Associations of thiocyanate, nitrate, and perchlorate exposure with dyslipidemia: a cross-sectional, population-based analysis

The aim of this study was to assess the associations of urinary thiocyanate, nitrate, and perchlorate concentrations with dyslipidemia, individually and in combination, which has not previously been studied. Data from the 2001–2002 and 2005–2016 National Health and Nutrition Examination Surveys (NHANES) were analyzed in this cross-sectional study. The dependent variables were continuous serum lipid variables (triglycerides [TG], total cholesterol [TC], low-density lipoprotein cholesterol [LDL-C], high-density lipoprotein cholesterol [HDL-C], non-HDL-C, and apolipoprotein B [Apo B]) and binary serum lipid variables, with the latter reflecting dyslipidemia (elevated TG, ≥ 150 mg/dL; elevated TC, ≥ 200 mg/dL; elevated LDL-C, ≥ 130 mg/dL; lowered HDL-C, < 40 mg/dL in men and < 5 0 mg/dL in women; elevated non-HDL-C, ≥ 160 mg/dL; and elevated Apo B, ≥ 130 mg/dL). Multivariate logistic, linear, and weighted quantile sum (WQS) regression analyses were used to explore the associations of thiocyanate, nitrate, and perchlorate with the continuous and binary serum lipid variables. The linearity of the associations with the binary serum lipid variables was assessed using restricted cubic spline (RCS) regression. A total of 15,563 adults were included in the analysis. The multivariate linear and logistic regression analyses showed that thiocyanate was positively associated with multiple continuous (TG, TC, LDL-C, non-HDL-C, and Apo B, but not HDL-C) and binary (elevated TG, TC, LDL-C, and non-HDL-C) serum lipid variables, whereas perchlorate was negatively associated with elevated LDL-C. Multivariate RCS logistic regression revealed a linear dose–response relationship between thiocyanate and elevated TG, TC, LDL-C, non-HDL-C, and Apo B, but a nonlinear relationship with lowered HDL-C (inflection point = 1.622 mg/L). WQS regression showed that a mixture of thiocyanate, nitrate, and perchlorate was positively associated with all binary serum lipid variables except for Apo B. Our findings indicate that urinary thiocyanate, nitrate, and perchlorate concentrations, individually and in combination, were associated with dyslipidemia.


Introduction
Cardiovascular disease (CVD) is the leading cause of mortality in every region of the world (Roth et al. 2020). Dyslipidemia is prevalent in general population, which increases the development and progression of atherosclerotic CVD (Virani et al. 2020). There is a significant correlation between serum total cholesterol (TC) and atherosclerotic CVD (Arnett et al. 2019). Low-density lipoprotein cholesterol (LDL-C) is the primary atherogenic type of cholesterol (Arnett et al. 2019). LDL is primarily responsible for transporting triglycerides (TG), and LDL-C can induce atherosclerosis (AS) (Arnett et al. 2019). Similarly, research suggests a relationship between apolipoprotein B (Apo B) and the development of AS (Marston et al. 2022). In contrast, high-density lipoprotein cholesterol (HDL-C) reduces the progression of AS (Barter and Genest 2019). Moreover, it is proposed in several dyslipidemia management guidelines that lowering non-HDL-C should be the primary therapeutic target (Pearson et al. 2021;Stone et al. 2014;Visseren et al. 2021). In addition to atherosclerotic CVD, dyslipidemia is related to hypertension, diabetes, valvular heart disease, chronic renal disease, metabolic syndrome, and nonalcoholic fatty liver disease (Bloomgarden 2004;Kane et al. 2021;Katsiki et al. 2016;Kotsis et al. 2018;Mata et al. 2021;Slinin et al. 2012). To successfully control the progression of AS and CVD, the detection and treatment of dyslipidemia at an early stage are crucial (Virani et al. 2020).
Thiocyanate, nitrate, and perchlorate are common in nature. Thiocyanate is a pseudohalide widely found by people (Chiang et al. 2016), which is a primary metabolite in cigarette smoke and is extensively utilized in furniture as well as electronic manufacturing (Jain 2016). Nitrates are employed as precursors of nitrites, which are widely used as agricultural fertilizers and food preservatives (Sindelar and Milkowski 2012). In human diets, drinking water and vegetables account for > 80% of the exposure to nitrate (Salehzadeh et al. 2020). Perchlorate is both a manufactured pollutant and a naturally occurring chemical in the environment (Dasgupta et al. 2005;Wan et al. 2015), which are widely used for military and industrial purposes and has been found in milk as well as vegetables (Dasgupta et al. 2005;Wan et al. 2015). There is emerging evidence that the general public may be exposed to these three toxins through many environmental and dietary sources. The concentration of urinary thiocyanate, nitrate, and perchlorate is extensively utilized as a well-established biomarker for biomonitoring and assessing individuals' internal exposure status (Liu et al. 2017;Mervish et al. 2016;Yu et al. 2022b;Zhu et al. 2021).
A growing body of in vivo and in vitro evidence shows that thiocyanate, nitrate, and perchlorate may have negative impacts on thyroid functions (García Torres et al. 2022;King et al. 2022;Leung et al. 2014;Willemin and Lumen 2017), cancers (Carvalho et al. 2021;Picetti et al. 2022;Said Abasse et al. 2022;Shiue 2015;Wang et al. 2022aWang et al. , 2018Zhang et al. 2018), oral health (Yu et al. 2022a, b), as well as asthma and allergic reactions (Bose et al. 2017;Nadif et al. 2014;Zhu et al. 2021). Furthermore, thiocyanate, nitrate, and perchlorate have been linked to CVD (Bondonno et al. 2021;Wang et al. 2022b) as well as its risk factors (obesity Ghasemi and Jeddi 2017;Zhu et al. 2019), high fasting glucose, glycated hemoglobin A1c, and insulin resistance (Liu et al. 2017;Parslow et al. 1997)). Environmental contaminants may play a significant role in the development of dyslipidemia, according to a growing number of studies (Gaio et al. 2019;He et al. 2018;Zhu et al. 2022). However, there is currently no evidence linking perchlorate, nitrate, or thiocyanate exposure to dyslipidemia risk. We hypothesized a relationship between thiocyanate, nitrate, and perchlorate exposure and dyslipidemia. We analyzed the associations of urinary thiocyanate, nitrate, and perchlorate concentrations with continuous and binary serum lipid variables (with the latter reflecting dyslipidemia) in the general adult population from the 2001-2002-2016.

