Association Between Levels of Urine Di-(2-ethylhexyl)phthalate Exposure, a Potentially Harmful and Invisible Chemical, and Heart Rate Variability in Young Adults

Phthalate exposure is associated with cardiovascular risk. Among the various phthalates, di-(2-ethylhexyl) phthalate (DEHP) is the most important plasticizer in our daily lives. This study investigated the association between DEHP exposure and the alteration of heart rate variability (HRV). During 2017-2019, we recruited 974 young adults to investigate the effects of living environments and dietary habits on cardiometabolic disorders in Taiwan. We quantitatively analyzed urinary metabolites of phthalates, including mono-(2-ethylhexyl) phthalate (MEHP), mono-(ethyl-5-hydroxyhexyl) phthalate (MEHHP), and mono-(2-ethyl-5-oxohexyl) phthalate (MEOHP). A continuous electrocardiogram was recorded to obtain a 5-minute ECG. Time-domain and frequency-domain HRV analyses were performed. Multiple linear regression showed that urinary oxidized DEHP metabolites MEHHP and MEOHP were associated with decreased HRV after controlling for associated cardiovascular risk factors. A higher MEHHP level was associated with a lower TINN (triangular interpolation of NN interval histogram), very-low-frequency (VLF), and low frequency/high frequency (LF/HF) ratio. A higher MEOHP level was associated with a decreased LF/HF ratio. In addition, trend analysis showed that higher MEHHP and MEOHP quantiles were signicantly associated with a decreased LF/HF ratio. The DEHP metabolites MEHHP and MEOHP are associated with decreased HRV, indicating an unfavorable autonomic balance in young adults in Taiwan. one-way analysis of variance (ANOVA) was performed to test linear trend of HRV parameters among the four groups.


Introduction
Phthalate esters are colorless and odorless chemicals widely used as plasticizers to add exibility and resilience to plastic products [1]. Phthalates are essential to make cosmetics, medical devices, plastic, and rubber products. Phthalates are hydrophobic and bound to polymers with noncovalent bonding. Thus, phthalates readily leach into their environment. Phthalates are invisible chemicals hard to detect. Human exposure to phthalates is possible through dietary ingestion, air inhalation, or direct contact. Among various phthalates, di-(2-ethylhexyl) phthalate (DEHP) is the most important plasticizer we encounter in daily life [2].
After entering the human body, DEHP is rapidly degraded to its monoester, mono-(2-ethylhexyl) phthalate (MEHP), which is further metabolized by various hydroxylation and oxidation reactions [3,4]. Two of the major secondary oxidized DEHP metabolites are mono-(2-ethyl-5-oxohexyl) phthalate (MEOHP) and mono-(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP) [5]. DEHP is eliminated from urine in the form of its metabolites. Although phthalate excretion is rapid with a half-life of less than 24 hours, continuous daily exposure and ingestion of phthalates still cause persistent physiological effects due to steady concentrations in the human body.
Exposure to DEHP is associated with cardiovascular risk factors such as increased blood pressure [6], insulin resistance [7,8], and diabetes mellitus [9]. Phthalates are also associated with increased in ammation markers of absolute neutrophil counts, alkaline phosphatase, ferritin levels, and C-reactive proteins [10,11]. Our previous studies demonstrated that urinary phthalate metabolites were associated with apoptotic microparticles from endothelial cells and platelets, insulin resistance, and subclinical atherosclerosis in terms of increased carotid intima-media thickness [12,13]. Urinary phthalates have been associated with a higher risk of stroke in the U.S. [14], and we have also demonstrated that increased DEHP metabolite exposure may be linked to patients with coronary heart diseases [15]. DEHP accelerates atherosclerosis by disturbing cholesterol homeostasis, and the in ammatory response has been ascribed to a pathogenic mechanism in animal models [16]. Our recent study also con rmed that the global DNA methylation marker 5 mdC/dG may mediate the association between DEHP exposure and subclinical atherosclerosis [17]. However, these attributable risk factors could not fully explain the underlying adverse effects of DEHP on the cardiac and cardiovascular systems [15,18].
Heart rate variability (HRV) measures the uctuation of the heart rate. HRV is a noninvasive method to analyze autonomic nervous system effects. For the general population, decreasing heart rate variability is associated with an increased incidence of cardiac events [19] and mortality [20,21]. Decreased HRV was associated with coronary artery disease [22], heart failure [23], pulmonary hypertension [24], and impaired renal function [25]. For patients with congestive heart failure [26] or after myocardial infarction [27][28][29], decreased HRV is associated with a higher mortality rate. All this evidence suggests that HRV is a useful tool to evaluate autonomic dysfunction and cardiac adverse effects. In animal models, DEHP-treated mice had decreased HRV, enhanced cardiovascular reactivity, and prolonged blood pressure recovery [30]. This animal model suggested that DEHP may cause adverse cardiac effects through the mechanism of decreasing HRV due to autonomic dysfunction.
However, the association between phthalate exposure and HRV in humans is still unclear. This study was designed to investigate the association between DEHP metabolites and alterations in HRV in young adults with low cardiovascular risk factors.

