Blood Test Breakthrough Could Detect Hidden Artery Disease Years Earlier

An international research team has uncovered four distinct plasma proteomic signatures that could help detect hidden atherosclerosis before it leads to heart attacks or strokes. Using machine learning on data from nearly 45,000 UK Biobank participants, the scientists aimed to find a blood-based alternative to imaging for spotting early-stage artery disease. Their findings suggest these protein patterns may improve risk prediction beyond current clinical tools.

Atherosclerosis often develops silently for years before causing life-threatening events like myocardial infarction or stroke. While imaging remains the gold standard for detecting plaque buildup, no blood biomarker has reliably measured this burden—until now. The team applied CatBoost machine learning to plasma proteomes from 44,788 UK Biobank volunteers, identifying four key signatures: WholeProteome, Genetic, Mechanistic, and Arterial.

Over a median follow-up of 13.7 years, all four signatures showed strong links to future major cardiovascular events—even in participants without baseline atherosclerosis. The WholeProteome signature, for instance, carried a hazard ratio of 1.70 per standard deviation increase. These models also outperformed the established SCORE2 risk assessment, achieving higher discrimination (C-index 0.74–0.77 vs. 0.70) and better calibration, particularly in high-risk individuals. The signatures proved highly accurate at distinguishing between people with known atherosclerotic disease and healthy controls, with ROC-AUC values reaching 0.91. Unlike traditional risk scores, they offered significant improvements in reclassifying patients' risk levels, potentially enabling earlier intervention.

The study highlights plasma proteomics as a promising, scalable method for detecting subclinical atherosclerosis without relying on imaging. These protein-based signatures could refine cardiovascular risk prediction, helping clinicians identify high-risk patients sooner. Further validation may pave the way for their use in routine clinical practice.