Pharmacogenomics polygenic risk score for drug response prediction using PRS-PGx methods.

Area of Science:

• Genetics
• Pharmacogenomics
• Biostatistics

Background:

• Polygenic risk scores (PRS) are used for disease prediction.
• Current practice applies disease PRS to pharmacogenomics (PGx) settings, assuming prognostic and predictive effects are equivalent.
• This assumption is often violated, limiting prediction accuracy.

Purpose of the Study:

• To develop and evaluate novel pharmacogenomic polygenic risk score (PRS-PGx) methods.
• To address the limitations of applying disease-derived PRS to drug response prediction.
• To introduce a Bayesian regression approach (PRS-PGx-Bayes) for simultaneous modeling of prognostic and predictive effects.

Main Methods:

• Development of a series of PRS-PGx methods, including PRS-PGx-Bayes.
• Simulation studies to compare PRS-PGx methods against disease PRS methods.
• Application of PRS-PGx methods to PGx Genome-Wide Association Study (GWAS) data from the IMPROVE-IT trial.

Main Results:

• PRS-PGx methods generally outperform disease PRS methods in prediction accuracy.
• PRS-PGx-Bayes demonstrated superior performance compared to other PRS-PGx methods in simulations.
• Application to IMPROVE-IT data showed PRS-PGx-Bayes significantly improved prediction of treatment-related LDL cholesterol reduction.

Conclusions:

• Shifting from disease PRS to PGx PRS approaches is crucial for accurate drug response prediction.
• The proposed PRS-PGx methods, particularly PRS-PGx-Bayes, offer substantial improvements in prediction accuracy.
• PRS-PGx-Bayes effectively captures treatment-specific predictive effects, advancing pharmacogenomic research.