Non-monotonic Association Between Chlorinated Polyfluorinated Ether Sulfonic Acids Exposure and the Risk of Overweight/Obesity Status in Adults

Chlorinated polyfluorinated ether sulfonic acids (Cl-PFESAs, including 6:2 Cl-PFESA and 8:2 Cl-PFESA), one of poly- and perfluoroalkyl substances (PFAS), used as perfluorooctane sulfonate (PFOS) alternatives in electroplating industry. Evidence in vivo and in vitro indicates that Cl-PFESAs might disrupt lipid metabolism. However, the association between Cl-PFESAs exposure and the prevalence of overweight/obesity in human is unknown. We conducted a cross-sectional study to investigate associations of serum 6:2 Cl-PFESA and 8:2 Cl-PFESA exposure with overweight/obesity status in adults. We quantified four perfluoroalkyl substances (PFAS), including 6:2 Cl-PFESA, 8:2 Cl-PFESA, PFOS, and perfluorooctanoic acid (PFOA) in 1275 Chinese adults from the Isomers of C8 Health Project in China study. Characteristics of participants were gathered from interviewer-administered questionnaires and anthropometric measurements. We classified overweight/obesity based on body mass index (BMI) according to WHO. Participants were categorized into normal weight group (BMI < 25 kg/m2) and overweight/obesity group (BMI ≥ 25 kg/m2). The detection proportion of 6:2 Cl-PFESA was 100% among the participants in this study. Adjusted for potential confounders, BMI in the second quartile of each ln-ng/mL greater concentration of 6:2 Cl-PFESA and 8:2 Cl-PFESA was 0.45 [95% confidence interval (CI): 0.08, 0.82], and 0.39 (95% CI:0.03, 0.76) significantly higher than the lowest quartile, respectively. Cl-PFESAs displayed inverted U-shaped associations with the risk of overweight/obesity, and the inflection point of 6:2 Cl-PFESA and 8:2 Cl-PFESA was 1.80 ng/mL, 0.01 ng/mL, respectively. For example, The risk of overweight/obesity increased (OR= 1.94; 95%CI: 1.24, 3.01) until around 1.80 ng/mL of predicted 6:2 Cl-PFESA concentration and then decreased (OR= 0.69; 95%CI: 0.39, 1.21). PFOS was associated with waist circumference (WC) but not BMI in each quartile. For PFOA, the associations with outcomes were linearly positive (P for trend < 0.05). This study reports the first observations on non-monotonic associations between serum 6:2 Cl-PFESA and 8:2 Cl-PFESA concentrations and the prevalence of overweight/obesity in adults. Our findings suggest that Cl-PFESAs may have endocrine disrupting characteristics, and this exposure-outcome association is a challenge for risk assessment of Cl-PFESAs. But more epidemiological investigations are required to confirm the observed associations.


Introduction
Obesity is a global public-health problem that increases mortality, decreases life expectancy, and elevates risk for many comorbidities including type two diabetes, metabolic syndrome, reproductive abnormalities, hypertension, coronary heart disease, cancers and others (Flegal et al. 2013;Haslam and James 2005;Riaz et al. 2018). The trends in obesity prevalence have been growing over the past 40 years, with 39% of the global adult population becoming overweight Chu Chu, Qiu-Ling Fang, Xin-Xin Cui and Peng-Xin Dong have contributed equally to this work and should be listed as first authors. and 13% becoming obese in 2016 (Jaacks et al. 2019;WHO, Obesity and overweight, https:, , www. who. int, news-room, fact-sheets, detail, obesity-and-overweight, in 2021). The etiology of obesity is intricate. It is not simply an energy imbalance, but is also related to ubiquitous obesogens, such as poly-and perfluoroalkyl substances (PFAS) (Heindel and Blumberg 2019). PFAS are a series of highly fluorinated aliphatic compounds with wide distribution, bio-accumulative properties and extreme environmental persistence (Dhore and Murthy 2021), among which perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) are the most commonly studied and used (Sunderland et al. 2019). Previous literature suggests the associations of exposure to PFOS and PFOA with increasing body mass index (BMI) and overweight/obesity (Jain 2014;Tian et al. 2019), though conclusions are not consistent (Chen et al. 2019;Lin et al. 2009;Timmermann et al. 2014).
