Septic patients with T2DM were prone to prolonged recovery and unfavorable prognoses. Thus, this study aimed to pinpoint potential genes related to sepsis with T2DM and develop a predictive model for the disease. The candidate genes were screened using protein-protein interaction networks (PPI) and machine learning algorithms. The nomogram and receiver operating characteristic curve were developed to assess the diagnostic efficiency of the biomarkers. The relationship between sepsis and immune cells was analyzed using the CIBERSORT algorithm. The biomarkers were validated by qPCR and western blotting in basic experiments, and differences in organ damage in mice were studied. Three genes (MMP8, CD177, and S100A12) were identified using PPI and machine learning algorithms, demonstrating strong predictive capabilities. These biomarkers presented significant differences in gene expression patterns between diseased and healthy conditions. Additionally, the expression levels of biomarkers in mouse models and blood samples were consistent with the findings of the bioinformatics analysis. The study elucidated the common molecular mechanisms associated with the pathogenesis of T2DM and sepsis and developed a gene signature-based prediction model for sepsis. These findings provide new targets for the diagnosis and intervention of sepsis complicated with T2DM.
Keywords: bioinformatics; immune; inflammation; machine learning; sepsis; type 2 diabetes mellitus.
© 2024 Federation of American Societies for Experimental Biology.