Polygenic risk score translation across diverse populations

Area of Science:

• Genetics and Genomics
• Cardiovascular Medicine
• Precision Public Health

Background:

• Polygenic risk scores (PRSs) are valuable for cardiovascular and cardiometabolic disease risk stratification.
• Current PRSs are primarily derived from European-ancestry datasets, limiting their effectiveness in diverse and admixed populations.
• This ancestry bias stems from variations in allele frequencies, linkage disequilibrium, and effect sizes across populations.

Purpose of the Study:

• To review methodological and translational advancements in PRS development across diverse populations.
• To emphasize progress in coronary artery disease (CAD), blood pressure, hypertension, type 2 diabetes, obesity, and atrial fibrillation.
• To highlight the importance of multi-ancestry and ancestry-aware models for equitable PRS application.

Main Methods:

• Review of recent literature on PRS development and validation in diverse populations.
• Examination of single-ancestry versus multi-ancestry PRS frameworks.
• Analysis of emerging methods for admixed genomes, including ancestry deconvolution and local-ancestry-aware models.

Main Results:

• Broader discovery resources and ancestry-aware methods improve PRS predictive performance across traits.
• Coronary artery disease (CAD) shows the most mature evidence for clinical gains from multi-ancestry PRSs.
• Progress is uneven across phenotypes, with blood pressure, hypertension, and type 2 diabetes facing calibration and implementation challenges; obesity and atrial fibrillation are rapidly advancing but less translationally ready.

Conclusions:

• Admixed and underrepresented populations are essential for developing robust and generalizable PRS models.
• Future precision cardiovascular medicine requires PRS models that are calibrated, interpretable, and clinically useful across diverse populations.
• Addressing ancestry bias is critical for the equitable clinical translation of PRSs in cardiovascular risk assessment.