Assessment of Alzheimer's disease polygenic risk score on longitudinal amyloid accumulation in cognitively intact older adults.

Background: We investigated whether an Alzheimer's disease (AD) polygenic risk score (PRS) could predict longitudinal brain amyloid accumulation in cognitively intact older adults participating in the Flemish Prevent AD Cohort KU Leuven (F‐PACK). Previously published data have highlighted associations between AD susceptibility loci and AD‐related brain changes. Method: Sixty‐one cognitively healthy F‐PACK participants (median age: 68 (56‐79), 29 females, 29 APOE ε4 carriers) received an [18F]Flutemetamol‐PET scan and structural MRI at baseline and follow‐up (average time interval 4.6 years (3.4‐8.6)). Amyloid rate of change was defined as absolute change divided by time interval. Participants were genotyped using the Illumina GSA and imputed using the Michigan server and HRC reference panel. Data were quality controlled using PLINK(v1.9, http://pngu.mgh.harvard.edu/purcell/plink/, Purcell et al., 2007) (minor allele frequency >0.01). PRS calculation was performed using PRSice(Choi & O'Reilly, 2019) with the Stage 1 summary statistics (containing SNP p‐values and corresponding effect sizes) from the GWAS performed by Kunkle et al.,(2019). European individuals from the 1000 Genomes dataset were used for clumping (N=503). We calculated PRSs at three thresholds for SNP inclusion: pT=0.5(Escott‐Price et al., 2015); 1x10‐5; 5x108. A Mann‐Whitney U‐test was used to determine amyloid rate of change differences between APOE ε4 carriers and non‐carriers. Linear models were used to assess if the PRS could predict changes in amyloid, with age, sex and the first three principal components from the PRS calculation as covariates, at each of the p‐value thresholds for SNP inclusion. Result: APOE ε4 carriers showed a trend towards a significantly higher rate of change than non‐carriers (U=459.5, p=0.056). When predicting amyloid rate of change, the linear model outputs were similar when the thresholds for SNP inclusion were pT=1x10‐5 and 5x10‐8. Although models did not reach significance, there was a significant effect of PRS on amyloid rate of change: pT=1x10‐5: β=0.0056 (CI: 0.0000785‐0.011), p<0.05; pT=5x10‐8: β=0.0054 (CI: 0.000042‐0.011), p<0.05. PRS was not a significant predictor for amyloid accumulation when the threshold for SNP inclusion was pT=0.5: β=0.0013 (‐0.004‐0.007), p>0.05. Conclusion: PRSs calculated with a SNP inclusion threshold of pT=1x10‐5 or 5x10‐8 explain a small but significant amount of variance in amyloid accumulation in cognitively intact older adults. [ABSTRACT FROM AUTHOR]