Postprandial Negatively Charged Very-Low-Density Lipoprotein Is a Novel Biomarker for Atrial Remodeling in Metabolic Syndrome

Background: Negatively charged very-low-density lipoprotein (VLDL-χ) in metabolic syndrome (MetS) patients exerts cytotoxic effects on endothelial cells and atrial myocytes. Atrial cardiomyopathy, manifested by atrial remodeling with a dilated diameter, contributes to atrial brillation pathogenesis and predicts atrial brillation development. The correlation of VLDL-χ with atrial remodeling is unknown. This study investigated the association between VLDL-χ and left atrial dilatation. Methods: We evaluated 87 MetS and 80 non-MetS individuals between 23 and 74 years old (50.6% men) without overt cardiovascular diseases. Blood samples were collected while fasting and postprandially (at 0.5, 1, 2, and 4 hours after a unied meal). VLDL was isolated by ultracentrifugation; the percentile concentration of VLDL-χ (%) was determined by ultra-performance liquid chromatography. The correlations of left atrial diameter (LAD) with variables including VLDL-χ, LDL-C, HDL-C, triglycerides, glucose, and blood pressure, were analyzed by multiple linear regression models. A hierarchical linear model was conducted to test the independencies of each variable’s correlation with LAD. Results: The mean LAD was 3.4 ± 0.5 cm in non-MetS subjects and 3.9 ± 0.5 cm in MetS patients (P< 0.01). None of the fasting lipid proles were associated with LAD. VLDL-χ, BMI, waist circumference, hip circumference, and blood pressure were positively correlated with LAD (all P<0.05) after adjustment for age and sex. We observed signicant interactions between VLDL-χ and blood pressure, waist circumference, and hip circumference. When adjusted for obesity- and blood pressure-related variables, 2-hour postprandial VLDL-χ (mean 1.30 ± 0.61%) showed a positive correlation with LAD in MetS patients. Each 1% VLDL-χ increase was estimated to increase LAD by 0.62 BP, blood pressure; LA, left atrium; LVEDD, left ventricle end-diastolic dimension; LVEDV, left ventricle end-diastolic volume; LVEF, left ventricular ejection fraction; E/E’ (med), the ratio of mitral peak velocity of early lling (E) to early diastolic mitral annular velocity (E') at medial mitral ring; E/E’ (lat), the ratio of mitral peak velocity of early lling (E) to early diastolic mitral annular velocity (E') at the lateral mitral ring. mass (BMI), waist circumference, hip circumference, systolic blood pressure (BP), diastolic BP, heart rate, triglycerides, very-low-density lipoprotein (VLDL)-χ, high-density lipoprotein cholesterol (HDL-C), left ventricular end-diastolic dimension (LVEDD), left ventricular end-diastolic volume (LVEDV), ejection fraction (EF), ratio of mitral ow E velocity and tissue Doppler E’ velocity of the medical and lateral mitral ring, i.e., E/E’ med and E/E’ lat.

