Integrated Bioinformatics and Machine Learning Analysis Identifies Inflammation-Related Biomarkers and Immune
1Department of Biochemistry and Molecular Biology, College of Pharmacy, Hebei University of Chinese Medicine, Shijiazhuang 050000, China.
Abstract:
Background: Atherosclerosis (AS) is a chronic inflammatory vascular disease lacking reliable biomarkers for early diagnosis and risk stratification. This study aimed to identify hub genes with diagnostic potential and characterize immune microenvironment remodeling in AS. Methods: GSE43292 and GSE100927 were integrated as the training cohort (n = 168), while GSE41571, GSE120521, and GSE28829 served as independent validation cohorts (n = 48). Batch effects were corrected using the ComBat algorithm. Differentially expressed genes (DEGs) were identified using limma, followed by GO/KEGG enrichment analysis. LASSO regression and Random Forest analysis were performed to identify hub genes. A logistic regression diagnostic model was constructed and evaluated using ROC analysis, 5-fold cross-validation, and external validation. Immune infiltration was assessed using ssGSEA, and correlations between hub genes and immune cells were analyzed using Spearman correlation. Results: A total of 1349 DEGs (870 upregulated and 479 downregulated) were identified. GO and KEGG analyses demonstrated significant enrichment of immune- and inflammation-related biological processes and pathways. Seven hub genes (IBSP, XAF1, SCAMP5, SAMD9L, MYBL1, PCDH12, and CDH19) were identified through the combined application of LASSO regression and Random Forest analysis. The 7-gene logistic model achieved excellent performance in the training cohort (AUC = 0.992, 95% CI: 0.981-1.000), with a mean 5-fold cross-validation AUC of 0.984 ± 0.014, and maintained robust performance in three independent validation cohorts (AUC range: 0.952-1.000). Immune infiltration analysis revealed extensive immune microenvironment remodeling, with significantly increased infiltration of 22 of the 24 immune cell types, particularly monocytes, macrophages, and myeloid cells. Spearman correlation analysis demonstrated strong associations between hub genes, particularly SAMD9L and IBSP, and immune cell infiltration. Conclusions: This study identified a robust 7-gene diagnostic signature for AS and revealed its close association with the inflammatory immune microenvironment, providing potential biomarkers for early diagnosis and risk stratification.
