Around 30-50% of women with history of gestational diabetes (GDM) develop type 2 diabetes (T2D) within years after parturition. However, the role of polygenetic score on prediction of T2D in these women is still unclear. We aimed to investigate the utility of global extended polygenic scores (gePS) in predicting incident T2D using GDM cohorts from 1) UK Biobank (N=941) , 2) Seoul National University Hospital (SNUH, N=357) , and 3) Korean Genome and Epidemiology Study (KoGES, N=379) . We calculated gePS for T2D using Bayesian regression and continuous shrinkage priors, with summary statistics from representative genome-wide association studies in Europeans and East Asians. We performed time-to-event analysis for incident T2D with gePS and assessed improvement of C-statistic when gePS was added to clinical risk factors. In addition, risk of T2D was estimated for those who were in the top percentile of gePS compared to the remainders. The incidence of T2D was 19.8% in UK Biobank, 28.3% in SNUH, and 23.5% in KoGES cohort with median follow-up duration of 28, 4, and 28 years, respectively. The normalized mean PRS for those who developed T2D was higher in each cohort: 0.25±1.vs. -0.06±0.99 in UK Biobank, 0.22±1.vs. -0.09±0.97 in SNUH, and 0.52±0.96 vs. -0.16±0.96 in KoGES (all P<0.01) . When meta-analyzed, adding gePS to clinical risk factors improved the C statistic from 0.715 (95% confidence interval [CI] 0.658-0.766) to 0.743 (95% CI 0.698-0.784) . The standardized gePS was associated with incident T2D in women with history of GDM when the results of the three cohorts were meta-analyzed: hazard ratio 1.68 (95% CI 1.52-1.86, P<0.001) . Furthermore, those who were in the top percentile had 3.52-fold increased risk of T2D: (95% CI 2.68-4.62, P<0.001) . To sum up, genetic information in women with previous GDM pregnancy could be used to identify those who are at high risk of future development of T2D and provide the opportunity for tailored prevention.


J.Choi: None. H.Lee: None. M.Han: None. D.Choi: None. H.Jang: None. K.Park: None. S.Kwak: None.

Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at