Nevertheless, nonfasting triglyceride values have been shown to better predict cardiovascular disease risks [16,17]. The capacity to retain hemostasis upon food intake has been proposed to be important in normal human lipid and glucose metabolism [18]; a sequence of responses are elicited from the gastrointestinal system involving hormones and digestion enzymes, followed by nutritional absorption and delivery to organs and tissues. Abnormalities in postprandial lipid catabolism have been shown to be able to effectively detect early MetS [19].
Furthermore, VLDL remnants in the postprandial state have higher a nity for VLDL receptors [20]. VLDL receptors are expressed abundantly in the heart and are associated with hypertension-related ventricular hypertrophy [21]. Regarding atrial remodeling, we are the rst group to examine VLDL-induced atrial remodeling and related mechanisms [22] [9,[23][24][25]. Of note, the aforementioned changes can be induced only by VLDL in MetS patients but not VLDL in non-MetS subjects. On the basis of electrical charge (i.e., the abundance of negative charge), our colleagues developed a method for strati cation of lipoproteins and found much higher cytotoxicity associated with VLDL in MetS patients with the highest abundance of negative charge [23,26]. Here, we tested the hypothesis that VLDL may be attributed to atrial remodeling, more so in the postprandial state than in the fasting state. We also hypothesized that negatively charged VLDL (referred to as VLDL-χ) may be associated with atrial remodeling in MetS patients.
Left atrial (LA) diameter, a simple echocardiographic parameter, predicts AF recurrence risk after electrical cardioversion and is the rst reported index of structural remodeling and a widely accepted predictive marker for AF [27]. On the other hand, noninvasive electrocardiography P wave indexes, such as the P wave duration, PR interval, and terminal force of P waves, have been reported as indicators of atrial electrical dysfunction [28][29][30]. The goals of the current study were to determine the quantitative changes in lipids and glucose levels before and after meal intake and to evaluate the associations of VLDL-χ and markers of MetS with indexes of LA remodeling. The differences in postprandial variation in VLDL-χ between patients with and without LA remodeling were also evaluated.

Study participant enrollment
In this study, 167 participants were enrolled (age range, 28-74 years old). Among these patients, 87 and 80 had MetS and did not have MetS, respectively. All study participants with MetS were seen by cardiologists or endocrinologists at Kaohsiung Medical University Hospital (Kaohsiung, Taiwan). The study protocol was approved by the Kaohsiung Medical University Hospital Institutional Review Board (KMUHIRB-E(I)-20170256). An informed consent form was signed and returned by all participants before joining the study and undergoing plasma collection. This study adhered to the principles of the Declaration of Helsinki.

Sample and demographic data collection
All study participants were instructed to fast before beginning at midnight and came to the hospital at 8 AM. Each participant underwent two venous blood draws (20 mL for each sampling, in BD VACUETTE® EDTA Blood Tubes (Becton, Dickinson and Company, Franklin Lakes, NJ, USA)); one blood draw was performed in a fasting state, and the other was at a postprandial time of 0.5 h, 1 h, 2 h, or 4 h, which was randomly selected. In addition to blood sample collection, each participant underwent measurements of height, body weight, abdominal and hip circumferences, blood pressure and heart rate measurements.
The medical record, if available, was reviewed, and data related to medical history and medication use were recorded.
Laboratory testing for biochemical indicators and the quanti cation of negatively charged low-density lipoprotein (LDL-χ) and negatively charged VLDL (VLDL-χ) The analysis of biochemical parameters was performed in the Department of Laboratory Medicine at Kaohsiung Medical University Hospital according to the standard operating procedures. As described previously, pairs of plasma samples (in fasting and postprandial states) were obtained from 167 participants. Plasma samples were immediately supplemented with the following after collection to prevent bacterial contamination and oxidation: protease inhibitor cocktail (Roche Diagnostics, Indianapolis, IN), 1% penicillin/streptomycin/neomycin mixture (Invitrogen, Carlsbad, CA), and 0.5 mM EDTA. Plasma LDL and VLDL were isolated by using sequential potassium bromide density-gradient ultracentrifugation between a density range of 1.006 and 1.063 g/mL. Fractions of LDL and VLDL samples that were isolated by density were resolved into subfractions with most negatively charged lipoproteins, i.e., LDL-χ and VLDL-χ, respectively, by increasing the negative charge on UnoQ12 columns (BioRad, Hercules, CA) in the ion-exchange fast-protein liquid chromatography system (FPLC, GE Healthcare, Chicago, IL), as described previously [26,31]. In short, the columns were rst equilibrated with buffer A (0.02 M Tris-HCl, pH 8.0; 0.5 mM EDTA). With a multistep linear gradient of buffer B (1 M NaCl in buffer A) at a ow rate of 2 mL/min under observation at 280 nm, the LDL-χ and VLDL-χ subfractions were eluted, separately concentrated by using Centriprep lters (YM-30; EMD Millipore Corp., Billerica, MA) and sterilized by being passed through 0.22-µm lters. The protein concentration of the LDL and VLDL samples was measured by using the Lowry method [31,32].

