Protein-Predicted Obesity Phenotypes and Cardiovascular Events: A Secondary Analysis of UK Biobank Proteomics Data

Proteomic profiling may improve the understanding of obesity and cardiovascular risk prediction. This study explores the use of protein-predicted scores for body mass index (PPS_BMI), body fat percentage (PPS_BFP), and waist-hip ratio (PPS_WHR) to estimate risk for major adverse cardiovascular events (MACEs).

We used data from the UK Biobank with proteome profiling. PPS_BMI, PPS_BFP, and PPS_WHR were derived using the LASSO algorithm. The association between these protein scores and incident MACEs was evaluated using a competing risk model.

Strong to moderate correlations were observed between protein-predicted obesity phenotypes and their measured counterparts (R²: BMI = 0.78, BFP = 0.85, WHR = 0.63). Each standard deviation increment of PPS_BFP and PPS_WHR, but not PPS_BMI, was associated with greater risk of MACEs (hazard ratio [HR] 1.25, 95% CI 1.14-1.38, p < 0.0001; HR 1.15, 95% CI 1.06-1.24, p = 0.001, respectively). For predicting MACEs, compared with the PREVENT equation (C statistic 0.694), the models adjusted for only age, sex, current smoking, and protein scores showed comparable performance (C statistics 0.684-0.688).

Protein-predicted scores of obesity showed strong independent associations and predictive performance for MACEs, suggesting they may capture additional biological risk beyond anthropometry. These scores may complement existing risk models by providing a biologically informed approach to assessing obesity-related cardiovascular risk and improving risk stratification.