Epicardial Adipose Tissue and Left Ventricular Hypertrophy in Hypertensive Patients With Preserved Ejection Fraction: A Multicenter Retrospective Cohort Study

ABSTRACT

This study aimed to investigate the correlation of the increased volume index of epicardial adipose tissue (EAT) and left ventricular hypertrophy (LVH) in patients with Hypertension (HTN). A total of 209 HTN patients and 50 healthy controls, who underwent cardiovascular magnetic resonance (CMR) at two medical centers in China between June 2015 and October 2024, were enrolled for this study. Postprocessing and imaging analysis were conducted and EAT measurements were performed. Restricted cubic splines (RCS) were used to explore the potential relationship with LVH. Binary logistic regression models and mediation analyses were employed to evaluate the association between EAT volume and CMR parameters as well as LVH. Hypertensive patients with LVH exhibited larger indexed EAT volumes, more pronounced diffuse fibrosis, and reduced left ventricular strain compared to hypertensive patients without LVH (all p < 0.001), with results remaining stable after adjusting for confounding factors. The variables that were significant in the univariate regression were included in the multivariate logistic regression model, indicating that indexed EAT volume (p = 0.001), extracellular volume (ECV) (p = 0.012), and global longitudinal strain (GLS) (p = 0.024) were independently associated with LVH. These associations remained stable after adjusting for confounding factors. Mediation analysis further revealed that the relationship between increased EAT volume and LVH was mediated by ECV, native T1, GLS, global circumferential strain (GCS), and global radial strain (GRS) (p < 0.05). These findings imply that EAT is independently linked to LVH in hypertensive patients. The association between EAT and LVH in hypertensive patients may be mediated by myocardial fibrosis or dysfunction.

1. Introduction

Hypertension (HTN) is the leading risk factor for morbidity and mortality worldwide [1]. The left ventricle (LV) is the main target of HTN‐induced end‐organ damage. Patients with long‐standing or poorly controlled HTN face an elevated risk of developing left ventricular hypertrophy (LVH) and impaired diastolic function. In addition to being a manifestation of end‐organ response, LVH also acts as an independent risk factor for cardiovascular disease (CVD) [2]. Inducing regression of LVH can effectively mitigate the incidence of major cardiovascular events and enhance the prognosis for patients with HTN.

Epicardial adipose tissue (EAT) is a unique fat depot situated between the myocardium and the visceral layer of the epicardium [3]. EAT plays a pivotal role in the progression and development of illness such as coronary artery disease (CAD), atrial fibrillation (AF), and heart failure. Studies have demonstrated that EAT exerts proinflammatory effects, thereby contributing to the development of coronary atherosclerosis [4]. Moreover, it induces myocardial fibrosis and subsequent abnormalities in the cardiac conduction system, ultimately leading to AF development [5, 6]. Furthermore, there exists a strong association between EAT and the progressive deterioration in left ventricular diastolic and systolic function in heart failure patients [7, 8]. The accumulation of EAT has been associated not only with ventricular hypertrophy, diastolic dysfunction, and increased cardiac filling pressures but also with endothelial dysfunction and AF, all of which are highly prevalent in HTN. However, few studies have investigated the role of EAT in LVH among hypertensive patients.

Previous studies have shown that individuals with HTN exhibit increased EAT volume compared to normal controls. Increased diffuse fibrosis in the myocardium has been associated with LVH in HTN [9]. However, the precise factors influencing LVH remain unclear, necessitating further investigation into whether EAT plays a role in its development among patients with HTN. As a modifiable risk factor, EAT may provide a new target for pharmacological intervention in hypertensive heart disease [10].

2. Materials and Methods

2.1. Subjects

The study was approved by the Ethics Review Board, and all subjects provided written informed consent. A total of 133 subjects with HTN LVH, 76 subjects with HTN non‐LVH, and 50 normotensive controls were enrolled at two medical centers in China (Figure 1). Patients with a history of HTN were included in the study. Exclusion criteria were as follows: (1) patients with other causes of LVH, (2) patients with concomitant CAD or significant valvular disease, (3) patients with concomitant renal impairment with glomerular filtration rate <45 mL/min/1.73 m², and (4) patients with reduced systolic function (LV ejection fraction [EF] < 45%) [11]. Subjects meeting the criteria for HTN including systolic blood pressure (SBP) ≥ 130 mm Hg and/or diastolic blood pressure (DBP) ≥80 mm Hg on at least two office readings [12], or those taking one or more medications for HTN were included. Subsequently, subjects were classified as having LVH if their left ventricular mass indexed by body surface area (LVMI) measured by cardiac magnetic resonance imaging exceeded >81 g/m² for men or >61 g/m² in women based on the methodology employed by Kuruvilla et al. [11]. Hypertensive subjects not meeting criteria for LVH as defined in the preceding text were included in the HTN non‐LVH group. Fifty age‐ and sex‐matched healthy subjects were selected as controls, satisfying the following criteria: normal physical examination, optimal blood pressure levels (SBP < 130 mm Hg and DBP < 80 mm Hg), unremarkable electrocardiogram (ECG) findings, absence of chest pain or dyspnea history, no diabetes or hyperlipidemia, and normal results from both 2D echocardiography and Doppler examinations. None of the participants were taking any medications. Any potential subjects displaying evidence of heart disease, HTN, or other systemic disorders were excluded from this study.

