key: cord-0774553-n3upa0xn
authors: Ji, Junjun; Song, Lingyan; Wang, Jing; Yang, Zhiyun; Yan, Haotian; Li, Ting; Yu, Li; Jian, Lingyu; Jiang, Feixiang; Li, Junfeng; Zheng, Jinping; Li, Kefeng
title: Association Between Urinary Per- and Poly-fluoroalkyl Substances and COVID-19 Susceptibility
date: 2021-03-19
journal: Environ Int
DOI: 10.1016/j.envint.2021.106524
sha: a5a07ba8dd89a49333673b21b1637f024441b09e
doc_id: 774553
cord_uid: n3upa0xn

BACKGROUND AND OBJECTIVE: The growing impact of the COVID-19 pandemic has heightened the urgency of identifying individuals most at risk of infection. Per- and poly-fluoroalkyl substances (PFASs) are manufactured fluorinated chemicals widely used in many industrial and household products. The objective of this case-control study was to assess the association between PFASs exposure and COVID-19 susceptibility and to elucidate the metabolic dysregulation associated with PFASs exposure in COVID-19 patients. METHODS: Total 160 subjects (80 COVID-19 patients and 80 symptom-free controls) were recruited from Shanxi and Shandong provinces, two regions heavily polluted by PFASs in China. Twelve common PFASs were quantified in both urine and serum. Urine metabolome profiling was performed by liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS). RESULTS: In unadjusted models, the risk of COVID-19 infection was positively associated with urinary levels of perfluorooctanesulfonic acid (PFOS) (Odds ratio: 2.29 [95% CI: 1.52 - 3.22]), perfluorooctanoic acid (PFOA) (2.91, [1.95 - 4.83], and total PFASs (∑ (12) PFASs) (3.31, [2.05 - 4.65]). After controlling for age, sex, body mass index (BMI), comorbidities, and urine albumin-to-creatinine ratio (UACR), the associations remained statistically significant (Adjusted odds ratio of 1.94 [95% CI: 1.39 - 2.96] for PFOS, 2.73 [1.71 - 4.55] for PFOA, and 2.82 [1.97 - 3.51] for ∑ (12) PFASs). Urine metabolome-PFASs association analysis revealed that 59% of PFASs-associated urinary endogenous metabolites in COVID-19 patients were identified to be produced or largely regulated by mitochondrial function. In addition, the increase of PFASs exposure was associated with the accumulation of key metabolites in kynurenine metabolism, which are involved in immune responses (Combined β coefficient of 0.60 [95% CI: 0.25 to 0.95, P = 0.001]). Moreover, alternations in PFASs-associated metabolites in mitochondrial and kynurenine metabolism were also correlated with clinical lab biomarkers for mitochondrial function (serum growth/differentiation factor-15) and immune activity (lymphocyte percentage), respectively. CONCLUSION: Elevated exposure to PFASs was independently associated with an increased risk of COVID-19 infection. PFASs-associated metabolites were implicated in mitochondrial function and immune activity. Larger studies are needed to confirm our findings and further understand the underlying mechanisms of PFASs exposure in the pathogenesis of SARS-CoV2 infection.

The coronavirus disease 2019 caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has rapidly evolved into a global pandemic after its first report in Wuhan, China, in December 2019 . There is clear evidence of high heterogeneity in COVID-19 susceptibility (probability of being infected) (Viner et al., 2020) . The growing impact of the pandemic intensifies the need for identifying individuals most at risk of infection. Such information will not only be useful for public health experts, but also for individuals seeking to assess their own personalized risk. Prior studies have hypothesized that differences in susceptibility are associated with age , sex (Haitao et al., 2020) , BMI (Jung et al., 2020) , genetic factors (Anastassopoulou et al., 2020) and certain comorbidities such as obesity, type 2 diabetes, cardiovascular diseases (CVDs) (Aung et al., 2020) . However, COVID-19 is a new disease, and more work is urgently needed to determine if there are other factors that increase a person's risk to contract the infection. Little information is available regarding the impact of environmental exposures on COVID-19 susceptibility.

