Predictive capabilities of polygenic scores in an East-Asian population-based cohort: the Singapore Chinese health...

Polygenic scores (PGS) are derived primarily from European population studies. It is unclear how these perform in risk predictions among East-Asians. We generated 2173 PGSs from 519 traits and assessed their associations with 58 baseline phenotypes in the Singapore Chinese Health Study, a prospective cohort of 23,622 Chinese adults residing in Singapore. PGS performances were evaluated through explained variance (r²) and area under the receiver operating characteristic curve (AUC) in linear and logistic regression models, respectively. Traits with higher heritability exhibited stronger PGS associations, while behavioural traits, like sleep duration, showed weaker associations. Height and type 2 diabetes (T2D) exhibited largest SNP-based heritability with the largest increments in explained variance and AUC. We explored the effect of T2D risk factors on the association between the T2D PGS (PGS003444) and incident T2D. PGS associations were significantly mediated and modified by hypertension (P_indirect = 1.56 × 10^-18, P_interaction = 2.10 × 10^-3) and BMI (P_indirect = 1.25 × 10^-36, P_interaction = 1.11 × 10^-6). Prediction ability of PGS003444 for incident T2D was stronger among non-overweight individuals without hypertension (AUC = 0.774) than in overweight individuals with hypertension (AUC = 0.709). Our study demonstrates the divergent ability of PGSs in predictions of complex traits. For certain traits, such as T2D, PGSs may have the potential for improving risk prediction and personalized healthcare.