Chlorinated polyfluorinated ether sulfonic acids (Cl-PFESAs, commercially named F-53B), including 6:2 Cl-PFESA and 8:2 Cl-PFESA, have been used as a mist suppressant in China for over 40 years as PFOS alternatives. Because of an oxygen atom in the perfluoroalkyl chain, Cl-PFESAs were presupposed less persistent environmentally (Brase et al. 2021;Wang et al. 2013). Due to global actions to phase out PFOS and PFOA, many alternatives are being used in increasing quantities. However, there is increasing evidence that the health risks and toxicity of Cl-PFESAs may be greater than PFOS (Chu et al. 2020;Cui et al. 2018;Shi et al. 2016;Zhang et al. 2018).  reported that exposure to Cl-PFESAs was significantly positively associated with cholesterol, low-density lipoprotein cholesterol and triglycerides. Evidence in vitro and in vivo indicates that Cl-PFESAs might disrupt lipid metabolism, as Cl-PFESAs affected both metabolic transcription and organismal metabolic phenotype in fertilized zebrafish embryos , and elevated the relative triglyceride content in mouse 3T3-L1 preadipocyte (Li et al. 2018). However, there are few epidemiological studies currently to assess the risk of adiposity exposure to Cl-PFESAs in human beings.
To address this research gap, we explored the associations between exposure to Cl-PFESAs and overweight status in Isomers of C8 Health Project Study among Chinese adults. We hypothesized that exposure to 6:2 Cl-PFESA and 8:2 Cl-PFESA would be positively associated with BMI, waist circumferences (WC), and risks for being overweight/obese.

Study Population
The Isomers of C8 Health Project in China is a cross-sectional study which recruited residents of Shenyang city, Liaoning province from July 2015 to October 2016. We explored the associations between PFAS exposure and health outcomes among people generally exposed to high PFAS concentrations (Yeung et al. 2006). We enrolled government employees, including retirees, from Shenyang city. Furthermore, to assess different PFAS exposure sources across city sectors, 500 residents from 5 geographical zones (Central, North, East, South, and West) of Shenyang city were randomly selected. Following the inclusion criteria, people of aged ≥ 35 years with ≥ 5 years residency at their current address were enrolled. Serum PFAS concentrations and proportions of male and female participants were similar between government employees and communitydwellers, therefore, were combined for statistical analysis (Supplementary Material, Table S1). A total of 1612 participants were included, of those 1275 (response rate: 79.1%) completed the questionnaire assessment, anthropometric measurements, and provided blood samples. Details about participants enrollment and data collection have been described elsewhere (Bao et al. 2017;Zeeshan et al. 2020). The local Institutional Review Board of Sun Yat-Sen University Research Ethics Committee approved our study, and study procedures followed the principles of the Helsinki Declaration.

Outcome Measurements
Outcomes included standing height (centimeters, cm), weight (kilograms, kg) and WC (centimeters, cm) measured by physicians following the anthropometric measurements method from the Chinese Ministry of Health (CMoH, Chinese Ministry of Health. Anthropometric measurements method in health surveillance 2013). Standing height was measured to the nearest 0.1 cm using a stadiometer with a vertical backboard and a sliding horizontal head piece. Digital weight scales calibrated by calibration weights were used to measure weight to the nearest 0.1 kg. Procedures included positioning the measuring tape around a horizontal plane perpendicular to the bilateral midaxillary line at the midpoint between the lowermost of costal margin and the uppermost lateral border of the ilium, then recording the WC measurement values to the nearest 0.1 cm at the end of participants' normal expiration. We calculated BMI as weight divided by height (kg/m 2 ). BMI between 25.0 and 29.9 kg/m 2 corresponds to overweight, and BMI of ≥ 30.0 kg/m 2 is obesity for adults, according to the World Health Organization (WHO, Obesity and overweight, https:, , www. who. int, news-room, fact-sheets, detail, obesity-andoverweight, in 2021). Participants were categorized into an overweight/obesity group (BMI ≥ 25 kg/m 2 ) and normal group (BMI < 25 kg/m 2 ).

