Genome-Wide Polygenic Risk Score Predicts Incident Type 2 Diabetes in Women With History of Gestational Diabetes

Abstract

Objective: Women with a history of gestational diabetes mellitus (GDM) are at increased risk of developing type 2 diabetes (T2D). It remains unclear whether genetic information improves prediction of incident T2D in these women.

Research design and methods: Using five independent cohorts representing four different ancestries (n = 1,895), we investigated whether a genome-wide T2D polygenic risk score (PRS) is associated with increased risk of incident T2D. We also calculated the area under the receiver operating characteristics curve (AUROC) and continuous net reclassification improvement (NRI) following the incorporation of T2D PRS into clinical risk models to assess the diagnostic utility.

Results: Among 1,895 women with previous history of GDM, 363 (19.2%) developed T2D in a range of 2 to 30 years. T2D PRS was higher in those who developed T2D (-0.08 vs. 0.31, P = 2.3 × 10-11) and was associated with an increased risk of incident T2D (odds ratio 1.52 per 1-SD increase, 95% CI 1.05-2.21, P = 0.03). In a model that includes age, family history of diabetes, systolic blood pressure, and BMI, the incorporation of PRS led to an increase in AUROC for T2D from 0.71 to 0.74 and an intermediate improvement of NRI (0.32, 95% CI 0.15-0.49, P = 3.0 × 10-4). Although there was variation, a similar trend was observed across study cohorts.

Conclusions: In cohorts of GDM women with diverse ancestry, T2D PRS was significantly associated with future development of T2D. A significant but small improvement was observed in AUROC when T2D PRS was integrated into clinical risk models to predict incident T2D.

Graphical abstract

Figure 1

Risk of incident T2D according to genetic risk. ORs of T2D PRS in each group are shown. ORs were calculated for a 1-SD increment of T2D PRS using a logistic regression model. The model was adjusted for the statistically significant first 10 PCs that reflected ancestry. A: We compared participants with incident T2D to participants without incident T2D. B: We compared participants whose T2D PRS ranked in the top 10% to all other remaining participants.

Figure 2

Risk of incident T2D according to genetic risk and follow-up time. HRs of T2D PRS in each group are shown. HRs were calculated for a 1-SD increment of T2D PRS using Cox proportional hazards regression model regarding follow-up time. The model was adjusted for the statistically significant first 10 PCs that reflected ancestry. A: We compared participants with incident T2D to participants without incident T2D. B: We compared participants whose T2D PRS ranked in the top 10% to all other remaining participants.