P3-233: Correlation between lipid profile and APOE genotype in cognition in an elderly Korean population

Abstract

statistical power is degraded by the inherent heterogeneity of samples ascertained using clinical exams. The problem of heterogeneity may be alleviated by the use of quantitative endophenotypes. Cerebrospinal fluid (CSF) amyloid(A ) levels have recently emerged as promising endophenotypes for LOAD. Objective: To leverage the strengths of an endophenotype based approach to test the hypothesis that genetic variation in candidate genes for LOAD and A metabolism are associated with CSF A levels. Methods: Our CSF series consists of 313 individuals, with CSF collected at 8am after overnight fasting. CSF levels of 40 amino acid A (A 40) and 42 amino acid A (A 42) were measured using ELISA. We selected 39 candidate genes for LOAD and A metabolism based on published data. For each gene region we selected tagSNPs based on LD bins (r 0.8), putative functional polymorphisms, and SNPs from pub lished association studies (where applicable). We genotyped a total of 1553 SNPs in 39 candidate genes. Genotypes were tested for association with CSF total A levels and A 42/A 40 ratio using an analysis of covariance after adjusting for associated covariates. Multiple test correction was performed using the variance of the Eigenvalues to determine the effective number of tests. Corrections were calculated for each gene separately. Results: We identified SNPs in SOAT1 (rs4652362 p 0.0024; rs10753191 p 0.0014) and TFAM (rs10826177) which show significant association with total A levels. Rs12964454 in CNDP1 shows significant association with CSF A 42/A 40 ratio (p 0.0003). Conclusions: These results high light the strength of this endophenotype-based approach and suggest that genetic variation in CNDP1, SOAT1 and TFAM may influence risk for LOAD by modulating CSF A levels. Support: AG16208, AG03991, AG05681, AG026276, Ford Foundation.