Background Previous research indicated that isomers and alternatives of per- and polyfluoroalkyl substances (PFAS) probably disturb glucose metabolism; however, current epidemiological evidence on the associations of PFAS with fasting blood glucose is inconsistent. Besides, studies on the joint association of multiple components of PFAS and fasting blood glucose as well as the key component are scarce.

Objective To evaluate the associations of PFAS isomers and alternatives with fasting blood glucose and their joint effects, as well as identify the key component among population without glucose metabolism problems.

Methods We selected 923 adults without glucose metabolism problems or missing data from the Isomers of C8 Health Project in China (2015—2016). Serum PFAS isomers and alternatives and fasting blood glucose were measured using ultra-performance liquid chromatography coupled with tandem mass spectrometry (UPLC-MS/MS) and automatic biochemical analyzer. We applied multiple linear regression to explore the associations of 16 pollutants which were detected among over 80% participants with fasting blood glucose. Meanwhile, we utilized qgcomp and Bayesian kernel machine regression (BKMR) models to explore the joint effects of PFAS isomers and alternatives mixture on target outcome indicators and identify the key component.

Results The average age among the 923 participants in this study was (62.4±13.8) years old, including 472 men (51.1%) and 451 women (48.9%). Among selected PFAS isomers and alternatives, the highest serum concentration was ∑3+4+5m-PFOS (perfluoro-3/4/5-methylheptanesulfonate) with a median concentration of 10.20 ng·mL⁻¹. The concentrations of linear perfluorooctane sulfonate (n-PFOS, 9.61 ng·mL⁻¹), perfluorooctanoic acid (PFOA, 4.55 ng·mL⁻¹), linear perfluorohexane sulfonic acid (n-PFHxS, 2.48 ng·mL⁻¹), 6:2 chlorinated polyfluorinated ethersulfonic acid (6:2 Cl-PFESA, 1.90 ng·mL⁻¹), perfluoro-6-methylheptanesulfonate (iso-PFOS, 1.85 ng·mL⁻¹), perfluorobutanoic acid (PFBA, 1.81 ng·mL⁻¹), perfluorinated n-nonanoic acid (PFNA, 1.39 ng·mL⁻¹), and perfluoro-1-methylheptanesulfonate (1m-PFOS, 1.27 ng·mL⁻¹) were higher than 1.00 ng·mL⁻¹. After being adjusted for selected confounders, PFAS isomers and alternatives were positively associated with fasting blood glucose. With 1 ln unit concentration increment of 6:2 Cl-PFESA and PFNA, the estimated changes of fasting blood glucose were 0.18 (95%CI: 0.13, 0.23) mmol·L⁻¹ and 0.24 (95%CI: 0.18, 0.30) mmol·L⁻¹, respectively. The multi-pollutant models indicated a joint association of PFAS isomers and alternatives mixture with fasting blood glucose. The BKMR models reveals that as the quantiles of mixture elevated from the 50^th to the 75^th percentile, the values of fasting blood glucose increased 0.25 (95%CI: 0.21, 0.30) mmol·L⁻¹, and the posterior inclusion probability of PFNA was 0.92, implying that PFNA was the key component.

Conclusion PFAS isomers and alternatives are positively associated with fasting blood glucose. PFNA is the key component of the joint association.