Microglia‐specific Alzheimer's disease polygenic risk score is associated with amyloid‐β, tau, and microglial activation.

Background: Genome‐wide association studies (GWAS) of Alzheimer's disease (AD) have implicated microglia in AD pathogenesis, and microglia is a promising cellular therapeutic target in AD. However, it remains unclear when and how AD genetic risk localizing to microglia contributes to AD pathophysiology. Method: Participants are from two community‐based cohorts (Religious Orders Study, the Rush Memory and Aging Project [ROSMAP]; n = 299 deceased; age 88.9±7.1, female 67%) and the screening phase of an AD prevention trial (the Anti‐Amyloid Treatment in Asymptomatic Alzheimer's study [A4]; n = 2919 cognitively unimpaired; age 71.4±4.8, female 60%). We quantified post‐mortem Aβ and tau in ROSMAP, and used PET to quantify Aβ (florbetapir cortical composite standardized uptake value ratio [SUVR]) and tau (flortaucipir temporal composite SUVR; subset n = 302) in A4. In a subset of ROSMAP participants (n = 31), we evaluated the proportion of activated microglia (PAM) in post‐mortem neocortex. We combined AD GWAS summary statistics with cellular gene expression profiles from human brain single nucleus sequencing, to derive cell‐type‐specific ADPRS (excluding APOE region) in ROSMAP and A4. We assessed the association of each (standardized) cell‐type‐specific ADPRS with AD pathology (linear regression; adjusting for APOE genotypes, age, sex, and first 10 genotype principal components), and examined the correlation between the microglial ADPRS and PAM (Pearson's correlation). We used Bonferroni‐corrected p‐value threshold, p<6.25×10−3. Result: Microglial, astrocytic, and endothelial ADPRS were associated with higher Aβ and tau in ROSMAP (Figure 1A‐B). In A4, microglial ADPRS showed the strongest association with Aβ (beta = 0.017, p = 1.6×10−7; Figure 1C), while the association of all other cell‐type‐specific ADPRS with Aβ were no longer significant once adjusted for microglial ADPRS (all p>0.05). Microglial ADPRS showed a suggestive association with temporal composite tau (beta = 0.033, p = 0.012; Figure 1D), even after additionally adjusting for Aβ (beta = 0.024, p = 0.048). Microglial ADPRS was positively correlated with PAM in ROSMAP (Figure 2). Conclusion: Genetic risk of AD localizing to microglial genes may contribute to Aβ and tau accumulation and microglial activation, likely starting from the preclinical (cognitively unimpaired) stage. These findings directly connect aggregate microglial AD genetic risk with Aβ and tau accumulation, and provide a genetic rationale to test microglia‐specific interventions in AD prevention trials. [ABSTRACT FROM AUTHOR]