Screening of Inflammatory Biomarkers in Atherosclerosis Basedon WGCNA Analysis and Machine Learning

Introduction: Inflammatory response-related signaling pathways are associated with Atherosclerosis (AS), yet the particular inflammation-related genes underpinning this process are still not fully characterized. Methods: In this study, we integrated two independent transcriptomic datasets with a manually curated inflammation-related gene collection. To identify inflammation-related genes with altered expression, we intersected inflammation-related genes (IRGs), differentially expressed genes (DEGs) between atherosclerosis (AS） and controls, and inflammation score–associated genes (ISRGs) obtained from weighted gene coexpression network analysis (WGCNA) for downstream analyses. The shared genes among these sets were designated as differentially expressed inflammation-related genes (DEIRGs). We then used the DEIRGs to assemble a protein–protein interaction (PPI) network and screen for potential hub genes. Univariate logistic regression, together with least absolute shrinkage and selection operator (LASSO) regression, was used to screen candidate inflammation-related biomarkers. Their expression was then validated by reverse transcription quantitative polymerase chain reaction (RTqPCR). On the basis of the validated biomarkers, we constructed a diagnostic model. As a final analysis, Ingenuity Pathway Analysis (IPA) was used to clarify the major signaling pathways and regulatory networks associated with these biomarkers. Results: Nineteen DEIRGs were obtained as the intersection of the IRG, DEG, and ISRG gene sets. Then, 16 critical genes were identified following PPI analysis. Next, 13 feature genes were filtered using logistic analysis, of which 11 were protective elements, and 2 were risk elements. Four biomarkers (PIK3R5, ADM, RGS16, and B7RP1) were screened by LASSO analysis and RT-qPCR. IPA showed that these inflammation-related biomarkers are mainly enriched in humoral immunity and pathogen-related signaling pathways. Meanwhile, analysis of the regulatory network indicated that STAT3 and KLF6 activated ADM expression and IFNA2 activated IFNAR1 expression. Conclusion: In this study, we identified four inflammation-related biomarkers of AS (ADM, PIK3R5, B7RP1 and RGS16) that may serve as novel indicators for the diagnosis of patients with atherosclerosis.