Predictive potentials of glycosylation-related genes in glioma prognosis and their correlation with immune infiltration

Glycosylation is currently considered to be an important hallmark of cancer. However, the characterization of glycosylation-related gene sets has not been comprehensively analyzed in glioma, and the relationship between glycosylation-related genes and glioma prognosis has not been elucidated. Here, we firstly found that the glycosylation-related differentially expressed genes in glioma patients were engaged in biological functions related to glioma progression revealed by enrichment analysis. Then seven glycosylation genes ( BGN , C1GALT1C1L , GALNT13 , SDC1 , SERPINA1 , SPTBN5 and TUBA1C ) associated with glioma prognosis were screened out by consensus clustering, principal component analysis, Lasso regression, and univariate and multivariate Cox regression analysis using the TCGA-GTEx database. A glycosylation-related prognostic signature was developed and validated using CGGA database data with significantly accurate prediction on glioma prognosis, which showed better capacity to predict the prognosis of glioma patients than clinicopathological factors do. GSEA enrichment analysis based on the risk score further revealed that patients in the high-risk group were involved in immune-related pathways such as cytokine signaling, inflammatory responses, and immune regulation, as well as glycan synthesis and metabolic function. Immuno-correlation analysis revealed that a variety of immune cell infiltrations, such as Macrophage, activated dendritic cell, Regulatory T cell (Treg), and Natural killer cell, were increased in the high-risk group. Moreover, functional experiments were performed to evaluate the roles of risk genes in the cell viability and cell number of glioma U87 and U251 cells, which demonstrated that silencing BGN , SDC1 , SERPINA1 , TUBA1C , C1GALT1C1L and SPTBN5 could inhibit the growth and viability of glioma cells. These findings strengthened the prognostic potentials of our predictive signature in glioma. In conclusion, this prognostic model composed of 7 glycosylation-related genes distinguishes well the high-risk glioma patients, which might potentially serve as caner biomarkers for disease diagnosis and treatment.