Associations of Perfluoroalkyl Substances with Prevalence of Metabolic Syndrome in Highly Exposed Young Adult Community Residents—A Cross-Sectional Study in Veneto Region, Italy

Background: Studies on the association between perfluoroalkyl substances (PFAS) and metabolic syndrome (MetS) are limited, and results are inconsistent. We aimed to examine the associations between PFAS serum levels and the prevalence of MetS among highly exposed young adults (ages 20–39) residents of a large area of the Veneto Region (North-Eastern Italy) primarily stemming from PFAS water contamination before September 2013. A total of 15,876 eligible young adult residents living in the investigated municipalities were enrolled in the study from January 2017 to July 2019. Methods: MetS was defined by using a modified harmonized definition requiring the presence of 3 of the following: obesity (body mass index ≥30), elevated triglyceride (TG), reduced high-density lipoprotein cholesterol, elevated blood pressure, and hemoglobin A1c ≥ 6.1% or self-reported diabetes mellitus or drug treatment for hyperglycemia. Multivariable generalized additive models were performed to identify the associations between four serum PFAS, including perfluorooctane sulfonic acid (PFOS), perfluorooctanoic acid (PFOA), perfluorohexane sulfonic acid (PFHxS), and perfluorononanoic acid (PFNA), and risk of MetS controlling for potential confounders. Results: A total of 1282 participants (8.1%) met the criteria of MetS with a higher prevalence among men. PFOA, PFHxS, and PFNA were not associated with the risk of MetS, whereas PFOS showed a consistent protective effect against the risk of MetS (OR 0.76, (95% CI: 0.69, 0.85) per ln-PFOS). However, we found statistically significant positive associations between PFAS serum levels and individual components of MetS, mainly elevated blood pressure and elevated TG. Conclusion: Our results did not support a consistent association between PFAS and MetS and conflicting findings were observed for individual components of MetS.


Introduction
Perfluoroalkyl and polyfluoroalkyl substances (PFAS) are a group of more than 4700 man-made chemicals containing an aliphatic fluorinated carbon chain [1]. Due to their

Participants
The study population was recruited from an ongoing population-based health surveillance program approved with the Regional Government Deliberation n. 2133 on 23 December 2016. The health surveillance comprises individuals from more than twenty municipalities with PFAS-contaminated waterworks, so called "Red area" in Veneto region, Italy, due to the activities of a local manufacturing plant that started PFAS production since the late-1960s [11]. Clean water was provided in the Red area from September 2013. To be eligible, participants were required to live in the Red area, and be willing to follow to the health surveillance program. Information on this ongoing health surveillance program including study design, survey administration, participants enrollment as well as characteristics and a map of the Red area have been previously described in details [11]. Overall, 16,224 young adult residents, born between 1978 and 1999 (aged 20-39 years at the recruitment) and living in the investigated municipalities, were enrolled in the surveillance plan from January 2017 to July 2019. Participation rate for this population was 61%. Pregnant participants (n = 327) and individuals with missing information on the metabolic syndrome components (approximately 0.1% of the studied population, n = 21) were excluded leaving a total of 15,876 participants included in the statistical analysis.
A structured software-based questionnaire was administered collecting information on socio-demographics, personal characteristics, anthropometry, and an abbreviated health history at the enrolment.
Blood pressure was measured during the medical exam at the time of enrolment according to the European Society of Hypertension recommendations [23]. Non-fasting blood samples were collected from participants and shipped for biochemistry markers testing related to MetS at three laboratories within the Red Area (Arzignano, San Bonifacio, Legnago). In the present study, laboratory data included triglycerides (TG), high density lipoprotein cholesterol (HDL-C), and glycated hemoglobin (HbA1c) from whole blood. Blood sample collection and processing were described previously [11].