Serum PFAS Measurement
Serum PFAS methods were described in detail in a previous study (Chu et al. 2020). In brief, PFAS including novel and legacy in 0.2 mL serum were extracted by solid phase extraction and detected by an Agilent ultra-performance liquid chromatography (UPLC) 1290 attached to an Agilent 6495B triple-quadrupole tandem mass spectrometer (MS/MS) (Agilent Technologies, Palo Alto, CA, USA). We purchased PFAS standards from Wellington Laboratories (Guelph, ON, Canada). Our detailed analytic approach, including methods for quality control, is described in the Supplementary Material. The limit of detection (LOD) of PFAS was set as the minimum detectable concentration in PFAS samples requested to attain a signal-to-noise ratio of 3 (S/N = 3). Table S2 provides abbreviation and detection parameters of the studied PFAS. PFAS concentrations were replaced with LOD divided by the square root of 2, when lower than the LOD (Hornung 1990).

Statistics
Distributions were characterized for demographic features, socioeconomic characteristics, behavioral habits, outcomes and exposures. We used Q-Q plots and Shapiro-Wilk tests to evaluate the normality for continuous variables. Due to right-skewed distributions, serum PFAS concentrations were natural log-transformed. To explore associations of BMI, WC and overweight/obesity with PFAS, log10-transformed outcome variables were entered into restricted cubic spline (RCS) with three knots at the 10th, 50th, and 90th percentiles to modify models. We found possible non-linear associations between PFAS exposure and outcomes except for PFOA ( Figure S1), so we calculated categorical quartiles of PFAS concentrations in models. We conducted generalized additive models (GAMs) to estimate correlations of PFAS quartiles as predictors with BMI and WC as outcomes. Logistic regression models were used to estimate odds ratios (OR) and 95% confidence intervals (CI) for overweight/obesity as functions of PFAS quartiles. P-values for trend were estimated by converting each corresponding PFAS into an ordinal variable in the model. We also used the R "segmented" package to analyze the inflection points for the non-linear associations and then conducted a 2-piecewise binary logistic regression model. We assessed associations between PFAS and outcomes stratified by sex for sensitivity analyses, due to sex-specific effects reported in previous studies (Hales et al. 2017). To evaluate the robustness of our results, we conducted additional analyses by excluding participants for smoking (Audrain-McGovern and Benowitz 2011;Carreras-Torres et al. 2018;Chiolero et al. 2008), alcohol consumption (Sim 2015), and physical inactivity (Basterfield et al. 2014;Hankinson et al. 2010) given changes for BMI, WC and being obese in these groups.
Smoking was defined as at least one cigarette per day that lasted for a year. Drinking was defined as ≥ 8 mL of alcohol (either wine, beer or spirits) per day. Physical activity was defined as self-reported exercise greater than or equal to 60 min/day for the past year, otherwise physical inactivity (Tian et al. 2019).