Per-and poly-fluoroalkyl substances (PFASs) are synthetic chemicals containing fluorinated carbon chains with different functional groups. Due to their excellent thermal and chemical stability, hydrophobic and oleophobic properties, PFASs have been widely used as coatings in many consumer products such as disposable food packaging, cookware, carpets, furniture, and more (Susmann et al., 2019) . The widespread use of PFASs has resulted in the ubiquitous detection of these chemicals in various environments, including drinking water, air, soil, crops, seafood, and wildlife Liu et al., 2019; Wang et al., 2020) . Depending on the perfluoroalkyl chain length and trophic position, some PFASs species such as perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) have long biological half-lives (3.5 years for PFOA and 4.8 years for PFOS in humans) (Perez et al., 2013) . It was reported that over 85% of people in the United States and China have detectable PFOS and PFOA in their blood (Li et al., 2013; Pan et al., 2010; Pelch et al., 2019; Tian et al., 2018) .

Epidemiological studies have indicated that PFASs exposure is linked to the increased risks of cancer, liver damage, and diabetes Pelch et al., 2019) . Additionally, toxicological studies underscore the immunotoxic potential of PFASs, which may adversely affect the immune responses to infectious diseases and vaccination (Chang et al., 2016; DeWitt et al., 2019; Zeng et al., 2019) . For instance, several studies found that exposure to PFASs was associated with increased risks of respiratory tract infections, otitis media, respiratory syncytial virus (RSV), and pneumonia (Ait Bamai et al., 2020; Granum et al., 2013; Impinen et al., 2018) . Decreased antibody responses to booster, measles, mumps, and rubella (MMR) vaccination had also been observed in individuals exposed to PFASs (Grandjean et al., 2012; Kielsen et al., 2016; Timmermann et al., 2017) .

Metabolites in biological fluids are the intermediate or end products of metabolism, which may reflect the final consequences of the functional changes in response to external exposures . Identifying these metabolic changes by next-generation metabolomics might provide insights into the mechanisms underlying exposure-related diseases (Han et al., 2018) . Regarding PFASs, the recent metabolomic analysis revealed that PFASs exposure might cause metabolic perturbations in phospholipid and amino acid metabolism, thereby contributing to increased risk and pathogenesis of diabetes and other metabolic disorders (Alderete et al., 2019; Kingsley et al., 2019; McGlinchey et al., 2020) . In addition, metabolites play important roles in the pathogenesis of infectious diseases and host immune responses. On the one hand, small molecules from the host metabolism are essential for viral infection and replication because they provide building blocks that a rapidly proliferating virus requires to assemble its nucleic acids, proteins (including capsid proteins), and membrane (Gonzalez Plaza et al., 2016; Thomas et al., 2020) . Dysregulation of amino acid metabolism, tricarboxylic acid cycle (TCA) and fatty acid oxidation had been characterized in patients with RSV and dengue virus infection (Shahfiza et al., 2017; Turi et al., 2018) . In a recent study, COVID-19 infection was reported to be associated with increased tryptophan catabolism . On the other hand, metabolites such as organic acids, short-chain fatty acids, bile acids, and tryptophan metabolites fuel and regulate the maturation of immune responses (Ganeshan and Chawla 2014; Michaudel and Sokol 2020) . Collectively, results from these studies suggested that PFASs exposure might be associated with perturbations in metabolic pathways which are implicated in viral infection.

In this cohort study of 160 participants from two regions with high PFASs pollution in China, we, ( ⅰ ) quantified urinary levels of PFASs in these subjects and examined the associations between PFASs exposure and COVID-19 infection, ( ⅱ ) conducted metabolomic analysis to examine the metabolic differences in urine metabolome between symptom-free controls, and COVID-19 patients, (ⅲ) performed a metabolome-wide association study coupled with pathway enrichment analysis to find out the metabolic abnormities in COVID-19 patients associated with urinary PFASs concentrations (Fig. 1a for the experiment design). We hypothesized that PFASs exposure would be associated with alternations in key metabolic pathways involved in mitochondrial metabolism and immune responses, which may be associated with increased COVID-19 risk.

The brief methods were described below, and the details can be found in the supplemental materials.

COVID-19 patients were recruited from January 7, 2020 to March 5, 2020 in Shanxi and Shandong, two provinces with high PFASs exposure and incidence of COVID-19 in China Xie et al., 2013; Zhang et al., 2020a) . Details regarding PFASs and incidence of COVID-19 in Shandong and Shanxi provinces were described in Supplemental Methods. The diagnosis of COVID-19 was made by SARS-CoV-2 nucleic acid testing of nasopharyngeal swabs in the COVID-19 testing centers at Heping Hospital Affiliated to Changzhi Medical College, Shanxi, and Yantai Yuhuangding Hospital, Shandong. COVID-19 disease severity was judged according to the guidelines for the diagnosis and management of COVID-19 patients (7 th edition) by the National Health Commission of China. All COVID-19 patients in our study had mild clinical symptoms such as fever, fatigue, dry cough, muscle pain, and sore throat, and no signs of pneumonia on chest CT imaging. They were quarantined and given general supportive treatment including bed rest, adequate nutrition, water and electrolyte balance, and intensively monitoring of vital signs. SARS-CoV-2 negative controls were recruited from the aggressive contact tracing for all potential contacts of the confirmed COVID-19 patients, which is part of the authorities' COVID-19 surveillance (Xing et al., 2020 