MetS Definition
We adopted the definition of MetS from the Joint Interim Statement on MetS definition (JIS) [24]. The JIS definition is not hierarchical and does not require demonstration of any specific criteria such as insulin resistance or obesity per se; that is, every combination of three or more of the following criteria fulfils the diagnosis of MetS [24]: elevated waist circumference (population-and country-specific definitions); elevated TG: ≥150 mg/dL or drug treatment for elevated triglycerides; reduced HDL-C: <40 mg/L (male) or <50 mg/L (female), or drug treatment for reduced HDL-C; hypertension ≥130 mmHg systolic blood pressure (SBP) and/or ≥85 mmHg diastolic blood pressure (DBP) or antihypertensive drug treatment or self-declared hypertension; and elevated fasting glucose or drug treatment of elevated glucose.
Waist circumference was not measured in the present study therefore body mass index (BMI) was used instead. AHA/NHLBI recommended Waist circumference cut points of ≥102 for men and ≥88 cm for women for people of European origin. The latter values are corresponding to the definitions of abdominal obesity (BMI ≥ 30 kg/m 2 ) recommended by the World Health Organization (WHO), American Association of Endocrinology (AAE), and US National Program for Cholesterol Education (NCEP ATP III) [25].
In the absence of information on fasting blood glucose, HbA1c ≥ 6.1% (≥43 mmol/mol; as a screening tool for detection of type 2 diabetes) [26] or self-reported diabetes or drug treatment of elevated glucose (hypoglycemic medicines) were used as alternate indicators. Due to using non-fasting measurements, the cut-off value for elevated TG was set at 175 mg/dL [27]. Drug treatment for elevated TG and reduced HDL-C was limited to fibrates and statin, respectively. Furthermore, metabolically healthy obesity (MHO) was established when obese subjects had none of the MetS components, and had no previous diagnosis of cardiovascular disease (self-reported history of acute myocardial infarction, stroke, angina pectoris, or cardiovascular intervention) [28].

Statistical Analysis
Demographic characteristics were presented using frequencies and proportions for categorical variables and the mean ± standard deviation (SD) for continuous variables among study population classified according to MetS status stratified by gender. Each characteristic of the participants was compared between MetS and non-MetS by using t-test for continuous variables or chi-square test for categorical variables. In order to adjust the skewed distributions, serum PFAS levels were natural log transformed (ln-PFAS). We also analyzed the Spearman correlation between serum PFAS concentrations.
The shape of possible dose-response associations between continuous PFAS serum levels and MetS prevalence was modelled through thin plate spline smooth terms for the exposures and continuous covariates. Degree of smoothing was selected by Generalized Cross Validation as implemented in the R package mgcv [29]. Since the spline analysis showed associations compatible with a linear relationship on the natural-log transformed PFAS, linear regression coefficient (β) and 95% confidence intervals (CI) were reported. The PFAS quartiles (IQ, IIQ, IIIQ, IVQ), were examined in relation to MetS as a dichotomous outcome with the first PFAS quartile as the reference, and a p-value for trend across the quartiles was calculated.
We estimated the odds ratios (ORs) of MetS using Multivariable Generalized Additive Models (GAMs) with a binomial link function to identify associations between serum PFAS levels and MetS (yes or no for having at least 3 components of MetS) and the presence of MetS components (yes/no for a given component), controlling for several sociodemographic covariates.
Potential confounders were determined a priori based on associations between PFAS and metabolic outcomes from previous studies [17,30,31]. Thereafter a Directed Acyclic Graph (DAG) was used (Supplementary Figure S1) to identify set of a minimally sufficient set of variables to control for confounding.
The final model for adjustment was Model 1 including gender, age, country of birth, smoking status (current smoker/former smoker/never smoked), diet (meat, fish, cheese, eggs, fruits and vegetables, sweets, snacks and sugary drinks, salt intake), alcohol consumption (categorized in 0, 1-2, 3-6, 7+ alcohol units per week), time-lag between enrollment and the beginning of the study, number of deliveries, clinical centers where the blood pressure have been measured and the questionnaire have been filled, education and physical activity (light/moderate/heavy). Model 2 was performed only for individual components of MetS, including all covariates of model 1 plus all other components of MetS.
