Prediction of incident coronary artery disease in individuals with zero coronary artery calcium using a novel multi-ancestry, label-free polygenic risk score framework

Background A coronary artery calcium (CAC) score of 0 is widely considered to indicate low short- to intermediate-term risk for coronary artery disease (CAD) and is frequently used to defer lipid-lowering therapy. However, a subset of individuals with CAC=0 still experience events, highlighting residual risk not captured by imaging alone. Polygenic risk scores (PRS) quantify lifelong inherited susceptibility, but conventional approaches rely on predefined ancestry labels despite human genetic diversity existing along a continuum. To address this limitation, we developed 8 Billion, a novel, label-free framework that models genetic similarity without pre-labeling individuals by ancestry. We evaluated whether a CAD PRS derived using this approach identifies clinically meaningful residual risk among individuals with baseline CAC=0.

Methods We analyzed participants from the Multi-Ethnic Study of Atherosclerosis (MESA) with baseline CAC=0. The 8 Billion framework estimates individualized PRS by anchoring each participant to a genetically similar reference neighborhood rather than discrete ancestry groups. Multivariable Cox proportional hazards models assessed associations between PRS-defined risk groups and incident CAD, adjusting for principal components of genetic variation (PC1–PC4), age, sex, smoking status, systolic blood pressure, total and high-density lipoprotein cholesterol, diabetes, and antihypertensive medication use. Two classifications were evaluated: (1) a Top 5% group defined by the highest 5% of PRS-derived odds ratios in the cohort; and (2) an individualized high-risk group defined using a personalized threshold derived from the 8 Billion framework. Ten-year absolute risk estimates were derived from adjusted models.

Results Despite CAC=0 at baseline, polygenic burden was independently associated with incident CAD. Individuals in the Top 5% PRS group had higher risk of CAD events compared with the remainder (hazard ratio [HR], 3.12; 95% CI, 1.05–9.31; P=0.041). The individualized high-risk group defined through 8 Billion was similarly associated with increased CAD risk (HR, 2.52; 95% CI, 1.12–5.66; P=0.025). Estimated 10-year ASCVD risk among high-PRS individuals exceeded the 7.5% threshold commonly used to guide initiation of lipid-lowering therapy, despite CAC=0. In fully adjusted models, conventional risk factors were not statistically significant within this subset.

Conclusions Among individuals with CAC=0 in a multi-ethnic cohort, a label-free, ancestry-continuum PRS approach identified subgroups at significantly increased risk of incident CAD and at guideline-relevant 10-year treatment thresholds. Integration of polygenic risk with CAC imaging refines preventive decision-making beyond imaging alone.