Cell death-related signature associated with prognosis and immune infiltration in glioma

Abstract

Abstract Background: Accumulating evidence suggests that a wide variety of cell deaths are deeply involved in cancer immunity. However, their roles remain unexplored in glioma. Methods: Logistic regression with shrinkage regularization (LASSO) Cox was conducted to develop a scoring system based on the cell deaths patterns (cuproptosis, ferroptosis, pyroptosis, apoptosis, necrosis) in The Cancer Genome Atlas (TCGA) cohort. A nomogram for overall survival was developed and validated, whose discrimination was evaluated by ROC and calibration curves, respectively. Cell-type identification was estimated by CIBERSORT and ssGSEA methods. Hub genes associated with the prognostic model were screened by machine learning. The expression pattern and clinical significance of MYD88 were investigated by immunohistochemistry (IHC). Results: Cell death score represents an independent prognostic factor of poor outcomes in glioma patients. A nomogram performed well in predicting outcomes by time-dependent ROC and calibration plots. In addition, the high-risk score has a significant relationship with high expression of immune checkpoints and dense infiltration of pro-tumor cells, including macrophage M2. Based on machine learning and differential expression analysis, MYD88 was a hub gene associated with a cell death-based prognostic model. Up-regulated MYD88 was associated with malignant phenotypes and undesirable prognosis by IHC. Furthermore, high-expression MYD88 was associated with poor clinical outcomes, and positively related to CD163, PD-L1, and Vimentin expression in the in-horse cohort. Conclusions: Cell death score provides a precise stratification and immune status for glioma. MYD88 was found to be an outstanding representative that might play an important role in glioma.