Associations described in the results as statistically significant use the criterion p < 0.05. The statistical software STATA/SE version 13.0 (Stata Corp LP, College Station, TX, USA) and R (R Development Core Team 2010, R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL: http://www.R-project.org/) were used for statistical analyses.

Sensitivity Analyses
We assessed the robustness of the results in various sensitivity analyses. A sensitivity analysis was conducted to consider estimated glomerular filtration rate (eGFR) as an additional covariate in the analyses since eGFR value < 90 mL/min/1.73 m 2 has been associated with increased risk of metabolic diseases, obesity, hypertension, type 2 diabetes, CVD, and non-alcoholic fatty liver disease (NAFLD) [32,33].
Furthermore, a sensitivity analyses was performed defining MetS by International Diabetes Federation (IDF) criteria [34]. The major difference between these two definitions is that the IDF criteria regards central obesity as an essential and core criterion of MetS with cut point of waist circumference (WC) ≥94 cm for men and ≥80 cm for women in Europeans which corresponds to BMI ≥25 kg/m 2 ("overweight").
Lastly, a sensitivity analyses was carried out restricting our population to nonsmokers because unhealthy lifestyle, such as smoking is linked to higher prevalence of MetS [35].

Population Characteristics and MetS Components
The study population consisted of 15,876 individuals that were almost evenly distributed by gender (48.61% men, and 51.39% women) with a mean age of 30.0 years (SD 5.8 for both). The overall prevalence of the MetS was 4.68% (n = 743) based on the JIS definition, with a significant higher prevalence in men than women (Table 1). The most commonly reported components were elevated blood pressure (33.9%) followed by reduced HDL-C (15.2%). The mean level of triglycerides was 104.9 ± 73.6 mg/dL and elevated triglycerides (≥175 mg/dL) was shown in 11.48% of participants.
When stratified by gender, more men than women met MetS criteria for almost all the components of MetS, particularly for elevated TG, elevated blood pressure and obesity. In contrast, more women showed reduced level of HDL-C (Table 1). The median HDL-C level was 58.64 mg/dL, indicating that 16.5% of women were below the cut off for reduced HDL-C (<50.0 mg/dL) (Table 1).
However, among obese population (BMI ≥ 30: n = 1383, 8.71%), a total of 303 subjects (21.91%) did not have any metabolic abnormalities, as well as no previous diagnosis of cardiovascular disease, as defined by the MHO phenotype. The prevalence of individuals with MHO was found higher in women (n = 199, 29.05%) than in men (n = 104, 14.90%) (Supplementary Table S2).
Significant differences in demographic and lifestyle characteristics according to MetS status were observed among participants ( Table 2). Compared to the non-MetS group, the MetS group had significantly higher proportions of current and former smoking status, lower level of education (elementary-middle school), higher meat and salt consumption, lower fruits and vegetables consumption, and origin from HMPC countries. The proportion of not drinkers as well as heavy drinkers (more than seven alcoholic units per week) was significantly higher in the MetS group than in the non-MetS group. The prevalence of women with two or more deliveries was significantly higher in the MetS group (Table 2). Serum PFAS concentrations of both MetS and non-MetS participants is presented in Table 2 Table S3). Additionally, Spearman's rank correlation analysis showed that all PFAS compounds were moderately to strongly correlated with one another (p < 0.05). The most highly correlated compounds are PFHxS with PFOA, the two PFAS showing highest average concentrations (rs = 0.907); the least correlated are PFNA with PFOA (rs = 0.397). The Spearman correlation coefficients of PFOS with PFOA and PFHxS were 0.634 and 0.680, respectively; and the correlation coefficients of PFNA with PFHxS was 0.403.