Subjects
From 1992-2000, a nationwide urine screening program for early renal disease was conducted for Taiwanese children between 6 and 18 years of age [31]. Among over 103,756 students who underwent the screening, 303 children with hypertension and 486 children with normal blood pressure living in the Taipei area joined the YOung TAiwan cohort (YOTA) study between 2006 and 2008 under informed consent and parental agreements in National Taiwan University Hospital [12,13].
During 2017-2019, we recruited 980 young adults to investigate the effects of living environments and dietary habits on cardiometabolic disorders in Taipei, Taiwan. There were 542 YOTA participants, and 438 young adults of similar age were recruited as the New YOTA cohorts. Among all 980 participants, 4 participants were excluded due to being less than 20 years old or more than 45 years old. Two participants were excluded from the New YOTA study due to no measurable urine sample under hemodialysis status. Thus, 974 participants were enrolled in this study.

Assessment of Clinical Information and Risk Strati cation
Basic cardiovascular risk factors, including age, sex, weight, height, diet, smoking habits, alcohol drinking, and exercise habits, and the personal living environment characteristics and dietary habits of each participant were collected. The arterial pressure waveform was recorded by a cuff sphygmomanometer using an oscillometric BP device (DynaPulse 200 M, Pulse Metric Inc., San Diego, CA) [32]. The arterial pressure waveform was measured from left and right hands after at least 5 min of rest in a sitting position in a quiet room. Hypertension was diagnosed if the mean systolic blood pressure was greater than 140 mmHg, diastolic pressure was greater than 90 mmHg, or the patient was taking anti-hypertension medications.
Blood samples were obtained via the antecubital vein of each participant after overnight fasting for 10-14 h. The serum cholesterol, triglyceride, and low-and high-density lipoprotein cholesterol (LDL-C and HDL-C, respectively) levels and plasma glucose were measured using an autoanalyzer (Toshiba, TBA-200FR; Toshiba, Tokyo, Japan). Biochemical examination for each participant was performed according to standard lab protocols/methods. Diabetes mellitus was diagnosed according to the American Diabetes Association criteria, and those whose fasting glucose levels were equal to or greater than 126 mg/dL (7.0 mmol/L) had diabetes. We measured the weight and height of the participants by standard methods. Body mass index (BMI) was calculated using weight (in kilograms) divided by the square of the height (in meters).
Finally, urine mixtures were quantitatively analyzed by liquid chromatography with a tandem mass spectrometric (LC-MS/MS) system.
Regarding the quality assurance and control of DEHP metabolites, we prepared blank samples for each batch of samples during sample preparation. Internal quality control was performed using pooled quality control urine samples, with precision ranging from 6-26%, depending on the metabolite. Alongside pooled urine samples for each batch, low-concentration (20 mg/L) and highconcentration (50 mg/L) quality control materials were also analyzed. The method detection limits of MEHP, MEHHP, and MEOHP were 0.7, 0.1, and 0.1 µg/L, respectively. External quality assurance was assessed using the German External Quality Assessment Scheme for Biological Monitoring (G-EQUAS) [13].