Paired blood and morning urine samples were collected after the subjects were tested positive or negative for SARS-CoV-2. Urine samples were collected and centrifuged at 1500 g for 10 min to remove any cellular debris. Whole blood samples were drawn in lithium heparin tubes.

Additionally, blood samples were collected into 4 mL BD vacutainer serum separator tubes (SSTs) for serum. Serum was obtained by allowing whole blood to coagulate for 30 min at room temperature before centrifugation at 1500 g for 10 min at 4°C. All whole blood samples were refrigerated until analyzed. Urine and sera samples were aliquoted into screw-capped cryovials and stored at -80 °C before analysis.

Lymphocyte percentage (LYP, %) in whole blood samples was analyzed using an automated hematology analyzer XE-5000 (Sysmex, Japan). Serum growth and differentiation factor 15 (GDF-15) was quantified using human GDF-15 ELISA kits (Abcam, CA, USA) according to the manufacturer's instructions. Urinary creatinine was measured based on a modified Jaffe reaction using urinary creatinine assay kits obtained from Cayman Chemical (MI, USA). Urinary albumin was determined by the immunoturbidimetry method using human albumin ELISA kits (Abcam, by LC-MS/MS. Chromatographic separation was conducted on a 100 mm × 4.6 mm, 3.5 µm XBridge amide column (Waters, USA) held at 25 °C. A total of 330 of the 462 targeted metabolites were measurable in all urine samples. Metabolomic data were normalized using urinary creatinine concentration for comparative analysis (Khamis et al., 2018) . The quality control and analytical performance were described in the supplemental materials.

The continuous variables are mean ± standard deviation (SD) or as the median and interquartile range (IQR) (Skewed data). The categorical characteristics are described as numbers (%). PFASs levels were log2 transformed prior to statistical analysis. Missing data due to lower limit of detection (LLOD) were estimated using multiple imputation (m = 5) in IVEware v0.3 and replaced with random values between 0 and LLOD with the same probabilistic distribution of the observed PFASs concentrations (Table 2 for the detection rate of PFASs) (Harel et al., 2014; Lubin et al., 2004) . Differences for continuous variables between controls, and COVID-19 patients were analyzed using either Student's t test or Mann-Whitney U test (Skewed data). The categorical data were analyzed using chi-square test. P < 0.05 was statistically significant. The relationship between PFASs in serum and the matched urine samples was calculated using Spearman's rank order correlation.

PFASs levels were log 2 transformed and standardized by z-scores before regression analysis.

The logistic regression analyses were performed on COVID-19 patients (all mild cases) and controls to examine the associations between urinary PFASs and COVID-19 susceptibility. Both univariate and multivariate models were used to calculate odds ratios and 95% confidence intervals (CIs) of COVID-19 susceptibility associated with a 1-SD increment in PFASs levels (ng/g creatinine). Confounding factors for inclusion in multivariable models were selected based on univariable logistic regression analysis (P < 0.1), correlations with PFASs or the evidence from the literature . Finally, age, gender, body mass index (BMI), diabetes, cardiovascular diseases (CVDs), and kidney health biomarker UACR were included in the multivariable model.

Metabolomic data were log2 transformed before metabolomic analysis. Partial least squares discriminant analysis (PLS-DA) was performed in MetaboAnalyst 5.0. The validation of the PLS-DA model was performed via the leave-one-out cross-validation (LOOCV) method. Metabolites with PLS-DA variable importance in projection (VIP) scores > 1.5 were statistically significant.

Metabolic pathway analysis was performed using a custom Python script with an in-house library for human metabolism. The associations between PFASs and urinary endogenous metabolites were evaluated using correlation network analysis with Spearman's rank correlation in Cytoscape 3.8.2.

