Association between organophosphate flame retardant exposure and lipid metabolism: data from the 2013–2014 National Health and Nutrition Examination Survey

Organophosphate flame retardants (OPFRs) are emerging environmental pollutants that can be detected in water, dust, and biological organisms. Certain OPFRs can disrupt lipid metabolism in animal models and cell lines. However, the effects of OPFRs on human lipid metabolism remain unclear. We included 1,580 participants (≥20 years) from the 2013–2014 National Health and Nutrition Examination Survey (NHANES) to explore the relationship between OPFR exposure and lipid metabolism biomarkers. After adjusting for confounding factors, results showed that one-unit increases in the log levels of diphenyl phosphate (DPhP) (regression coefficient = −5.755; S.E. = 2.289; p = 0.023) and log bis-(1-chloro-2-propyl) phosphate (BCPP) (regression coefficient = −4.637; S.E. = 2.019; p = 0.036) were negatively associated with the levels of total cholesterol (TC) in all participants. One-unit increases in the levels of DPhP (regression coefficient = −2.292; S.E. = 0.802; p = 0.012), log bis (1,3-dichloro-2-propyl) phosphate (BDCPP) (regression coefficient = −2.046; S.E. = 0.825; p = 0.026), and log bis-2-chloroethyl phosphate (BCEP) (regression coefficient = −2.604; S.E. = 0.704; p = 0.002) were negatively associated with the levels of high-density lipoprotein cholesterol (HDL-C). With increasing quartiles of urine BDCPP levels, the mean TC levels significantly decreased in all participants (p value for trend = 0.028), and quartile increases in the levels of DPhP (p value for trend = 0.01), BDCPP (p value for trend = 0.001), and BCEP (p value for trend<0.001) were negatively corelated with HDL-C, with approximately 5.9, 9.9, and 12.5% differences between the upper and lower quartiles. In conclusion, DPhP, BDCPP, and BCEP were negatively related to HDL-C concentration, whereas DPhP and BCPP levels were negatively associated with TC level. Thus, exposure to OPFRs may interfere with lipid metabolism.


Introduction
Cardiovascular disease (CVD) is a major leading cause of morbidity and mortality worldwide, and dyslipidemia is an established risk factor for CVD (1).Dyslipidemia is characterized by elevated serum total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), or triglyceride (TG) levels (2) and reduced serum highdensity lipoprotein cholesterol (HDL-C) concentrations.Data from the 2007-2018 National Health and Nutrition Examination Survey (NHANES) showed that the prevalence rates of hypercholesterolemia (TC values ≥ 240 mg/dL) and hypertriglyceridemia (TG levels ≥ 200 mg/dL) were 11.5 and 10.4%, respectively (3).Dyslipidemia can originate from familial disorders (primary) or an alternative underlying etiology, such as metabolic disorder (diabetes, hypothyroidism), medications, unhealthy diet, and poor lifestyle regimen (4).
Organophosphate flame retardants (OPFRs) are ubiquitous in various environmental media because they are physically rather than chemically bound to a material, allowing these compounds to be easily released into the environment (5).Few toxicologic studies have demonstrated that OPFR exposure might interfere with lipid metabolism.TCP exposure might disturb the homeostasis and fluidity of lipid in in cerebrum, spinal cord and sciatic nerve (6).Evidence showed that the meta-isomer of TCP could alter hepatocytes lipid metabolism of seabream through interacting between liver X receptor α and proliferator-activated receptors (PPARs) proteins or modulating the expression levels of micro ribonucleic acids (7).Furthermore, TCP exposure could lead to increased lipid content and alter the fatty acid profile in human hepatocarcinoma (HepG2) cells through activation the pregnane X receptor pathway along with the deficient FA β-oxidation and enhanced lipogenesis (8).Triphenyl phosphate (TPhP, parent compound of diphenyl phosphate) inhibits specific liver carboxylesterases (CEs), altering hepatic lipid metabolism, inducing serum hypertriglyceridemia, and increasing very-low-density lipoprotein (VLDL) and LDL masses in mice (9).TPhP treatment significantly increases blood TC and TG concentrations and induces large lipid droplets in the livers of zebrafish possibly by inhibiting cholesterol utilization and liver lipid transfer (10).Lipid metabolism pathways, such as the fatty acid elongation pathway, are also significantly affected by TPhP exposure (10).TG levels increase and cholesterol levels significantly increase in hepatocytes exposed to high concentrations of tri-m-cresyl phosphate, one of the major isomers of commercial tricresyl phosphate (TCP), in gilthead sea bream (7).Le et al. (11) found that aryl-OPFRs (TPhP and TCP) and chlorinated-OPFRs, such as tris (1,3-dichloro-2-propyl) phosphate (TDCPP), cause lipid accumulation in mouse hepatic cells, accompanied with reduced mitochondrial (mito)-networks/cell, biased mitoATP/ glcoATP rate, and expanded mito-area/cell.
Due to dyslipidemia being one of the major risk factors for cardiovascular diseases, toxicological studies have also indicated that exposure to OPFRs may interfere with lipid metabolism.OPFRs have been identified in various environments, including air, dust, water sources, soil, and sediments.Furthermore, traces of OPFRs have been detected in human samples and biotic organisms (12).However, the relationship between OPFRs exposure and lipid metabolism in humans remains unclear.Our study aims to investigate these associations through the analysis of the National Health and Nutrition Examination Survey (NHANES) database.