FIGURE 1.

Flowchart shows the selection process of patients with hypertension based on inclusion and exclusion criteria. CMR, cardiac MR; LVEF, left ventricular ejection fractions.

2.2. Cardiovascular Magnetic Resonance (CMR) Protocol

The CMR examination was performed using a 3.0‐T scanner (MAGNETOM Prisma, Siemens Healthcare) equipped with an 18‐channel phased‐array body coil. Standard protocols were utilized to perform CMR examinations [13], which included breath‐hold cine imaging, pre‐ and post‐enhancement T1mapping imaging, and steady‐state free precession (SSFP). The following were typical parameters for the cine picture protocol: Flip Angle: 80°, repetition time (TR)/echo time (TE) = 1.43/3.26 ms, slice thickness: 7 mm. Ten minutes following an intravenous dose of 0.2 mmol/kg gadopentetate dimeglumine (North Road Pharmaceutical Co., Ltd.), CMR scans were performed again. Three short‐axis levels (basal, middle, and apical) of myocardial T1 values were recorded under breath holding and ECG gating using a modified Look‐Locker inversion recovery procedure. Before and after administration, T1 mapping was carried out using the 5(3 s)3 and 4(1 s)3(1 s)2 procedures, respectively. The scanning parameters for T1 mapping before contrast agent injection were as follows: a bSSFP single‐shot readout with a 35° excitation flip angle, rate‐2 parallel imaging, matrix size of 256 × 164, pixel size of 1.3 mm × 1.3 mm, slice thickness of 8 mm, minimum inversion time (TI) of 189 ms, and incremented by 80 ms, TE/TR echo spacing of 1.15 ms/2.77 ms. After the contrast agent injection, the T1 imaging parameters were modified to include a bSSFP single‐shot readout with a reduced excitation flip angle of 20°, rate‐2 parallel imaging, matrix size of 192 × 164, pixel size of 1.9 mm × 1.9 mm, slice thickness of 8 mm, minimum TI of 100 ms with 80 ms increments, and TE/TR echo spacing of 1.01 ms/2.44 ms [14, 15].

2.3. Post‐Processing and Imaging Analysis

Two experienced CMR physicians, who were blinded to clinical information, independently contoured and analyzed all CMR images. The commercial post‐processing software CVI42 (version 5.12) was utilized for processing the CMR images (A–D of Figure 2). Manual delineation of endocardial and epicardial contours was performed for the basal, mid‐, and apical segments of the LV. Raw and post‐contrast T1 values were calculated from the excluded apex region based on the 17‐segment American Heart Association (AHA) model [16, 17]. Every slice's papillary muscle, moderator bands, and epicardial boundary were carefully removed [18, 19, 20]. Before calculating extracellular volume (ECV), hematocrit levels were obtained, and the blood pool was manually drawn [21, 22]. ECV was calculated using the following equation [23]:

$$\begin{array}{ccc}& & MyocardialECV=(1-Hematocrit)\hfill \\ & & \times \frac{\left(\frac{1}{\mathrm{Post}\mathrm{contrast}\mathrm{myocardial}\mathrm{T}1}-\frac{1}{\mathrm{Native}\mathrm{myocardial}\mathrm{T}1}\right)}{\left(\frac{1}{\mathrm{Post}\mathrm{contrast}\mathrm{blood}\mathrm{T}1}-\frac{1}{\mathrm{Native}\mathrm{blood}\mathrm{T}1}\right)}\hfill \end{array}$$

FIGURE 2.

A 68‐year‐old male with HTN non‐LVH (Left images) and a 52‐year‐old male with HTN LVH (right images). The top row images show measurement of EAT volume and extracellular volume (ECV) fraction. HTN non‐LVH group had less EAT volume and lower ECV fraction, and HTN LVH group had higher EAT volume and higher ECV fraction. The bottom row images illustrates the process by which EAT induces LVH by promoting myocardial fibrosis.

The global longitudinal strain (GLS), global circumferential strain (GCS), and global radial strain (GRS) of the LV were determined using feature tracking methods on SSFP cine sequences. The tissue signal intensity module was utilized to measure the EAT volume. EAT was manually delineated on end‐diastolic short‐axis slices, starting from the most basal slice around the atria and progressing toward the most apical slice surrounding the ventricles (Figure 2).