Study population
NHANES are cross-sectional surveys with 2-year cycles; they use a multi-stage, cluster-sampling design to ensure nationally representative samples (https:// wwwn. cdc. gov/ nchs/ nhanes/ defau lt. aspx). The NHANES research protocols were approved by the National Center for Health Statistics (NCHS) Research Ethics Review Board, and all participants provided written informed consent.
We analyzed descriptive data from the 2001-2002 and 2005-2016 NHANES. We enrolled eligible participants with complete data on urinary thiocyanate, nitrate, and perchlorate. Exclusion criteria were as follows: (1) missing data on serum lipid variables, (2) taking prescribed medicine for lipids, (3) aged < 18 years, and (4) pregnancy.

Assessment of urinary thiocyanate, nitrate, and perchlorate
Urine samples were stored at -20 °C or less until examination. Thiocyanate, nitrate, and perchlorate concentrations in the urine were quantified using electrospray ion chromatography-tandem mass spectrometry. The limits of detection (LOD) of urinary thiocyanate, nitrate, and perchlorate were 0.05 μg/L, 0.70 mg/L, and 0.02 mg/L, respectively. All concentrations below the LOD were replaced by an LOD value divided by the square root of 2 (LOD/√2). The NHANES website provides details on the analytical processes employed (https:// wwwn. cdc. gov/ nchs/ data/ nhanes/ 2015-2016/ labme thods/ PERNT-PERNTS-I-MET-508. pdf).

Covariates
In NHANES, data collection was carried out using a standardized questionnaire during a household interview, two 24-h recall interviews (for assessing energy intake), and a medical evaluation (including urine tests based on urine spot samples and blood tests). We selected the following potential confounding covariates linked to dyslipidemia: age, gender, race, education level, smoking, alcohol use, poverty, body mass index (BMI), physical activity, energy intake, urinary creatinine, uric acid, estimated glomerular filtration rate (eGFR), diabetes mellitus, and hypertension.
The categories of race/ethnicity were "Mexican American," "other Hispanics," "non-Hispanic White," "non-Hispanic Black," and "others." Education levels were categorized as "below high school," "high school," and "above high school." Poverty was assessed based on the poverty income ratio (PIR), which was defined as PIR < 1 for a family. Individuals with a cotinine level > 14 ng/mL were classified as smokers (Jarvis et al. 1987). Individuals who consumed ≥ 12 alcoholic drinks in a single calendar year were considered alcohol users. Energy intake was calculated by averaging the above two values based on two 24-h recall interviews. According to the Compendium of Physical Activities (Ainsworth et al. 2000), physical activity was measured by metabolic equivalent (MET) levels and classified as inactive (no leisure-time physical activity), insufficiently active (moderate activity 1-5 times per week with MET 3-6 or vigorous activity 1-3 times per week with MET > 6), and active (those who had more moderate or vigorous activity than above) (Pate et al. 1995). Detailed information on the definition of physical activity is described in Supplementary Material. Uric acid in blood samples was measured via laboratory tests. The eGFR was computed using the chronic kidney diseaseepidemiology collaboration (CKD-EPI) equation (Levey et al. 2009). Descriptions of each variable are presented at https:// wwwn. cdc. gov/ Nchs/ Nhanes/ conti nuous nhanes/.