Electrocardiographic parameters
Twelve-lead electrocardiography was performed by experienced medical technicians. The parameters that were measured and recorded by one experienced technician who was blinded to the other data and clinical information included P wave durations, PR intervals, QRS width, QTc intervals, and the terminal force of P waves in lead V1 [33]. Cardiac rhythms were interpreted, and any rhythm other than regular sinus rhythm (such as AF and/or utter, pacemaker rhythm, ventricular tachycardia, supraventricular tachycardia, and second-or third-degree atrial-ventricular block) was discarded before further analyses.

Echocardiographic assessment
Echocardiography was performed by one experienced cardiologist using a transthoracic cardiac probe (Vivid 7; General Electric Medical Systems, Horten, Norway), with the participant in the left decubitus position. Two-dimensional and two-dimensional-guided M-mode images were obtained, and LA diameter, left ventricle (LV) size, and LV function were assessed according to the standards of the American Society of Echocardiography [34]. LV ejection fraction (LVEF) was derived by the modi ed Simpson's method [34].
The raw data were measured and recorded while the researchers were blinded to the other data.

Statistical analysis
All continuous variables are presented as the mean ± standard deviation. For all parameters examined in this study, the Shapiro-Wilk normality test was used to determine whether a random sample of values followed a normal distribution. To compare differences between the non-MetS and MetS groups, Student's t test was used for continuous data, and a chi-square test or Fisher's exact test was used for binary data. The associations of LA diameter and electrocardiogram (ECG) parameters with body mass index (BMI), waist circumference, hip circumference, systemic blood pressure, pulse rate, fasting and postprandial plasma glucose, triglycerides, HDL-C, LDL-χ and VLDL-χ were evaluated by using the Pearson or Spearman (nonparametric) correlation. To evaluate the determining and confounding factors for LA diameter, simple linear regression and hierarchical multivariate linear regression with the stepwise method in the MetS group, non-MetS group, and total group were used to analyze all the variables, including age, BMI, waist circumference, hip circumference, heart rate, systolic blood pressure, diastolic blood pressure, VLDL-χ, triglycerides, high-density lipoprotein cholesterol, left ventricular size, LVEF, E/E' med, and E/E' lat. Results were considered statistically signi cant based on a P value < 0.05. Statistical analyses were performed by using the statistical package in GraphPad Prism (version 8; GraphPad Software, Inc., San Diego, CA, USA) software system and SPSS statistical software (version 22).

