Plasma proteomic signatures strongly predicted incident MACE (HR for Q4 vs. Q1: 3.18; 95% CI: 2.8-3.61) and significantly improved risk discrimination beyond SCORE2 (ΔC-index: +0.044; p<0.0001).
Cohort (n=41,200)
Yes
Do plasma proteomic signatures improve the detection of atherosclerotic burden and prediction of MACE beyond standard risk scores in asymptomatic individuals?
Plasma proteomic signatures can effectively capture atherosclerotic burden and significantly improve cardiovascular risk prediction beyond traditional risk scores in asymptomatic individuals.
Hazard Ratio: 3.18 (95% CI 2.8–3.61)
AIMS: Atherosclerosis is currently evaluated by imaging, but scalable circulating biomarkers to detect its presence and quantify overall burden are lacking. We sought to define plasma proteomic signatures that reflect the systemic burden of atherosclerosis. METHODS AND RESULTS: In UK Biobank, we trained machine-learning models on plasma proteomics in a nested, propensity-score-matched case-control sample (1,666 cases; 1,666 controls; Olink Explore 3072, 2,920 proteins) to derive four AtheroBurden signatures: one based on the entire proteome (all 2,920 proteins) and three biologically informed subsets-genetically anchored (n = 402), atherogenesis-related (n = 680), and artery-enriched (n = 248). We then computed individual-level signatures in 41,200 disease-free participants and validated their performance through 1) correlation with carotid plaque burden via ultrasound (n = 1,712); 2) prospective major adverse cardiovascular events (MACE) prediction over 13.7-year follow-up using multivariable Cox models and incremental prediction over Systematic COronary Risk Evaluation version 2 (SCORE2) (ΔC-index; net reclassification improvement NRI); as well as in the external Cooperative Health Research in the Region of Augsburg (KORA) S4 (n = 1,361) and Age1 (n = 796) cohorts; and 3) longitudinal trajectories analysis in 1,210 participants with three serial proteomic measurements.All four AtheroBurden signatures robustly discriminated prevalent atherosclerotic disease (ROC-AUC up to 0.91, 95% CI: 0.89-0.93) with their levels significantly increasing with the number of clinically affected vascular beds. We found significant correlations with carotid ultrasound-measured plaque burden. In prospective analysis, each signature exhibited strong associations with incident MACE (HR per SD increase in whole proteome signature: 1.75, 95% CI: 1.68-1.83; HR for Q4 vs. Q1: 3.18, 95% CI: 2.8-3.61), providing significant improvements in risk discrimination (ΔC-index: +0.044; p < 0.0001) and reclassification (NRI: 0.12 (95% CI: 0.09-0.15) at a 10% risk threshold) beyond SCORE2. These results were replicated in the KORA S4 and KORA-Age1 cohorts, where the signatures were associated with risk of future myocardial infarction and stroke. Furthermore, longitudinal analyses demonstrated steeper annual increases in proteomic signature among individuals with higher burden of vascular risk factors and individuals who experienced MACE over follow-up. CONCLUSIONS: Plasma proteomic signatures effectively capture atherosclerotic burden and improve cardiovascular risk prediction in asymptomatic individuals. They may complement existing risk stratification by serving as a scalable and accessible blood-based screening tool to identify individuals more likely to have subclinical atherosclerosis and may benefit from confirmatory imaging and earlier prevention strategies.
Zhang et al. (Mon,) conducted a cohort in Atherosclerosis (n=41,200). Plasma proteomic signatures (AtheroBurden) vs. Lower proteomic signature levels (Q1) and SCORE2 was evaluated on Incident major adverse cardiovascular events (MACE) (HR 3.18, 95% CI 2.8-3.61). Plasma proteomic signatures strongly predicted incident MACE (HR for Q4 vs. Q1: 3.18; 95% CI: 2.8-3.61) and significantly improved risk discrimination beyond SCORE2 (ΔC-index: +0.044; p<0.0001).