2.4. Statistical Analysis

Statistical analyses were performed using SPSS statistical software (version 26.0) and R (version 4.0.2). All tests were two‐tailed, with a significance level of p < 0.05. The distribution of variables was assessed using the Shapiro–Wilk test. Variables were expressed as mean ± SD or median with IQR and compared using either a two‐sample Student's t‐test or Wilcoxon rank‐sum test based on their distribution characteristics. Categorical variables were presented as frequencies with percentages and analyzed by Chi‐square test for comparison purposes. To determine the appropriate sample size, we performed a priori power analysis using GPower 3.1. A binary logistic regression model was employed with an odds ratio of 3.47, an alpha level of 0.05, a statistical power (1−β) of 0.8, R ² other X = 0.345, and X param π = 0.64.

Restricted cubic splines (RCS) were used to explore the potential associations among indexed EAT volume, cardiac magnetic resonance parameters, and LVH in patients with HTN. Binary logistic regression analysis was employed to evaluate the independent impact of CMR‐based parameters, clinical characteristics, and EAT volume on clinical outcomes. Morphological, functional parameters, and EAT volume were included as continuous in univariable analysis. Significant variables (p < 0.05) from univariate analysis were included in multivariate models, with the following adjustments:

Model 1: Adjusted for age, sex, BMI, SBP, and DBP.

Model 2: Adjusted for history of diabetes, smoking, drinking, and family history of HTN.

Model 3: Adjusted for antihypertensive and lipid‐lowering medication history.

Model 4: Adjusted for triglycerides, HDL, LDL, waist and hip circumference.

Model 5: Employed backward selection with all variables.

Subgroup analyses were used to test the robustness of the results. Furthermore, a mediation analysis was conducted to examine whether any cardiac measures or clinical characteristics influenced the association between indexed EAT volumes and clinical outcomes. The mediation analysis was performed using R software version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria), with p value < 0.05 considered statistically significant. Firstly, the clinical characteristics or cardiac measures that simultaneously correlated with both indexed EAT volume and LVH were selected. Next, the shortlisted indicators were included as mediating variables in the mediation analysis [24], with indexed EAT volume as the independent variable and LVH as the dependent variable. Intra‐ and inter‐observer variability of T1 values, ECV, and indexed EAT volumes were estimated using the intraclass correlation coefficient (ICC).

3. Results

3.1. Patient Characteristics

The baseline characteristics of our study cohort are presented in Table 1. Our patient cohort consisted of 209 patients with HTN, including 157 men (75%). The median age [IQR] was 53 [37,63] years old. We also included 50 sex‐ and age‐matched healthy controls for comparison. No significant differences were observed regarding BMI and clinical history between the HTN patients and healthy controls. However, it is worth noting that both SBPs and DBPs were significantly higher in the HTN LVH group compared to the HTN non‐LVH group and control subjects. Additionally, lipid levels were generally higher in HTN LVH patients than in HTN no LVH patients, consistent with clinical patterns.

TABLE 1.

Baseline clinical characteristics: Healthy controls and patients with Hypertension.

Patients with hypertensive heart disease
	Hypertensive Non‐LVH (n = 76)	Hypertensive LVH (n = 133)	P*	All patients with Hypertensive (n = 209)	Healthy controls (n = 50)	P¥
Age (years)	57 [39, 64]	51 [37, 61]	0.214	53 [37, 63]	50 [35, 63]	0.557
Male sex, n (%)	60 (79)	97 (73)	0.333	157 (75)	33 (66)	0.190
BMI (kg/m2)	27.06 [24.69, 29.64]	27.01 [24.69, 30.80]	0.486	27.01 [24.69, 30.12]	26.12 [24.84, 30.45]	0.741
Systolic BP (mm Hg)	160 [146, 170]	160 [149, 180]	0.052	160 [148, 178]	115 [103, 123]	<0.001
Diastolic BP (mm Hg)	92 [86, 100]	97 [89, 110]	0.021	95 [87, 103]	72 [64, 77]	<0.001
Heart rate (beats/min)	70 [64, 77]	68 [62, 76]	0.317	69 [63, 76]	70 [58, 78]	0.828
Diabetes, n (%)	15 (20)	23 (17)	0.660	37 (19)	—	—
Smoking, n (%)	31 (41)	51 (38)	0.728	76 (40)	—	—
Drinking, n (%)	30 (39)	56 (42)	0.710	81 (43)	—	—
Family history, n (%)	10 (13)	26 (20)	0.239	34 (18)	—	—
ACEI/ARB (n, %)	36 (47)	67 (50)	0.676	93 (49)	—	—
Bta‐blocker, n (%)	31 (41)	52 (39)	0.810	75 (39)	—	—
Diuretic (n, %)	19 (25)	47 (35)	0.122	63 (33)	—	—
Statins (n, %)	28 (37)	64 (48)	0.114	82 (43)	—	—
Calcium channel blocker (n, %)	20 (26)	39 (29)	0.642	54 (28)	—	—
Plasma triglycerides (mmol/L)	1.34 [0.92, 2.38]	1.83 [1.27, 2.84]	0.005	1.73 [1.18, 2.74]	—	—
Total cholesterol (mmol/L)	3.20 [2.26, 4.01]	3.92 [3.04, 4.70]	<0.001	3.62 [2.68, 4.59]	—	—
HDL (mmol/L)	1.26 [1.06, 1.52]	1.17 [0.94, 1.32]	0.008	1.22 [0.98, 1.35]	—	—
LDL (mmol/L)	2.52 [2.12, 3.06]	2.95 [2.52, 3.27]	0.003	2.89 [2.33, 3.27]	—	—
Waist circumference (cm)	82 [75, 91]	82 [75, 92]	0.772	82 [75, 91]	—	—
Hip circumference (cm)	97 [92, 105]	97 [92, 104]	0.758	97 [92, 104]	—	—