Study population
This study utilized the dataset from the 2013-2014 NHANES dataset in the United States.NHANES is a comprehensive, nationwide, population-based survey initiated in 1999 to evaluate the health and nutritional status of the U.S. population.The 2013-2014 NHANES was in review approved by the US National Center for Health Statistics Research Ethics Review Board (Continuation of Protocol #2011-17), and informed consent was obtained from all participants.The dataset and detailed survey protocols are provided on the NHANES website (13).In the 2013-2014 NHANES, one-third of the participants aged ≥6 years were randomly selected for the measurement of OPFR profiles in stored spot urine samples, and lipid profiles were examined in those who were ≥ 6 years of age and provided serum specimens.In our study, adult participants (≥ 20 years of age) of the 2013-2014 NHANES who had available urinary OPFR profiles and serum lipid data were enrolled for the analysis (n = 1,580, Figure 1).We selected the 2013-2014 NHANES data as it includes measurements of both Organophosphate Flame Retardants (OPFRs) and indicators of lipid metabolism.

Measurement of lipid profiles
At each study site, trained personnel followed the standardized protocol outlined on the NHANES website to collect blood specimens.Detailed procedures for specimen collection are provided in the NHANES Laboratory Procedures Manual (13).The data on TC, LDL-C, HDL-C, and TG were included for analysis in the present study.

Collection of baseline characteristics
In accordance with the NHANES protocols, sociodemographic profiles were collected during household interviews by well-trained interviewers using standardized questionnaires and a computerassisted personal interview system.Body measurement data were recorded by trained health technicians in NHANES mobile examination centers following standardized procedures.Other procedure details are provided on the NHANES website (13).The age, sex, ethnicity, household income, smoking status, alcohol consumption, and body mass index (BMI) of all participants were recorded.