The correlation threshold was set as 0.2, P value < 0.05 and false discovery rate (FDR) < 20%. We also performed the multiple linear regression analysis to further confirm the associations between the exposure of PFASs and changes in metabolism. The effect sizes (Beta) and P values were adjusted for confounders described in the multivariate logistic regression model. The summary effect of PFASs exposure on a metabolic pathway was meta-analyzed using the random-effects model. Spearman's rank correlation was conducted to determine the correlations between metabolite levels and clinical laboratory markers of mitochondrial metabolism (serum GDF-15) and immune function (lymphocyte percentage, %).

All general statistical analyses were performed in GraphPad Prism 9.0 (San Diego, CA) unless specified in the methods and legends.

All the participants in this study are Chinese from Shanxi (n = 80) and Shandong (n = 80)

provinces. None of them worked in higher-risk COVID-19 occupations such as frontline workers or occupational workers in fluorochemical plants. The urinary PFASs levels in the subjects from Shandong were not statistically different from those in the subjects from Shanxi (Fig. S1 ). The characteristics of the participants for symptom-free healthy controls (n = 80) and COVID-19 patients (n = 80) are shown in symptoms before COVID-19 testing and sample collection was 2 days (IQR: 0 -4). The brief experiment design for this study is shown in Fig. 1a .

We first quantified the urinary levels of PFASs in all the samples by LC-MS/MS. The targeted

PFASs compounds, their retention times (RTs), and MS/MS parameters are listed in Table S1 . Out of 12 PFASs measured, perfluorooctanesulfonic acid (PFOS) and perfluorooctanoic acid (PFOA)

were detected in the urine samples of all participants (Table 2) . Other PFASs were detected in 5.00%

-86.3% of HC samples, and the detection frequency was not statistically different between HC and COVID-19 groups (all P > 0.05) ( Table 2 ).

The urinary levels of PFASs and total PFASs (∑ (12) PFASs) in HC and COVID-19 patients are shown in Fig. 1b -1d and controls and 67.6 ng/g creatinine (IQR: 42.5 -103.7) in COVID-19 patients (P < 0.05) (Fig. 1b) .

Similarly, COVID-19 patients had significantly higher median levels of PFOA and total PFASs in urine than those in symptom-free controls (P < 0.05) ( Fig. 1c and 1d ).

Correlations between urinary and serum PFOS and PFOA levels: Previous animal studies have demonstrated that urine was the primary elimination route for PFOA and PFOS in rats (Cui et al., 2010) . We then analyzed the correlations between urinary levels of PFOS and PFOA and their corresponding serum concentrations. The median serum PFOS was 19.1 ng/mL (IQR: 14.4 -26.1) and the median serum PFOA was 3.87 ng/mL (IQR: 2.97 -5.32 ) A good correlation was observed for PFOA between urine and serum concentrations (Spearman's correlation coefficient r = 0.51, P < 0.001) (Fig. 1e) . PFOS levels in urine were also significantly correlated with concentrations in serum (Spearman's correlation coefficient r = 0.67, P < 0.001) (Fig. 1f) . These results suggested that urinary PFASs might be good indicators of the internal dose and used for evaluating human exposure to PFASs.

PFASs showed moderate to strong pairwise correlations, with the strongest correlation observed between PFOA and PFOS (Spearman's correlation coefficient r = 0.73) (Fig. 1g) . We also found a weak correlation between PFOA and gender (Spearman r = -0.22, P < 0.05) (Fig. 1f) . Females tend to have lower levels of PFOA and PFOS than those in males. In addition, the number of subjects with diabetes was positively correlated with urinary PFOS and PFOA levels.

We next evaluated the associations between PFASs exposure and COVID-19 infection using both unadjusted and multivariate adjusted models. (Fig. 1h ).

Adjusted models: Confounding factors including age, gender, BMI, and diabetes were selected for adjusted models based on univariate logistic regression analysis (P < 0.1), or Spearman's rank correlations with urinary PFASs (P < 0.05). Urine albumin-to-creatinine ratio (UACR) is a welldocumented biomarker for renal function (Lopez-Giacoman and Madero 2015). Since renal function may affect the excretion of PFASs (Lin et al., 2021) , we also added UACR in the adjusted models. The adjusted associations of PFASs exposure with COVID-19 susceptibility are presented in Fig.1h and (Fig. 1h) . Other

PFASs were not significantly associated with COVID-19 susceptibility after adjustment for confounders (Table 3) .

We then performed metabolomic analysis to explore the differences in urine metabolome between control and COVID-19 patients. Out of 462 metabolites targeted, 335 were detected in all samples. The median intra-day and inter-day coefficient of variation (CVs) of metabolomics quality control (QC) samples were 10.2% and 11.5%, respectively, which indicated excellent reproducibility for the measured metabolites (Supplemental materials and methods).

