Int Immunopharmacol. 2026 Jul 3;186:117073. doi: 10.1016/j.intimp.2026.117073. Online ahead of print.
ABSTRACT
BACKGROUND: Sepsis-induced acute respiratory distress syndrome (ARDS) is a life-threatening form of acute diffuse lung injury caused by sepsis. However, the role of mitotic catastrophe-related genes (MCRGs) in its development remains poorly understood. This study investigated the potential roles of MCRGs in the progression of sepsis-induced ARDS.
METHODS: The study used publicly available mRNA expression data (Gene Expression Omnibus: GSE66890, GSE32707). Differentially expressed genes (DEGs) were identified and intersected with MCRGs to obtain potential candidate genes. To screen potential key genes, least absolute shrinkage and selection operator regression was conducted via 5-fold cross-validation to select genes with nonzero coefficients at the optimal model error, and support vector machine recursive feature elimination was performed via 5-fold cross-validation to screen feature genes based on accuracy ranking. The intersection of these two sets yielded potential key genes, which were subsequently validated in the validation set for consistent differential expression. Furthermore, their putative biological functions were explored. A prediction nomogram was constructed. Immune infiltration analysis, drug prediction, exploratory molecular docking, and network pharmacology analysis were conducted. Subsequently, reverse transcription-quantitative polymerase chain reaction (RT-qPCR) was employed to validate the expression of key genes.
RESULTS: Overall, 863 DEGs associated with sepsis-induced ARDS were selected and intersected with 900 MCRGs, yielding potential 92 candidate genes. Three potential key genes (ATM, PTGS2, and PSME4) were identified using machine learning algorithms and RT-qPCR. These three key genes were mainly enriched in immune-related pathways such as the Toll-like receptor signaling pathway. The nomogram exhibited robust predictive performance for the risk of sepsis-induced ARDS (area under the ROC curve = 0.81). Comparative analysis revealed seven differentially abundant immune cell types (including eosinophils) in patients with ARDS versus controls (p < 0.05), with eosinophils exhibiting a significant negative correlation with PSME4 (r = -0.38, p < 0.05). These potential key genes were targeted by multiple drugs (e.g., N-acetyl-l-cysteine) exhibiting binding energies of <-5.0 kcal/mol. Additionally, the active ingredients of traditional Chinese medicines (e.g., the active ingredient of prepared aconite root) demonstrated strong binding capacities (≤-5 kcal/mol) with their corresponding target genes (e.g., ACHE). RT-qPCR indicated that ATM expression is significantly lower in patients with sepsis-induced ARDS than in those with sepsis without ARDS (p < 0.05). The directional trends for PTGS2 and PSME4 were consistent with those of the bioinformatic analyses, but statistical significance was not reached.
CONCLUSION: Differential expression and bioinformatic analysis suggest that ATM, PTGS2, and PSME4 are potential candidate genes associated with sepsis-induced ARDS, advancing our understanding of the potential roles of MCRGs in the progression of sepsis-induced ARDS.
PMID:42398170 | DOI:10.1016/j.intimp.2026.117073
