Little is known about the population-based mismatch between phenotypic and genetic BMI (BMI-PGM) and its association with type 2 diabetes. We therefore used data from the China Kadoorie Biobank and UK Biobank and calculated BMI-PGM for each participant as the difference between the percentile for adjusted BMI at baseline and the percentile for adjusted polygenic risk score for BMI. Participants were categorized into discordantly low (BMI-PGM< the 1st quartile), concordant (the 1st quartile ≤BMI-PGM<the 3rd quartile), and discordantly high (BMI-PGM≥the 3rd quartile) groups. We calculated adjusted hazard ratios (aHRs) for the association of BMI-PGM and type 2 diabetes using Cox proportional hazard models in each cohort, and combined HRs using random-effects meta-analyses. During a median follow-up of 12 years for both cohorts, BMI-PGM was associated with the risk of type 2 diabetes, with the discordantly low group showing reduced risk and the discordantly high group showing elevated risk compared to the concordant group, independent of BMI and other conventional risk factors. In addition, normal-weight individuals with discordantly high BMIPGM faced a higher risk of type 2 diabetes than overweight individuals. These findings suggest that BMI-PGM may play a potential role in reassessing the risk of type 2 diabetes, particularly among normal-weight populations.
© 2024 by the American Diabetes Association.