1477-P: Improved Diabetes Risk Prediction Using Ancestry-Specific Polygenic Score—Results from the Asian Indian Diabetic Heart Study/Sikh Diabetes Study

Background: Genome-wide polygenic risk scores (PRS) have shown high specificity and sensitivity in predicting type 2 diabetes (T2D) risk in Europeans. However, PRS-driven information on other ethnic groups is sparse, including Asian Indians (AI), who have a heightened risk for T2D. We examined the predictive efficacy of ancestry-derived (PRS_AI) and European-derived (PRS_EU) with a clinical risk score (CRS) of T2D.

Methods: Weighted PRSs were computed using 2921 variants (common and rare) from (Punjabi/Sikh) genome-wide association studies and 432 variants from European meta-analyses studies in 4,602 individuals (2,574 cases and 2,028 controls) of the Asian Indian Diabetes Heart Study/Sikh Diabetes Study. Both Ancestry-specific PRS_AI and European PRS_EU were validated on 9372 UK Biobank South Asian individuals (1943 cases/7429 controls).

Results: Ancestry-specific PRS_AI showed 36.2% improved efficacy than PRS_EU in Punjabi Asian Indians_. Similarly, Ancestry-specific PRS_AI was 18.6% superior to PRS_EU in Asian Indians from UKBB based on Nagelkerke’s R². Sensitivity analysis explained enhanced sensitivity (area under the curve-AUC) of ancestry-specific PRS over PRS_EU in predicting T2D risk.

Conclusions: Our data suggest expanding genetic studies in diverse ethnic groups to exploit the full potential of PRS to improve health outcomes. More genetic evaluations, specifically in understudied/underrepresented populations, would help increase the transferability of genome-wide polygenic scores across racial and ethnic groups to make their integration into clinical practice easier.