Abstract
BACKGROUND: Osteosarcoma (OS) is an aggressive and fast-growing malignant tumor associated with high mortality. Early diagnosis and prompt treatment can markedly enhance prognosis and increase survival rates. Constructing prognostic models can effectively predict OS progression, assist in patient diagnosis, and provide personalized treatment plans. In this study, we identified OS-related prognostic genes using the weighted gene co-expression network analysis (WGCNA) method to construct and validate a robust prognostic model, providing guidance for patient risk assessment and clinical treatment. METHODS: Clinical data for OS samples were collected from the Gene Expression Omnibus (GEO) and the Therapeutically Applicable Research to Generate Effective Treatments (TARGET) databases. Statistical analyses, including enrichment analysis, cluster analysis, and model construction, were performed using the R programme. RESULTS: The WGCNA method was used to identify genes which were important to OS development and progression, screening for those relevant to prognosis to build a reliable and widely applicable model. To enhance the model's applicability to diverse OS patient populations, we initially conducted a clustering analysis based on the identified prognostic-related key genes. We then identified differentially expressed genes (DEGs) between clusters and used these genes to subtype OS patients, assessing their ability to distinguish among different patient populations. Subsequently, we selected prognostic-related DEGs to establish the prognostic model, resulting in a risk scoring method utilizing the expression of creatine kinase, mitochondrial 2 (CKMT2) and cell growth regulator with EF-hand domain 1 (CGREF1). We validated the predictive capability of the constructed prognostic model, confirming its robust predictive performance. Finally, based on our prognostic model, we analyzed the immune infiltration and drug sensitivity of OS patients, aiding in evaluating responses to immunotherapy and optimizing treatment plans. CONCLUSIONS: A predictive model based on OS-related prognostic genes was constructed to accurately evaluate risk and guide treatment in OS patients, and CKMT2 and CGREF1 were identified as potential therapeutic targets.