This study investigates the optimal approach for calculating polygenic risk scores (PRS) to identify individuals at high and low risk of Alzheimer's disease (AD). The researchers found that the best prediction accuracy was achieved using a model with two predictors (APOE and PRS excluding the APOE region), selecting SNPs with a p-value threshold of <0.1. While prediction accuracy was similar across different PRS approaches, individual scores and rankings varied. Standardizing PRS against the population mean, rather than the sample mean, improved comparability between studies. The study offers recommendations for best practices in polygenic profiling for AD risk assessment.

Ganna Leonenko, Emily Baker, Joshua Stevenson-Hoare, Annerieke Siersma, Mark Fiers, Julie Williams, Bart de Strooper, Valentina Escott-Price