Deviation From Genetically Predicted BMI and All-Cause Mortality: A Cohort Study in the UK Biobank.

Abstract

The relation between genetically predicted BMI (gBMI) and actual BMI may have health effects. This study examines the relationship between deviations from gBMI and all-cause mortality in 208,146 UK Biobank participants. We derived gBMI from polygenic risk scores, with deviations calculated as the difference between observed and predicted BMI. Cox proportional hazards models are adjusted for confounders and current BMI. Downward deviations (> 2 SD below gBMI) were associated with significantly increased mortality (HR: 1.25, 95% CI: 1.01-1.55), whereas upward deviations (> 2 SD above) showed no significant effect (HR: 1.10, 95% CI: 0.93-1.29). The mortality exhibited the known nonlinear J-shaped association with observed BMI, here lowest at BMI ~22 kg/m², but this nadir varied by genetic predisposition; thus, for individuals with high gBMI, lowest mortality occurred at higher observed BMI (24-26 kg/m²), while those with low or medium gBMI showed sharper increases in mortality at higher BMI. These findings highlight the possible importance of aligning current BMI to genetic predisposition, and future research should examine BMI deviations and their long-term health effects. This perspective may inform personalized obesity management strategies to optimize health outcomes.