Statistical analysis
Categorical variables are presented as numbers (n) with percentages, and continuous variables are presented as medians with interquartile ranges.To identify baseline covariates associated with lipid profiles and further subgroup analysis, we stratified the study population into subgroups according to age (20-50 vs. >50 years), sex, ethnicity, household income (<4,500 vs. ≥4,500 USD/year), smoking status, alcohol consumption (<12 vs. ≥12 drink/year), and BMI (<25 vs. 25-30 vs. ≥30 kg/m 2 ).The lipid profiles were compared between subgroups via the Mann-Whitney U test (for two subgroups) or Kruskal-Wallis H-test (for three or more subgroups).Baseline covariates with a p-value of <0.05 in univariate analyses were included in further multivariate analyses for adjustment.We conducted multiple linear regression analyses in the complex samples to explore the relationships between urinary OPFR metabolite concentrations and lipid profiles.The analyses were adjusted for baseline covariates, and sampling weights were applied in accordance with the National Center for Health Statistics Analytic Guidelines.Due to the non-normal distribution, urinary OPFR metabolite concentrations were subjected to logarithmic transformation and quartile stratification before the linear regression analysis.Statistical significance was set at a p < 0.05.Statistical Product and Service Solutions (version 22.0; IBM, Armonk, NY, United States) was used for all analyses.
The adjusted regression coefficients (S.E.) for the differences in TG, TC, LDL-C, and HDL-C relative to a one-unit increase in log-transformed BDCPP are summarized in Table 3.We found that a one-unit increase in the log BDCPP level was negatively associated with the levels of HDL-C (regression coefficient = −2.046;S.E.= 0.825; The adjusted regression coefficients (S.E.) for the differences in TG, TC, LDL-C, and HDL-C relative to a one-unit increase in the log-transformed BCPP level are summarized in Table 4.A one-unit increase in the log BCPP level was negatively associated with TC levels (regression coefficient = −4.637;S.E.= 2.019; p = 0.036).Subgroup analysis also showed the negative association of BCPP level with the levels of TG (regression coefficient = −13.286;S.E.= 5.626; p = 0.032) and TC (regression coefficient = −9.410;S.E.= 3.448; p = 0.016) in the female group.
The adjusted regression coefficients (S.E.) for the differences in TG, TC, LDL-C, and HDL-C relative to a one-unit increase in the log-transformed BCEP level are summarized in Table 5.A one-unit increase in the log BCEP level was negatively associated with the levels of HDL-C (regression coefficient = −2.604;S.E.= 0.704; p = 0.002),   The adjusted regression coefficients (S.E.) for the differences in TG, TC, LDL-C, and HDL-C levels relative to a one-unit increase in the log-transformed DnBP level (μg/L) are summarized in Table 6.
The association between a one-unit increase in the log DnBP level and TG, TC, LDL-C, and HDL-C levels did not achieve statistical significance in the overall group.
After adjusting for potential covariates in multiple regression analysis, the correlations between the quartiles of each OPFR and TC, as well as HDL-C in all participants and different sexes are listed in Figure 2.With increasing quartiles of urine BDCPP levels, the mean TC levels significantly decreased in all participants (p value for trend = 0.028) and the female group (p value for trend<0.001),whereas the mean differences in TC levels between the upper and lower quartiles of BDCPP in all participants and the female group were 3.4 and 5.8%, respectively.Furthermore, the quartile increase in urine BCEP level was negatively related to TC levels (p value for trend = 0.016) in the female group, with approximately 5.6% difference between the upper and lower quartiles.Quartile increases in the levels of DPhP (p value for trend = 0.01), BDCPP (p value for trend = 0.001), and BCEP (p value for trend<0.001)were negatively corelated with HDL-C, with approximately 5.9, 9.9, and 12.5% differences between the upper and lower quartiles.Conversely, we also observed gender differences in the impact of OPFRs on HDL-C.In males, DPhP (p value for trend = 0.017) and BDCPP (p value for trend = 0.019) levels were negatively correlated with HDL-C levels, with a decrease of approximately 10.4 and 9.3%, respectively, in the highest quartile of HDL-C compared with the lowest quartile.In contrast to the male group, the female group showed negative correlations of DPhP (p value for trend = 0.025), BDCPP (p value for trend = 0.009), and BCEP (p value for trend = 0.01) with HDL-C.The highest quartile of DPhP, BDCPP, and BCEP levels in the females was associated with approximately 5.2, 7.0, and 12.1% reductions in HDL-C, respectively, compared with the lowest quartile.

