Genetic risk score for Alzheimer's disease in the Amish highlights differences in the genetic architecture compared to other European ancestry populations.

Alzheimer's disease is the most prevalent type dementia and has a strong genetic component. Much of AD genomic research has focused on identifying risk variants, primarily in European ancestry populations. After immigration to the United States, the Amish experienced a genetic bottleneck, making it likely that their underlying genetic architecture is different from the broader European ancestry population. Here, we compare the genetic risk for Alzheimer's disease in an Amish and general European ancestry population using genetic risk scores (GRSs).1,981 adults were recruited from Amish communities in Indiana and Ohio. A non-Amish population of 2,470 adults were recruited from clinics in Tennessee, North Carolina, and Florida as a comparison group. Subjects were screened for cognitive impairment and further evaluated for AD and dementia. Genotype data were collected using Illumina genotyping chips. Imputation was performed based on a 1,000 Genomes reference panel. GRSs were generated using genome-wide significant variants from Kunkle et al., (2019), weighted by their log-odds ratios. We compared the GRSs by source population and case status. Further, we evaluated their predictive ability on AD status with and without sex, age, and APOE genotype covariates.The results indicated that there exists less variation in GRSs among the Amish population. This is evident in how the four highest GRSs belonged to non-Amish individuals. The median GRS of Amish cases was also moderately lower overall among Amish cases (0.0163) compared to non-Amish cases (0.0171). In the Amish population, area under curve (AUC) of the GRS model was strengthened considerably from 0.541 to 0.872 by inclusion of the sex and age covariates and slightly (AUC=0.883) by further inclusion of APOE genotype. In the non-Amish group, inclusion of sex and age improved AUC from 0.583 to 0.697 and to 0.817 by subsequent inclusion of APOE genotype.These results suggest that the Amish have a different genetic architecture than a general European-ancestry population, manifesting itself in less overall variation but also a lower relative importance of APOE. Since the prevalence of dementia in the Amish is similar to the general European population, this suggests the Amish may harbor unique AD genetic risk variants.