Integrative Analysis of Necroptosis-Related Signature for Predicting the Prognosis of Osteosarcoma

Abstract

Background Osteosarcoma (OS) is the most common and malignant bone tumor among children and adolescents worldwide. Over decades, clinical treatment for osteosarcoma has proven to be intractable. Novel approaches, such as immunotherapy, face immune escape. Thus, exploring potential therapeutic targets for osteosarcoma is an urgent need. Method Gene expression data and clinical information were downloaded from Therapeutically Applicable Research to Generate Effective Treatments (TARGET), Gene Expression Omnibus (GEO), and Univariate Cox regression analysis was used to identify prognostic necroptosis-related genes (NRGs). A non-negative matrix factorization algorithm (NMF) was used to cluster patients into various molecular subgroups with NRGs. We dealt with multi-collinearity with the least absolute shrinkage and selection operator (LASSO). Multivariate Cox regression was used to construct the prediction model that divided OS patients into two risk groups. The model's validity was assessed by time-dependent receiver operating characteristic (ROC) analysis. Different expression genes (DEGs) between these two groups were conducted for functional analysis, including gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses, Gene Set Enrichment Analysis (GSEA), and Gene Set Variation Analysis (GSVA). Eight algorithms were carried out to evaluate the tumor microenvironment. These marker genes on the single-cell transcriptome were further labeled to explore whether their expression was cell-specific. Results Based on the model constructed by 5 NRGs (TLR4, STAT5A, IFNGR1, PYGM, CHMP4C), the patients were divided into two risk groups. Patients in the high-risk group suffered a poorer prognosis than those in the low-risk group. The nomogram was constructed and integrated with clinical features and gene signatures, demonstrating better predictive ability in training and testing cohorts. Immune cell infiltrations were highly associated with the risk score generated by Multivariate Cox. All 5 NRGs can be successfully marked on the feature plot of single-cell RNA-Seq, and two NRGs were associated with cell-specific genes of osteosarcoma pluripotency with statistical significance. Conclusion This study can provide a reference for diagnosing molecular subtyping and treating patients with OS.