Discussion
In the present study, we observed a statistically significant association between OPFRs and lipid profiles.After adjusting for confounding factors, the DPhP level was negatively associated with TC and HDL-C levels, the BDCPP level was negatively associated with HDL-C levels, the BCPP level was negatively associated with TC levels, and the BCEP level was negatively associated with HDL-C levels.Furthermore, quartile increases in the levels of DPhP, BDCPP, and BCEP were negatively correlated with HDL-C, with approximately 5.9, 9.9, and 12.5% differences between the upper and lower quartiles.
Dyslipidemia is one of major risk factors for cardiovascular and cerebrovascular diseases, leading to an increased risk of atherosclerotic cardiovascular disease (16).Among lipid metabolites, HDL-C has been found to be associated with mortality.Li et al. followed 7,766 older adults individuals aged ≥65 years, and found that the group with HDL-C < 61 mg/dL had higher rates of all-cause mortality and cardiovascular-related mortality (17).Another research also indicated that for each 1 mg/dL increase in HDL-C, there is a 3.7 to 4.7% decrease in the rate of cardiovascular mortality (18).In our study, we observed a negative correlation between the levels of DPhP, BDCPP, and BCEP with HDL-C.This may suggest that populations with higher levels of DPhP, BDCPP, and BCEP could potentially have an increased risk of cardiovascular diseases.Further research is needed to clarify this association.
Recently, a few studies have investigated the effects of exposure to OPFRs on fatty acid metabolism.Hu et al. ( 19) found that the exposure of RAW264.7 macrophage cells to TPhP increases endoplasmic reticulum (ER) stress and inflammation, which further downregulate and decrease fatty acid saturation.Lpcat3 is one of the factors that regulate carbohydrate metabolism and adipocyte differentiation.Another study using Alpha mouse liver 12 cells found that exposure to two aryl-OPFRs (TCP and TPhP) and three chlorinated OPFRs (TDCPP, TCPP, and TCEP) causes intracellular lipid accumulation at relatively low concentrations (<10 μmol/L) for TCP, TPHP, and TDCPP.They also observed intracellular lipid accumulation at concentrations >10 μmol/L for TCPP and TCEP.This study also found that OPFRs increase oxidative stress and alter mitochondrial membrane potential in liver cells, thereby interfering with ATP metabolism and causing lipid accumulation (11).Meanwhile, CEs are responsible for hydrolyzing xenobiotic or endogenous compounds that contain ester, thioester, or amide groups (20).In the liver, CEs are responsible for metabolizing TGs and fatty acids in lipid droplets and resynthesizing them into VLDL in the ER, which is then released into the bloodstream, further affecting the metabolism of carbohydrates and esters and promoting insulin resistance (21).In a previous study, exposure to TPhP inhibits CE activity in the liver of mice, resulting in increased concentrations of LDL-C and VLDL in the serum (9).The reason for the inhibition of CEs may be that OPFRs irreversibly bind to the activation site of CEs, thereby inhibiting their function.Another study using Atlantic cod liver found that exposure to TCPP, 2-ethyldiphenyl phosphate, or their mixture downregulates the expression of genes involved in cholesterol synthesis and affects subsequent lipid metabolism (22).Cholesterol is a precursor for steroid hormones, such as follicle-stimulating hormone, luteinizing hormone, total testosterone, and total estradiol, and interference of cholesterol synthesis might lead to endocrine disruption.However, limited studies focused on the relationship between OPFR exposure and human lipid metabolism.The results of the present study showed that DPhP, BDCPP, and BCEP were negatively related to HDL-C, whereas DPhP and BCPP were negatively associated with TC.In addition, OPFRs exerted differential effects on lipid metabolism interference in males and females.Specifically, the negative correlation of DPhP, BDCPP, and BCEP with TC was more pronounced in females than in males.However, the association between OPFRs exposure and HDL-C showed less gender difference.Further investigation is warranted to clarify the reasons for this gender difference.
OPFRs are low-cost and effective flame retardants widely used in various consumer products, building materials, textiles, and electronics.They have been used to replace polybrominated diphenyl ethers (PBDEs) owing to the persistence, bioaccumulation, and toxicity of the latter.By 2011, OPFRs accounted for 20% of the global   flame retardant market (23).They are also used as plasticizers for epoxy resins, coatings, engineering thermoplastics, and floor polishes.The consumption of OPFRs reached 83,000 tons in Europe and 72,000 tons in the United States in 2007, and the usage has grown at a rate of 3.7% annually from 2007 to 2012 (24).However, OPFRs are physically rather than chemically bound to the products, allowing them to easily detach from the products during use and enter the surrounding