Association of exposure to multiple perfluoroalkyl and polyfluoroalkyl substances and glucose metabolism in National Health and Nutrition Examination Survey 2017–2018

• DOI: 10.3389/fpubh.2024.1370971
• Publication Date: 2024-04-03T04:46:44Z
• Authors: Qinghua Tian; Yutong Yang; Qi An; Yang Li; Qingyao Wang; Ping Zhang; Yue Zhang; Yingying Zhang; Lina Mu; Lijian Lei
• Keywords: Bayesian kernel machine regression; glucose metabolism; perfluoroalkyl and polyfluoroalkyl substances; National Health and Nutrition Examination Survey; Public Health; least absolute shrinkage and selection operator; Public aspects of medicine; RA1-1270; 01 natural sciences; 0105 earth and related environmental sciences

ObjectiveTo investigate the relationships between perfluoroalkyl and polyfluoroalkyl substances (PFASs) exposure and glucose metabolism indices.MethodsData from the National Health and Nutrition Examination Survey (NHANES) 2017–2018 waves were used. A total of 611 participants with information on serum PFASs (perfluorononanoic acid (PFNA); perfluorooctanoic acid (PFOA); perfluoroundecanoic acid (PFUA); perfluorohexane sulfonic acid (PFHxS); perfluorooctane sulfonates acid (PFOS); perfluorodecanoic acid (PFDeA)), glucose metabolism indices (fasting plasma glucose (FPG), homeostasis model assessment for insulin resistance (HOMA-IR) and insulin) as well as selected covariates were included. We used cluster analysis to categorize the participants into three exposure subgroups and compared glucose metabolism index levels between the subgroups. Least absolute shrinkage and selection operator (LASSO), multiple linear regression analysis and Bayesian kernel machine regression (BKMR) were used to assess the effects of single and mixed PFASs exposures and glucose metabolism.ResultsThe cluster analysis results revealed overlapping exposure types among people with higher PFASs exposure. As the level of PFAS exposure increased, FPG level showed an upward linear trend (p < 0.001), whereas insulin levels demonstrated a downward linear trend (p = 0.012). LASSO and multiple linear regression analysis showed that PFNA and FPG had a positive relationship (>50 years-old group: β = 0.059, p < 0.001). PFOA, PFUA, and PFHxS (≤50 years-old group: insulin β = −0.194, p < 0.001, HOMA-IR β = −0.132, p = 0.020) showed negative correlation with HOMA-IR/insulin. PFNA (>50 years-old group: insulin β = 0.191, p = 0.018, HOMA-IR β = 0.220, p = 0.013) showed positive correlation with HOMA-IR/insulin, which was essentially the same as results that obtained for the univariate exposure-response map in the BKMR model. Association of exposure to PFASs on glucose metabolism indices showed positive interactions between PFOS and PFHxS and negative interactions between PFOA and PFNA/PFOS/PFHxS.ConclusionOur study provides evidence that positive and negative correlations between PFASs and FPG and HOMA-IR/insulin levels are observed, respectively. Combined effects and interactions between PFASs. Given the higher risk of glucose metabolism associated with elevated levels of PFAS, future studies are needed to explore the potential underlying mechanisms.