Association between PFAS and MetS and Its Components
Tables 3 and 4 display the prediction of serum PFAS on odds ratios (ORs) with 95% confidence interval (CI) for MetS and its components as binary outcomes per natural-log unit (ng/mL) increase in levels of PFAS as well as in categorical (based on quartiles of exposure). As shown in Table 2, the odds of MetS were not significantly associated with any of PFOA, PFHxS and PFNA concentrations. Serum ln-PFOS concentrations were significantly associated with a lower prevalence of MetS (OR 0.76, (95% CI: 0.69, 0.85) per ln-PFOS) in covariate-adjusted models. When stratified by gender, the effect estimates were similar ( Table 3). The tests for trend by treating PFAS level in quartiles showed only a significant trend for PFOS. The OR for the highest compared with the lowest quartile of PFOS was 0.55 (95% CI: 0.43-0.70, p-trend < 0.01). The negative association between PFOS concentrations and the risk of MetS remained consistent with increasing PFOS quartiles and these associations were more pronounced in women. Table 3. Generalized Additive Models (GAM) models for the association between perfluoroalkyl substances (PFAS) (ln µg/L) and metabolic syndrome (MetS), stratified by gender, adjusted by several covariates a , using the continuous and PFAS quartiles.  The associations between the serum PFAS levels and individual components of the MetS are summarized in Table 4. After adjustment by other components of the MetS in addition to other cofounders, increased serum PFAS concentrations were associated with a higher prevalence of hypertension (PFOA OR 1.05 (95% CI 1.01-1.08); PFOS OR 1.10 (95% CI 1.03-1.17); and PFHxS OR 1.07 (95% CI 1.03-1.12)). Increased serum PFOA and PFHxS concentrations were also correlated with a higher prevalence of elevated triglyceride as defined in MetS criteria (PFOA OR 1.10 (95% CI 1.05-1.16)); and PFHxS OR 1.09 (95% CI 1.02-1.16)). PFNA was positively but non-significantly associated with increased risk of hypertension (OR 1.10 (95% CI: 0.99, 1.21)) and with higher risk of higher triglyceride level above the MetS criteria (OR = 1.06 (95% CI: 0.92, 1.23)).

PFAS
In contrast, increased serum levels of PFOA, PFOS, PFHxS, and PFNA were significantly associated with decreased risk of other components, with the most consistent pattern seen for PFOS (Table 4).
Sensitivity analysis were conducted to assess the constancy of the association between investigated PFAS and the prevalence of MetS while adding eGFR in the models as possible confounder. Adjusting for eGFR did not change the results for PFOS, PFOA and PFHxS, whereas, the null association turn out to be significant between PFNA and the prevalence of MetS.
Considering the IDF definition, about 4.7% (n = 743) of the participants met the IDF criteria for MetS, with higher prevalence in men (6.6%) than women (2.9%). Overall, 3.55% (based upon JIS definition) and 6.08% (based upon IDF definition) of the survey participants had at least three of the MetS components present (Supplementary Table S4), whereas less than 0.1% met the criteria for all five components. In the total study population, 668 (4.21%) of the participants had met the criteria of MetS by both JIS and IDF definitions, with rates of 5.75% and 2.75% among men and women, respectively.
The association between PFAS and the risk of MetS remains consistent in the models while using the IDF definition. Serum PFOS concentrations were significantly associated with a lower risk of MetS defined by the IDF criteria with OR = 0.70, (95% CI: 0.61, 0.79) per ln-PFOS. Modelling PFAS by quartiles yielded consistent results.
Moreover, the results of the sensitivity analysis restricted to non-smokers were consistent with the main result that none of the PFAS was positively associated with MetS and the negative associations between serum PFOS were robust.

Discussion
To our knowledge, this is the first study investigating the association between PFAS exposure and MetS in a highly exposed community of residents in Europe.
The observed prevalence of MetS was 8.1% among young adult population aged at 20 to 39 years old, comorbidities conveying high risk of both cardiovascular disease and type 2 diabetes.
