These results are unsurprising in the light of results for a range of complex genetic disorders published since this study was conceived in 2007.46 Results from GWAS of psychiatric disorders47 have, on the whole, yielded replicated associations only for low prevalence/high heritability disorders, that is, autism, bipolar disorder and schizophrenia, with tobacco dependence being the exception.48 As a high prevalence/low heritability disorder, the challenge for MDD was always going to be harder. From Yang et al.,49 we estimate that sample sizes 2.4-fold greater are needed for GWAS of MDD (prevalence 0.15) compared with schizophrenia (prevalence 0.007) to identify a variant that explains the same proportion of phenotypic variance to liability of these disorders (Supplementary File 1). This result reflects the smaller mean difference in phenotypic liability between cases and controls for MDD compared with schizophrenia. Hospital-based MDD cohorts may represent a more extreme phenotype, with lower prevalence and higher heritability.50 Using a prevalence for such clinical samples to be 0.06 (the average across sexes)51 still requires a sample size ∼1.8 times greater for a case–control study of MDD compared with one for schizophrenia.
