P1‐028: POLYGENIC RISK SCORE DISTRIBUTIONS OF ALZHEIMER'S DISEASE WITHIN DIFFERENT POPULATIONS

The utility of Polygenic Risk Score (PRS) is gaining increasing attention for estimating individual genetic risk profile and disease risk prediction in Alzheimer's Disease (AD). The accuracy of PRS depends on the discovery cohort, with the majority of studies mainly conducted in European-ancestry (EUR) populations. The latter along with the variability of linkage disequilibrium (LD), even when the causal variants are the same, results in the poor generalizability of PRS analysis among different ancestries. In the current study, we aimed to assess the differences in AD PRS estimates due to the differences in LD and allele frequencies in EUR, East-Asia (EAS) and Africa (AFR) populations. Our study utilized AD associated SNPs reported by the International Genomics of Alzheimer's Project (IGAP) consortium as the discovery cohort to calculate AD PRS in different populations. We used data from 1000 Genomes Phase 3 for the EUR, EAS and AFR populations. As the aim of the study was to assess the systematic differences in PRS due to genomic structure (LD and allele frequencies), equal effect sizes (European based) were assumed for causal variants across the different populations. The investigation of mean and standard deviation of PRS distributions indicated substantial differences among the different populations due to the differences in the effect allele frequencies and in LD that drives differences in effect size estimates even when causal variants are the same. The latter was visually inspected by plotting the LD structure of statistically significant AD-associated SNPs in the three populations. Our results indicate that the genetic score estimates based on one population cannot be applied to another population without taking into account both populations’ structures. The predictive ability of the PRS in non-European populations is currently limited by the lack of genetic research performed in those populations. The implementation of genetic studies across all ethnicities and the development of statistical methods that can appropriately account for effect allele frequency and LD is of great importance if PRS is destined to be applied in a clinical environment.