Genetic Moderation of the Association of β-Amyloid With Cognition and MRI in Alzheimer Disease.

Abstract

There is considerable heterogeneity in the association between increasing amyloid β (Aβ) pathology and early cognitive dysfunction in preclinical Alzheimer's disease (AD). At this stage, some individuals show no signs of cognitive dysfunction, while others show clear signs of decline. The factors explaining this heterogeneity are particularly important for understanding progression in AD but remain largely unknown. In this study, we examine an array of genetic variants that may influence the relationships among Aβ, brain structure, and cognitive performance in two large cohorts. In 2953 cognitively-unimpaired participants from the A4 study, interactions between genetic variants and 18F-Florbetapir PET standardized uptake value ratio (SUVR) to predict the Preclinical Alzheimer's Cognitive Composite (PACC) were assessed. Genetic variants identified in the A4 study were evaluated in the Alzheimer's Disease Neuroimaging Initiative (ADNI, N=527) for their association with longitudinal cognition and brain atrophy in both cognitively unimpaired participants and those with mild cognitive impairment. In A4, four genetic variants significantly moderated the association between Aβ load and cognition. Minor alleles of three variants were associated with additional decreases in PACC scores with increasing Aβ SUVR (rs78021285, β =-2.29, SE=0.40, pFDR=0.02, nearest gene ARPP21; rs71567499, β =-2.16, SE=0.38, pFDR=0.02, nearest gene PPARD; rs10974405, β =-1.68, SE=0.29, pFDR=0.02, nearest gene GLIS3). The minor allele of rs7825645 was associated with less decrease in PACC scores with increasing Aβ SUVR (β =0.71, SE=0.13, pFDR=0.04, nearest gene FGF20). The genetic variant rs76366637, in linkage disequilibrium with rs78021285, was available in both A4 and ADNI. In A4, rs76366637 was strongly associated with reduced PACC scores with increasing Aβ SUVR (β =-1.01, SE=0.21, t=-4.90, p<0.001). In ADNI, rs76366637 was associated with accelerated cognitive decline (χ2=15.3, p=0.004) and atrophy over time (χ2=26.8, p<0.001), with increasing Aβ SUVR. Patterns of increased cognitive dysfunction and accelerated atrophy due to specific genetic variation may explain some of the heterogeneity in cognition in preclinical and prodromal AD. The genetic variant near ARPP21 associated with lower cognitive scores in A4 and accelerated cognitive decline as well as brain atrophy in ADNI may help to identify those at the highest risk of accelerated progression of Alzheimer's disease.