Metabolomic profiling, followed by multivariate PLS-DA, revealed dramatic metabolic differences in urine metabolome between HC and COVID-19 patients (Fig. 2a) . Since PLS-DA is a supervised model, we further verified the reliability of the PLS-DA model using the leave-oneout cross-validation (LOOCV) method (Fig. S2) . Variable importance in projection (VIP) analysis showed the significant discriminating urinary metabolites between the control and COVID-19

group. The top 25 discriminating metabolites are shown in Fig. 2 , and the full list of 54 differential metabolites between the two groups is in Table S2 . 

Correlation network analysis was then conducted to identify what metabolic alterations in urine metabolome of COVID-19 patients were associated with PFASs exposure. Out of 54 differential urinary metabolites in COVID-19 patients (HC versus COVID-19) (Table S2) , 49 were significantly correlated with urine concentrations of PFOA, PFOS, and ∑ (12) PFASs at coefficient r > 0.2 or < -0.2, FDR ≤ 20%, and P < 0.05 (Fig. 3a) . Pathway enrichment analysis revealed that the metabolites correlated with PFASs exposure were primarily from mitochondrial metabolism, kynurenine, eicosanoid, glycolysis, niacin, bioamine and neurotransmitter metabolism (Fig. S3) .

We next used multiple linear regression models to further confirm the relationship between PFASs exposure and metabolic disturbances in COVID-19 patients. After controlling for the confounding variables, the associations between ∑ (12) PFASs and urine endogenous metabolites from the metabolic pathways of kynurenine and eicosanoid and mitochondrial metabolism remained statistically significant (Fig. 3b) . Similar associations were observed for PFOS and PFOA with urine metabolites in COVID-19 patients ( Fig.S3 and S4 ). In detail, we observed three major patterns: ( ⅰ ) 13 PFASs-associated urine metabolites were identified to be produced or largely regulated by mitochondrial function (Fig. 3b) PFASs after confounders adjustment) (Fig. 3b, Fig. S4 and S5 ). The interconnections of metabolites and their metabolic pathways associated with PFASs exposure in COVID-19 patients are summarized in Fig. 3c . Interestingly, metabolic dysregulation of PFASs-associated metabolites and biochemical pathways were also reported to be connected to disease pathogenesis and the host defense response to COVID-19 infection (Ayres 2020).

We next analyzed whether urinary metabolites associated with PFASs were also correlated with lab biomarkers for mitochondrial function and immune activity. Serum GDF-15 is a known biomarker of mitochondrial function (Montero et al., 2016) . We found that the levels of urinary metabolites in mitochondrial metabolism such as L-Acetylcarnitine, Aconitic acid, and hydroxypropionic acid correlated well with the concentration of serum GDF-15 (Spearman's rank correlation, all P values < 0.01 or 0.05) (Fig. 4a -4c ). Lymphocyte percentage (LYP, %) in whole blood is one of the indicators for the changes in immune function in COVID-19 patients . There were significant negative correlations between urinary metabolites from kynurenine pathway and LYP (Fig. 4d -4f) .

The potential interactions between environmental toxicants exposure and COVID-19 remain largely unknown. This study represented the first attempt to explore the associations between PFASs in urine and COVID-19 susceptibility. Our results indicate increased risks for COVID-19

infection with high urinary PFASs after adjusting for potential confounding factors including age, gender, number of diabetes, CVDs, and urine albumin-to-creatinine ratio. Using metabolome-wide association analysis, we found that urinary endogenous metabolites associated with PFASs are involved in essential metabolic pathways underlying the pathophysiology of COVID-19 infection including mitochondrial metabolism, eicosanoids, and kynurenine pathways (Fig. 3c ).

Since 2014, China has become the exclusive producer and supplier of PFOS globally, and the largest producer of PFOA in the world. Both Shandong and Shanxi provinces had relatively high industrial emission of PFASs in China. It is not a surprise that we detected PFOS and PFOA in the urine of all subjects. There is no previously published data for PFASs exposure from the same regions. However, the urinary levels of PFOS and PFOA we measured in our study were dramatically higher than the values reported in 2014 in the general population of Tianjin (Zhang et al., 2015) . Out of 12 common PFASs measured in our study, the associations with COVID-19

infection are significant for PFOS, PFOA, and total PFASs after adjustment for confounders.