Heart Rate Variability Analysis
After blood sample collection and a 15-min rest for each participant, we immediately performed the resting electrocardiogram (ECG) examination in the supine position during the daytime (9:00 a.m. to 12:00 p.m.) using an HRV+ (BeneGear, Taipei, Taiwan) with a sampling rate of 250 Hz (4 ms). A complete 5-min segment of the N-N interval was taken for HRV analysis.
Time-domain and frequency-domain analyses were analyzed for heart rate variability. All analyses were performed according to the recommendations of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology [34].
The time-domain measurements of HRV included the mean of the R-R intervals, the standard deviation of the normal-to-normal intervals (SDNN), root mean square successive differences (RMSSDs), between adjacent normal-to-normal intervals and percentage of absolute differences in normal RR intervals greater than 50 ms (pNN50), and TINN (triangular interpolation of the NN interval histogram), indicating the baseline width of the RR interval histogram. The frequency-domain measurements of HRV included very-low-frequency (VLF, < 0.04 Hz), low frequency (LF: 0.04-0.15 Hz), high frequency (HF: 0.15-0.40 Hz), and LF/HF ratio, which were calculated by Welch's averaged periodogram of the normal-to-normal intervals. These parameters represented the modulation of sympathetic and parasympathetic activity to heart rate variability. The details of translating ECG wave complexes to HRV indices are given in our previous study [35].

Statistical Analysis
We performed statistical analysis using IBM SPSS Statistics for Windows, Version 26.0. (IBM Corp., Armonk, NY, US). Three urine DEHP metabolites (MEHP, MEHHP, and MEOHP) were log-transformed to t a normal distribution, as con rmed by the Kolmogorov-Smirnov test. Samples of urine phthalate metabolites below the detection limit were recorded as half of the detection limit. The concentration of urine phthalate metabolites was calibrated by urine creatinine. Phthalate metabolite concentrations are expressed as the mean ± standard deviation. The Student's t-test was performed to compare the concentrations of urinary phthalate metabolites in different subgroups. The HRV parameters following a nonnormal distribution were also log-transformed.
A multiple linear regression model was used to evaluate the dose-response relationships between HRV parameters and urine phthalate metabolites. To adjust the effects of possible confounders, covariates of age, sex, BMI z-score, systolic BP, fasting blood sugar, and LDL-C were added into the multiple linear regression model. The estimated coe cient of urinary phthalate metabolites and its 95% con dence interval (95% CI) were calculated to measure the effect of 1-unit increase of phthalates metabolites to HRV parameters after adjusting the covariates. We considered an estimate was statistically signi cant when the p value was less than 0.05. To con rm the association between the concentration of urinary phthalate metabolites and HRV parameters, we divided the participants into four exposure groups according to the quantiles of phthalate metabolite concentrations. Trend analysis in oneway analysis of variance (ANOVA) was performed to test linear trend of HRV parameters among the four exposure groups.

Participants Characteristics
Among the 974 participants recruited in this study, there were 407 men and 567 women. The mean age was 31.9 years old. The geometric means and standard deviations of the creatinine-adjusted urinary DEHP metabolites of different characteristic subgroups are listed in Table 1. Women had statistically signi cantly higher levels of MEHHP and MEOHP than men. Participants aged between 18 and 32 had a higher level of MEHP and a lower level of MEOHP. Participants with normal LDL-C had a higher level of MEOHP. Participants with higher education had signi cantly lower levels of MEHHP and MEOHP. Other clinical characteristics such as hypertension, diabetes mellitus, body mass index, and current smoking had no signi cant association with urinary DEHP metabolites (Table 1). Heart Rate Variability Analysis Univariable analysis of the time domain showed that higher levels of MEHHP and MEOHP were associated with decreased TINN ( Table 2). Frequency domain analysis revealed that higher levels of MEHHP and MEOHP were associated with a lower VLF, LF, and LF/HF ratio, as shown in Table 3.  Univariable analysis of HRV showed that older age; being a woman; higher BMI z-score; and higher BP, blood sugar, and LDL-C were negatively associated with time-domain HRV in the lower mean RR interval, SDNN, RMSSD, pNN50, and TINN. In addition, we also showed a negative association between cardiovascular risk factors and frequency domain HRV for the VLF, LF, HF, and the LF/HF ratio (Supplemental Table 1 and Supplemental Table 2).

Multiple Linear Regression Models
In the multiple linear regression analysis of the HRV parameters after adjusting for age, sex, BMI z-score, blood pressure, fasting blood glucose, and LDL-C, we demonstrated that a higher MEHHP level was associated with a signi cantly lower TINN, with an     The trend analysis in one-way ANOVA showed that exposure to higher MEHHP and MEOHP quantiles were signi cantly associated with a lower LF/HF ratio. The p values for trend were 0.014 for MEHHP and 0.001 for MEOHP (Table 6 and Fig. 1).

