Identification of glioblastoma immune subtypes and immune landscape based on a large cohort

Glioblastomas (GBM) are the most common primary brain malignancy and also the most aggressive one. In addition, GBM have to date poor treatment options. Therefore, understanding the GBM microenvironment may help to design immunotherapy treatments and rational combination strategies. In this study, the gene expression profiles and clinical follow-up data were downloaded from TCGA-GBM, and the molecular subtypes were identified using ConsensusClusterPlus. Univariate and multivariate Cox regression were used to evaluate the prognostic value of immune subtypes. The Graph Structure Learning method was used for dimension reduction to reveal the internal structure of the immune system. A Weighted Correlation Network Analysis (WGCNA) was used to identify immune-related gene modules. Four immune subtypes (IS1, IS2, IS3, IS4) with significant prognosis differences were obtained. Interestingly, IS4 had the highest mutation rate. We also found significant differences in the distribution of the four subtypes at immune checkpoints, molecular markers, and immune characteristics. WGCNA identified 11 co-expressed module genes, and there were significant differences among the four subtypes. Finally, CD1A, CD1E, and IL23R genes with significant prognostic significance were selected as the final feature genes in the brown module. Overall, this study provided a conceptual framework for understanding the tumor immune microenvironment of GBM.