Notes: P*‐value comparing patients with left ventricular hypertrophy and without left ventricular hypertrophy, P¥‐value comparing all patients with hypertensive and healthy controls.

Abbreviations: BMI, body mass index; BP, blood pressure; HTN, hypertension; LVH, left ventricular hypertrophy.

p < 0.05 indicates statistical significance.

3.2. Association of Cardiac Magnetic Resonance Parameters and LVH

The CMR characteristics of both HTN patients and healthy controls are summarized in Table 2. Patients with HTN showed larger LV volume and mass, as well as lower LVEF, compared to the control group. Similarly, HTN patients with LVH demonstrated greater LV volume and mass, along with lower LVEF, when compared to those without LVH.

TABLE 2.

CMR characteristics: Healthy controls and patients with hypertension.

Patients with hypertensive heart disease
	Hypertensive	Hypertensive	P*	All patients with hypertensive	Healthy controls	P¥
	Non‐LVH	LVH
	(n = 76)	(n = 133)		(n = 209)	(n = 50)
Native T1 (ms)	1249.13 ± 45.72	1274.09 ± 49.07	<0.001	1265.01 ± 49.36	1233.12 ± 41.85	<0.001
Extracellular volume	28.03 ± 2.96	30.77 ± 3.58	<0.001	29.77 ± 3.62	26.22 ± 2.07	<0.001
GRS (%)	25.84 [20.44, 30.08]	20.10 [15.80, 24.72]	<0.001	21.70 [16.81, 27.10]	35.16 [30.90, 38.81]	<0.001
GCS (%)	−16.55 ± 3.72	−14.87 ± 4.03	0.003	−15.48 ± 4.00	−19.62 ± 2.65	<0.001
GLS (%)	−14.84 ± 3.01	−12.70 ± 3.51	<0.001	−13.48 ± 3.49	−16.32 ± 2.10	<0.001
Index EAT volume (mL)	64.95 ± 15.70	79.05 ± 17.16	<0.001	73.92 ± 17.98	51.66 ± 13.01	<0.001
LVEF (%)	58 [49, 67]	51 [46, 60]	0.002	53 [47, 64]	64 [60, 72]	<0.001
LVEDVi (mL/m2)	70.19 [55.50, 88.99]	83.51 [67.63, 112.09]	<0.001	76.84 [62.64, 105.28]	64.12 [56.74, 79.23]	<0.001
LVESVi (mL/m2)	29.39 [20.28, 43.50]	43.72 [27.06, 64.02]	<0.001	36.13 [22.94, 58.22]	23.86 [17.94, 32.83]	<0.001
LVCI (L/min/m2)	2.85 [2.20, 3.47]	2.70 [2.15, 3.38]	0.533	2.74 [2.19, 3.43]	2.82 [2.43, 3.17]	0.676
LV mass index (g/m2)	61.30 [54.79, 71.86]	102.82 [87.53, 122.82]	<0.001	86.88 [67.60, 109.67]	54.06 [46.34, 60.60]	<0.001

Notes: P*‐value comparing patients with left ventricular hypertrophy and without left ventricular hypertrophy, P¥‐value comparing all patients with hypertensive and healthy controls.

Abbreviations: EAT, epicardial adipose tissue; ECV, extracellular volume; GCS, global circumferential strain; GLS, global longitudinal strain; GRS, global radial strain; LVCI, left ventricular cardiac index; LVH, left ventricular hypertrophy; LVEF, left ventricular ejection fractions; LVEDVi, left ventricular end‐diastolic volume index; LVESVi, left ventricular end‐systolic volume index; LVMI, left ventricular mass index.