environment through volatilization, dissolution, deposition, and infiltration.OPFRs can be detected in various environmental and biological matrices, such as air (25), soil (26), water (27), fish (28), and even breast milk (29).Humans can be exposed to OPFRs through skin contact, inhalation, and ingestion (30).OPFRs with low logarithmic octanol-air coefficient values (Log K oa ) mainly exist in the gas phase; hence, the contribution of these compounds in the air is greater than that in the dust.Dyslipidemia is an important risk factor for atherosclerotic CVD and ischemic cerebrovascular accident (CVA).Insulin resistance, which is associated with metabolic syndrome, increases plasma TG and LDL-C levels and reduces HDL-C levels, thereby increasing the risk for atherosclerotic CVD, CVA, and peripheral artery disease.High-density lipoproteins are involved in delaying the formation of atherosclerotic lesions through several mechanisms, such as removal of cholesterol from macrophages within the arterial wall and transportation to the liver for excretion (35,36).Observational studies found that a 1 mg/dL (0.026 mmol/L) increase in HDL-C is associated with a 3% risk reduction of coronary heart disease in women and 2% risk reduction in men, irrespective of age, body mass index, smoking habit, blood pressure, and LDL-C level (16).In a nationwide, community-based, prospective cohort study in the US, the risk for all-cause mortality was significantly higher in the group with HDL-C concentrations <61 mg/dL than in the group with HDL-C concentrations ranging from 61 to 87 mg/dL among older adults (aged ≥65 years).Repeatedly measured low HDL-C levels (defined as <40 mg/dL for men and < 50 mg/dL for women) have been associated with the risk of thyroid cancer, and this correlation is stronger in metabolically unhealthy Korean persons (37).Data from a large German primary care provider database showed that low HDL-C concentrations (<40 mg/dL) are positively associated and elevated TC levels (>200 mg/dL) are negatively associated with cancer, irrespective of diabetes, obesity, age, and sex.By contrast, serum levels of TG and LDL pose no impact on cancer risk (38).In the present study, exposure to DPhP, BDCPP, and TCEP were negatively associated with HDL-C.However, whether these negative associations result in adverse health outcome merits further investigation.
The majority of total cellular cholesterol is localized on the plasma membranes and interacts with the adjacent lipids to regulate the rigidity, fluidity, and permeability of the cell membrane.Cholesterol could bind to numerous transmembrane proteins and either maintain or alter their conformation.It can also interact with several transport proteins that facilitate cholesterol trafficking and regulate the subcellular distribution.In addition to their roles in membrane structure and function, derivatives of cholesterol are engaged in various biological processes, such as steroid hormone generation and bile acid production.The homeostasis of cholesterol is determined by de novo biosynthesis, uptake, export, and storage (39).Negative associations of DPhP and BCPP levels with TC levels were disclosed in our study.TBEP exposure in Tm3 Leydig cells increases oxidative stress, decreases cell viability, disrupts hormone generation (40), and induces abnormal sperm morphology and testicular histopathology in male rats (41).Moreover, TPhP and TDCPP can cause endocrine disruption, alter thyroid hormone levels (42), and decrease semen quality in men (43).DPhP downregulates the expression of genes involved in lipid/cholesterol and glucose/fatty acid metabolism (44).An animal study revealed that exposure to DPHP causes metabolic disturbance in the organism possibly because of its interfering effects on estrogen and mineralocorticoids (45).Thyroid hormone is an important regulator of serum cholesterol levels and hepatic cholesterol metabolism, including synthesis, endocytosis by the (LDL)-receptor, and peripheral uptake and hepatic excretion by reverse cholesterol transport.The disruption of cholesterol metabolism by OPFRs might further interfere with thyroid hormone synthesis.There are several limitations about our study.First, the composition and concentration of different OPFRs might be varied in different regions, hence, the results might not be applied to other countries.Second, several kinds of chemicals such as phthalates and polybrominated diphenylethers are co-existing in the environment.These chemicals might interfere with OPRFs which lead to different impacts on human health.The interactions between different environmental toxicants and its effects on human health could not be further clarified in our study.Third, the mechanisms of lipid metabolism might vary between different persons biochemically, therefore, the disturbance from OPFRs on lipid metabolism might also be different.The concomitant medical illness and medications might also exert different degrees of influence of lipid metabolism which could not be delineated in our study.

