2.1 Study population
Participants were randomly enrolled from August 2018 to November 2018 at the Health Management Center of Wuhan Tongji Hospital. The criteria for participant inclusion were established as follows: 1) residents in the study area; 2) adults ≥18 years old; 3) no history of occupational exposure to OPEs. A sample of 2103 individuals was enrolled, and both urine and blood samples were collected during their clinic visit. 347 individuals who met the diagnostic criteria for MetS were selected as the case group, while 347 participants were randomly selected as the control group by matching the case on gender and age (±5 years). A total of 694 participants (347 pairs) were ultimately enrolled in this study. Of these, 592 participants (294 in the case group and 298 in the control group) were examined for inflammation biomarkers, after excluding 102 individuals who did not have sufficient blood samples. The research received approval from the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology (NO. [2018] IEC (S329)).
2.2 Measurement of urinary OPE metabolites
The concentration of urinary OPE metabolites was measured by solvent-induced phase transition extraction coupled with ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) following the methods described in previous studies (Hu et al., 2019). In brief, 1 mL of urine was incubated at 37°C for 12h with mixed internal standard, ammonium acetate buffer solution, and β-glucuronidase. Acetonitrile and methyl tert-butyl ether were added and the upper organic phase was isolated and subjected to evaporation until complete dryness. Then, 100 μL of a methanol/water (1:9, v/v) mixture was used for reconstitution of the solutions before UPLC-MS/MS analysis. Quality control was performed by measuring blank samples and quality control (QC) samples (0.2 ng/mL, 2 ng/mL) in each batch of 30 samples. The recoveries of the 16 OPE metabolites ranged from 71.34% to 117.63%, with intra-day relative standard deviations (RSD) less than 18.3% and inter-day RSD less than 22.5%. The limit of detections (LODs) for 16 OPE metabolites varied between 0.012 to 0.121 ng/mL. More comprehensive details on standards and internal standards for the 16 OPE metabolites were described in Supplementary Materials. Concentrations below the LODs are replaced with the value of LOD/.
2.3 Definition of MetS and its components
MetS was determined in accordance with the International Diabetes Federation's criteria for Chinese people. Individuals with central obesity (waist circumference ≥90 cm in men and ≥80 cm in women) and two or more of the following criteria were defined as having MetS: (1) Triacylglycerol (TG): fasting serum TG ≥1.7 mmol/L; (2) high-density lipoprotein cholesterol (HDL-C): fasting serum HDL-C <1.04 mmol/L for men, <1.29 mmol/L for women; (3) blood glucose: fasting blood glucose (FBG) ≥5.6 mmol/L or have been diagnosed with type II diabetes; (4) blood pressure (BP): systolic blood pressure (SBP) ≥130 mmHg, diastolic blood pressure (DBP) ≥85 mmHg or are undergoing blood pressure-lowering therapy. Trained physicians carried out physical examinations, comprising height, weight, and waist circumference, and used a calibrated electronic sphygmomanometer to measure the resting BP of each participant three times. Biochemical indexes including HDL-C, FBG, and TG were analyzed by a HITACHI auto-analyzer (Cobas 8000, Roche Diagnostics, Basel, Switzerland).
2.4 Measurements of oxidative stress and inflammation biomarkers
Methods for determining urinary levels of 8-iso-prostaglandin F2a (8-isoPGF2a), 8-hydroxy-2-deoxyguanosine (8-OHdG), and 4-hydroxy-2-nonenal-mercapturic acid (HNE-MA) referenced to previous studies (Wang et al., 2019b). In brief, 100 μL of urine samples were subjected to solid phase extraction (SPE) after adding 1.5 mL of deionized water and 50 μL of internal standard. The resulting eluate, upon evaporation to dryness, was re-dissolved in 100 μL of 10% methanol in water. Three oxidative stress biomarkers were measured using LC-MS/MS. Each batch of urine samples consists of 1 blank sample and 2 QC samples. The spiked recoveries of the three oxidative stress biomarkers varied between 85.6% to 110.9%, with intra-day RSD less than 10.5% and inter-day RSD less than 15.6%. The LODs of 8-isoPGF2a, 8-OHdG, and HNE-MA were 0.02 ng/mL, 0.03 ng/mL, and 0.01 ng/mL, respectively.
Serum concentrations of inflammation biomarkers, including interleukin-6 (IL-6), tumor necrosis factor-α (TNF-α), and C-reactive protein (CRP), were measured using an enzyme-linked immunosorbent assay (ELISA) kit (Link Bio, China). The protocol was executed under the guidelines provided by the manufacturer. To minimize systematic bias and inter-assay variation, all samples were processed identically and analyzed randomly. The LODs for IL-6, TNF-α, and CRP were 0.02 pg/mL, 0.16 pg/mL, and 0.4 mg/dL, respectively.
