P2‐031: A VARIANT IN STK24 ACHIEVES GENOME‐WIDE SIGNIFICANCE IN AFRICAN AMERICANS USING A LIABILITY MODEL

Le résumé fourni par la source

Late onset Alzheimer's disease (LOAD) risk is influenced by multiple genetic, clinical and environmental factors. Although AD case-control studies frequently collect data on non-genetic risk factors, they usually are not included as covariates in genetic analyses because doing so reduces power to detect an association. However, a liability model can be used to extract risk information from known genetic and non-genetic factors without a reduction in power. A recent genome wide association study (GWAS) by Reitz et al. found that, in addition to the APOEε4 allele, a variant in the ABCA7 gene (rs115550680) was significantly associated with LOAD at the genome-wide level in African Americans.In the current study, we conducted a GWAS in African Americans, employing a liability model that included non-genetic and these genetic risk factors. Subjects included 247 well-characterized African American AD cases and 292 cognitively normal African American controls from the MIRAGE family-based study. First, we used logistic generalized estimating equations (GEE) to identify a set of genetic and non-genetic factors associated with AD status. Next, we derived a liability score (Pearson residual) from the significant and previously established predictors to serve as a quantitative continuous phenotype for GWAS. We evaluated association of the liability score with a genome-wide set of 6 million imputed markers using a linear GEE model that included principal components of population substructure. After variable selection, the liability model included: age, sex, APOEε4 status, rs115550680, history of depression, history of head injury, smoking history and education level. We obtained genome-wide significant evidence of association (p=2.57x10 -8) between the liability score and a SNP in STK24, a gene that encodes a serine/threonine protein kinase that has a role inapoptosis. In our previous GWAS of AD, we used a traditional case-control model to analyze a larger dataset of African Americans that included the MIRAGE sample. One of the most significant association findings was observed with a variant in STK24, but this result did not achieve genome-wide significance. The current study suggests a liability model can improve association signals in GWAS. We plan to apply this approach to a larger, combined African American data-set.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.