Conclusion
DPhP, BDCPP, and TCEP levels were negatively related to the concentrations of HDL-C, whereas DPhP and BCPP levels were negatively associated with the levels of total cholesterol.Furthermore, the mean differences in TC levels between the upper and lower quartiles of BDCPP in all participants and the female group were 3.4 and 5.8%, respectively.Conversely, quartile increases in DPhP, BDCPP, and BCEP levels were negatively corelated with HDL-C levels, with approximately 5.9, 9.9, and 12.5% differences between the upper and lower quartiles.The findings of the current study may suggest that exposure to OPFRs could potentially interfere with lipid metabolism and have associated health effects.

FIGURE 1
FIGURE 1Participant flow chart algorithm.

TABLE 1
The median (25 and 75 percentile) of lipid profile in different subgroups.

TABLE 2
Adjusted regression coefficients (S.E.) for differences in lipid profile relative to a one-unit increase in log10-transformed diphenyl phosphate (DPhP), with results weighted for sampling strategy.

TABLE 5
Adjusted regression coefficients (S.E.) for differences in TG, cholesterol, LDL, and HDL relative to a one-unit increase in log10-transformed biomarkers of bis-2-chloroethyl phosphate (BCEP), with results weighted for sampling strategy.

TABLE 6
Adjusted regression coefficients (S.E.) for differences in triglycerides, cholesterol, HDL-cholesterol, and LDL-cholesterol relative to a one-unit increase in log10-transformed di-n-butyl phosphate (DnBP), with results weighted for sampling strategy.
(31)sure to volatile OPFRs such as TCPP and TCEP usually occurs through air inhalation.By contrast, OPFRs with high Log K oa are primarily in the particulate phase and settled on dust.Therefore, dust ingestion is the important exposure pathway for less volatile OPRFs, such as tris (2-butoxyethyl) phosphate (TBEP), TCP, and TPhP.A study from Vietnam reported that the total estimated daily intakes of ΣOPFRs via dermal absorption, air inhalation, and dust ingestion for toddlers and adults under medial exposure are 160 ngkg −1 day −1 and 36.7 gkg −1 day −1 , respectively(31).The value is approximately 4-5 times greater in toddlers than in adults.Dermal absorption is the major exposure pathway for toddlers and adults (accounting for 45.1 and 49.5% ΣOPFRs, respectively), followed by air inhalation (contributing to 40 and 46.5% ΣOPFRs, respectively).Cr for BBOEP, respectively.Universal exposure to OPFRs was proved by growing evidence, and the disturbance of OPFR exposure on lipid metabolism might result in more and more adverse health impacts.