Congener-Specific Associations of Per- and Polyfluoroalkyl Substances With Obesity-Metabolic Phenotypes: A Multimodel Analysis.

Abstract

This study aimed to investigate the associations between per- and polyfluoroalkyl substances (PFAS) and distinct obesity-metabolic phenotypes and explore underlying mechanisms. A total of 8692 adults from NHANES 2003-2018 were classified into four phenotypes as follows: metabolically healthy obesity (MHO), metabolically unhealthy obesity (MUO), metabolically healthy non-obesity (MHNO), and metabolically unhealthy non-obesity (MUNO). Multiple statistical models were used to evaluate the associations between PFAS exposure and distinct phenotypes. Network toxicology and molecular docking were employed for mechanistic exploration. Logistic regression analyses showed congener-specific associations of PFAS with MHO, MUO, and MUNO. PFUA was inversely associated with all three phenotypes (OR = 0.60, 0.55, and 0.79, respectively; p < 0.05). Weighted quantile sum (WQS), quantile g-computation (Qgcomp), and Bayesian kernel machine regression (BKMR) analyses consistently indicated the negative associations of PFAS mixture exposure with MHO (OR_WQS = 0.67, log[OR]_Qgcomp = -0.48; p < 0.001) and MUO (OR_WQS = 0.81, log[OR]_Qgcomp = -0.24; p < 0.001). Stratified BKMR suggested the inverse PFAS-BMI relationship was independent of metabolic status. The peroxisome proliferator-activated receptor signaling pathway was predicted to link PFAS and obesity-metabolic phenotypes. Although PFAS mixture exposure may not increase the risks of obesity and metabolic abnormalities, congener-specific effects on distinct phenotypes indicate the necessity of analyzing specific compositional profiles of PFAS exposure when evaluating endocrine toxicity.