Association Between Diverse Cell Death Patterns Related Gene Signature and Prognosis, Drug Sensitivity, and Immune Microenvironment in Glioblastoma

Abstract

AbstractGlioblastoma (GBM) is the most invasive type of glioma and is difficult to treat. Diverse programmed cell death (PCD) patterns have a significant association with tumor initiation and progression. A novel prognostic model based on PCD genes may serve as an effective tool to predict the prognosis of GBM. The study incorporated 11 PCD patterns, namely apoptosis, necroptosis, pyroptosis, ferroptosis, cuproptosis, entotic cell death, netotic cell death, parthanatos, lysosome-dependent cell death, autophagy-dependent cell death, alkaliptosis, and oxeiptosis, to develop the model. To construct and validate the model, both bulk and single-cell transcriptome data, along with corresponding clinical data from GBM cases, were obtained from the TCGA-GBM, REMBRANDT, CGGA, and GSE162631 datasets. A cell death-related signature containing 14 genes was constructed with the TCGA-GBM cohort and validated in the REMBRANDT and CGGA datasets. GBM patients with a higher cell death index (CDI) were significantly associated with poorer survival outcomes. Two separate clusters associated with clinical outcomes emerged from unsupervised analysis. A multivariate Cox regression analysis was conducted to examine the association of CDI with clinical characteristics, and a prognostic nomogram was developed. Drug sensitivity analysis revealed high-CDI GBM patients might be resistant to carmustine while sensitive to 5-fluorouracil. Less abundance of natural killer cells was found in GBM cases with high CDI and bulk transcriptome data. A cell death-related prognostic model that could predict the prognosis of GBM patients with good performance was established, which could discriminate between the prognosis and drug sensitivity of GBM.