Integrated pathway analysis identifies prognostically relevant subtypes of glioblastoma characterized by abnormalities in multi‐omics

Abstract Background Gene expression‐based molecular subtypes in glioblastoma from The Cancer Genome Atlas Network (TCGA‐GBM) unraveled the pathological origins by identifying tumour cell driver genes. However, the causal inference between molecular subtype origins and their therapeutic efficacy remains obscure. Methods We integrated TCGA‐GBM multi‐omics (DNA, mRNA, and protein profiles) using correlation analysis to identify cis‐ regulation. We analyzed the exposure‐mediated base substitution‐level mutations and their potential triggers. Importantly, we performed Consensus Clustering based on the MSigDB database with Silhouette‐correction to identify prognostically relevant pathway‐based MSig subtypes. The tumour driver mutations (co‐occurrence mutation pattern), aberrant pathways (tumour hallmarks), immune microenvironment ( xCell ), and pseudo‐time analysis ( dyno ) were used to characterize the MSig subtype landscape. Furthermore, we evaluated potential drug sensitivities across MSig subtypes using the Genomics of Drug Sensitivity in Cancer database. Results We classified five MSig subtypes, characterized by neural‐like, tumour‐driving, low tumour evolution, immune‐inflamed, and classical tumour features. We observed several key features in ‘tumour‐driving’ GBM patients: (1) mutual exclusivity between prognostic factors TP53 and EGFR; and (2) IDH1 mutations co‐occurring with TP53, which account for the protective role of IDH1 in TP53 mutant patients. The immune‐inflamed GBM, characterized as a ‘hot’ tumour, exhibited upregulation of immune‐related pathways, including PD‐1 and IFN‐γ signalling responses. DNA methylation landscape revealed 14 MGMT CpG‐rich regions regulating expression. Evolutionary trajectories revealed progression from a primary tumour state (close to normal tissue) to two distinct endpoints (tumour‐driving and immune‐inflamed subtypes). Conclusions Our findings reveal interactions between tumour cells and their surrounding immune environment, classifying GBM into two newly identified subtypes: (1) the tumour‐driving subtype is characterized by multiple oncogenic mutations, while (2) the immune‐blockade subtype is marked by a high presence of immune cells. We highlight the importance of integrating multi‐type data (somatic mutations, DNA methylation, and RNA transcripts, etc.) to decipher GBM biology and potential therapeutic implications. Highlights We report the interaction between tumor cells and environmental immune cells, classifying GBM into two main subtypes: 1) The tumor‐driving subtype is characterized by multiple oncogenic mutations, while 2) the immune‐blockage subtype is marked by a high presence of immune cells. We used integrated multidimensional analyses of somatic mutations, DNA methylation, and RNA transcripts to gain a deeper understanding of GBM biology and potential therapeutic implications.