Biomedical Technology · Published 2025-06-01 · DOI 10.1016/j.bmt.2025.100087
Background: Aortic valve calcification (AVC) represents a progressive, age-associated disorder characterized by substantial mortality, yet effective early diagnostic markers for AVC complicated by periodontitis, a common inflammatory condition linked to systemic inflammation, remain elusive. Our investigation sought to uncover immune-specific molecular indicators for AVC in patients with periodontitis using bioinformatics and machine learning. Methods: Gene expression data for AVC (utilizing datasets GSE153555, GSE148219, GSE51472) and periodontitis (from dataset GSE16134) underwent analysis. We identified differentially expressed genes (DEGs) and determined the overlapped genes between AVC and periodontitis. The study included functional enrichment, protein-protein interaction (PPI) network construction, and immune infiltration analyses. To screen potential target genes, four machine learning models were developed (SVM, RF, XGB, GLM), with validation performed using an external dataset and clinical specimens via qRT-PCR. Results: A total of 30 intersecting genes between AVC and periodontitis were identified. Four key genes—CXCL12, HCST, ITGA4, and GZMK—were selected through machine learning. The nomogram model combining these genes demonstrated high diagnostic accuracy, with an AUC of 0.985 in the training set and AUC values of 0.8, 0.72, 0.88, and 0.76 for HCST, ITGA4, CXCL12, and GZMK, respectively, in the external validation using the GSE51472 dataset. qRT-PCR validation in clinical samples confirmed that these genes were significantly upregulated in AVC patients with periodontitis. These genes were also correlated with immune cell infiltration, suggesting their potential role in AVC pathogenesis. Conclusion: These findings provide new clinical molecular diagnostics, treatment related molecular markers for AVC in patients with periodontitis and may facilitate further basic research into biological functions.
Abstract from DOAJ. Public domain (CC0 1.0).
Read the article at the publisher →