Plasma proteomics are linked to intravascular imaging–derived plaque characteristics and clinical outcomes in patients with coronary artery disease

Background Intravascular imaging enables in vivo characterization of coronary plaque morphology and composition. While several plaque features are known to independently predict adverse outcomes, the underlying pathobiology remains incompletely defined. Plasma proteomic studies may contribute to its elucidation. Purpose We aimed to relate plasma proteomic profiling to plaque characteristics, and to identify clinically relevant coronary artery disease (CAD) subphenotypes. Methods We examined 581 CAD patients who underwent diagnostic coronary angiography/percutaneous coronary intervention. Blood samples were drawn prior to the procedure. Proteins were quantified by Slow Offrate Modified Aptamer assay, yielding relative fluorescence unit measurements for 1500 pre-selected plasma proteins involved in cardiovascular and inflammatory processes. All patients underwent virtual histology intravascular ultrasound, and a subset of 273 patients underwent near-infrared spectroscopy. We assessed associations of individual proteins with plaque characteristics, 5-year major adverse cardiac events (MACE) and 12-year all-cause mortality. Using the significant proteins, we performed Reactome pathway enrichment analysis. To identify distinct patient subphenotypes, we used principal component analysis, uniform manifold approximation and projection, and then HDBSCAN clustering. Subphenotypes were subsequently related to clinical outcomes and plaque characteristics, using linear/logistic regression and Cox proportional hazards models, respectively (subphenotype vs. the remainder of the cohort). Results Median age was 61.5 years, 75% were males and 55% of patients were diagnosed with acute coronary syndrome, while 45% presented with chronic coronary syndrome. Over a median follow-up of 4.3 years 151 MACEs occurred, and over a median of 12.8 years 179 patients died. Figure 1 presents pathway enrichment analyses for proteins associated with plaque components (A-B) and for those linked to MACE and all-cause mortality (C). Overall, inflammatory, hemostatic, and metabolic pathways were overrepresented in relation to plaque vulnerability and adverse outcomes. Figures 2A-B show the results of the clustering algorithm, yielding four distinct subphenotypes. In particular, Cluster 4 displayed significant protective associations with both all-cause mortality (HR=0.55 [95%CI: 0.35–0.88], padj=0.013) and MACE (HR=0.54 [0.33–0.87], padj=0.011) as well as associations with several plaque components: dense calcium percentage, fibrous tissue, and Lipid Core Burden Index. Conclusion This study identified plasma proteins and pathways associated with plaque components and morphological features. Moreover, unsupervised clustering of high-dimensional proteomic data delineated distinct, clinically relevant CAD subphenotypes. Together, these findings motivate future mechanistic studies and may support improved risk stratification and personalized therapeutic strategies.