Global and local ancestry modulate APOE association with Alzheimer's neuropathology and cognitive outcomes in an admixed sample

Background Dementia affects more Black individuals, likely due to a combination of environmental and biological factors 1,2,3 . APOE ε4 allele risk of dementia is different between individuals with European (EUR) and African (AFR) ancestries 4,5,6 . It is unclear what drives these differences in Alzheimer’s disease (AD) neuropathology among patients with cognitive decline, both outcomes influenced by different sets of biological and environmental factors. We studied the interaction between global and local APOE African ancestry, e4 allele, burden of neuritic plaques (NP) and neurofibrillary tangles (NFT), and cognition in a postmortem sample of admixed individuals with AD pathology. Method Community‐dwelling autopsied brains (n=400) without neuropathological changes or with NP scored using CERAD and NFT scored by Braak stage but lacking others comorbid neuropathologies. Subject’s cognitive function prior to death as assessed using Clinical Dementia Rating sum of boxes (CDR‐SOB). APOE genotypes were assessed with allele‐specific amplification. Global ancestry was inferred using Admixture and local ancestry using RFMix, derived from panels of variants interrogated by microarray. Result When dichotomously ascertained at 2% of AFR, and adjusting for age, sex, educational attainment and APOE ε4 carrier status, the admixed group is significantly protected of neuritic plaques (NP, beta=‐0.21, p=0.04) but not neurofibrillary tangles (NFT, beta=0.137, p=0.39). However, CDR‐SOB was associated with continuously distributed AFR (Table 1) only among individuals with severe levels of NFT (beta=7.04, p<0.001) and NP (beta=3.9, p<0.001). Stratifying the same analysis between APOE status, only APOE ε4 noncarriers presented significant associations between AFR and CDR‐SOB, but APOE ε4 carriers lost significance. APOE local ancestry inference of showed that non‐European admixed APOE ε4 noncarriers had lower NP burden outcomes (Table 2, beta=‐0.28, p=0.032 and beta=‐0.27, p=0.015 with and without adjusting for AFR, respectively), but worst cognitive decline compared to European APOE of the same genotype group when adjusting by NP levels (beta=1.01, p=0.041, without adjusting for AFR). Lastly, APOE ε has a significant effect on neuropathology, but not cognition, only within European local ancestry contexts (Table 3, beta=0.61, p<0.001). Conclusion Our results demonstrate a complex relation between APOE genotypes and local ancestry, AFR global ancestry and AD‐related neuropathology and cognition outcomes.