Overall and individual associations between per- and polyfluoroalkyl substances and liver function indices and the metabolic mechanism

Per- and polyfluoroalkyl substances (PFAS) can disrupt liver homeostasis. Studies have shown that a single exposure to PFAS may provoke abnormal liver function; however, few studies have investigated the overall effect of PFAS mixtures. We aimed to investigate associations between exposure to PFAS mixtures and liver function indices and explore the relevant mechanisms. This study included 278 adult males from Guangzhou, China. Serum metabolite profiles were analyzed using untargeted metabolomics. We applied weighted quantile sum (WQS) regression as well as Bayesian kernel machine regression (BKMR) to analyze the association of nine PFAS mixtures with 14 liver function indices. PFAS mixtures were positively associated with apolipoprotein B (APOB) and gamma-glutamyltransferase (GGT) and negatively associated with direct bilirubin (DBIL) and total bilirubin (TBIL) in both the WQS and BKMR analyses. In addition, Spearman's correlation test showed individual PFAS correlated with APOB, GGT, TBIL, and DBIL, while there's little correlation between individual PFAS and other liver function indices. In linear regression analysis, PFHxS, PFOS, PFHpS, PFNA, PFDA, and PFUdA were associated with APOB; PFOA, PFDA, PFOS, PFNA, and PFUdA were associated with GGT. Subsequently, a metabolome-wide association study and mediation analysis were combined to explore metabolites that mediate these associations. The mechanisms linking PFAS to APOB and GGT are mainly related with amino acid and glycerophospholipid metabolism. High-dimensional mediation analysis showed that glycerophospholipids are the main markers of the association between PFAS and APOB, and that (R)-dihydromaleimide, Ile Leu, (R)-(+)-2-pyrrolidone-5-carboxylic acid, and L-glutamate are the main markers of the association between PFAS and GGT. In summary, overall associations between PFAS and specific indices of liver function were found using two statistical methods; the metabolic pathways and markers identified here may serve to prompt more detailed study in animal-based systems, as well as a similar detailed analysis in other populations.