Compared to healthy controls, patients with HTN exhibited longer native T1 values, ECV fraction, and indexed EAT volume, lower GRS, GCS, and GLS (All p < 0.001). Likewise, patients with HTN LVH had longer native T1 values, ECV fraction and indexed EAT volume (Figure 3), lower GRS, GCS, and GLS compared to those with only HTN (GCS p = 0.003, other p < 0.001). Figure 4 illustrate the results of the RCS analysis, demonstrating a linear relationship between the cardiac magnetic resonance parameters and LVH. However, as EAT, native T1, ECV increased and GRS, GCS, GLS were impaired, the risk of LV hypertrophy in patients tended to rise.

FIGURE 3.

Bar graph of the proportion of patients with high or low index EAT volume in HTN non‐LVH and HTN LVH. Bar graphs for “above median” (high) versus "below median (low) index EAT volume are shown in orange and blue. HTN, hypertension; LVH, left ventricular hypertrophy; EAT, epicardial adipose tissue.

FIGURE 4.

RCS curve of CMR parameters and index EAT volume. ECV, extracellular volume; EAT, epicardial adipose tissue; GRS, global radial strain; GCS, global circumferential strain; GLS, global longitudinal strain.

3.3. Univariate and Multivariate Logistic Regression of Variables for Magnetic Resonance Parameters and LVH

The power analysis indicated that the sample size of 205 individuals could detect a moderate main effect of LV hypertrophy in hypertensive patients at a significance level (α) of 0.05 or less and a statistical power (1−β) of 0.8. As indicated in Table 3, after adjusting for common risk factors, indexed EAT volume (p < 0.001), native t1 (p = 0.001), ECV (p < 0.001), GRS (p = 0.003), GCS (p = 0.023), GLS (p < 0.001), LVEDVI (p = 0.001), and LVESVI (p = 0.001) demonstrated a significant association with LVH. Subsequently, variables that showed a significant association with LVH group in the univariate analysis (p < 0.05) were included in the multivariate model. The results presented in Table 4 demonstrate that in the multivariate logistic regression model, indexed EAT volume (p = 0.001), ECV (p = 0.012) and GLS (p = 0.024) continue to be significantly associated with LVH. ECV (p = 0.007), indexed EAT volume (p < 0.001) and GLS (p = 0.029) remained associated with LVH after adjusting for age, sex, BMI, SBP, and DBP (model 1). Furthermore, after adjusting for history of diabetes, smoking, drinking, and family history of HTN (model 2), ECV (p = 0.011), indexed EAT volume (p = 0.001) and GLS (p = 0.019) were also linked to the LVH group. With adjustments for baseline HTN medications (model 3), ECV (p = 0.019), indexed EAT volume (p = 0.002), and GLS (p = 0.034) remained significantly associated with the LVH. The results remained stable after adjustment for plasma triglycerides, HDL, and LDL (model 4) (ECV: p = 0.032, indexed EAT volume: p = 0.001, GLS: p = 0.019). In the backward selection model (model 5), ECV (p = 0.011), indexed EAT volume (p = 0.001), GLS (p = 0.012) and LVEDVi (p = 0.008) maintained a significant association with the LVH.

TABLE 3.

Associations of cardiac measures with LVH.

	OR	95%CI	p value	OR_adjusted※	95% CI_adjusted※	p value_adjusted※
nativeT1	1.011	[1.005, 1.018]	0.001	1.014	[1.006, 1.023]	0.001
ECV	1.283	[1.162, 1.418]	<0.001	1.315	[1.168, 1.481]	<0.001
GRS	0.923	[0.887, 0.961]	<0.001	0.930	[0.887, 0.976]	0.003
GCS	1.114	[1.034, 1.199]	0.004	1.110	[1.014, 1.214]	0.023
GLS	1.206	[1.102, 1.319]	<0.001	1.222	[1.098, 1.360]	<0.001
Index EAT volume	1.054	[1.033, 1.076]	<0.001	1.062	[1.035, 1.090]	<0.001
HR	0.986	[0.960, 1.013]	0.308	0.977	[0.948, 1.007]	0.135
LVEF	0.961	[0.935, 0.989]	0.006	0.971	[0.938, 1.005]	0.095
LV cardiac index	0.922	[0.676, 1.256]	0.606	0.972	[0.666, 1.419]	0.884
LVEDV index	1.017	[1.007, 1.028]	0.001	1.021	[1.009, 1.033]	0.001
LVESV index	1.024	[1.011, 1.038]	<0.001	1.025	[1.010, 1.040]	0.001

Notes: ※Adjusted for age, sex, body mass index, systolic blood pressure, diastolic blood pressure, history of diabetes smoking, drinking, family history of HTN, history of taking drugs (ACEI/ARB, Bta‐blocker, diuretic, statins, calcium channel blocker), triglycerides, HDL, LDL, waist, and hip circumference.

Bold values indicate p < 0.05 after adjusting for confounding factors.

