Integrating clinical, proteomics, and polygenic scores to improve cardiovascular risk prediction: a prospective...

Traditional cardiovascular disease (CVD) risk assessment relies on clinical scores like SCORE2 and AHA PREVENT equations. However, these tools often miss dynamic molecular shifts preceding clinical events, leaving substantial residual risk undetected. High-dimensional proteomics provides a real-time snapshot of physiological states that may bridge the gap between genetic predisposition and clinical outcomes.

This study aimed to evaluate if high-dimensional proteomics enhances cardiovascular risk stratification beyond established clinical standards and polygenic risk scores, identifying significant residual risk through an integrated multi-omics framework.

We analyzed 44,431 CVD-free participants from the UK Biobank Pharma Proteomics Project. Plasma levels of 1,460 proteins were quantified via the Olink Explore platform. A Protein Risk Score (ProtRS) was developed using LASSO-penalized Cox regression with stability selection (1,000 resamples). We evaluated the incremental value of ProtRS when added to a baseline model comprising clinical scores (SCORE2, PREVENT, FRS), standard biomarkers (CRP, NT-proBNP), and a CVD-specific PRS.

The 117-protein ProtRS demonstrated superior individual discrimination (C-index 0.769) over SCORE2 (0.667), PREVENT (0.645), and FRS (0.672). Adding ProtRS to the baseline clinical model improved predictive performance, increasing the C-index from 0.669 to 0.774 (Delta C = 0.105). ProtRS effectively captured the prognostic information of CRP and NT-proBNP. The integrated multi-omics model achieved a significant total Net Reclassification Improvement of 3.1% and an Integrated Discrimination Improvement of 0.018. Decision Curve Analysis confirmed the highest clinical net benefit for the integrated model. Functional enrichment showed ProtRS captures a critical immune-metabolic nexus involving cytokine signaling and lipid handling.

Integrating dynamic proteomics with genetic and clinical factors offers predictive power far superior to traditional standards. This multi-omics approach identifies significant residual risk, supporting a shift toward molecular phenotyping for precision cardiovascular prevention.