Subject characteristics
Among 80 non-MetS and 87 MetS subjects participating in this study, sex was matched between groups ( Table 1). The MetS group was older than the non-MetS group (52.7 ± 11.5 years and 44.4 ± 10.2 years). Markers of obesity, including BMI, waist circumference, and hip circumference, were greater in the MetS group. Similarly, blood pressure and heart rate were higher in the MetS group. In regard to biochemistry data, plasma hemoglobin A1c and fasting glucose were higher in the MetS group. While creatinine was similar in both groups, aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were higher in the MetS group. In regard to lipid pro les, while the MetS group had higher fasting triglycerides and HDL-C, total cholesterol and LDL-C were similar in both groups. In addition, the MetS group had a higher plasma level of uric acid. Changes in glucose and triglycerides from fasting to after a standard meal in the MetS and non-MetS groups Glucose and triglycerides were higher in the MetS group under fasting conditions, while glucose levels decreased 4 h after breakfast, and triglyceride levels were still elevated in the MetS group (Fig. 1A & B).
Changes in HDL-C, LDL-C, LDL-χ and VLDL-χ from fasting to after a standard meal in the MetS and non-MetS groups HDL-C was lower in the MetS group at fasting and exhibited a decreasing trend after the meal at 2 h and 4 h (Fig. 1C), whereas the postprandial decrease in HDL-C was not present in the non-MetS group. Unlike the results for HDL-C, the plasma LDL-C at fasting and the change after a meal were similar in both groups (Fig. 1D). While LDL-C did not change after a meal, the negatively charged LDL and LDL-χ in the MetS group decreased 4 h after the meal (Fig. 1E). The decrease at 4 h in LDL-χ did not occur in the non-MetS group. At fasting (0 h, Fig. 1F), the negatively charged VLDL-χ was similar in both groups but was signi cantly higher in the MetS group than in the non-MetS group 0.5 h after the meal. VLDL-χ decreased 4 h after the meal in both groups.
LA dilation as a marker for atrial structural remodeling Among 167 non-MetS and MetS subjects participating in this study, we recruited subjects aged between 40 and 60 years to undergo 12-lead electrocardiography and standard echocardiography. The characteristics of 62 non-MetS and 68 MetS subjects who had completed the heart examinations are shown in Table 2. Similar to all participants, the MetS group was more obese, had higher blood pressure, and had a higher heart rate than the non-MetS group. Regarding electrocardiographic parameters, the PR interval was wider in the MetS group than in the non-MetS group (167.6 ± 20.0 msec and 156.2 ± 24.9 msec), whereas the P wave duration, terminal force of P waves, QRS width and QTc interval were similar in both groups. Regarding echocardiography, LV function was similar in both groups (LVEF, MetS 67.3 ± 7.1% and non-MetS 63.9 ± 11.3%), whereas the diameter of the aorta, LA and size of LV were larger in the MetS group ( Table 2). The ratios of mitral peak velocity of early lling (E) to early diastolic mitral annular velocity (E') as markers for diastolic function and indicators for LV lling pressure were higher in the MetS group ( Table 2). The aforementioned results showed evidence of LA and LV remodeling in MetS participants in this study, and this remodeling was associated with diastolic dysfunction of the LV without overt systolic dysfunction.

LA dilation was associated with central obesity and hypertension
The LA diameter was shown to signi cantly correlate with BMI and waist and hip circumferences in both the non-MetS ( Fig. 2A-C) and MetS (Fig. 2D-F) groups. The LA diameter was also positively correlated with blood pressure (Fig. 3) and with diastolic blood pressure in the MetS group (Fig. 3E).

Correlation of LA size with fasting and postprandial VLDL-χ
The results of the analysis of the correlation between LA diameter and fasting and 0.5-h, 1-h, 2-h, and 4-h postprandial VLDL-χ are shown in Fig. 4. In the MetS group, the LA size was signi cantly correlated with fasting ( Fig. 4F) and 2-h postprandial VLDL-χ (Fig. 4I). Similarly, the LA size was also signi cantly correlated with 2-h postprandial VLDL-χ in the non-MetS group (Fig. 4D).

Two-hour postprandial VLDL-χ is correlated with P wave duration
Markers of electrical remodeling on P waves were measured and analyzed (Fig. 5). The P wave duration was signi cantly correlated with 2-h postprandial VLDL-χ in the MetS group (Fig. 5A), and the PR interval tended to be correlated with VLDL-χ ( Fig. 5B) but without a signi cant correlation in the non-MetS group (Fig. 5D-F).

Different kinetics of VLDL-χ in MetS subjects with larger LA size
The changes in VLDL-χ from fasting to 2 h after the meal are shown in Fig. 6. The pattern of changes was different between the non-MetS and MetS groups (Fig. 6A). In the MetS group with dilated LA (diameter ≧ 3.8 cm), VLDL-χ increased after a meal compared with the level in the fasting state, and VLDL-χ remaining steady in those with LA diameters smaller than 3.8 cm (Fig. 6B).
There was a signi cant correlation between VLDL-χ and blood pressure and waist and hip circumference. After adjusting for variables of obesity and blood pressure, VLDL-χ showed an independent positive correlation with LA diameter in the MetS group (Table 3). Each 1% increase in VLDL-χ was estimated to increase the LA diameter by 0.62 cm. The other independent determining factor of LA diameter is waist circumference. For all subjects in this study, the univariate analysis revealed four factors that were signi cantly correlated with LA diameter, including waist circumference, LVEDD, left ventricular end-diastolic volume (LVEDV), and E/E' med. Among those factors, waist circumference and LVEDD remained independent determinants of LA diameter in multivariate analysis (see suppl. Table). Among non-MetS subjects, age, BMI, and LVEDD were signi cantly and independently correlated with LA diameter (see suppl. Table).

