Basic Science and Pathogenesis.

Background: Currently, it is unclear to what extent late-onset Alzheimer's disease (AD) risk variants contribute to early-onset AD (EOAD). One method to clarify the contribution of late-onset AD genetic risk to EOAD is to investigate the association of AD polygenic risk scores (PRS) with EOAD. We hypothesize that in the Longitudinal Early-Onset Alzheimer's Disease Study (LEADS), EOAD participants will have greater PRS than early-onset amyloid-negative cognitively-impaired participants (EOnonAD) and controls, and investigate the association of AD PRS with age of disease onset (AoO) and cognitive performance.

Methods: GWAS data was generated for LEADS participants, including those with EOAD, EOnonAD, and controls, with the Illumina Global Screening Array. A PRS was calculated using the 31 SNPs and weights published previously by Desikan et al. (2017) for LEADS participants with imputed GWAS data (N = 369). Logistic regression models including age, sex, PRS, and genetic ancestry principal components were tested to identify predictors of EOAD (N = 210) vs. EOnonAD (N = 69) and controls (N = 89). ANCOVA models were used to assess group differences in PRS scores. Kaplan-Meier regression was used to assess differences in EOAD AoO for tertile-binned PRS groups. Within EOAD, pre-calculated cognitive domain scores for speed and attention, working memory, episodic memory, language, and visuospatial performance were assessed for correlation with PRS.

Results: The AD PRS was a predictor of EOAD (p = 0.014), with the model explaining 10.5% of variance (X2 = 40.971, p<0.001). EOAD participants had higher PRS scores (mean = 0.0012, standard deviation (SD) = 0.015) compared to EOnonAD and controls (mean = -0.0018, SD = 0.015) (F = 6.602, p = 0.011). Survival analysis indicated no significant differences in EOAD AoO between PRS groups (X2 = 3.396, p = 0.183). In the EOAD group, PRS was associated with cognitive scores for speed and attention (r = 0.204, p = 0.007), language (r = 0.230, p = 0.002), and visuospatial performance (r = 0.166, p = 0.037).

Conclusions: In the LEADS cohort, AD PRS is a predictor for EOAD, and is associated with cognitive performance, but does not predict EOAD AoO. This suggests that while late onset AD-associated genetic variants contribute to disease risk and processes, they do not account for a large portion of disease risk, and do not explain differences in disease AoO in the LEADS cohort.