Abstract
Background Genetic and environmental factors contribute to Alzheimer's disease (AD) risk. Understanding gene‐environment interactions may provide insight into unexplained AD heritability. Higher educational attainment is associated with lower AD risk, but the mechanism remains unclear. We conducted a genome‐wide association study (GWAS) to explore genetic‐education‐related associations with AD through SNP‐education interaction and education‐stratified analyses. Method Educational attainment data were available and analyzed among 23,642 non‐Hispanic white (NHW; 10,272 cases) and 3,461 African American (AFA; 1064 cases) participants from the AD Genetic Consortium and the Framingham Heart Study. Educational attainment was dichotomized by median years of education across cohorts, which equated to completing four years of college. Across 35 datasets, we conducted separate GWAS: 1) including a SNP‐by‐education interaction term and 2) stratifying by median education status. MAGEE was used to estimate SNP‐by‐education interaction effects and SAIGE was used to estimate SNP effects in stratified analysis. GWAS models adjusted for age, sex, and principal components for population structure. METAL was used for inverse‐variance weighted within‐ancestry fixed‐effects meta‐analysis and METASOFT was used to estimate cross‐ancestry effects. Top GWAS hits were further analyzed for association with longitudinal trajectories of harmonized memory, language, and executive function factor scores in education‐stratified linear mixed effects models. Result Stratified GWAS identified a genome‐wide significant association among participants with lower educational attainment in CR1, a well‐known AD‐associated locus on chromosome 1 (top SNP: rs12037841; lower educational attainment: MAF=0.19, OR=1.33, p = 3.1x10‐10; higher educational attainment: MAF=0.19, OR=1.09, p = 0.03; interaction‐model: βsnpXedu=‐0.18, p = 0.0018). Effects among those with lower educational attainment were present in both ancestries (NHW: MAF=0.19, OR=1.30, p = 1.8x10‐9; AFA: MAF=0.03, OR=1.64, p = 0.02). In analysis with neuropsychological factor scores, rs12037841 was associated with faster decline in memory and language among participants with lower educational attainment (memory: βsnpXtime=‐0.010, 95% CI:[‐0.016,‐0.004], p = 0.0019; language: βsnpXtime=‐0.006, 95% CI:[‐0.011,‐0.002], p = 0.0083). Weaker, non‐significant effects were observed among participants with higher educational attainment. Conclusion In educational attainment‐stratified GWAS of AD, we identified stronger association of known AD‐related gene CR1, among those with lower educational attainment. The finding implicating CR1, a complement pathway gene, suggests that the risk education confers on AD may be moderated by immune‐related mechanisms.