Relationship between epicardial fat tissue, endothelial function, and coronary flow reserve in coronary microvascular disease patients

Abstract

Background

Coronary microvascular disease (CMVD) is a major cause of hospitalization for a considerable number of patients with ischemia with non-obstructive coronary arteries (INOCA), with coronary flow reserve (CFR) reduction being the main feature of CMVD. However, imaging methods for CMVD evaluation are still lacking. This study aimed to investigate associations among epicardial fat thickness (EFT), endothelial function assessed by flow-mediated dilation (FMD), and CFR in patients with CMVD, and to evaluate their predictive value for CMVD diagnosis.

Methods

CFR was measured by regadenoson stress echocardiography, while EFT and FMD were measured by transthoracic echocardiography and brachial artery vascular ultrasound in 66 subjects with INOCA. Multivariable logistic regression and receiver operating characteristic analysis were performed.

Results

Twenty-nine patients were categorized into the CFR decreased group (CFR < 2.0), and thirty-seven patients were separated into the CFR normal group (CFR ≥ 2.0). EFT was significantly higher in the CFR decreased group (6.063 ± 1.732 mm vs. 5.95 ± 1.718 mm, P = 0.003), while FMD was significantly lower in the CFR decreased group [4.05 (3.10, 5.40) vs. 5.30 (4.85, 7.30), P = 0.011]. EFT was an independent predictor for CMVD [odds ratio: 5.084, 95% confidence interval: 1.498–17.251, P = 0.009]. A cut-off value of EFT > 5.51 mm predicted CMVD with 90% sensitivity and 71% specificity. FMD was not an independent predictor for CMVD, but when FMD < 4.65%, it predicted CMVD with 60% sensitivity and 83% specificity.

Conclusions

EFT and FMD may serve as potential evaluation methods for CMVD in patients with INOCA. EFT was independently associated with the presence of CMVD in this cohort, offering significant clinical diagnostic insights. EFT is a novel independent predictor of CMVD in this specific population, while FMD provides complementary diagnostic value for risk stratification. These non-invasive markers may help to improve risk stratification in patients with microvascular dysfunction.

Graphical abstract

Introduction

Coronary microvascular disease (CMVD) refers to a subset of structural and/or functional disorders of coronary microcirculation that lead to impaired coronary blood flow, decreased coronary flow reserve (CFR), and ultimately myocardial ischemia [1]. CMVD has emerged as a prevalent cause of a broad spectrum of cardiovascular disease, including obstructive and non-obstructive coronary artery disease (CAD), cardiomyopathy, and heart failure [2–4]. CMVD is caused by the interaction of multiple factors and mechanisms, including microvascular structural abnormalities, micro-thrombosis, coronary spasm, oxidative stress, endothelial damage, and inflammatory stimulation [5, 6]. Moreover, it was shown that the presence of CMVD identifies patients at increased risk of death and myocardial infarction [7, 8].

CFR is defined as the ratio of coronary blood flow during maximal hyperemia to the flow under resting conditions. It is reflective of the global coronary atherosclerotic burden, endothelial function, and overall state of the microvasculature. CFR serves as a comprehensive indicator of flow through both large epicardial arteries and the coronary microcirculation. Notably, in the absence of epicardial obstructive CAD, decreased CFR is a marker of CMVD [1]. Vasodilator stress echocardiography is an effective method to evaluate CFR, which has been widely used in clinical practice [9, 10]. A CFR value < 2.0 currently indicates microvascular dysfunction. Furthermore, a decreased CFR serves as a crucial and reliable predictor of adverse outcomes and persistent symptoms in angina patients, irrespective of the presence of obstructive CAD [11].