Abbreviations: EAT, epicardial adipose tissue; ECV, extracellular volume; GCS, global circumferential strain; GLS, global longitudinal strain; GRS, global radial strain; HR, heart rate; LVCI, left ventricular cardiac index; LVEF, left ventricular ejection fractions; LVEDVi, left ventricular end‐diastolic volume index; LVESVi, left ventricular end‐systolic volume index; LVMI, left ventricular mass index.

TABLE 4.

Multivariable logistic regression.

	OR (95%CI)		OR (95%CI)		OR (95%CI)		OR (95%CI)		OR (95%CI)		OR (95%CI)
	Unadjusted	p value	Model 1	p value	Model 2	p value	Model 3	p value	Model 4	p value	Model 5	p value
NativeT1	0.997 (0.988–1.006)	0.502	0.996 (0.986–1.006)	0.222	0.996 (0.987–1.005)	0.608	0.999 (0.989–1.008)	0.788	0.997 (0.988–1.006)	0.514
ECV	1.174 (1.038–1.338)	0.012	1.210 (1.056–1.397)	0.007	1.183 (1.043–1.352)	0.011	1.163 (1.028–1.324)	0.019	1.156 (1.015–1.323)	0.032	1.178 (1.041–1.343)	0.011
Index EAT volume	1.041 (1.018–1.066)	0.001	1.050 (1.025–1.077)	0.000	1.045 (1.020–1.072)	0.001	1.038 (1.014–1.063)	0.002	1.041 (1.017–1.067)	0.001	1.043 (1.019–1.070)	0.001
GRS	0.948 (0.865–1.037)	0.246	0.952 (0.860–1.051)	0.328	0.959 (0.871–1.053)	0.382	0.952 (0.867–1.043)	0.293	0.949 (0.866–1.039)	0.263
GCS	0.838 (0.694–1.000)	0.059	0.852 (0.699–1.029)	0.852	0.845 (0.696–1.011)	0.076	0.841 (0.696–1.003)	0.063	0.845 (0.698–1.009)	0.072	0.904 (0.787–1.030)	0.139
GLS	1.183 (1.026–1.337)	0.024	1.182 (1.020–1.381)	0.029	1.197 (1.034–1.400)	0.019	1.173 (1.016–1.368)	0.034	1.196 (1.035–1.398)	0.019	1.210 (1.048–1.414)	0.012
LVEDVi	1.008 (0.985–1.038)	0.572	1.013 (0.988–1.048)	0.404	1.010 (0.986–1.040)	0.475	1.008 (0.984–1.040)	0.586	1.010 (0.986–1.042)	0.501	1.016 (1.005–1.030)	0.008
LVESVi	1.008 (0.968–1.043)	0.684	1.001 (0.958–1.038)	0.948	1.006 (0.967–1.042)	0.741	1.009 (0.968–1.046)	0.652	1.006 (0.965–1.042)	0.770

Notes: Model 1 adjusted for age, sex, body mass index, systolic blood pressure, diastolic blood pressure. Model 2 adjusted for history of diabetes, smoking, drinking, and family history of HTN. Model 3 adjusted for history of taking drugs (ACEI/ARB, Bta‐blocker, diuretic, statins, calcium channel blocker). Model 4 adjusted for triglycerides, HDL, LDL, waist and hip circumference. Model 5 adjusted for all variables in backward selection model.

Bold values indicate p < 0.05 in the multivariate model.

Abbreviations: EAT, epicardial adipose tissue; ECV, extracellular volume; GCS, global circumferential strain; GLS, global longitudinal strain; GRS, global radial strain; LVEDVi, left ventricular end‐diastolic volume index; LVESVi, left ventricular end‐systolic volume index.

To assess the robustness of the multivariate model results, we categorized patients according to age, sex, BMI, clinical history, and specific antihypertensive medications, indicating a consistent link between the ECV, indexed EAT volume and LVH remained consistent across different patient groups (p values for interaction > 0.05). However, subgroup analysis revealed that the direct correlation between GLS and LVH was attenuated in patients on antihypertensive medications. The connections between EAT, ECV, GLS, and LVH were all diminished in diabetic patients (Figure S1).

3.4. Mediation Analysis

The prespecified parameters of cardiac magnetic resonance were evaluated for their association with the risk of HTN LVH and indexed EAT volume before conducting mediating analyses (Table S1). After adjusting for confounding factors, among the five indicators (native T1, ECV, GRS, GCS, GLS) that were simultaneously related to the risk of HTN LVH and indexed EAT volume. ECV exhibited a mediating effect on the relationship between indexed EAT volume and LVH (OR [95% CI], 1.023 [1.017–1.030], p < 0.05). Consequently, the indirect influence of ECV further strengthened the association between larger EAT volume and LVH (OR [95% CI], 1.048 [1.021–1.077], p < 0.001), resulting in an overall greater impact (OR [95% CI], 1.062 [1.036–1.091], p < 0.001). Additionally, native T1, GRS, GCS, GLS also exhibited a modest mediating effect between the index of EAT volume and LVH (OR [95% CI], native T1: 1.017 [1.012–1.023], GRS: 1.007 [1.004–1.011], GCS: 1.004 [1.002–1.008], GLS: 1.007 [1.004–1.012], p < 0.05) (Table S2, Figure 5).

