Machine learning identifies mitochondria-related biomarkers for Staphylococcus aureus osteomyelitis with validation in mouse and human bone tissues

Background Staphylococcus aureus (SA)-induced osteomyelitis (OM) is a severe orthopedic infection characterized by complex host–pathogen interactions and limited molecular markers for lesion assessment. Mitochondrial dysfunction is closely linked to infection-induced inflammatory and metabolic remodeling, but mitochondria-related signatures in SA-induced OM remain incompletely defined. This study aimed to identify mitochondria-related candidate biomarkers for SA-induced OM and evaluate their associations with immune infiltration and regulated cell death. Methods Peripheral-blood transcriptomic datasets GSE16129 and GSE6269 were used as the training cohort, and whole-blood dataset GSE30119 was used as an external evaluation cohort. Differentially expressed genes (DEGs) were identified using limma and intersected with MitoCarta3.0 to obtain mitochondria-related DEGs (MRDEGs). GO and KEGG enrichment analyses were performed. Core MRDEGs were selected using LASSO, SVM-RFE, and Random Forest, and diagnostic models were trained and evaluated using eight machine-learning algorithms. Immune infiltration was assessed by CIBERSORT, and cell death-related pathway activity was analyzed using ssGSEA. The expression of core MRDEGs was further validated in a SA-induced OM mouse model by RT-qPCR and Western blotting, and in human bone tissue specimens by Western blotting. Results Twenty MRDEGs were identified, and five core genes, QTRT1, ABCB6, PPIF, ACSL1, and GLRX5, were selected. The LASSO-based model showed the best performance in the external evaluation cohort, with an AUC of 0.879. These genes were associated with neutrophil, macrophage, and monocyte infiltration, as well as cell death-related pathways, including ferroptosis, apoptosis, and NETosis. Mouse and human bone tissue validation further supported disease-associated expression changes of the five core genes. Conclusion This study identifies a mitochondria-related five-gene signature associated with SA-induced OM. These genes may serve as tissue-based adjunctive molecular indicators for SA-induced OM lesion assessment and provide a framework for linking mitochondrial dysfunction, immune remodeling, and regulated cell death in osteomyelitis.