However, PFOA, PFHxS, and PFNA failed to show positive significant associations with MetS and depending upon the specific PFAS and MetS component under investigation, either null, deleterious or protective associations were observed with borderline to statistical significance. Further, multiple models demonstrated that PFOS was associated with decreased risk of MetS and consistently for all but one of its components (hypertension). For individual components of MetS, all the investigated PFAS were positively associated with hypertension but with decreased risk of reduced HDL-C.
In other epidemiological studies, inconsistent findings exist regarding the association between PFAS and MetS. Overall, no association was observed between PFOA, PFOS, PFHxS, and MetS in different cross-sectional studies, which is concordant with our analysis [15][16][17][18]21,30,36]. Only two studies one on a small scale (n = 148) of Chinese adult men (19-60 years old) and the other of 1501 adults from a cross-sectional study, the "Isomers of C8 Health Project in China" found a significant association between linear PFOA and the risk of MetS with ORs of 29.4 (95% CI, 2.90-299.7), after adjustment only for age [19] and 1.99 (95% CI, 1.40, 2.83) after adjustment for age, sex, income status, smoking status, alcohol consumption, regular exercise, seafood consumption, and eGFR [20]. A protective (negative) significant association between PFOS concentrations and the risk of MetS is only reported in the present study and the study by Leary et al. (2018), but in the latter the association disappeared after adjusting for confounders [16]. In the study conducted by Ye et al. (2020) an isomeric-specific association between PFOS and MetS was found, in which was statistical significance for branched PFOS isomers but null for linear PFOS [20]. Different structure of perfluoroalkyl branched-chain between linear and branched PFAS may influence their pharmacokinetic characteristics (e.g., binding affinity to protein, elimination half-lives) and their toxicity. For PFNA, results are inconsistent across the studies. Recently, Christensen et al. examined the data from the National Health and Nutrition Examination Survey (NHANES) 2007-2014 and reported that the prevalence of MetS defined by JIS definition was 37% among U.S. adult population (≥20 years old) and only PFNA was associated with an increased risk of MetS, when controlling for multiple PFAS [21]. Likewise, a cross-sectional study on 122 adult participants (44-56 years old) of the Croatian island on 2007-2008 identified a 2-3 times increased risk of MetS as defined by the ATP III criteria for a natural log unit increase of PFNA [30]. In the cross-sectional study conducted by Yang et al. PFNA serum levels above the median were associated with 10.9-fold (95% CI, 2.00-59.1) increased risks of MetS, while after adjustment for age, the association was not robust [19]. In the present study, a significant positive association was found between PFNA and MetS after extra adjustment for eGFR. In contrast, increased serum PFNA concentrations were associated with a lower prevalence of the MetS according to NHANES 1999-2000 and 2003-2004 data and borderline but not significant based on subsequent analysis of NHANES 2013-2014 data [16,36]. The PFAS quantified at the highest serum concentration in this population was PFOA followed by PFHxS. For PFOA, the median was 35.9 µg/L, which is about 12-32 times higher than other European studies with background exposure [30,31] as well as Canadian and U.S. general populations [15][16][17]21,36], even 4-5 years after eliminating the exposure via contaminated drinking water in the Veneto region. Overall, the PFAS concentrations found in total serum in the young adult population was around five times higher for PFOA and PFHxS than the average burden reported for Europe 7.7 µg/mL and 0.67 µg/mL, respectively [9]. In contrast, the PFOS median concentration (3.7 µg/L) was significantly lower in our study population than the other studies [18,20] but in a roughly similar range to the median PFOS concentration of a Chinese adult population (3 µg/L) [19]. Nevertheless, the medians of serum PFOS concentration here were almost two times higher than the median concentrations in serum/plasma in the general adult population in Europe (1.9 µg/mL) [9]. In general, the observed pattern of PFAS levels, relating among participants defined by MetS criteria, were comparable with the results of the four cross-sectional studies. Overall, no association was observed between PFOA, PFOS, PFHxS, and MetS in different cross-sectional studies, which is concordant with our analysis [17][18][19][20]. For PFNA, the level in this study was relatively lower than the other studies. Nevertheless, concentrations alone do not explain the discrepancy in findings between the different studies. Moreover, serum PFAS concentrations were strongly correlated with each other in all the studies and therefore, effects of specific PFAS cannot be distinguished.
