Functionally Informed Polygenic Risk Scores for Alzheimer’s Disease

Abstract

AbstractBackgroundAlzheimer’s Disease (AD) is a heritable, neurodegenerative disease. Most AD cases can be categorized as Late Onset Alzheimer’s Disease (LOAD) cases, which are polygenic. Genome Wide Association Studies (GWAS) of LOAD have identified several associated single nucleotide polymorphisms (SNPs). These variants are in 75 genetic loci, and have implicated a diverse array of genes, cell types, and biological pathways in LOAD. Additionally, many associated variants are not in protein coding regions of the genome. However, it is difficult to determine if all associated variants are functionally relevant and contribute to the etiology or progression of the disease. Here, we present CoRE‐BED, a variant annotation tool that can categorize SNPs into functional classes, such as promoters or enhancers, for a given cell type. These functional variants in each cell type are subsequently used to determine functional variant contributions to LOAD heritability and risk.MethodWe annotated a recently published GWAS of AD (Bellenguez et al 2022) using CoRE‐BED. CoRE‐BED leverages 19 functional assays across 33 distinct cell types from EpiMap to classify variants into functional categories for each supported cell type. Annotated variants were used to estimate cell type specific partitioned heritability using LD Score Regression. Finally, annotated variants were used to build cell type specific Polygenic Risk Scores (PRS). PRS was tested using multiple PRS software and utilized the annotated GWAS as a training dataset and independent cases and controls derived from Vanderbilt University’s biobank, BioVU.ResultCoRE‐BED provided functional classifications for all variants in the GWAS for individual tissue types. Variants that pass genome wide significance (P < 5×10−8) are enriched for functional annotations. Immune related cell types contribute the greatest proportion of partitioned heritability. Standard PRS methods outperform cell type informed PRS measures.ConclusionCoRE‐BED is a robust annotation tool that can determine functional annotations at the SNP level. AD GWAS significant associations overlap with functionally annotated SNPs in multiple cell types. Partitioned heritability indicates that immune related cell types may play an important role in the heritability of AD. However, conventional methods for PRS still provide the most robust means to predict AD risk.