A GWAS of 1,020 recurrent early-onset MDD cases and 1,636 screened controls did not detect genome-wide significant evidence of association. This is consistent with other GWAS results for common, genetically complex diseases38, 39: the genotypic relative risks (GRR) of significant findings have typically been in the range of 1.1–1.2, often requiring samples 10,000–20,000 cases obtained by combining multiple samples (in each of which the evidence for association can be quite modest). The odds ratios listed in Table 2 are much higher. They could represent a combination of false positive results and of true associations whose GRRs have been over-estimated by selecting the best results in one study (the “winner’s curse” effect), and particularly with an underpowered sample. A companion paper17 provides details of meta-analyses combining the GenRED, STAR*D and GAIN-MDD samples. We are also participating in an effort to carry out larger meta-analyses of MDD GWAS data through the Psychiatric GWAS Consortium.39, 40 But because findings with the strongest statistical evidence for association in each study are most likely to be true positive results, we briefly review here several interesting findings.
