Identification of DNA Damage Repair Gene Signature as a Novel Prognostic Marker in Glioblastoma Multiforme

Abstract

The therapeutic efficacy of Glioblastoma multiforme (GBM) is limited. The effect of the DNA damage repair is an important factor. The expression data was downloaded from the TCGA (training dataset) and GEO (external dataset) database. The univariate cox regression analysis and Least absolute shrinkage and selection operator (LASSO) was used to constructed a DNA damage repair (DDR) gene signature. The receiver operating characteristic (ROC) analysis and Kaplan-Meier (KM) curve analysis were employed to estimate the prognostic value of the risk signature. Moreover, the consensus clustering analysis was used to investigate the potential subtypes of the GBM based on the DDR expression. We constructed a three DDR-related gene signature through the survival analysis. The K-M curve result suggesting that patients in the low risk group have a significantly better survival outcome than high risk group in the training dataset and external validation dataset. The ROC analysis result indicated that the risk model has a high prognostic value in the training and external validation dataset. Moreover, three stable molecular subtypes were identified and validated in the GEO database and TCGA database based on the expression of DNA repair gene. The microenvironment and Immunity of GBM was further investigated and discovered that Cluster 2 group have a higher immunity and immune score, compared with other two clusters. In conclusion, the DNA damage repair-related signature was an independent and powerful prognostic biomarker in GBM. The subtype for the GBM may have important implications in the sub-classification of GBM.