A Risk Classification System With Five-Gene for Survival Prediction of Glioblastoma Patients

Objective: Glioblastoma (GBM) is the most common and fatal primary brain tumor in adults. It is necessary to identify novel and effective biomarkers or risk signatures for GBM patients. Methods: Differentially expressed genes (DEGs) between GBM and low-grade glioma (LGG) in TCGA samples were screened out and weight correlation network analysis (WGCNA) was performed to confirm WHO grade-related genes. Five genes were selected via multivariate Cox proportional hazards regression analysis and were used to construct a risk signature. A nomogram composed of the risk signature and clinical characters (age, radiotherapy, and chemotherapy experience) was established to predict 1, 3, 5-year survival rate for GBM patients. Results: One hundred ninety-four DEGs in blue genemodule were found to be positively related to WHO grade via WGCNA. Five genes (DES, RANBP17, CLEC5A, HOXC11, POSTN) were selected to construct a risk signature for GBM via R language. This risk signature was identified to independently predict the outcome of GBM patients, as well as stratified by IDH1 status, MGMT promoter status, and radio-chemotherapy. The nomogram was established which combined the risk signature with clinical factors. The results of c-index, ROC curve and calibration plot revealed the nomogram showing a good accuracy for predicting 1, 3, or 5-year survival of GBM patients. Conclusion: The risk signature with five genes could serve as an independent factor for predicting the prognosis of patients with GBM. Moreover, the nomogram with the risk signature and clinical traits proved to perform better for predicting 1, 3, 5-year survival rate.