Plasma metabolites associated with exposure to perfluoroalkyl substances and risk of type 2 diabetes – A nested case-control study

Tessa Schillemans,Lin Shi,Carolina Donat-Vargas,Kati Hanhineva,Andreas Tornevi,Ingegerd Johansson,Jani Koponen,Hannu Kiviranta,Olov Rolandsson,Ingvar A Bergdahl,Rikard Landberg,Agneta Åkesson,Carl Brunius

doi:10.1016/j.envint.2020.106180

Abstract

Perfluoroalkyl substances (PFAS) are widespread persistent environmental pollutants. There is evidence that PFAS induce metabolic perturbations in humans, but underlying mechanisms are still unknown. In this exploratory study, we investigated PFAS-related plasma metabolites for their associations with type 2 diabetes (T2D) to gain potential mechanistic insight in these perturbations.We used untargeted LC-MS metabolomics to find metabolites related to PFAS exposures in a case-control study on T2D (n = 187 matched pairs) nested within the Västerbotten Intervention Programme cohort. Following principal component analysis (PCA), six PFAS measured in plasma appeared in two groups: 1) perfluorononanoic acid, perfluorodecanoic acid and perfluoroundecanoic acid and 2) perfluorohexane sulfonic acid, perfluorooctane sulfonic acid and perfluorooctanoic acid. Using a random forest algorithm, we discovered metabolite features associated with individual PFAS and PFAS exposure groups which were subsequently investigated for associations with risk of T2D.PFAS levels correlated with 171 metabolite features (0.16 ≤ |r| ≤ 0.37, false discovery rate (FDR) adjusted p < 0.05). Out of these, 35 associated with T2D (p < 0.05), with 7 remaining after multiple testing adjustment (FDR < 0.05). PCA of the 35 PFAS- and T2D-related metabolite features revealed two patterns, dominated by glycerophospholipids and diacylglycerols, with opposite T2D associations. The glycerophospholipids correlated positively with PFAS and associated inversely with risk for T2D (Odds Ratio (OR) per 1 standard deviation (1-SD) increase in metabolite PCA pattern score = 0.2; 95% Confidence Interval (CI) = 0.1–0.4). The diacylglycerols also correlated positively with PFAS, but they associated with increased risk for T2D (OR per 1-SD = 1.9; 95% CI = 1.3–2.7). These results suggest that PFAS associate with two groups of lipid species with opposite relations to T2D risk.

Highlights

Perfluoroalkyl substances (PFAS) comprise a large group of chem icals used in a variety of technological applications like textile or leather treatment, surfactants and food packaging to make them e.g. non-stick and stain repellent
Results did not differ greatly between males and females, males had a slightly higher Odds ratio (OR) of type 2 diabetes (T2D) for PC1, whereas females had a higher OR for PC2 (Supplementary Fig. 3). This population-based prospective nested case-control study is the first to use metabolomics data to examine the relationship between PFAS and risk of T2D, providing valuable information about potential un derlying mechanisms
Modelling of PFAS exposures by the metabolome in the random forest algorithm resulted in strikingly strong models and 171 metabolite features associated with PFAS, from which 35 associated with T2D risk

Summary

Introduction

Perfluoroalkyl substances (PFAS) comprise a large group of chem icals used in a variety of technological applications like textile or leather treatment, surfactants and food packaging to make them e.g. non-stick and stain repellent. Fore most, in vitro studies show that PFAS are ligands for peroxisome proliferator-activated receptors (PPARs) α, γ (Wolf et al, 2008b; Rosen et al, 2017; Vanden Heuvel et al, 2006) and β/δ with different po tencies depending on carbon chain length and terminal functional group (Li et al, 2019; Wolf et al, 2008a). Some animal data support more of a potential beneficial effect of PFAS on glucose and lipid homeostasis (Yan et al, 2015; Bijland et al, 2011; Kees et al, 1992), the epidemiological evi dence is limited and inconsistent, showing direct (He et al, 2018; Sun et al, 2018), null or even inverse associations with T2D (Donat-Vargas et al, 2019a; MacNeil et al, 2009; Conway et al, 2016; Karnes et al, 2014). Our previous study, based on the present study population, assessing PFAS exposures with T2D risk without including the metab olome, found nominal inverse associations of PFAS with insulin resis tance and T2D risk, possibly explained by PFAS activation of PPARs (Donat-Vargas et al, 2019a)

Methods

Results

Conclusion