Prognostic model of osteosarcoma based on telomere-related genes and analysis of immune characteristics

Osteosarcoma is a malignant tumor with significant challenges in treatment and prognosis. Telomeres play a crucial role in genetic stability and tumor development, and telomere-related genes (TRGs) have shown considerable prognostic potential in various cancers. However, the prognostic significance of TRGs in osteosarcoma and their involvement in the tumor immune microenvironment (TIME) remain poorly understood.

This study initially identified 2086 TRGs from the TelNet database as candidate genes. Using RNA sequencing and clinical data from osteosarcoma patients available in the TARGET and GEO public databases, we developed a TRG-based prognostic scoring model through univariate, LASSO regression, and multivariate Cox regression analyses, with its predictive performance subsequently validated. Unsupervised clustering was performed on TRGs associated with prognosis. To investigate the TIME, we utilized several algorithms including ESTIMATE, CIBERSORT, xCELL, and ssGSEA to analyze the immune landscape associated with TRG patterns. Additionally, functional enrichment analysis of different subtypes was conducted using KEGG, GO, and GSVA approaches. We also performed single-cell localization and drug sensitivity analysis on the prognostic model genes. Finally, the predictive results were preliminarily validated through a series of in vitro experiments.

Differential expression analysis revealed 841 TRGs with significant changes in osteosarcoma, where P-value < 0.05 and |logFC| ≥ 1. Based on the prognostic relevance of these TRGs, we successfully identified two subtypes with distinct clinical and immune characteristics. Immune infiltration levels between Cluster 1 and Cluster 2 were significantly different, as assessed by multiple algorithms. Furthermore, we constructed a prognostic scoring model based on TRGs, which demonstrated excellent predictive performance, with AUC values for 1-year, 3-year, and 5-year ROC curves being 92.43 %, 87.08 %, and 84.34 % in the training cohort, respectively, and 74.49 %, 87.77 %, and 94.52 % in the validation cohort, indicating good stability of the model. Notably, functional enrichment analysis highlighted a strong association between immune dysfunction and poor prognosis, while drug sensitivity analysis offered personalized chemotherapy recommendations for osteosarcoma patients with different subtypes. We observed that Fludarabine had a higher IC50 value in the high-risk group compared to the low-risk group, and it showed a strong correlation with the prognostic model genes, with all P-values less than 0.05.

This study successfully constructed a prognostic risk prediction model for osteosarcoma by systematically analyzing the expression patterns of TRGs. Fludarabine may represent a promising therapeutic option for patients with osteosarcoma.