Molecular subtypes and prognostic analysis of cytokine-related genes in gliomas

Abstract Objective: This study aimed to delineate the features of cytokine-associated genes within glioma and formulate a corresponding prognostic model. Methods: We utilized mRNA expression and clinical data from glioma patients within TCGA and CGGA, along with mRNA expression data of normal brain tissue from GTEx. Cytokine-associated prognostic genes were identified through differential gene expression analysis, univariate Cox analysis, and LASSO regression, culminating in a risk score model. Subsequently, glioma patients were stratified into high and low-risk groups based on their risk scores. Kaplan-Meier survival and receiver operating characteristic curve analyses were conducted to assess the prognostic utility of the risk score model in TCGA and CGGA cohorts. We performed relevance analyses of risk score distributions across various subgroups defined by age, gender, WHO grade, IDH1 mutation status, MGMT promoter methylation status, and 1p/19q co-deletion status. Furthermore, we developed a column line plot model employing the risk scoring mechanism and validated its predictive accuracy in the TCGA and CGGA cohorts. Additionally, gene set enrichment analysis identified the predominant signaling pathways and pathological processes in the high-risk group. Lastly, we examined the tumor microenvironment, focusing on immune infiltration and immune checkpoint dynamics in relation to the risk score. Results: Integration of CGGA and GTEx data, comprising 325 glioma and 105 normal brain tissues, yielded 186 differentially expressed cytokine-related genes (DE-CRGS). KEGG analysis highlighted the cytokine-receptor interaction, JAK-STAT, and chemokine signaling pathways as most significant. Protein-protein interaction analysis segregated the 186 DE-CRGS into six modules, pinpointing core pathways such as HIF-1α, chemokine, p53, MAPK, JAK-STAT, and TNF signaling. Based on 186 DE-CRGS expression, the 325 CGGA glioma samples were bifurcated into clusters 1 and 2. Survival analysis indicated poorer prognosis for cluster 1, with CIBERSORT revealing higher quantities of M2 macrophages, activated mast cells, and centrocytes. Notably, cluster 1 correlated with individuals over 40, presenting wild-type IDH1, WHO grade IV, unmethylated MGMT promoters, and absence of 1p/19q co-deletions.We pinpointed prognostic features of three cytokine-associated genes (GPR17, CASP1, CYLD) and devised a risk score model. Elevated risk scores correlated with parameters such as age above 45, wild-type IDH1, WHO grade IV, unmethylated MGMT promoter, and absence of 1p/19q co-deletions. The model demonstrated proficiency in forecasting 1-, 3-, and 5-year overall survival rates for glioma patients in both TCGA and CGGA cohorts, as substantiated by ROC and KM analyses. The line plot model underscored its predictive potential for patient survival, with high-risk scores notably associated with increased expression of certain immune cells and checkpoints. Conclusions: We established a validated cytokine-associated gene risk scoring system, employing TCGA and CGGA data for enhanced prognosis and risk stratification. The constructed column chart model accurately predicts 1-year, 3-year, and 5-year survival, contributing to our understanding of glioma pathology. Additionally, analyses of tumor immune infiltration and immune checkpoints indicate cytokine involvement in aspects like tumorigenesis, progression, tumor microenvironment, and immune evasion, offering potential therapeutic avenues for glioma management.