Urinary PFOS and PFOA are highly correlated, which could likely lead to the positive associations observed for both PFOS and PFOA on COVID susceptibility. This observation highlighted the needs to treat PFASs as a class of chemicals rather than individual chemical (Cousins et al., 2020; Kwiatkowski et al., 2020) .

Urine is the biological fluid of choice in our study because of the ease of collection, and unlike blood, it contains rich information on the functionality of organs at the metabolic level and provides an integrated estimate of exposure overtime, making it valuable for a broad range of biological investigations (Miller et al., 2019) . Regarding PFASs, according to the outcomes of animal studies, urine is the major excretion route for PFOA and PFOS in rats (Cui et al., 2010) . A human study reported that for all PFASs except PFUnA, levels in urine correlated positively with levels in the blood . Consistent with the previous study, we also identified significant correlations between serum and the corresponding urine PFOS and PFOA concentrations in our cohort. Together, these findings suggest that urinary concentrations might be good indicators of the internal dose for PFASs, and this less invasive strategy can therefore be used in future epidemiological and biomonitoring studies. However, it is worth mentioning that urinary PFASs might be influenced by kidney function. Those who had higher and longer exposure might have impaired kidney function and thus cause larger variation in urinary PFASs levels than those in serum (Lin et al., 2021) . Our urine metabolomics, in combination with metabolomewide association analysis, identified a strong urine metabolic signature associated with PFASs exposure in COVID-19 patients, which can be traced to altered mitochondrial regulation of cellular redox, signaling, and energy. Mitochondria have been considered as a nexus point for the convergence of environmental stress signaling (Youle and van der Bliek 2012). In our study, we found that both Krebs cycle-related metabolites (succinic acid, pyruvic acid, and aconitic acid) Our study demonstrated positive associations between PFASs exposure and several key metabolites (L-Kynurenine, 3-hydroxyanthranilic acid, hydroxykynurenine, and Nformylkynurenine) in the kynurenine pathway, independent of confounders. The kynurenine pathway is highly regulated in the immune system, where it promotes immunosuppression in response to inflammation or infection (Cervenka et al., 2017) . For example, the accumulation of kynurenine and hydroxykynurenine inhibits T-cell proliferation at G1 phase and induces natural killer cells apoptosis (Wu et al., 2018) . The immunotoxic effects of PFASs had been previously implicated in several epidemiological studies of childhood immunization (Grandjean et al., 2012; Grandjean et al., 2017) , in which high PFASs exposure was correlated with decreased immune responses to childhood vaccines. Importantly, alteration of kynurenine metabolism was also observed in a recent serum metabolomic analysis of COVID-19 patients .

Eicosanoids, including prostaglandins, thromboxanes, and isoprostanes, are bioactive lipid mediators derived from polyunsaturated fatty acids (PUFAs). Eicosanoid signaling, similar to cytokine signaling, regulates a diverse set of inflammatory processes, primarily as a proinflammatory component of the innate immune responses (Dennis and Norris 2015) . Here, we observed, for the first time, that increased levels of eicosanoids were associated with elevated PFASs exposure.

Our study has several limitations. First, our study is limited by the small sample size. Future studies are warranted to confirm the findings in large cohorts. Second, even though the controls were tested negative for SARS-CoV-2 through nasopharyngeal swab nucleic acid amplification, we can not completely exclude the potential false negative results of PCR testing. Third, the observational nature of the study precludes inferences of causality. PFAS-associated metabolic signatures in urine should be interpreted as potential biomarkers rather than causal mechanisms of PFASs exposure. Finally, COVID-19 is a new public concern. The possibility of residual confounding from unknown variables cannot be excluded.

In summary, we observed significantly higher risks of SARS- analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

The authors declared no conflict of interest.

Supplementary data to this article can be found online at Tables   Table 1 Summary of characteristics of the study participants. 3 (3.75%) 4 (5.00%) The continuous variables are mean and standard deviations (mean ± SD) or median and interquartile range (IQR). The categorical characteristics are described as numbers (%). Differences between groups were analyzed by Student's t test, Mann-Whitney U test or chi-square test (compare the proportions). All CVOD-19 patients had mild clinical symptoms such as fever, fatigue, dry cough, muscle pain and sore throat and no signs of pneumonia on chest CT imaging. Abbreviations: HC: Healthy control; BMI: Body mass index; WBC: White blood cell; hpf: High power field; RBC: Red blood cell; UACR: urine albumin-to-creatinine ratio. for PFHxS represent the risks of SARS-CoV-2 infection (susceptibility) per standard deviation (SD) of PFHxS increment (ng/g urinary creatinine).The multiple logistic regression models were adjusted for potential covariates including age, gender, body mass index (BMI), diabetes, cardiovascular diseases (CVDs), and urine albumin-to-creatinine ratio (UACR). Fig. 1 . Elevated exposure of perfluorinated alkyl substances (PFASs) is significantly associated with higher risks of SARS-CoV-2 infection. 

