From genes to survival: A novel cholesterol-based risk model for osteosarcoma

Background Osteosarcoma (OS) prognosis remains challenging due to limited biomarkers. While cholesterol metabolic reprogramming is recognized in cancer, its prognostic and therapeutic implications in OS are poorly characterized. This gap is particularly pronounced in OS due to its heterogeneous nature and the historical focus on conventional prognostic factors, leading to limited dedicated studies on metabolic vulnerabilities. This study establishes a cholesterol-centric gene signature to stratify OS outcomes, interrogates its immune microenvironment associations, and validates mechanistic roles through functional assays. Methods Multi-omics analysis integrated transcriptomic (TARGET-OS cohort, n = 86) and single-cell RNA-seq (GSE162454) data. Prognostic cholesterol-related genes (CholRGs) were identified via Cox regression, and a 3-gene risk model was constructed using LASSO regression and validated for survival prediction. Immune infiltration was assessed via CIBERSORT/ESTIMATE. In vitro functional validation employed siRNA/overexpression in OS cell lines (143B, U2OS), with proliferation (CCK-8), apoptosis (flow cytometry), migration/invasion (Transwell), and protein analyses (Western blot). Results Twelve CholRGs significantly correlated with OS prognosis. A 3-gene signature (ANXA13, FADS2, SEMA3B) categorized patients into high-/low-risk groups with distinct survival (AUC 3-year: 0.820 training, 0.704 validation). Low-risk patients exhibited prolonged survival and enhanced chemosensitivity to agents like AZD5991 and linsitinib, suggesting potential avenues for personalized therapeutic interventions. High-risk patients demonstrated M1 macrophage enrichment and immunosuppression. Single-cell analysis localized FADS2 predominantly in macrophages/osteoblasts. In vitro, FADS2 knockdown suppressed OS cell proliferation, migration, invasion, and anti-apoptotic capacity, while overexpression exacerbated oncogenicity via EMT pathway modulation exacerbated oncogenicity, as evidenced by significantly enhanced migratory and invasive capabilities, which are hallmarks of EMT. Conclusion This cholesterol metabolism-derived risk model robustly predicts OS survival and therapeutic response. FADS2 is mechanistically implicated in OS malignancy, offering a potential therapeutic target. The integrated approach, by enabling precise patient stratification and identifying a novel therapeutic target, significantly advances personalized OS management.