Discussion
The most important nding of this study is that urinary phthalate metabolites were associated with decreased HRV in young adults. This is the rst report to show an association between DEHP exposure and decreased autonomic balance in a human study.
Previous studies have shown that phthalate DEHP exposure alters the autonomic nervous system and decreases HRV in animal models [30]. DEHP exposure also increased the expression of the genes encoding endothelin-1 and angiotensin-converting enzyme in heart tissue [30]. Other studies showed that DEHP-treated cardiomyocytes had increased gene expression of calcium handling genes [36] and subsequently markedly reduced cardiac network synchronicity of DEHP-treated cardiomyocytes [37].
These were possible mechanisms of decreased HRV caused by DEHP exposure.
Endothelial function was associated with heart rate variability in animal studies [38]. Endothelial dysfunction was also associated with decreased HRV in healthy adults and patients with stable coronary artery disease or diabetes mellitus [39][40][41]. Increased carotid intima-media thickness was associated with decreased HRV in previous studies [42,43]. Our previous studies showed that urinary phthalate metabolites were associated with endothelial dysfunction, insulin resistance and increased carotid intima-media thickness [12,13,44]. Thus, exposure to DEHP may mediate impaired HRV through these mechanisms.
Our study showed that higher concentrations of urinary MEHHP and MEOHP were associated with decreased HRV, primarily in frequency domain HRV. We did not identify this association in MEHP, which is the monoester of DEHP after the rst step of hydrolysis metabolism. Both MEHHP and MEOHP were two of the major secondary and oxidized urinary metabolites of DEHP. The possible explanations for the results of our study are as follows. First, the formation of MEHP from DEHP is possible through abiotic hydrolysis during urinary sample collection, storage, and processing. MEHHP and MEOHP are secondary oxidized DEHP metabolites in the liver and thus are less likely to be contaminated during sampling handling [5]. Second, urinary MEHP accounts for less than 10% of DEHP intake [45]. Urinary MEHP also has the shortest half-life (approximately 5 hours) compared to urinary MEHHP or MEOHP (both approximately 10 hours) [4]. Urinary MEHP could thus be more easily in uenced by prolonged sample handling during urine collection at the study site. For example, it took approximately 4-5 hours before urine was sent to the refrigerator to freeze in this study. Thus, urinary MEHP might not be able to re ect the actual DEHP exposure and its pathophysiological effects on autonomic dysfunction.
HRV analysis includes time-domain analysis, frequency-domain analysis, and nonlinear regression. We used time-domain and frequency-domain HRV analysis in our study as studies showed that linear and nonlinear HRV analyses yielded similar conclusions [46,47]. Multivariable linear regression was performed to control for the possible confounding factors such as age, sex, BMI z-score, SBP, fasting glucose, and LDL. Thus, the association between decreased HRV and urinary DEHP oxidized metabolites was less likely due to other confounding factors.
This study has many strengths. First, most cardiovascular risk factors such as older age; a higher BMI z-score; and higher BP, blood sugar, and LDL-C were mostly associated with lower time-domain and frequency-domain HRV, which also corroborated the reliable measurement of HRV. Second, measurements of phthalate metabolites are compatible with our previous study in which women had higher levels of DEHP metabolites than men [13]. Third, even after controlling for most associated confounders, DEHP metabolites were still strongly linked to reduced HRV parameters. Fourth, decreased HRV was noted in various physiological and pathophysiological statuses, such as major cardiovascular risk factors. This evidence also corroborates that the validity of HRV measurements. We recruited young adults with low cardiovascular risk and prevented these possible confounding factors.
This study also has several limitations.

ACKNOWLEDGMENTS
We thank all participants who were recruited in the National Health Research Institute grant-supported study during 2017-2019 (NHRI-EX106-10629PI), with the grant proposal "Effects of living environments and dietary habits on cardiometabolic disorders in young adults". We thank associate professor Cheng-Chih Hsu for the measurements of urinary phthalate metabolites by the Mass Spectrometry, Analytical Chemistry Laboratory. We thank Dr. Kuo-Tong Huang and Nan-Chang Clinical Laboratory for the study space and laboratory analysis support. We also thank the 3rd core laboratory of National Taiwan University Hospital.

AUTHOR CONTRIBUTION STATEMENTS
Ching-Way Chen: data management and analysis, data interpretation and manuscript drafting  Figure 1 Means plot of log LF/HF ratio among four exposure groups categorized by the four quartiles of MEOHP measurements. Abbreviations: MEOHP: mono-(2-ethyl-5-oxohexyl) phthalate, LF: low-frequency, HF: high frequency