Supplemental materials and methods.

The targeted PFASs, their retention times (RTs), and MS/MS parameters.

The differential urinary metabolites between HC and COVID-19 patients revealed by PLS-DA followed by VIP analysis. 

 The growing toll of the COVID-19 pandemic intensifies the need for identifying individuals who are more vulnerable to be infected.

 PFASs exposure was associated with an increased risk of COVID-19 infection.

 PFASs-associated metabolites were implicated in mitochondrial stress signaling and immune responses.

 PFASs-associated urine metabolites were correlated with serum GDF-15 and blood lymphocyte percentage.

Effect of prenatal exposure to per-and polyfluoroalkyl substances on childhood allergies and common infectious diseases in children up to age 7 years: The Hokkaido study on environment and children's health

Increased urinary excretion of carnitine and acylcarnitine by mercuric chloride is reversed by 2,3-dimercapto-1-propanesulfonic acid in rats

Perfluoroalkyl substances, metabolomic profiling, and alterations in glucose homeostasis among overweight and obese Hispanic children: A proof-of-concept analysis

Human genetic factors associated with susceptibility to SARS-CoV-2 infection and COVID-19 disease severity

Causal Inference for Genetic Obesity, Cardiometabolic Profile and COVID-19 Susceptibility: A Mendelian Randomization Study. Front Genet

A metabolic handbook for the COVID-19 pandemic

Kynurenines: Tryptophan's metabolites in exercise, inflammation, and mental health

A critical review of perfluorooctanoate and perfluorooctanesulfonate exposure and immunological health conditions in humans

Impact of short-term exposure to fine particulate matter air pollution on urinary metabolome: A randomized, double-blind, crossover trial

The high persistence of PFAS is sufficient for their management as a chemical class

Excretion of PFOA and PFOS in male rats during a subchronic exposure

Eicosanoid storm in infection and inflammation

Exposure to per-fluoroalkyl and polyfluoroalkyl substances leads to immunotoxicity: epidemiological and toxicological evidence

Metabolic regulation of immune responses

Role of metabolism during viral infections, and crosstalk with the innate immune system

Serum vaccine antibody concentrations in children exposed to perfluorinated compounds

Estimated exposures to perfluorinated compounds in infancy predict attenuated vaccine antibody concentrations at age 5-years

Pre-natal exposure to perfluoroalkyl substances may be associated with altered vaccine antibody levels and immune-related health outcomes in early childhood

COVID-19 and sex differences: mechanisms and biomarkers

Next-generation metabolomics in the development of new antidepressants: using albiflorin as an example

The use of multiple imputation for data subject to limits of detection. Sri

Prenatal exposure to perfluoralkyl substances (PFASs) associated with respiratory tract infections but not allergy-and asthma-related health outcomes in childhood

Association between body mass index and risk of COVID-19: A nationwide case-control study in South Korea

Comparative analysis of creatinine and osmolality as urine normalization strategies in targeted metabolomics for the differential diagnosis of asthma and COPD

Antibody response to booster vaccination with tetanus and diphtheria in adults exposed to perfluorinated alkylates

Metabolomics of childhood exposure to perfluoroalkyl substances: a crosssectional study

Scientific basis for managing PFAS as a chemical class

Can nail, hair and urine be used for biomonitoring of human exposure to perfluorooctane sulfonate and perfluorooctanoic acid?

Occurrence of perfluorinated compounds in agricultural environment, vegetables, and fruits in regions influenced by a fluorine-chemical industrial park in China

Per-and polyfluoroalkyl substances and kidney function: Follow-up results from the Diabetes Prevention Program trial

Risk factors associated with COVID-19 infection: a retrospective cohort study based on contacts tracing

Multiple crop bioaccumulation and human exposure of perfluoroalkyl substances around a mega fluorochemical industrial park, China: Implication for planting optimization and food safety

Biomarkers in chronic kidney disease, from kidney function to kidney damage

Epidemiologic evaluation of measurement data in the presence of detection limits

Prenatal exposure to perfluoroalkyl substances modulates neonatal serum phospholipids, increasing risk of type 1 diabetes

