In the discovery stage, we performed a GWAMA in 70,017 participants of European ancestry by combining the PGC MDD (7) and CHARGE GWAMA (14). We applied an LD score regression to the summary statistics from each study to compute the SNP-based heritabilities and the genetic correlation. As reported previously (35), the SNP-based liability scale heritability of MDD was 0.2 (standard error 0.02) for 20% of prevalence. The lambda was 1.1 and the regression intercept was 1.0 (standard error 0.01). The SNP-based heritability of depressive symptoms was 0.04 (standard error 0.01). The lambda was 1.1 and the regression intercept was 1.0 (standard error 0.01). The SNP-based heritability of the broad depression phenotype was 0.3 (standard error 0.04). MDD and depressive symptoms showed significant co-heritability (1.001, standard error 0.2, Z-score= 4.6, P=4.6×10−6). This result supports the contention of a continuum between depressive symptoms and MDD. However, the genetic correlation should be interpreted carefully as LD regression is quite sensitive to environmental confounding and like twin studies often lacks precision. Also, different evaluation methods of the depression phenotypes might cause different genetic correlation estimates that cannot easily be compared.