We adjusted essential potential confounding variables in the regression models as factors related to PFAS exposure and outcomes or their precursors. These included age (years, continuous), sex (male/female), education (≤ high school/ > high school), occupation (blue collar/white collar), family income (< 30,000, 30,000-100,000 and > 100,000 Yuan/year), smoking (yes/no), drinking (yes/no), and physical activity (yes/no). These socio-economic factors and behavioral habits were considered to affect either the sources or the distribution and clearance of PFAS exposure (Brantsaeter et al. 2013;Christensen et al. 2016;Eriksen et al. 2011), that may also affect BMI and WC by changing calorie intake, lipid metabolism, and energy balance (Audrain-McGovern and Benowitz 2011;Carreras-Torres et al. 2018;Chiolero et al. 2008;Sim 2015). Directed acyclic graphs (DAGs) were used to characterize minimally sufficient covariables for reducing confounding bias ( Figure S2). The statistical analyses were conducted in R (Version 4.1.0; R Foundation for Statistical Computing, Vienna, Austria) and SAS (version 9.4, SAS Institute Inc., Cary, NC, USA).

Results
The characteristics of 1275 participants, including demographic characteristics, behavioral habits, outcomes, and selected PFAS concentrations stratified by group, are presented in Table1. The population was 62 years old on average, principally white-collar workers, high school educated, and 72.6% from middle-income family (30,0000-100,000 Yuan/year). People in the overweight/obesity group were younger (59 years vs. 62 years) and more of them were men (84.6% vs. 43.9%) compared to normal weight group. There was a significantly greater proportion of participants who smoked (P = 0.014) and drank alcohol (P = 0.001) in the overweight/obesity group. Average BMI was 24.6 kg/ m 2 , with an average WC of 87.5 cm overall. Except for 8:2 Cl-PFESA (70.12%), PFAS in all participants' serum were measured above the LODs (Table S2). The dominant PFAS in serum was PFOS (24.67 ng/mL). Table S3 lists the Spearman correlation coefficients between PFAS. Table S4 shows the characteristic of participants by PFAS quartiles.
After adjusting confounders, participants with serum concentrations of 6:2 Cl-PFESA and 8:2 Cl-PFESA in the second quartile had higher BMI compared with those who had serum concentrations in the lowest quartile in models (Table 2). For example, participants' BMI in the second quartile of 6:2 Cl-PFESA were 0.45 (95% CI: 0.08, 0.82) kg/m 2 , significantly higher than those in the reference group. The continuous trends for associations of Cl-PFESAs exposure with BMI and WC were null (P for trend > 0.05). PFOS was not associated with BMI but was associated with WC in each quartile. For PFOA, the associations with outcomes were linear (P for trend < 0.05). Table 2 shows the adjusted ORs with 95% CIs of overweight/obesity with categorical serum PFAS concentrations. We detected 1.80-fold (95% CI: 1.23, 2.63) greater odds of overweight/obesity in the second quartile of 6:2 Cl-PFESA compared to the reference group, which had not been found significantly in the higher quartile. Figure. 1A and B also described the non-monotonic relationship and the inflection points of 6:2 Cl-PFESA and 8:2 Cl-PFESA exposure with overweight/obesity, respectively.
The results of crude regression models were generally consistent with the main analysis (Table S5). There were slightly different associations between PFAS and overweight/obesity status stratified by sex (Table S6-7). When the participants who smoked and drank were excluded, results were similar (Table S8-9), but the associations among PFAS and outcomes in the subgroup of physical activity were weaker than in the main analysis (Table S10).

