Prognostic value of DNA repair genes based on stratification of glioblastomas

Abstract

Abnormal expression of DNA repair genes is frequently associated with cancerogenesis of many tumors, however, the role DNA repair genes play in the progression of glioblastoma remains unclear. In this study, taking advantage of large scale of RNA-seq data, as well as clinical data, the function and prognosis value of key DNA repair genes in glioblastoma were analyzed by systematically bioinformatic approaches. Clustering was performed to screen potentially abnormal DNA repair genes related to the prognosis of glioblastoma, followed by unsupervised clustering to identify molecular subtypes of glioblastomas. Characteristics and prognosis differences were analyzed among these molecular subtypes, and modular driver genes in molecular subtypes were identified based on changes in expression correlation. Multifactor Cox proportional hazard analysis was used to find the independent prognostic factor. A total of 15 key genes, which were significantly related to prognosis, were identified and four molecular subtypes of disease were obtained through unsupervised clustering, based on these 15 genes. By analyzing the clinical features of these 4 molecular subtypes, Cluster 4 was found to be different from others in terms of age and prognosis level. A total of 5 key DNA repair genes, CDK7, DDB2, RNH1, RFC2 and FAH, were screened to be significantly related to the prognosis of glioblastomas (p = 9.74e⁻⁰⁵). In summary, the DNA repair genes which can predict the prognosis of patients with Glioblastoma multiforme (GBM) were identified and validated. The expression level of DNA repair genes shows the potential of predicting the prognosis and therapy design in targeting GBM.