Evaluating the genetic risk of Alzheimer's disease and related dementia

The estimated heritability of Alzheimer's disease and related dementia (ADRD) is ~60%-80%. Except AOPE, most ADRD genes have small effects but their effects can be collectively measured by using polygenic scores (PGS). For PGS having high predictability, they can be used to identify high-risk individuals before they show any symptoms thereby early intervention programs can be applied to prevent ADRD or delay the onset . In addition, for PGS with high predictability, genetic variants included in calculating them are likely to be ADRD related hence PGS provide another way to identify ADRD associated genes, especially those having small effects and thus are not genome-wide significant. The objectives of this study are deriving PGS for ADRD with high predictability and using PGS to understand the genetic mechanisms of ADRD in diverse populations. We will use publicly available large-scale genome-wide association studies (GWAS) of ADRD as the discovery datasets. The requested datasets will be used as the target datasets. Different PGS programs/pipelines, including those developed by us will be evaluated to identify the one with the best performance. Furthermore, we will use the requested datasets to develop methods to adjust for the genotyping platforms/batches effects in different datasets as well as population stratification, with the goal to harmonize PGS distributions and thus uniform PGS thresholds to define high/low risk in different datasets can be determined. Additionally, we will evaluate the effects of imputation quality on PGS calculation by comparing PGS calculated using imputation dosages and genotypes obtained by whole genome sequence. Cox proportional hazard model or logistic regression will be used with sex as a covariate. We will also perform sex stratified analyses as ADRD having different prevalence in males and females. For the PGS that have high predictability, we will perform fine mapping to identify genes impacted by genetic variants included in calculating PGS. Then network and pathway analyses will be performed to elucidate the genetic mechanism of ADRD.

Genetics factors play significant roles in Alzheimer's disease and related dementia (ADRD). Most ADRD genes have small effects and individually they cannot be used to predict ADRD risk. However, their effects can be collectively measured by using polygenic scores (PGS). Furthermore, if PGS have high predictability, then genes included in calculating PGS are likely to be ADRD related; therefore, PGS can also be used to identify ADRD genes and understand the genetic mechanism of ADRD. In this study, we will systematically evaluate the performance of different PGS programs to identify the one with the best performance. We will also develop methods to adjust for the genotyping platforms/batches effects and genetic ancestries, Additionally, as many genetic variants are imputed, we will evaluate the effects of imputation quality on PGS predictability. Our goals are to derive PGS that can be used to identify high risk individuals for ADRD hence early intervention can be applied, and to understand the genetic mechanisms of ADRD thereby novel therapeutic strategies can be developed.