Discussion
In this cross-sectional study, our results indicated that Cl-PFESAs exposure was associated with overweight/ obesity status. We found that serum concentrations of 6:2 Cl-PFESA and 8:2 Cl-PFESA in the second quartile had higher BMI compared with these in the lowest quartile, but not in the higher quartiles, suggesting the non-monotonic relationship between exposure to Cl-PFESAs and overweight/obesity prevalence rates. Cl-PFESAs displayed an inverse U-shape association with the prevalence of overweight/obesity. Our study also demonstrated that exposure to PFOA was linearly positively associated with each outcome, and PFOS was only positively associated waist circumference. This is the first study to report the nonmonotonic associations between Cl-PFESAs exposure and overweight status in human populations. Among 1275 Chinese adults' serum, 6:2 Cl-PFESA, PFOA, and PFOS were all detected and 8:2 Cl-PFESA was detected in 70% of participants. A recent review Fig. 1 Dose-response relationships of participants 6:2 Cl-PFESA (A) and 8:2 Cl-PFESA (B) exposure with obesity. The solid lines indicate adjusted log[odds ratios (OR)] for overweight/obesity, with the dashed lines indicate the 95% confidence intervals (CI) derived from restricted cubic spline. The inflection points for 6:2 Cl-PFESA and 8:2 Cl-PFESA are 1.80 ng/mL and 0.01 ng/ mL, respectively. Models are adjusted for sex, age, education, occupation, family income, smoking, drinking, and physical activity (n = 1275) summarized Cl-PFESAs concentrations in China, the means of which are 4.20, 102, 941 ng/mL in the general population, high fish consumers, and metal plating workers, respectively (Brase et al. 2021). In Shandong province of China, the 6:2 Cl-PFESA level of 977 residents living near a fluorochemical plant was 2.311 ng/mL ). These were slightly higher than our results. But the median value of 6:2 Cl-PFESA concentrations was 0.34 ng/mL that lower than ours among 519 pregnant women in Shanxi, China (Liu et al. 2020a). The concentrations of Cl-PFESAs varied regionally (Chu et al. 2020;Liu et al. 2020a;Chen et al. 2017;Pan et al. 2017). The serum PFOA and PFOS concentrations found by the current study were higher than those reported in most other national studies (Schulz et al. 2020).
To date, there are no human studies available to explore the associations between the levels of Cl-PEFSAs and overweight/obesity status, but several studies have reported the associations of exposure to PFOS and PFOA with obesity outcomes (Jain 2014;Chen et al. 2019;Lin et al. 2009;Timmermann et al. 2014;Christensen et al. 2016;Eriksen et al. 2011;Averina et al. 2021;Geiger et al. 2021;Liu et al. 2018;Nelson et al. 2010). We summarized them in Table S11. In line with our results,  reported greater PFOA exposure was associated with higher risk of overweight/obesity in US children during 1999-2012 (OR Q3 vs. Q1 = 2.22, 95%CI: 1.20, 4.13; OR Q4 vs. Q1 = 2.73, 95%CI: 1.10, 6.74). Also consistent with our studies, increased PFOA levels measured in 5591 US people aged 12 and older was associated with increases in BMI (P = 0.038) (Jain 2014). However, the relationship of human exposure and overweight/ obesity status is still controversial, given concerns about reverse causality and effective dose (Jain 2020). Due to different exposure levels, different modeling approaches and inconsistent confounding variables, the results of some studies were not consistent with ours (Chen et al. 2019;Lin et al. 2009;Timmermann et al. 2014;Christensen et al. 2016;Eriksen et al. 2011;Averina et al. 2021). For example, a cross-sectional study about the associations between PFAS concentrations in Norwegian adolescents and obesity, without adjustment for socioeconomic status, showed null associations for PFOS and PFOA (Averina et al. 2021).
Potential biological effects were described in previous studies that could reveal the hazard of Cl-PFESAs exposure, although the mechanisms promoting adiposity risk by Cl-PFESAs are not clear. 6:2 Cl-PFESA displayed toxic effects on human liver HL-7702 cell, and significantly up-regulated gene Cd36 expression that regulated long-chain fatty acids transportation through the adipocyte plasma membrane . Cl-PFESAs also affected osteogenic differentiation in human bone mesenchymal stem cells (hBMSCs) related to obesity and metabolic diseases . In several in vivo and in vitro investigations, Cl-PFESAs showed agonistic activity toward the peroxisome proliferator-activated receptors (PPARs) pathways related lipid metabolism (Li et al. 2018;Sheng et al. 2018;Shi et al. 2019). Cl-PFESAs also have the characteristics of endocrine disruptors, which might be associated with glucocorticoids and progestogens synthesis in neonates (Liu et al. 2020b), sex hormone disorders in adult men (Cui et al. 2020), and induced estrogenic effects in zebrafish (Xin et al. 2020). The endocrine system is important for energy balance, fat distribution and fat deposition. For example, sex hormones affect food intake and alter the balance of glucose and insulin, lipogenesis and lipolysis to cause obesity (Heindel and Blumberg 2019;Taxvig et al. 2012). Endocrine disrupting chemicals (EDCs) are considered as obesogens promoting obesity in humans or animals (Nadal et al. 2017). However, we need more data before identifying Cl-PFESAs as an obesogen.