FIGURE 5.

CMR parameters mediate the effect of EAT on the risk of HTN LVH. EAT, epicardial adipose tissue; ECV, extracellular volume; GRS, global radial strain; GCS, global circumferential strain; GLS, global longitudinal strain; HTN, hypertension; LVH, left ventricular hypertrophy. ※Adjusted for age, sex, body mass index, systolic blood pressure, diastolic blood pressure, history of diabetes smoking, drinking, family history of HTN, history of taking drugs (ACEI/ARB, Bta‐blocker, diuretic, statins, calcium channel blocker), triglycerides, HDL, LDL, waist, and hip circumference.

3.5. Intra‐ and Interobserver Variability

The reproducibility of native T1 values (intra: ICC = 0.96, inter: ICC = 0.98), ECV (intra: ICC = 0.93, inter: ICC = 0.96), and indexed EAT volume (intra: ICC = 0.94, inter: ICC = 0.97) demonstrated excellent intra‐ and inter‐observer agreement.

4. Discussion

LVH is a significant clinical outcome observed in patients with chronically uncontrolled HTN, and it is closely associated with poor prognosis. EAT has been shown to induce myocardial fibrosis and impaired myocardial strain through the release of proinflammatory and fibrotic cytokines. Therefore, elucidating the involvement of epicardial fat in the pathogenesis of LVH among hypertensive patients holds substantial implications. However, the specific role played by EAT in LVH remains unclear. To our knowledge, this multicenter study represents the first investigation exploring the relationship between EAT and LVH in individuals with HTN. The key findings from our research are as follows: (1) patients with HTN LVH exhibited significantly higher levels of native t1, ECV, indexed EAT volume, and lower levels of GRS, GCS, and GLS compared to HTN non‐LVH. The same relationship occurred in both patients with HTN and healthy controls; (2) binary logistic regression analysis demonstrated that even after adjusting for common confounding factors, ECV, EAT, and GLS still maintained their independent correlation with left ventricular remodeling; and (3) Furthermore, we found that ECV, native T1, GRS, GCS, GLS act as mediators in linking the EAT volume with ventricular remodeling. This suggests a potential role of EAT in promoting diffuse fibrosis within the ventricles and subsequently contributing to ventricular remodeling.

The present study revealed a significant increase in the volume of EAT in hypertensive patients with LVH compared to those without LVH. Similarly, the volume of EAT was significantly higher in hypertensive patients than in healthy controls, supporting previous findings by Austys et al. [25] that showed a greater accumulation of EAT in hypertensive individuals compared to normotensive subjects. Accurate assessment of the EAT pool could potentially aid in identifying and predicting the severity of HTN. Importantly, our study showed robust clinical features as there were no significant differences observed among control groups regarding age, sex ratio, BMI, waist and hip circumference, clinical history, and medication usage. Moreover, the findings revealed that patients with HTN and LVH displayed more diffuse fibrosis and impaired myocardial strain compared to those without LVH and controls; these results are consistent with prior studies [11]. Myocardial fibrosis is a ubiquitous consequence of various cellular and noncellular pathological mechanisms. Fibrosis within the myocardium and perivascular space, along with hypertrophy of intramyocardial coronary media and cardiomyocytes, contribute to adverse LVH in patients with HTN. Increased myocardial fibrosis results in heightened left ventricular stiffness, collectively contributing to a poor prognosis in HTN individuals. Hence, it is imperative to use ECV, native T1 and myocardial strain levels as novel targets for monitoring efficacy in hypertensive patients.

The results of both univariable and multivariable logistic regression analyses suggested that EAT, ECV, and GLS can significantly predict LVH in HTN patients. To evaluate the intricate relationships between elevated EAT volume, left ventricular diffuse fibrosis or impaired ventricular strain, and left ventricular remodeling, respectively, mediator variables such as ECV, native T1, GRS, GCS, and GLS were employed. In this study, higher levels of ECV or native T1 and lower levels of GRS, GCS, and GLS associated with thicker EAT mediated the elevated risk of LVH in the hypertensive population. This may be attributed to the fact that EAT promotes myocardial fibrosis or myocardial disfunction, which subsequently leads to LVH. Previous studies have shown that mild‐to‐moderate HTN is often accompanied by insulin resistance and visceral adiposity. Metabolic abnormalities and chronic inflammation may contribute to the accumulation of EAT in HTN patients [26]. The accumulation of EAT can potentially result in myocardial steatosis due to infiltration of adipocytes into adjacent myocardium [5]. In chronic inflammatory diseases, deranged epicardial adipogenesis occurs, leading to secretion of proinflammatory adipokines [27]. Since it shares microcirculation with adjacent myocardium, it can locally damage adjacent myocardium through paracrine mechanisms, resulting in microvascular dysfunction and fibrosis.

