GLOBAL CORTICAL [F18]FLORTAUCIPIR ASSOCIATION WITH THE TOP 20 ALZHEIMER’S DISEASE RISK GENES

Abstract

We previously explored the genetic associations of the top 20 genome-wide significant Alzheimer's disease (AD) risk genes with amyloid PET, FDG PET and MRI biomarkers. Continuing along the A/T/N framework for AD biomarkers, we investigated the association of these variants with [F18]Flortaucipir SUVR. Our sample included 128 cognitively normal (NC), 98 mild cognitive impairment (MCI), and 24 dementia (DEM) participants from ADNI with available [F18]Flortaucipir scans and imputed GWAS data. Scans were registered to nearest-visit MR and normalized to cerebellar crus, generating SUVR images. ROIs were defined using Freesurfer 6.0. Four step-wise regression models were run (pooled, NC, MCI and DEM) in SAS 9.4 with global cortical SUVR as an outcome variable and the genetic risk variants as predictors while controlling for age, sex, and APOE4 status using the Akaike Information Criterion (AIC) significance level of 0.157 for model selection. We reproduced the final step-wise regression models using voxel-wise analysis in SPM12 with a significance threshold of uncorrected p<0.01 and minimum cluster size of 50 voxels. Table 1 lists demographic characteristics and % minor allele carriers for variants that showed significant or trending associations with tau burden. The genetic major effects for each diagnostic group can be seen in Figure 1. FERMT2 rs17125944 and NME8 rs2718058 were associated with tau deposition in pooled sample. CASS4 rs7274581, PTK2B rs28834970 and SORL1 rs11218343 were associated with tau in NC, while ABCA7 rs3752246, ABCA7 rs3764650 and BIN1 rs7561528 were associated with tau in the MCI stage. CLU rs11136000 was associated with tau burden in both pooled and DEM samples. Finally, SLC24A4/RIN3 rs10498633 was associated with tau deposition in only the DEM group.