Paired plaque and plasma proteomics reveal molecular signatures of symptomatic atherosclerosis

Abstract

Background: Phenotyping of atherosclerotic plaque vulnerability has largely relied on histopathology that captures structural features, but does not fully account for clinical presentation. Proteomic profiling could uncover molecular readouts of vulnerability that refine plaque phenotyping and provide mechanistic insights. Yet, the proteomic signatures associated with plaque rupture and symptomatic presentation are poorly characterized. Methods: We profiled paired carotid plaque tissue and preoperative plasma from 88 patients undergoing carotid endarterectomy (51 symptomatic, 37 asymptomatic) using the Olink Explore 3072 platform. We related plaque protein abundance to symptomatic presentation and quantitative histopathological features, and compared the performance of histopathology- vs. proteomics-based models for discriminating symptomatic disease. Next, we developed proteomic signatures of cellular abundance and explored their associations with plaque phenotypes by using plaque single-cell RNA-sequencing (scRNA-seq) data. Finally, we assessed plaque-plasma concordance across 2,837 shared proteins. Results: Across 2,837 plaque proteins, 19 were differentially expressed in symptomatic plaques related to distinct clinical events, highlighting pathways related to neutrophil degranulation and innate immune system. FGFBP1 showed the strongest association with symptomatic presentation (log2 fold change = 1.14; P = 1.82 x 10^-6). Proteins associated with a composite vulnerability index based on histopathology were enriched for inflammatory pathways, including TNF signaling through NF{kappa}B, complement activation, and IL6-JAK-STAT3 signaling. Individual proteins also mapped to specific histopathological features, including CXCL8 associated with macrophage burden and lipid core size, and EPHB4 and PKN3 with neovascularization. A proteomics-based model discriminated symptomatic from asymptomatic plaques substantially better than a histopathology-based model (AUC 0.83 vs. 0.66; P = 0.026). Integration with scRNA-seq data enabled the development of cell-class signatures that correlated with histopathology readouts, including macrophage burden, smooth muscle cell content, and neovascularization. Plaque and plasma protein levels showed limited overall correspondence (median {rho}=0.11), although selected proteins, including FGFBP1, demonstrated concordant associations in plasma. Conclusions: Deep proteomic profiling of human carotid plaques identifies molecular signatures of symptomatic atherosclerosis that extend beyond conventional histopathology. These signatures implicate neutrophil activation and inflammatory signaling pathways as key determinants of plaque vulnerability. Although plaque and plasma proteomes are largely distinct, selected proteins may represent promising circulating biomarkers for future risk stratification.