Association study of cerebrospinal fluid amyloid levels with late-onset Alzheimers disease associated polymorphisms. J.S.K. Kauwe¹, A.M. Fagan², M.L. Spinner², A.R. Shah², K. Mayo¹, D.M. Holtzman^2,3, A. Grupe⁴, A. Goate^1,2,3. 1) Dept Psychiatry, Washington Univ, St Louis, MO; 2) Dept Neurology, Washington Univ, St Louis, MO; 3) Alzheimers Disease Research Center, St. Louis, MO; 4) Celera Diagnostics, Alameda, CA.
   Amyloid (A) peptides are normal products of -amyloid precursor protein processing and can be detected in cerebrospinal fluid (CSF). It has been suggested that the aggregation of amyloid peptide into insoluble plaques in the brain is a central feature of Alzheimers disease (AD) pathology. Reduced levels of CSF A42 in AD CSF are one of the most consistent findings in the AD biomarkers field. In addition, low levels of CSF A42 have recently been shown to correlate with the presence of brain amyloid. We hypothesize that the use of CSF A levels as an endophenotype will lead to the identification of novel genetic risk factors for AD. To test this hypothesis we have genotyped 12 SNPs from strong candidate genes in 120 individuals for whom we have CSF A40 and A42 measurements. We genotyped 12 SNPs in these samples. Included in this group of SNPs are 10 SNPs, which are significantly associated with AD in the meta-analysis of published data provided at Alzgene.org and 2 SNPs that have shown association with late onset AD in multiple datasets from our own studies. CSF A levels were adjusted for age, disease severity (based on CDR), and presence or absence of the APOE4 allele by regression. Due to the large effect of gender on the levels of both A40 and A42, the genetic analyses were stratified by gender (Males, N=50; Females, N=70). Genotypes at these SNPs were tested for association with levels of CSF A40, A42, as well as the ratio of A42/A40 using ANOVA. In males an uncorrected p-value of 0.004 was observed between rs4877365 (DAPK1) and the ratio of A42/A40. In females rs440446 (APOE) showed significant association with A40 levels (uncorrected p-value = 0.008). These findings suggest that these polymorphisms may affect risk for AD through an A related mechanism. We are currently genotyping these SNPs in additional samples to increase our power to detect association with CSF A levels.