Instability of high polygenic risk classification and mitigation by integrative scoring

Area of Science:

• Genetics and Genomics
• Personalized Medicine
• Bioinformatics

Background:

• Polygenic risk scores (PRS) are increasingly used in healthcare and commercial settings.
• The classification of high polygenic risk can vary significantly between different PRS.
• A need exists for stable and reliable methods to identify individuals at high genetic risk.

Purpose of the Study:

• To assess the association and classification performance of cataloged PRS for three complex traits.
• To compare the classification consistency of individual best PRS (Best(Pub n)) with an integrative scoring approach (ChronoAdd(Pub n)).

Main Methods:

• Chronologically ordered trait-related publications (Pub n) and identified the single best PRS (Best(Pub n)) for each.
• Utilized the PRSmix framework to generate ChronoAdd(Pub n) scores, integrating multiple PRS over time.
• Evaluated population-level association strengths and individual high-risk classification (top 10%) for both Best(Pub n) and ChronoAdd(Pub n) scores.

Main Results:

• Individual Best(Pub n) scores showed generally consistent population-level associations but wide variation in classifying individuals in the top 10% risk group.
• ChronoAdd(Pub n) scores demonstrated more consistent high-risk classification among themselves compared to Best(Pub n) scores.
• The integrative scoring approach improved the stability and reliability of high-risk individual identification.

Conclusions:

• An integrative polygenic risk scoring approach (ChronoAdd(Pub n)) provides more stable and reliable classification of high-risk individuals.
• This adaptable framework can incorporate new PRS as they emerge, aligning with current implementation strategies.
• The ChronoAdd(Pub n) method enhances the clinical utility of PRS by standardizing high-risk stratification.