To explore genetic contributions of Alzheimer's disease (AD) at the level of biological terms and pathways, we analyzed three Caucasian population-based genome-wide association study datasets (TGEN_ND, GeneADA and NIA_LOAD) using the Database for Annotation, Visualization and Integrated Discovery (DAVID). This analysis identified 4 annotation terms ("Fibronectin type III-like fold," "Cell adhesion," "Cell motion" and "Ig-like-C2-type 3") and 17 genes that associated with AD susceptibility in two or more of the GWAS datasets. Ten of these genes, have previously been identified as candidate AD liability genes in genetic association studies (ACT, COL11A1) or encode proteins that function in biological systems or pathways previously implicated in AD (BARHL2, CSF3R, DAB1, HMCN1, LEPR, PTPRF, PXDN, TNR). Among these, DAB1 (Dab, reelin signal transducer, homolog 1) was of particular interest, since it encodes a protein that functions downstream from reelin, a signaling pathway previously identified as protective in AD. Multiple linear regression analysis of correlations between brain DAB1 mRNA expression and SNP genotype using data from the "BrainCloud" database identified five SNPs within the DAB1 locus that correlated with mRNA expression in human dorsolateral prefrontal cortex. Analysis of predicted levels of DAB1 mRNA expression based on genotype combinations present in AD cases and controls vs. the log(10)-transformed odds ratios for AD diagnosis, revealed statistically significant correlations in one of the GWAS datasets (GenADA), with high DAB1 mRNA expression correlating with AD protection. Multidimensional scaling (MDS) analysis of cases and controls in the three GWAS, revealed genetic differences between GenADA and TGEN_ND/NIA_LOAD, which were similar to each other. To our knowledge, this study is the first to provide genetic evidence for DAB1 as a candidate AD liability/protection gene, although the strength of the contribution of DAB1 may differ among populations. (C) 2015 Elsevier Inc. All rights reserved.