EAT is influenced by various modifiable and nonmodifiable risk factors, including age, sex, blood pressure value, diabetes, chronic kidney disease, obesity, and metabolic syndrome [28]. Some studies have shown that obesity and other chronic inflammatory diseases can lead to the accumulation of EAT, which may contribute to cardiovascular events through inflammation [27]. These findings indicate a close association between LVH, EAT, obesity, metabolism, and blood pressure. To mitigate the influence of these confounding factors, we rigorously controlled for age, sex, BMI, SBP, DBP, clinical history, baseline HTN medications, serum lipid levels, waist and hip circumference. Multivariate analysis revealed that the index of EAT volume, ECV, and GLS were independently associated with LVH in hypertensive patients after controlling for these confounding factors. Various therapeutic strategies, such as diuretics, renin‐angiotensin‐aldosterone system (RAAS) inhibitors, and calcium channel blockers, have demonstrated efficacy in inducing remission of LVH [29, 30, 31, 32]. Similarly, ACEI/ARBs have been shown to have antifibrotic effects [33], a possibility also illustrated by the results of subgroup analysis in our cohort showing a weakened relationship between GLS and LVH in the group taking antihypertensive drugs. Additionally, emerging potential therapies targeting a reduction in EAT volume or thickness, including Glucagon Like Peptide 1 Receptor (GLP1R) agonists and sodium‐dependent glucose transporters 2 (SGLT2) inhibitors, have shown promising results in clinical trials [34, 35, 36, 37, 38]. We observed a weakened association between epicardial fat and LVH among patients with diabetes; nevertheless, it remains uncertain whether treatment‐induced lipolysis of epicardial fat confers cardiovascular benefits due to the limited number of diabetic patients included in our cohort and the infrequent use of these drugs. However, our study demonstrates that an increase in EAT volume is associated with LVH in hypertensive patients and may be partially mediated by the proinflammatory or profibrotic effects of EAT. Future prospective studies with large sample sizes are warranted to investigate the independent association between EAT‐targeted drugs, such as GLP‐1 receptor agonists and SGLT2 inhibitors, and LVH in patients with HTN. These studies should also aim to determine whether this association is mediated by the drugs' effects on reducing EAT volume. The potential for targeted therapy in EAT to regress LVH suggests a promising avenue for future research.

4.1. Study Limitation

The current studies still possess certain limitations. Firstly, the presence of multiple scanning equipment in this multicenter retrospective study may introduce potential issues. Secondly, we encountered challenges in matching the two case groups, which may have affected our experimental results. Some sample size was lost during the matched analysis, especially in the LVH group. This may be a major limitation to statistical power when the sample size in the positive group is too small. Therefore, we chose to consider multiple confounders in the multivariate analysis as a measure to mitigate their potential impact. Thirdly, the limited sample size of the control group may reduce statistical power, increasing the risk of both false positives and false negatives. Potential biases could also arise, affecting the validity of the conclusions. Furthermore, the small control sample may not adequately represent the broader population, limiting the ability to generalize the results. These issues persist despite our rigorous screening process, which matched healthy controls for sex, age, and BMI.

5. Conclusion

The present study demonstrates a significantly higher volume of EAT in patients with LVH compared to both hypertensive individuals without LVH and control subjects. Furthermore, it is suggested that the effect of increased EAT volume on LVH may be partially mediated by ECV, native T1 and LV strain. Considering EAT as a modifiable risk factor, interventions targeting EAT could potentially regress LVH and improve the prognosis of hypertensive patients.

Author Contributions

Runze Zhu and Wenxian Wang designed the study. Runze Zhu and Wenxian Wang interpreted the data and wrote the manuscript. Runze Zhu, Wenxian Wang, and Yan Gao analyzed the data and gave advice on data presentation. Runze Zhu, Wenxian Wang, Yan Gao, Bowen Li, and Rongxue Shan were responsible for collecting and sorting statistical data. Runze Zhu, Wenxian Wang, Yan Gao, Bowen Li, and Runjie Xue participated in editing and review of the manuscript. Technical support was provided by Junchuan Liu and Xianshun Yuan. Ximing Wang supervised the overall study, reviewed the manuscript, and provided financial assistance. Runze Zhu and Wenxian Wang contributed equally to this work. All authors read and approved the final manuscript.

Ethics Statement

This retrospective study was approved by the Institutional Review Board of Shandong Provincial Hospital Affiliated to Shandong First Medical University.

Consent

Written informed consent was obtained from all participants before enrollment.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information

Supporting Information

Data Availability Statement

The datasets analyzed during the current study are available from the corresponding author upon reasonable request.