Statistical analysis
In accordance with National Center for Health Statistics recommendations, weighted analyses were conducted with the R package "survey" to provide reliable national estimates (https:// wwwn. cdc. gov/ nchs/ nhanes/ tutor ials/ defau lt. aspx). The participant characteristics are presented as median (interquartile range [IQR]) or frequency (percentage) for continuous and categorical variables, respectively. Urinary thiocyanate, nitrate, and perchlorate concentrations had skewed distributions and so underwent log2 transformation (for the continuous dependent variable analyses). They were also divided into four quartiles (for the categorical dependent variable analyses). Correlation coefficients between log2-transformed thiocyanate, nitrate, and perchlorate were estimated using Spearman rank correlation analysis. All statistical analyses were conducted in R (version 4.1.0), and p < 0.05 (two sided) was deemed significant.
Three sequential multivariate linear regression models (model 1 covariates: age, gender, race, and urinary creatinine; model 2 covariates: model 1 plus education level, poverty, smoking, and alcohol use; and model 3 covariates: model 2 plus BMI, energy intake levels, physical activity, eGFR, uric acid, diabetes, and hypertension) with increasing levels of adjustment for confounding variables were employed to explore the associations of urinary thiocyanate, nitrate, and perchlorate concentrations with the following continuous serum lipid variables: TG, TC, LDL-C, non-HDL-C, and Apo B, and HDL-C. Three sequential multivariate logistic regression models (model covariates: age, gender, race, education level, poverty, smoking, alcohol use, body mass index, energy intake levels, physical activity, eGFR, uric acid, urinary creatinine, diabetes, and hypertension) were also employed to explore the associations of urinary thiocyanate, nitrate, and perchlorate concentrations with the following binary serum lipid variables: elevated TG, TC, LDL-C, non-HDL-C, and Apo B, and lowered HDL-C. Sensitivity analyses were performed using urinary creatinine-adjusted thiocyanate, nitrate, and perchlorate in order to correct for differences in urine dilution in the urine spot samples (Ko et al. 2014).
Restricted cubic spline (RCS) logistic regression was utilized to investigate the dose-response relationships between log2-transformed thiocyanate, nitrate, and perchlorate and the binary serum lipid variables, using three knots (10th, 50th, and 90th percentiles). The threshold inflection point of the linear relationship was calculated using a segmented regression to fit the piecewise-linear associations of the independent variables with the dependent variables, as described previously .
As there were significant correlations among urinary thiocyanate, nitrate, and perchlorate, weighted quantile sum (WQS) regression was conducted using the "gWQS" R package to assess the associations of the mixture of urinary thiocyanate, nitrate, and perchlorate with binary and continuous serum lipid variables. For this analysis, each chemical (thiocyanate, nitrate, and perchlorate) was given a weight within a WQS index that indicated how much of a contribution each chemical had made to the overall association in the regression. Specifically, we established the WQS index according to deciles of urinary thiocyanate, nitrate, and perchlorate concentrations. We used 40% of the data as the test set and the remaining 60% as the validation set, with a total of 10,000 bootstrap samplings.

Participant characteristics
A total of 71,975 individuals participated in NHANES in 2001-2002 and 2005-2016, 42,874 of whom had missing data on urinary thiocyanate, nitrate, and/or perchlorate. We subsequently excluded participants with missing serum lipid data (n = 2660), those taking prescribed medicine for lipids (n = 2992), those aged < 18 years (n = 7340), and those who were pregnant (n = 546). This left 15,563 participants for the analyses of HDL-C, TC, and non-HDL-C; 7236 participants for the analyses of TG and LDL-C; and 6732 participants for the analysis of Apo B (Fig. 1). Table 1 presents the survey-weighted characteristics of the enrolled participants. The weighted sample had a median age of 42.0 years and 51.2% of individuals were female. The proportion of individuals with elevated TG (≥ 150 mg/ dL), TC (≥ 200 mg/dL), LDL-C (≥ 130 mg/dL), non-HDL-C (≥ 160 mg/dL), and Apo B (≥ 130 mg/dL) and lowered HDL-C (< 40 mg/dL in men and < 50 mg/dL in women) was 23.5%, 44.2%, 34.2%, 31.5%, 8.0%, and 29.3%, respectively.

