Convergent genome wide association results for bipolar disorder and substance dependence

Twin studies document substantial heritability for substance dependence and bipolar disorder [Shih et al. (2004); Uhl et al. (2008a)]. Individuals with bipolar disorder display substance use disorders at rates that are much higher than those in the general population [Krishnan (2005)]. We would thus predict: 1) substantial overlap between different genome wide association (GWA) studies of bipolar disorder 2) significant overlap between results from bipolar disorder and substance dependence. Recent GWA studies [Baum et al. (2007); Sklar et al. (2008); Uhl et al. (2008a); Wellcome Trust Consortium (2007)] allow us to test these ideas, although 1) these datasets display difficult features that include use of differing sets of SNPs, likely polygenic genetics, likely differences in linkage disequilibrium between samples, heterogeneity both between and within loci and 2) several, though not all, reports have failed to identify any allele of any single nucleotide polymorphism (SNP) ("same SNP same allele") that is reproducibly associated with bipolar disorder with "genome wide" significance. We now report analyses that identify clustered, P < 0.05 SNPs within genes that overlap between the bipolar samples (Monte Carlo P < 0.00001). Overlapping data from at least three of these studies identify 69 genes. 23 of these genes also contain overlapping clusters of nominally-positive SNPs for substance dependence. Variants in these "addiction/bipolar" genes are candidates to influence the brain in ways that manifest as enhanced vulnerabilites to both substance dependence and bipolar disorder.