Genomic convergence in Alzheimer disease. M. Slifer¹, E. Martin¹, H. Munger¹, J. Haines², J. Gilbert¹, M. Pericak-Vance¹. 1) Ctr Human Genetics, Duke Univ Medical Ctr, Durham, NC; 2) Ctr Human Genetics Research, Vanderbilt Univ Medical Ctr, Nashville, TN.
   Late-onset Alzheimer disease (LOAD) is the most common cognitive disorder affecting more than four million Americans. Over the past two decades, hundreds of variations in candidate genes have been reported as associated with LOAD. However, only the apolipoproteinE4 allele has been widely accepted as a risk factor. The deluge of reported susceptibility alleles, limited resources, lack of follow-up, small sample sizes, varying datasets, and differing experimental designs have complicated the search for confirmed LOAD genetic variants. In this study, we use the genomic convergence approach to select candidate genes for systematic follow-up. A gene is declared convergent if significant effects are published from at least three independent experimental methods (e.g. genetic linkage, genetic association, and gene expression). Nineteen candidates (comprising 1.5 Mb of genomic DNA) meet full convergent criteria and have been interrogated in a large independent case-control dataset (n=1000). All cases meet NINCDS/ADRDA criteria for Alzheimer disease as agreed upon by consensus conferences. Controls have no history of cognitive impairment, no family history of LOAD in a first degree relative or and no evidence of dementia on psychometric testing. Using dense SNP genotyping (approximately 1 SNP per 2.5 kb) we examined candidate variants for association with LOAD. Nine of the nineteen convergent genes have SNPs significantly associated with LOAD (p<0.05) including multiple SNPs within 3 regions of the amyloid beta (A4) precursor protein gene (p=0.0002), and a region proximal to the 3 end of the major histocompatibility complex, class 1, A gene (p=0.0001).