For individual metabolic components, the literature is also not entirely consistent. Hepatic steatosis effects supported by convergent data from both the toxicology and epidemiology studies might be considered as a suggested contributing mechanism to the altered cholesterol following by PFAS exposure [9,33]. This association is consistently evident for altered total and LDL cholesterol but not for HDL-C and TG. Our findings of inverse associations between PFAS, triglycerides (mainly for PFOS) and HDL-C are in the same direction as other studies [17,31,36,37]. As for elevated blood pressure, positive associations are reported for some PFAS in recent studies [11,18,38].
To date, few studies have investigated the association between PFAS and overweight/obesity with inconsistent results [39][40][41]. A recent cross-sectional study among 1612 Chinese adults (1204 men and 408 women), ages 22-96 years old, indicated that increased serum concentrations of PFOS, PFOA, and PFNA were positively associated with overweight/obesity and PFNA and PFOA with increased WC [42]. A significant association was also observed between PFNA and BMI ≥ 25 kg/m2 but not remarkable association with obesity (BMI ≥ 30 kg/m 2 ) in Croatian island adult population at ages 44-56 years [30]. Statistically significant positive associations with BMI were reported for PFNA with serum concentrations above the medians in the study of Chinese men aged 19-60, after adjustment for age [19]. As for our data, the observed PFAS-BMI associations were mainly negative and for some PFAS null and therefore considered inconclusive. Among investigated community residents, there were a total of 5204 overweight and obese (BMI ≥ 25: 32.8%) participants with higher prevalence in men (40%) than women (26%), among which 1685 participants (32.38%) did not meet any of the criteria for MetS and did not have previous diagnosis of cardiovascular disease, so called MHO. The prevalence of MHO was higher in overweight and obese (BMI ≥ 25) women (n = 866, 41.40%) than in men (n = 819, 26.32%) (Supplementary Table S2). Whereas, considering overweight and obese based on IDF criteria for MetS did not alter the association between PFAS and MetS.
Although the directions of our relationships with MetS are consistent with most of other studies, it is notable that our study population was restricted to the age group between 20 and 39, while the most frequently investigated age group in other studies is that over 20 years old. According to the literature, aging contributes to increase prevalence of MetS since many predisposing conditions, such as hypertension, obesity, and insulin resistance increase in prevalence during aging. However, we do not know whether results may change considering older community residents.
Altogether, clear conclusions cannot be drawn on PFAS and MetS from our study and previous literature, due to the inconsistency in the results between different studies and the cross-sectional nature of the studies. Conflicting observations mostly are attributable to differences in cut-off points for criteria used for defining MetS associated with each PFAS, PFAS concentrations (occupational vs. community vs. environmental exposures), study design (cross-sectional vs. longitudinal studies) and age and gender-specific differences. In addition, any failure to fully adjust for potential confounders could be an alternative explanation for the conflicting results observed in human studies.