Discussion
This is the first study to investigate the associations of urinary thiocyanate, nitrate, and perchlorate concentrations with continuous and binary serum lipid variables (to explore associations with dyslipidemia) in a sample of 15,563 adults from the general population of the USA. Our two-tiered strategy first involved analyzing individual chemicals (thiocyanate, nitrate, and perchlorate) and then a mixture of the three. The latter analysis allowed us to determine the associations of co-exposure to thiocyanate, nitrate, and perchlorate with continuous and binary serum lipid variables. We discovered the following: (I) multivariate linear regression revealed that urinary thiocyanate was significantly positively associated with multiple continuous serum lipid variables (TG, TC, LDL-C, non-HDL-C, and Apo B) but not HDL-C; (II) multivariate logistic regression revealed that urinary thiocyanate was significantly positively associated with multiple binary serum lipid variables (elevated TG, TC, LDL-C, and non-HDL-C) but not Apo B or HDL-C, whereas urinary perchlorate was significantly negatively associated with elevated LDL-C; (III) multivariate RCS logistic regression revealed a linear dose-response relationship between the continuous urinary thiocyanate variable and multiple binary serum lipid variables (elevated TG, TC, LDL-C, non-HDL-C, and Apo B) and a nonlinear dose-response relationship with lowered HDL-C (IP = 1.622 mg/L); and (IV)WQS regression showed that the mixture of thiocyanate, nitrate, and perchlorate was significantly positively associated with all forms of dyslipidemia, except for elevated Apo B. Environmental and dietary exposures to thiocyanate, nitrate, and perchlorate are ubiquitous, and the relationships between these three chemicals and human health have been widely reported. The effect of the exposure to thiocyanate, nitrate, and perchlorate on thyroid functions in various populations has received widespread attention for decades (García Torres et al. 2022;King et al. 2022; Leung et al. Table 3 Multiple logistic regression associations of urinary perchlorate, nitrate, and thiocyanate quartiles with dyslipidemia in adults Model was adjusted as age, sex, race, education level, poverty, smoking, and alcohol use, body mass index, energy intake levels, physical activity, eGFR, uric acid, urinary creatinine, diabetes, and hypertension P values highlighted in bold are statistically significant HDL-C, high-density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride; LDL-C, low-density lipoprotein cholesterol; Non-HDL-C, non-high-density lipoprotein cholesterol; Apo B, apolipoprotein B; CI, confidence interval; Ref., reference * p < 0.05, ** p < 0.01, and *** p < 0.001  (Picetti et al. 2022;Said Abasse et al. 2022;Shiue 2015;Zhang et al. 2018), which is supported by several animal studies (Carvalho et al. 2021;Wang et al. 2022aWang et al. , 2018. High urinary perchlorate was linked to a higher prevalence of diabetes and an elevated level of risk factors (fasting glucose, HbA1c, insulin and homeostatic model assessment of insulin resistance) based on a sample of 11,443 individuals (Liu et al. 2017). A crosssectional study on adults in the USA shows that urinary nitrate was negatively linked to the prevalence of general and abdominal obesity, but urinary thiocyanate was positively associated with the prevalence of obesity (Zhu et al. 2019).
In addition, there is strong evidence that CVD (heart failure (Ferguson et al. 2021), hypertension (Kapil et al. 2020), myocardial infarction (Jackson et al. 2017), and stroke (Jackson et al. 2017;Lundberg et al. 2008)) and metabolic syndrome (Kapil et al. 2020;Liu et al. 2020) increase due to nitrates (both pharmacological and dietary). However, there are few general-population studies on the associations of the exposure of thiocyanate, nitrate, and perchlorate with continuous as well as binary serum lipid variables (the latter reflects dyslipidemia). As several of the above studies suggest that a long-term exposure to these three chemicals is a significant environmental risk factor for human health, they deserve further investigation. Our analysis revealed a positive linear relationship between thiocyanate and dyslipidemia. Thiocyanate is a Adjusted for age, gender, race, education level, poverty, smoking, alcohol use, energy intake, physical activity, body mass index, urinary creatinine, eGFR, uric acid, diabetes, and hypertension Table 4 WQS regression model to assess the association of the mixture of urinary perchlorate, nitrate, and thiocyanate with the prevalence of dyslipidemia in adults Model was adjusted as age, sex, race, education level, poverty, smoking, alcohol use, body mass index, energy intake levels, physical