Multi-Omics Analysis Reveals Novel Subtypes and Driver Genes in Glioblastoma

Abstract

Glioblastoma is the most lethal malignant primary brain tumor; nevertheless, there remains a lack of accurate prognostic markers and drug targets. In this study, we analyzed 117 primary glioblastoma patients’ data that contained SNP, DNA copy, DNA methylation, mRNA expression, and clinical information. After the quality of control examination, we conducted the single nucleotide polymorphism (SNP) analysis, copy number variation (CNV) analysis, and infiltrated immune cells estimate. And moreover, by using the cluster of cluster analysis (CoCA) methods, we finally divided these GBM patients into two novel subtypes, HX-1 (Cluster 1) and HX-2 (Cluster 2), which could be co-characterized by 3 methylation variable positions [cg16957313(DUSP1), cg17783509(PHOX2B), cg23432345(HOXA7)] and 15 (PCDH1, CYP27B1, LPIN3, GPR32, BCL6, OR4Q3, MAGI3, SKIV2L, PCSK5, AKAP12, UBE3B, MAP4, TP53BP1, F5, RHOBTB1) gene mutations pattern. Compared to HX-1 subtype, the HX-2 subtype was identified with higher gene co-occurring events, tumor mutation burden (TBM), and poor median overall survival [231.5 days (HX-2) vs. 445 days (HX-1), P-value = 0.00053]. We believe that HX-1 and HX-2 subtypes may make sense as the potential prognostic biomarkers for patients with glioblastoma.