African‐ancestry based polygenic risk scores improve Alzheimer disease risk prediction in individuals of African Ancestry
Alzheimer's & Dementia(2022)
Abstract
AbstractBackgroundPolygenic risk scores (PRS) may be a useful approach to predict the risk of the complex disease and to be an important clinical tool for early intervention. PRS studies in Alzheimer Disease (AD) have focused on individuals of European Ancestry resulting in a >75% prediction accuracy. PRS generated from genome wide data in one population often provides reduced predictive accuracy in other populations. This is particularly problematic for underserved groups. In this study, we assessed and compared the PRS prediction accuracy of AD in individuals of African Ancestry (AA) using both AA and non‐Hispanic White (NHW) Genome Wide Association (GWAS) studies.MethodAs part of the Research in African American Alzheimer Disease Initiative (REAAADI) and ADGC, two TOPMED imputed AA datasets were generated (REAAADI:AD=234, cognitively unimpaired (CU)=676 and ADC9: AD cases=109, CU=224). We assessed the PRS using the effect sizes from summary statistics from the NHW (Kunkle et al. 2019) and the AA (Kunkle et al. 2021) studies. To model the effect of APOE we excluded APOE region in PRS constructing and included APOE alleles as separate terms in the prediction model. First, we generated PRS scores on the REAAADI dataset, and validated our model in ADC9 dataset. To assess the PRS performance, we employed the logistic regression modeling (covariates‐only (age, sex, and PC1:3),PRS‐only, and full (PRS+APOE+covariates) model) to construct receiver operator (ROC) curves.ResultEuropean ancestry‐derived PRS has the poor prediction power (AUC=0.53) in the REAAADI dataset whereas the AA‐derived PRS predicts better (AUC= 0.87). Further validation of the AA PRS in ADC9 dataset using covariates‐only, PRS‐only and full modelsvalidated that inclusion of African ancestry derived PRS significantly improves the accuracy of AD prediction in AA individuals (AUCcovariates‐only=0.59; AUCPRS‐only=0.74 and AUCfull=0.81).ConclusionOur results showed that AA‐derived PRS significantly improves AD risk prediction in AA individuals over European ancestry‐derived PRS. Our findings demonstrate the importance of increasing the diversity in genetic studies to improve precision medicine approaches. Moreover, the development of more accurate PRS models that can detect the risk of AD in all in all groups paves the way for more accurate prevention, early detection, and intervention of AD.
MoreTranslated text
Key words
alzheimer disease risk prediction,polygenic risk scores,african‐ancestry african‐ancestry
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined