Integrative analysis of summary data from GWAS and eQTL studies predicts tissue‐specific gene targets for Alzheimer’s disease

Abstract

AbstractBackgroundGenome‐wide association studies (GWAS) have been successful in identifying risk variants for Alzheimer’s disease (AD). However, it is unclear how these genetic variants affect tissue‐specific transcriptomic alternations in AD. The summary data‐based Mendelian randomization analysis (SMR) is a method designed to address this problem. It integrates summary data from independent GWAS and expression quantitative trait locus (eQTL) studies to identify genes differentially expressed between phenotypic groups.MethodsWe performed SMR to predict gene targets differentially expressed in AD, using summary statistics of brain eQTL studies and a landmark AD GWAS. Significant eQTL findings (FDR p≤0.05) of thirteen brain tissues were downloaded from https://gtexportal.org. GWAS summary statistics were obtained from an AD meta‐GWAS: https://ctg.cncr.nl/software/summary_statistics. The linkage disequilibrium (LD) information, required by SMR, were estimated using the ADNI genotyping data (N=1,536). For each tissue, SMR was performed to identify the pleiotropic associations between gene expression data and the AD phenotype. Bonferroni method was used for multiple comparison correction.ResultsFigure 1 shows 64 significant AD‐gene associations identified by SMR, involving 14 genes and 13 brain tissues. Some findings have been studied as top AD genes (BIN1 and CR1), or located around the AD genes (APOC2, PVRIG, STAG3, GPC2). Two genes (HLA‐DQA2 and SNX32) are new candidates that have no SNPs achieving meta‐GWAS p<5.0E‐8 within 20 kb from the gene. HLA‐DQA2 is located in the region of AD risk gene HLA‐DRB1. SNX32 is a novel AD associated gene predicted by SMR. SNX32 is a family member of SNX proteins, and is a component of retromer which has been widely reported for its role in the AD etiology.ConclusionsTissue‐specific transcriptomic association analysis using GWAS summary statistics and eQTL data identified genes associated with AD due to pleiotropy. Multiple genes were identified as potential AD targets in various brain tissues, given the estimated pleiotropic associations between their expression levels and AD diagnosis. These identified genes warrant further investigation as potential targets for functional validation to help better understanding of molecular mechanisms of AD.