with independent prognostic values in GBMs. The DNA methylation profiles were downloaded from The Cancer Genome Atlas and Gene Expression Omnibus database. Differential methylated genes (DMGs) were screened from recurrent GBM samples using limma package in R software. Functional enrichment analysis was performed to identify major biological processes and signaling pathways. Furthermore, critical DMGs associated with the prognosis of GBM were screened according to univariate and multivariate cox regression analysis. A risk score-based prognostic model was constructed for these DMGs and prediction ability of this model was validated in training and validation data set. In total, 495 DMGs were identified between recurrent samples and disease-free samples, including 356 significantly hypermethylated and 139 hypomethylated genes. Functional and pathway items for these DMGs were mainly related to sensory organ development, neuroactive ligand-receptor interaction, pathways in cancer, etc. Five genes with abnormal methylation level were significantly correlated with prognosis according to survival analysis, such as ALX1, KANK1, NUDT12, SNED1, and SVOP. Finally, the risk model provided an effective ability for prognosis prediction both in training and validation data set. We constructed a novel prognostic model for survival prediction of GBMs. In addition, we identified five DMGs as critical prognostic biomarkers in GBM progression.