CELL INTELLIGENCE · 肿瘤细胞治疗研究

Osteosarcoma (OS) is a bone malignancy among adolescents with a poor prognosis. We aimed to explore the glycosylation-related features in OS.

A sum of 185 glycosylation-related genes (GRGs) and the RNA-seq data were obtained from the TARGET database. ConsensusClusterPlus package was applied for the various subtypes based on the significant prognostic factors. Subsequently, the limma R package performed the Lasso analysis for a RiskScore model. The survminer package conducted Kaplan-Meier (KM) analysis, and the timeROC package was used for the classifier efficiency. In addition, the ssGSEA method and MCPcounter performed the immune infiltration analysis. Finally, the expression of the selected key genes, as well as their migration and invasion capabilities, were evaluated through PCR, wound healing, and transwell assays, respectively.

High GRG score was associated with poor prognosis. Unsupervised clustering based on prognostic GRGs identified two subtypes (C1/C2), with C1 showing worse outcomes. A prognostic RiskScore model comprising CORT, LPAR5, CEBPA, MYH10, and MAGEA11 was constructed from subtype-specific differentially expressed genes. The model effectively stratified patients into high- and low-risk groups in both training and validation cohorts, with high-risk patients exhibiting significantly shorter survival. RiskScore is an independent prognostic factor, with lower infiltration levels of cells such as macrophages and NK cells observed in patients within the high-risk group. In external immunotherapy cohorts, high-risk patients had poorer survival and treatment response. Functionally, silencing the key model gene CORT impaired osteosarcoma cell migration and invasion in vitro.

We developed a useful risk model and this study provided new insights for OS therapy.