Leveraging Polygenic Scores to Predict Diabetes Risk After Gestational Diabetes: Insights from a 30-Year Cohort Study

Highlight

This cohort study demonstrates that polygenic scores (PGSs) for type 1 diabetes (T1D) effectively predict future type 1 diabetes in women with gestational diabetes mellitus (GDM). Combining T1D-PGS with clinical risk factors during pregnancy improves predictive accuracy. Conversely, the type 2 diabetes polygenic score (T2D-PGS) poorly predicts type 2 diabetes despite high clinical relevance of traditional risk factors.

Study Background

Gestational diabetes mellitus (GDM) complicates approximately 5-15% of pregnancies worldwide and is recognized as a major risk factor for future development of both type 1 and type 2 diabetes in affected women. Postpartum surveillance and early intervention strategies can reduce morbidity, but individual risk stratification remains challenging. Existing clinical predictors include maternal age, pre-pregnancy body mass index (BMI), fasting plasma glucose during pregnancy, insulin treatment requirement, and family history of diabetes. However, genetic predisposition may provide additional insight. Polygenic scores, which aggregate the cumulative effect of numerous genetic variants, offer a promising tool for more accurate personalized risk prediction if validated in longitudinal cohorts.

Study Design

This observational cohort study included 370 women diagnosed with GDM between 1978 and 1996. Patient data were obtained from national registries incorporating ICD-10 coded diabetes diagnoses until 2019, supplemented by oral glucose tolerance tests conducted in 2000-2002 to identify undiagnosed cases. Polygenic scores were calculated for type 1 diabetes (T1D-PGS) and type 2 diabetes (T2D-PGS) based on validated genetic risk variants. Clinical risk factors recorded at the index pregnancy were analyzed. The primary endpoint was development of overt diabetes during a median 30-year follow-up. The predictive ability of PGSs, alone and combined with clinical factors, was assessed using area under the receiver operating characteristic curve (AUC).

Key Findings

During follow-up, 200 (54.1%) women developed diabetes: 30 (8.1%) with type 1 diabetes, 157 (42.4%) with type 2 diabetes, and 13 (3.5%) with monogenic diabetes variants. Mean T1D-PGS was significantly higher in women who developed type 1 diabetes versus those who did not (13.6 vs. 11.5, p < 0.001). In contrast, T2D-PGS values were similar between women with and without type 2 diabetes (52.9 vs. 52.8, p = 0.16), indicating poor discriminatory power.

The T1D-PGS alone had good predictive accuracy for type 1 diabetes with an AUC of 0.788. Clinical risk factors during pregnancy alone yielded an AUC of 0.839 for type 1 diabetes prediction. Combining T1D-PGS with clinical predictors further improved discrimination to an AUC of 0.887, demonstrating synergistic potential.

For type 2 diabetes, clinical risk factors during pregnancy alone predicted diabetes with reasonable accuracy (AUC 0.745), but the T2D-PGS had no meaningful discriminative ability (AUC 0.539). Adding T2D-PGS to clinical variables modestly increased the model’s AUC to 0.752, indicating minimal benefit from genetic information.

Expert Commentary

This study highlights the heterogeneous genetic architecture of diabetes subtypes following GDM, underscoring that T1D and T2D have distinct predictive profiles with polygenic scores. The strong association and improved AUC suggest T1D-PGS could enable earlier and more targeted postpartum monitoring in women at risk of autoimmune diabetes, where early insulin therapy may prevent acute complications. The limited utility of T2D-PGS aligns with previous research suggesting that environmental and lifestyle factors profoundly modulate type 2 diabetes risk, complicating genetic profiling. It is notable that 3.5% of cases carried monogenic diabetes variants, affirming the importance of comprehensive genetic evaluation in select cases.

Study limitations include the relatively modest sample size for rare type 1 diabetes outcomes and the historical nature of GDM diagnosis criteria, which may affect generalizability to contemporary populations. Future research should explore integrating PGSs with biomarkers and continuous glucose monitoring to refine risk stratification further.

Conclusion

In summary, polygenic risk scoring offers a promising adjunct to clinical factors for predicting type 1 diabetes risk in women with prior GDM. This enhanced predictive capacity may facilitate individualized follow-up and timely intervention, improving clinical outcomes. However, for type 2 diabetes, polygenic scores do not meaningfully augment risk prediction beyond established clinical markers, highlighting the need to focus on comprehensive lifestyle and metabolic risk assessments. As genetic tools evolve, integrating multi-omic and environmental data will be essential for precision prevention in this high-risk population.

Funding and Registration

This study was supported by the Danish Diabetes Academy and relevant national funding bodies. Registry data were used with appropriate ethical approvals. Clinical trial registration details were not specified.

References

1. Jørgensen IL et al. Predicting future risk of type 1 and type 2 diabetes after gestational diabetes mellitus using polygenic scores: a cohort study. Diabetologia. 2026 Aug 4; PMID: 42550202.
2. American Diabetes Association. Gestational Diabetes Mellitus. Diabetes Care 2023;46(Suppl 1):S105-S121.
3. Klimentidis YC, et al. The polygenic nature of type 2 diabetes: biology and clinical implications. Nat Rev Endocrinol. 2020;16(9): 538-546.
4. McCarthy MI. Genomics, type 2 diabetes, and obesity. N Engl J Med. 2010;363(24):2339-50.