Interestingly, Cl-PFESAs displayed an inverted U-shaped relationship with the prevalence of overweight/obesity in our study. This unconventional dose-response relationship called non-monotonic dose-response (NMDR) relationship is normal in studies investigating the effects of EDCs (Lagarde et al. 2015;Vassilopoulou et al. 2017). Although there is less data about NMDR of Cl-PFESAs, several studies had reported that NMDR relationships occurred between PFAS and health outcomes. For example, Liao et al. found a J-shaped relationship of PFOA and PFNA with the risk of hypertension among adults in US (Liao et al. 2020). PFOS and progesterone levels displayed an inverse U-shape dose-response relationship in neonates (Liu et al. 2020b). NMDR relationship indicates the possibility of Cl-PFESAs with endocrine disrupting characteristics, and this exposureoutcome association is a challenge for risk assessment of Cl-PFESAs.
Our results were a community-based cross-sectional study with a relatively large population, which could reduce the occurrence of random errors. Moreover, we enrolled a comprehensive panel of potential confounders, consistent with previous studies, exploring the association between exposure to PFAS and overweight/obesity, including sociodemographic and behavior factors. Finally, we accounted for non-monotonic association of PFAS with the prevalence of overweight/obesity using a restricted cubic spline regression analysis and 2-piecewise binary logistic regression for statistical analyses.
However, several limitations should not be ignored in this study. First, there may be reverse causality in our crosssectional study design. However, it is difficult for PFAS to accumulate in adipose tissue with lipophobic properties (Aas et al. 2014;Fabrega et al. 2014;Wu et al. 2019). It is lipid mobilization, not fat, that could affect PFAS concentrations in blood and other tissues (Aas et al. 2014). Second, although we included key confounding variables in the models, we did not adjust for diets, which could change PFAS exposure due to different PFAS-containing food consumption, dietary quality, and energy intake (Christensen et al. 2016;Eriksen et al. 2011). Third, we did not consider the cocktail effect of various pollutants that may affect overweight/obesity, such as polychlorinated biphenyls (PCBs), phthalates and other POPs, as EDCs, which might act in a synergistic or antagonistic manner to impact PFAS metabolic disorder effects (Chamorro-Garcia and Veiga-Lopez 2021; Choi et al. 2021;Egusquiza and Blumberg 2020). Bayesian kernel machine regression (BKMR) models performed to analyze multiplechemical exposures will be necessary for further research in a larger population. Fourth, we conducted a common but controversial approach that imputed values below detection limits by LOD/√2 (Huynh et al. 2014;Richardson and Ciampi 2003;Schisterman et al. 2006), due to the detection proportion of 8:2 Cl-PFESA being 70.12%. However, our results were similar by using multiple imputing values below LOD for a repeat analysis (Table S12), although modestly stronger for the second quartile of 8:2 Cl-PFESA. Fifth, the population in this study was in China, hence future research conducted in other areas is needed.

Conclusion
Our study suggested that exposure to Cl-PFESAs displayed an inverted U-shaped association with the prevalence of overweight/obesity in Chinese adults. However, our results were based on a cross-sectional study with participants from only one city. Larger studies are needed to identify metabolic disorder effects and the active doses of Cl-PFESAs in the future. The alternatives to PFOS are not anticipated to be less toxic than PFOS, so the widespread use of Cl-PFESAs is a growing concern.