WGCNA and machine learning identify mitochondria programmed cell death-related genes as diagnostic biomarkers for septic shock

Septic shock (SS), the most severe stage of sepsis, has a high mortality rate. Mitochondria-mediated programmed cell death (MPCD) plays a key role in SS pathogenesis, but its diagnostic value remains unclear. This study integrated the GEO public dataset and used the WGCNA method to identify gene modules significantly associated with SS, and intersected them with MPCD-related genes to obtain the SS-MPCD gene set. LASSO regression and Boruta algorithms screened key genes, with their diagnostic performance verified in an independent queue. Through single-cell transcriptome analysis, immune infiltration score (ssGSEA), and regulatory network construction (TF miRNA, PPI), the immune regulatory characteristics of candidate genes were systematically analyzed. Clinical blood samples were collected, and the expression levels of candidate genes were validated through qRT-PCR. Finally, the CLP-induced SS mouse model was used for ACSL1 intervention, and its functional effects were evaluated by survival analysis, H&E, qRT PCR, WB, and IF. Four diagnostic genes were identified: ACSL1, BLOC1S1, SPTLC2, and TSPO. They were upregulated in SS patients and clinical samples, showed immune cell-specific expression, and were regulated by specific miRNAs/TFs. Immune profiling revealed a neutrophil- and Treg-dominated microenvironment. Inhibiting ACSL1 improved survival, reduced organ damage and inflammation, decreased apoptosis markers, and suppressed neutrophil/Treg infiltration. The study systematically identifies and preliminarily validates four MPCD-related diagnostic genes, offering insights into SS pathogenesis and potential early diagnosis.