activity, eGFR, uric acid, urinary creatinine, diabetes, and hypertension primary indirect marker of cyanide exposure (Bhandari et al. 2014). Myeloperoxidase (MPO)-catalyzed thiocyanate oxidation is an important endogenous source of cyanate (Bhandari et al. 2014). Cyanate has been shown to modify LDL and HDL through carbamylation (Holzer et al. 2011), which contributes to the development of AS, vascular endothelial dysfunctions, and CVD (Verbrugge et al. 2015). In addition, exposure to cyanate can affect liver functions and lipid metabolism (Okafor et al. 2002;Sokołowska et al. 2011). Cyanate induces oxidative stress damage by inhibiting the NF-E2-related factor 2 (Nrf2)/heme oxygenase-1 (HO-1) pathway, which increases the level of reactive oxygen species (ROS), impairs liver functions, and causes dyslipidemia (Hu et al. 2019). In a study based on mice models, the cyanate group (compared to the control group) presented significantly elevated liver function biomarkers (alanine aminotransferase [ALT], aspartate aminotransferase [AST], and alkaline phosphatase [ALP]), TC, and LDL-C as well as a significantly decreased HDL-C (Hu et al. 2019). It is believed that oxidative stress may be a significant mechanism involved in lipid metabolism (Chen et al. 2003;Tangvarasittichai 2015), and an impaired antioxidant capacity can accelerate the development of hyperlipidemia. The primary physiological functions of livers are to metabolize glucose and lipids, which play a crucial role in the regulation of insulin sensitivity, inflammatory responses, and oxidative stress (Alfaradhi et al. 2014). Moreover, the exposure to thiocyanate typically induces a hyperinflammatory response (White et al. 2018;Whitehouse and Jones 2009;Zhu et al. 2021). Population studies have shown that the exposure to thiocyanate induces allergic inflammation, which is associated with allergy-related symptoms (Zhu et al. 2021). In addition, elevated thiocyanate in smokers has been associated with impaired innate immune responses (White et al. 2018). In an animal study, the supplementation of thiocyanate increased inflammatory responses to multiple factors that cause arthritis/fibrotic inflammation (Whitehouse and Jones 2009). Increased inflammatory responses are thought to be associated with lipid metabolism (Esteve et al. 2005;van Diepen et al. 2013), which provides an insight into the potential causes of dyslipidemia associated with the exposure to thiocyanate. Nevertheless, further research is required to investigate the pathways between the exposure to thiocyanate and dyslipidemia. Our analysis also revealed that urinary perchlorate was negatively associated with an elevated LDL-C, but not other dyslipidemia indicators, which was interesting and somewhat unexpected. This was because in a recent animal experiment, mice fed with a high-fat diet and exposed to perchlorate exhibited no significant changes in LDL-C, though HDL-C was significantly increased (Wang et al. 2022c). Additionally, it was found in a study on zebrafish that the exposure to perchlorate (relative to control treatment) failed to alter liver or systemic lipid accumulation (Minicozzi et al. 2021). The exact mechanism through which perchlorate affects lipid metabolism in our study remains to be determined. In addition, consistent with previous findings (Des-Ormeaux et al. 2021;Khalifi et al. 2015), we found that creatinine-corrected nitrate was negatively associated with dyslipidemia (elevated LDL-C and non-HDL-C).
Our study also has some limitations. First of all, the causality regarding the associations of urinary thiocyanate, nitrate, and perchlorate with dyslipidemia was not explored because the temporal relationship among these factors could not be determined. Secondly, missing serum lipid data, especially data on TG, LDL-C, and Apo B, reduced the sample size and decreased the statistical power. Thirdly, although we adjusted our models for many confounders, residual confounders might affect the results. Fourthly, the NHANES data on thiocyanate, nitrate, and perchlorate were collected from a single urine spot sample per participant, which might not accurately reflect the long-term exposure status.

Conclusions
Our findings demonstrate that urinary thiocyanate is positively linearly associated with dyslipidemia (elevated TG, TC, LDL-C, and non-HDL-C) in the general adult population of the USA, whereas urinary perchlorate is negatively associated with an elevated LDL-C. Furthermore, the exposure to a mixture of urinary thiocyanate, nitrate, and perchlorate was positively associated with a higher prevalence of all types of dyslipidemia except for an elevated Apo B. For the validation and expansion of our findings, prospective and mechanistic research is necessary.