Additive Value of Polygenic Risk Score to Family History for Type 2 Diabetes Prediction: Results From the All of Us Research Database

Abstract

OBJECTIVE The goal of this study was to assess the additive value of considering type 2 diabetes (T2D) polygenic risk score (PRS) in addition to family history for T2D prediction. RESEARCH DESIGN AND METHODS Data were obtained from the All of Us (AoU) research database. First-degree T2D family history was self-reported on the personal family history health questionnaire. A PRS was constructed from 1,289 variants identified from a large multiancestry genome-wide association study meta-analysis for T2D. Logistic regression models were run to generate odds ratios (ORs) and 95% CIs for T2D. All models were adjusted for age, sex, and BMI. RESULTS A total of 109,958 AoU research participants were included in the analysis. The odds of T2D increased with 1 SD PRS (OR 1.75; 95% CI 1.71–1.79) and positive T2D family history (OR 2.32; 95% CI 2.20–2.43). In the joint model, both 1 SD PRS (OR 1.69; 95% CI 1.65–1.72) and family history (OR 2.06; 95% CI 1.98–2.15) were significantly associated with T2D, although the ORs were slightly attenuated. Predictive models that included both the PRS and family history (area under the curve [AUC] 0.794) performed better than models including only family history (AUC 0.763) or the PRS (AUC 0.785). CONCLUSIONS In predicting T2D, inclusion of a T2D PRS in addition to family history of T2D (first-degree relatives) added statistical value. Further study is needed to determine whether consideration of both family history and a PRS would be useful for clinical T2D prediction.