Angiogenesis related genes based prognostic model of glioma patients developed by multi-omics approach.

Introduction: Glioma, particularly glioblastoma (GBM), is a highly malignant brain tumor with poor prognosis despite current therapeutic approaches. The tumor microenvironment (TME), plays a crucial role in glioma progression by promoting invasion and drug resistance. Angiogenesis, the formation of new blood vessels, is a tightly regulated process involving endothelial cell activation, proliferation, and migration. In cancer, angiogenesis becomes dysregulated, leading to excessive blood vessel formation.

Methods: We enrolled bulk data of TCGA-LGG/GBM, CGGA-693, and CGGA-325 cohorts, scRNA data of GSE162631, GSE84465, and GSE138794 cohorts. Identification of malignant cells was conducted by "copycat" R package. The "AUCell" R package scored the activity of target gene set of each single cell. Consensus clustering was applied using the "ConsensusClusterPlus" R package, while tumor-infiltrating immune cells were determined using "IOBR" R package. To construct a prognostic model, we used LASSO and multiCOX algorithms based on the expression levels of the 15 hub genes, the efficacy of which was verified by KM and ROC analysis.

Results: We identified 4 different malignant cell subclusters in glioma and disclosed their distinct gene expression patterns and interactions within TME. We identified differentially expressed immune-related genes (DE-ARGs) in glioma and found 15 genes that were specifically expressed in the malignant glioma cell populations. Glioma cells with higher expression of these DE-ARGs were associated with gliogenesis, glial cell development, and vasculature development. We found that tumor-infiltrating monocytes were the main interacting cell type within glioma TME. Using the expression patterns of the 15 screened DE-ARGs, we categorized glioma samples into 2 molecular clusters with distinct immune features, suggesting a possible relationship between angiogenesis and immune activation and recruitment. We constructed a prognostic model based on the expression levels of the 15 DE-ARGs and evaluated its predictive ability for glioma patient outcomes, which displayed exceedingly high efficacy.

Conclusion: We characterized different malignant cell subclusters in glioma and investigate their gene expression patterns and interactions within TME. We constructed a prognostic model based on the expression levels of the 15 DE-ARGs and evaluated its predictive ability for glioma patient outcomes, which displayed exceedingly high efficacy.