GWAS of five gynecologic diseases and cross-trait analysis in Japanese

We performed genome-wide association studies of five gynecologic diseases using data of 46,837 subjects (5236 uterine fibroid, 645 endometriosis, 647 ovarian cancer (OC), 909 uterine endometrial cancer (UEC), and 538 uterine cervical cancer (UCC) cases allowing overlaps, and 39,556 shared female controls) from Biobank Japan Project. We used the population-specific imputation reference panel (n = 3541), yielding 7,645,193 imputed variants. Analyses performed under logistic model, linear mixed model, and model incorporating correlations identified nine significant associations with three gynecologic diseases including four novel findings (rs79219469:C > T, LINC02183, P = 3.3 × 10^-8 and rs567534295:C > T, BRCA1, P = 3.1 × 10^-8 with OC, rs150806792:C > T, INS-IGF2, P = 4.9 × 10^-8 and rs140991990:A > G, SOX9, P = 3.3 × 10^-8 with UCC). Random-effect meta-analysis of the five GWASs correcting for the overlapping subjects suggested one novel shared risk locus (rs937380553:A > G, LOC730100, P = 2.0 × 10^-8). Reverse regression analysis identified three additional novel associations (rs73494486:C > T, GABBR2, P = 4.8 × 10^-8, rs145152209:A > G, SH3GL3/BNC1, P = 3.3 × 10^-8, and rs147427629:G > A, LOC107985484, P = 3.8 × 10^-8). Estimated heritability ranged from 0.026 for OC to 0.220 for endometriosis. Genetic correlations were relatively strong between OC and UEC, endometriosis and OC, and uterine fibroid and OC (r_g > 0.79) compared with relatively weak correlations between UCC and the other four (r_g = -0.08 ~ 0.25). We successfully identified genetic associations with gynecologic diseases in the Japanese population. Shared genetic effects among multiple related diseases may help understanding the pathophysiology.