Acute coronary syndrome, driven by vulnerable plaque (VP) instability, is a major cause of cardiovascular mortality. Current diagnostic methods for VPs are limited by invasiveness or low specificity, highlighting the need for non-invasive biomarkers. Using single-cell RNA sequencing (scRNA-seq) of peripheral blood mononuclear cells (PBMCs) from coronary artery disease (CAD) patients with VPs and controls, we identified circulating T cell-platelet aggregates (TPAs) significantly enriched in VP patients and linked to plaque instability via pro-inflammatory pathways. Through high dimensional weighted gene co-expression network analysis, we discovered TPAs' hub genes and demonstrated their role in plaque destabilization. Furthermore, employing machine learning, including Boruta, least absolute shrinkage and selection operator (LASSO) regression and support vector machine-recursive feature elimination (SVM-RFE), we screened for five blood biomarkers that can serve as diagnostic indicators for VPs. Our study demonstrates that TPAs are critically involved in VPs formation. Furthermore, we identified EPHB6, STAT1, RPL23, IKZF3 and AHCY as potential circulating biomarkers for non-invasive detection of VPs.
Keywords: Coronary vulnerable plaque; Diagnostic biomarker; Machine learning; ScRNA-seq; T cell-platelet aggregates.
© 2026. The Author(s).