samples increased when more extreme cutoffs were applied; at the top 2% of the PRS distribution, an approximately three-fold increase in T2D risk was observed relative to the remaining 98% of the individuals (REGARDS: OR = 3.04, P value = 3.87E−10; GenHAT: OR = 2.70, P value = 8.44E−06; HyperGen: OR = 3.37, P value = 5.37E−04). The prevalence-adjusted PPVs ranged from 0.26 to 0.34 and all prevalence-adjusted NPVs were around 0.88. WPC had the lowest sample size among the four cohorts and had limited power to assess the tail discrimination of the trans-ancestry PRS, but the OR estimate at the top 2% of the PRS was comparable to the other three cohorts, though not statistically significant (OR = 2.7, 95% CI: 0.80–9.09, P value = 1.09E−01; Table 3; Additional File 2: Table S4).Table 3Prediction accuracy of the trans-ancestry T2D PRS in four Black cohortsCohortLiabilityR2Covariates-adjusted AUCOR per SD(95% CI)Top 2% PRSOR(95% CI)P valueSensitivitySpecificityAdjusted PPV*Adjusted NPV*REGARDS4.6%0.611.70(1.58, 1.84)3.04(2.15, 4.31)3.87E−100.040.990.300.88GenHAT3.6%0.611.85(1.70, 2.01)2.70(1.74, 4.18)8.44E−060.030.990.260.88HyperGen6.2%0.621.75(1.52, 2.02)3.37(1.69, 6.69)5.37E−040.050.990.340.88Warfarin1.5%0.571.37(1.13, 1.65)2.70(0.80, 9.09)1.09E−010.010.980.070.87*Adjusted PPV and NPV are calculated using 12.5% for the African population