2.5 Statistical analysis
Descriptive statistics were applied to examine the basic characteristics of participants. Differences in demographic characteristics, MetS components, urinary OPE metabolites, and levels of oxidative stress and inflammation biomarkers between MetS and non-MetS groups were assessed using either parametric or nonparametric methods. In this research, solely urinary OPE metabolites with detection rates above 60% were considered. The levels of urinary OPE metabolites, oxidative stress, and inflammation biomarkers below the LOD were identified as the square root of LOD values over 2. In addition, urinary OPE metabolites, oxidative stress, and inflammation biomarker concentrations were ln-transformed due to their skewed distributions. Spearman’s test was assessed to calculate the correlations between urinary OPE metabolites.
The correlations of urinary OPE metabolites, oxidative stress, and inflammation biomarkers with the MetS risk were assessed using multivariate logistic regression models with odds ratio (OR) and its 95% confidence interval (CI) as effect estimates. In addition, the correlations between urinary OPE metabolites and oxidative stress and inflammation biomarkers were assessed using multiple linear regression models with the corresponding percentage change (% change) and its 95% CI as the effect estimate and with the following equation: (% change) = 100%×[exp(β)-1], where β is the regression coefficient for urinary OPEs metabolites. Specifically, urinary OPE metabolites, oxidative stress, and inflammation biomarkers were modeled as both continuous and categorical (tertiles) variables in the above two models. Linear trend tests by tertiles of OPE metabolites, oxidative stress, and inflammatory biomarkers were executed by utilizing the median value of each tertile as a continuous variable in the above two models. Restricted cubic splines (RCS) were further applied to estimate the potential nonlinear correlations between the OPE exposure and MetS.
Weighted quantile sum (WQS) regression and Bayesian kernel machine regression (BKMR) were employed to further assess the relationship between OPE mixtures and MetS risk. The WQS analysis was employed to estimate the collective effect of OPE mixtures on the MetS and to identify the primary pollutants involved (Carrico et al., 2014). The WQS index was computed by utilizing the "gWQS" software package in R, where individual OPE concentrations were combined via a weighted sum. The WQS index, ranging from 0 to 1, indicates the weight of the seven urinary OPE metabolites and their relative importance. BKMR offers a flexible approach to modeling the individual and combined effects of exposure to chemical mixtures (Bobb et al., 2015). This method takes into account the potentially non-linear and non-cumulative dose-response relationships between these chemicals by utilizing kernel functions. To assess the overall joint effect of the OPEs on MetS, we estimated the difference in outcome levels at various percentiles for seven OPEs compared to their 50th percentile values. To fit the BKMR models, we employed a Markov chain Monte Carlo (MCMC) sampler, running for 20,000 iterations. Furthermore, we calculated posterior post-inclusion probabilities (PIPs) to determine the relative importance of each outcome.
Mediation analysis was applied to examine the potential mediating mechanisms involving oxidative stress and inflammation biomarkers in the association of urinary OPE metabolite concentrations with MetS (Tingley et al., 2014). The direct effect (DE) indicates the unmediated association between OPE metabolites and MetS, whereas the indirect effect (IE) quantifies the impact of OPE exposure on MetS risk mediated by oxidative stress and inflammation biomarkers. The mediation proportion arising from oxidative stress and inflammation biomarkers was computed as the ratio between the indirect effect and the total effect (TE).
Covariates were identified based on a combination of biological and statistical considerations (Yao et al., 2021a). The models incorporated the following covariates: urinary creatinine (continuous), age (continuous), gender (male, female), education (below high school, high school, college, and above), income (<6000, 6000-10000, ≥10000 yuan/month), smoking status (never smoker, former/current smoker), physical activity-metabolic equivalent (MET) (low, moderate, high), red meat intake (≤3 times/week, 4-6 times/week, ≥7 times/week), and vegetable intake (≤6 times/week, 7-12 times/week, ≥13 times/week). Details of physical activity-MET were described in Supplementary Material.
We conducted sensitivity analyses to evaluate the robustness of our findings. We re-examined the associations between creatinine-adjusted concentrations of OPE metabolites and the risk of MetS using multivariate logistic regression models, excluding creatinine as a covariate.
Statistical analyses were performed using SPSS version 25.0 (SPSS Inc., Chicago, IL, USA) and R version 4.1.1 (R Core Team 2021). P-values < 0.05 were considered to be statistically significant.