The Gut Microbiota at the service of immunometabolism

Real-time health monitoring through urine metabolomics

GDF-15 is elevated in children with mitochondrial diseases and is induced by mitochondrial dysfunction

Lipid-induced mitochondrial stress and insulin action in muscle

Metabolic signatures of the exposomequantifying the impact of exposure to environmental chemicals on human health

Concentrations of perfluorinated compounds in human blood from twelve cities in China

PFAS health effects database: Protocol for a systematic evidence map

Accumulation of perfluoroalkyl substances in human tissues

Metabolomics approach for multibiomarkers determination to investigate dengue virus infection in human patients

Proteomic and metabolomic characterization of COVID-19 patient sera

Dietary Habits Related to food packaging and population exposure to PFASs

Lymphopenia predicts disease severity of COVID-19: a descriptive and predictive study

COVID-19 infection alters kynurenine and fatty acid metabolism, correlating with IL-6 levels and renal status

Determinants of plasma concentrations of perfluoroalkyl and polyfluoroalkyl substances in pregnant women from a birth cohort in

Association between perfluoroalkyl substance exposure and asthma and allergic disease in children as modified by MMR vaccination

Using urine metabolomics to understand the pathogenesis of infant respiratory syncytial virus (RSV) infection and its role in childhood wheezing

Susceptibility to SARS-CoV-2 infection among children and adolescents compared with adults: A systematic review and meta-analysis

Review on perfluoroalkyl and polyfluoroalkyl substances (PFASs) in the Chinese atmospheric environment

Indoleamine 2, 3-dioxygenase regulation of immune response (Review)

Industrial source identification and emission estimation of perfluorooctane sulfonate in China

Rapid response to an outbreak in Qingdao

Mitochondrial fission, fusion, and stress

A positive/negative ion-switching, targeted mass spectrometry-based metabolomics platform for bodily fluids, cells, and fresh and fixed tissue

Assessing the human health risks of perfluorooctane sulfonate by in vivo and in vitro studies

PFOS and PFOA in paired urine and blood from general adults and pregnant women: assessment of urinary elimination

Comparison of spatiotemporal characteristics of the COVID-19 and SARS outbreaks in mainland China

Biomonitoring of perfluoroalkyl acids in human urine and estimates of biological half-life

Succinate accumulation induces mitochondrial reactive oxygen species generation and promotes status epilepticus in the kainic acid rat model

A novel coronavirus from patients with pneumonia in China

PFOA (c), and total PFASs (d) in HC (n = 80), and COVID-19 patients (n = 80). The box-whisker plots show the medians (middle line) and the first and third quartiles (boxes), whereas the whiskers are the maximum and minimum values. Mann-Whitney U test was performed. *P < 0.05. (e) PFOA levels in urine were significantly correlated with levels in serum. Data were log 2 transformed, and Spearman's rank correlation was performed

Spearman's rank-order correlation was conducted, and Spearman's correlation coefficients were plotted. (h) The associations between PFASs exposure and COVID-19 infection using unadjusted and multivariable adjusted models. PFASs values were log 2 transformed before regression analysis. Odds ratios represent the risks of COVID-19 infection per log 2 standard deviation (SD) of urinary PFASs increment. Log 2 SD was 1.02, 0.93, and 0.62 ng/g urinary creatinine, for PFOS, PFOA, and total PFASs, respectively. The multiple logistic models were adjusted for potential covariates including age, gender, body mass index (BMI), diabetes, cardiovascular diseases (CVDs), and urine albumin-tocreatinine ratio (UACR)

The combined effect of PFASs exposure on a metabolic pathway was meta-analyzed using the random-effects model. (c) The summary of metabolic dysregulations associated with PFASs exposure in COVID-19 patients. The red dots indicate the metabolites with significant positive associations with total PFASs in adjusted multiple linear regression models, while the pink dots were the metabolites without significant associations with PFASs. The gray dots are the metabolites that were not detected or targeted. Abbreviations: PFOS: Perfluorooctanesulfonic acid; PFOA: Perfluorooctanoic acid; 3-HAA: 3-Hydroxyanthranilic acid; N-MNA: N-Methylnicotinamide; 4-AHA: 4-Aminohippuric acid; VMA: Vanillylmandelic acid; Xao: Xanthosine; Xan: Xanthine

TCA: tricarboxylic acid cycle; FAO: fatty acid oxidation

Formal analysis

Funding acquisition

Writing -review & editing