Discussion
The main results from this study are as follows: (1)  and (5) MetS subjects whose VLDL-χ increased after a meal had a larger LA diameter than those whose VLDL-χ decreased.
The metabolism of VLDL is an important determinant of LDL formation and triglyceride levels. Both normal and hypertriglyceride VLDL molecules contain apoB-100 and apoE. Hypertriglyceridemic VLDL has a higher content of apoE, which facilitates its a nity to VLDL receptors [35]. The tissue distribution of VLDL receptors is quite different from that of LDL receptors. The VLDL receptor is abundant in tissue largely utilizing fatty acids as an energy source, including heart, muscle, and adipose tissue, and it mediates cellular catabolism and regulates the levels of lipoprotein lipase [36,37]. Whether the modulation of VLDL receptor expression is involved in VLDL-χ-related LA remodeling needs further study.
The importance of postprandial hyperlipidemia in the context of atherosclerosis was suggested early in 1979 [38] and was later increasingly explored in recent years [10,16,17]. In 2016, experts reached a consensus regarding the complementary value of nonfasting measurements of lipid pro les for clinical and laboratory medicine [39]. More recently, Nakajima et al. reported evidence to disclose the predictive role of remnant lipoproteins in coronary heart diseases [20]. They also reported that compared to fasting VLDL, postprandial VLDL has a higher a nity for VLDL receptors and easier internalization [40]. Accordingly, we suggest that the postprandial VLDL-χ has larger internalization and results in cytotoxicity in atrial tissue and ultimately signi cant remodeling of the atrium. To the best of our knowledge, our study is the rst to report the correlation of postprandial VLDL-χ with LA remodeling.
The association of markers of central obesity with LA size is consistent with results derived by other clinical studies [41,42] The pathogenesis of AF is complex and can involve a combination of etiological factors and metabolismassociated elements. AF is a common outcome of combining atrial dilatation with atrial brosis, triggered activities and electrical automaticity. In many patients, AF is not curable at the time of diagnosis due to long-lasting and signi cant atrial remodeling with extensive brosis. To intervene at the early stage of atrial remodeling, i.e., application of the "upstream therapy" approach, may be possible to prevent AF by targeting the cytotoxic fraction of postprandial VLDL that leads to the progression of atrial cardiomyopathy.

Conclusion
We observed a correlation between postprandial plasma VLDL-χ and LA remodeling. Despite other factors, such as hypertension and obesity, which also cause LA remodeling in MetS patients, plasma VLDL-χ is still attributed to LA remodeling. We suggest that the metabolism of VLDL and the characteristics of postprandial VLDL-χ are the major determinants of the progression of LA remodeling, especially in MetS patients.  Figure 1 Plasma concentrations of glucose and lipids before (0 h) and after breakfast (0.5-h, 1-h, 2-h, and 4-h) in the group of healthy controls (in blue) and the group of individuals with metabolic syndrome (in brown).

Figures
Signi cant differences between groups are shown as follows: P< 0.001 as ***, P< 0.01 as **, and P< 0.05 as *.      Correlation between electrocardiographic parameters that indicate electrical remodeling of the atria (P wave duration, PR interval and terminal force of P wave) and negatively charged very-low-density lipoprotein as a percentage (VLDL-χ) 2 h after consuming a meal in individuals without (in blue, A-C, n=15) and with metabolic syndrome (in brown, D-F, n=15). A signi cant correlation (with P< 0.05) is indicated with each correlation factor and P value.