Overall, MetS is a cluster of interrelated risk factors (obesity, insulin resistance, dyslipidemia, and hypertension) all known as risk factors of cardiovascular diseases (CVD), therefore different mode of action/adverse outcome pathways (AOPs) of PFAS could influence the final results [14,43]. For instance, the PFAS-to-lipid associations are explained in animal studies by transactivation of peroxisome proliferator-activated receptor α (PPARα) contributing to a liver metabolic pathway which plays an important role in the regulation of cholesterol and triglyceride levels; or oxidative stress has been proposed to be a key player in the pathogenesis of hypertension [44][45][46][47]. Other putative molecular initiating/key events for PFAS might be explained by mitochondrial dysfunction, interference of protein binding, partitioning into lipid bilayers (enterohepatic cycling of both bile acids and PFAS), altered calcium homeostasis, and inappropriate activation of molecular signals controlling cell functions [14,48]. As for CVD, few studies of occupationally exposed workers and nonoccupationally exposed population have reported associations between PFAS exposure and CVD [49]. However, the potential mechanisms beyond the potential association are still not clear and several hypotheses have been made. The results of a study investigating alteration of microRNA expression indicated that modulation of miRNA expression may be a mechanism of toxicity of PFAS and in silico functional analyses suggested potential links between PFAS concentrations in serum, miRNA expression, specific miRNA target genes and perhaps onset of health effects such as cardiovascular [50]. Moreover, a recent study on the role of impaired platelet aggregation in increased cardiovascular risk associated with exposure to PFOA provide evidence that PFOA distributed unevenly between blood cells, and altered membrane fluidity and, in turn, altered downstream signaling pathways regulating platelets' activation and aggregation. These findings could explain an association between PFAS exposure and CVD given that platelets play an important role in CVD both in the pathogenesis of atherosclerosis and in the development of acute thrombotic events [51].
This study was limited to cross-sectional analysis of serum PFAS levels at a certain period of time, which makes difficult to track the trends of PFAS levels in the study population or to pin-point significant exposure route(s). In addition, due to the cross-sectional design cause-and-effect associations have not been established for either MetS or any of individual components of MetS. Therefore, given the relatively long half-life of PFAS in the human body, long-term prospective studies (longitudinal studies) are urgently required to elucidate the putative causal relationships between PFAS and health effects. An imprecision in identifying MetS we also face is the lack of specific measurements for waist circumference and using self-reported height and weight that might cause inaccuracies in identify subjects with MetS. The study also lacked data on fasting blood glucose. Using more directly clinically relevant outcomes (e.g., elevated fasting glucose) would be preferable; however, because of using non-fasting blood to facilitate the recruitment process, this was not an option.
Furthermore, while the survey data allowed us to control for a wide number of sociodemographic and lifestyle risk factors, due to complexity and lack of understanding of the interplay between exposure to PFAS and personal behavior, we cannot rule out the possibility of both positive and negative confounding, that are unaccounted for in our study, such as consumption of junk food. We have not studied interactions of multiple PFAS and other potential environmental toxicants, nor the assessment of combined exposure to multiple PFAS.
Strengths of this study include examination of a large community residents, high quality of exposure measurements, detailed stratified analyses, adjusting a comprehensive profile of potentially confounding variables, and the ability to test for the robustness of our findings in various sensitivity analyses.

Conclusions
PFAS are suspected endocrine disruptors and human exposure is ubiquitous. In our large-scale cross-sectional study on young adult population, no associations were observed between legacy PFAS exposure and MetS, rather a negative association between PFOS and MetS. However, PFAS levels were significantly associated with some individual components of MetS and some showed protective associations, confirming results of previous researches and adding into the exciting evidence. Overall, further longitudinal follow-up studies on PFAS and their mixtures are warranted to clarify causal relationships.
Supplementary Materials: The following are available online at https://www.mdpi.com/1660-4 601/18/3/1194/s1, Table S1: Concentrations of perfluoroalkyl substances (PFAS; µg/L) in serum as per defined method detection limit for each compound; Table S2: Prevalence of metabolically healthy obesity (MHO) in the study population; Table S3: Distributions of serum PFAS concentrations (ng/mL) in the study population stratified by gender; Table S4: Number of components of the metabolic syndrome present among participants. Figure S1: Directed Acyclic Graph to identify set of a minimally sufficient set of variables to control for confounding.
Author Contributions: M.Z.J. wrote the manuscript, discussed and interpreted the results data. G.B. performed the statistical analysis. T.D.Z., A.S.C.F., and T.F., F.R., G.P. reviewed and edited the manuscript. F.D. performed the chemical analysis and reviewed the manuscript. C.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Regional (Veneto Region) Ethics Committee (24 maggio 2017 prot. n. 203638).

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data are not publicly available. The data presented in this study are available on request from the corresponding author.