The expected polygenic risk score (ePRS) framework: an equitable metric for quantifying polygenetic risk via modeling...

Area of Science:

• Genetics
• Population Genetics
• Statistical Genetics

Background:

• Polygenic risk scores (PRSs) are crucial for genetic risk prediction.
• Allele frequency differences across ancestral populations introduce bias in PRSs.
• This necessitates methods for equitable PRS application in diverse cohorts.

Purpose of the Study:

• To propose a novel framework for calibrating PRSs based on an individual's ancestral makeup.
• To introduce expected PRS (ePRS) and residual PRS (rPRS) metrics for bias correction.
• To ensure equitable genetic risk quantification across diverse populations.

Main Methods:

• Defined ePRS as the expected PRS value based on global/local admixture.
• Defined rPRS as the deviation of PRS from ePRS.
• Validated the framework using simulation studies and real-world datasets (TOPMed, All of Us).

Main Results:

• Adjusting for ePRS yields nearly unbiased PRS-outcome association estimates.
• rPRS effect size adjusted for ePRS mirrors PRS adjusted for genetic principal components (PCs).
• Consistent results across cardiovascular traits in TOPMed and All of Us datasets.

Conclusions:

• The ePRS framework effectively corrects for population stratification in PRS analysis.
• This approach provides an equitable strategy for quantifying genetic risk in diverse populations.
• Facilitates broader and fairer implementation of polygenic risk scores.