Evidence suggests that endothelial dysfunction (ED) in resistance coronary vessels plays a significant role in the development of CMVD. Several risk factors, including smoking, hyperlipidemia, and diabetes, are associated with ED, which in turn correlates with cardiac events [12]. Assessment of endothelial function in coronary arteries necessitates invasive techniques to evaluate both the epicardial coronary artery and microvascular beds [13]. Alternatively, non-invasive assessment using the Flow-Mediated Dilation (FMD) technique, a peripheral vascular approach, can indirectly evaluate coronary ED. FMD is primarily an endothelium-dependent and largely nitric oxide (NO)-mediated dilation of the conduit artery in response to increased blood flow and shear stress, and is related to coronary atherosclerotic burden [14]. In various studies, FMD has been demonstrated predictive value for cardiovascular events in patients with established CAD [15].

Epicardial adipose tissue (EAT) envelops approximately 80% of the cardiac surfaces and acts as a source of multiple endocrine and inflammatory mediators [16]. The accumulation of EAT can disrupt vascular homeostasis and contribute to ED, amplifying vascular inflammation, promoting intimal lesions, and exacerbating plaque progression through an outside-to-inside signaling mechanism [17].

Although both ED and CFR impairment are early indicators of subclinical CAD, the specific role of EAT in CMVD patients remains unclear. This study aims to explore the relationship among epicardial fat thickness (EFT), FMD, and CFR in patients with CMVD and to assess their predictive value for CMVD.

Materials and methods

Study population

Between March and August 2023, we conducted vasodilator drug stress echocardiography on 215 patients who were hospitalized at the First Affiliated Hospital of Harbin Medical University, who exhibited typical or atypical symptoms such as angina pectoris and chest tightness. We gathered comprehensive demographic, clinical, laboratory, and conventional echocardiographic data. Inclusion Criteria: (1) Patients admitted with typical or atypical symptoms of clinical myocardial ischemia, such as exertional angina; (2) Patients who have undergone coronary angiography within the past three months, with normal epicardial coronary arteries or stenosis < 50%; (3) Age between 18 and 80 years old; (4) Gender not restricted. Exclusion Criteria: (1) Patients with unstable vital signs, such as abnormally elevated blood pressure; (2) Patients with pulmonary diseases involving bronchial stenosis or bronchospasm; (3) Patients with a previous history of allergic reaction to regadenoson; (4) Patients with non-cardiac chest pain and other heart diseases (including acute coronary syndrome, previous old myocardial infarction, clinically ruled-out coronary artery spasm, hypertrophic cardiomyopathy, dilated cardiomyopathy, congenital heart disease, atrial fibrillation, arrhythmias such as second or third-degree atrioventricular block, and valvular heart disease); (5) Patients with pulmonary hypertension, hyperthyroidism, or metabolic diseases such as rheumatism; (6) Patients with poor ultrasound imaging conditions where image quality cannot be guaranteed; (7) Patients with incomplete data. Finally, based on the inclusion and exclusion criteria, 66 patients were included in the study, comprising 29 males and 37 females, with a mean age of 60.3 ± 11.2 years. All patients underwent regadenoson stress echocardiography, FMD technique, and EFT measurements. The flowchart is as follow (Fig. 1).

Fig. 1.

Inclusion Criteria & Exclusion Criteria

Echocardiographic image acquisition and data analysis

Echocardiographic examinations were performed using the PHILIPS EPIQ 7 C ultrasound imaging system (Philips Medical Systems, Andover, MA, USA) equipped with an S5-1 transducer operating at a frequency of 1–5 MHz and a frame rate of 50–70 frames per second. Patients were positioned on their left side and connected to a precordial lead lectrocardiogram during the exam. The dynamic image was set to 5 cardiac cycles. The parasternal long-axis view should be used for linear measurements of the left atrium and left ventricle. The left atrium anteroposterior diameter (LAD) was measured at end-systole using a leading edge–to–leading edge method. For the 2D technique, the caliper is placed at the level of the aortic root and extended to the leading edge of the posterior LA wall perpendicular to the assumed long axis of the atrial chamber. The left ventricular internal dimension diastole (LVIDd) was measured at the end of diastole, the measurement should be obtained at a level just below the MV leaflet tips. The left ventricular ejection fraction (LVEF) was calculated using the biplane Simpson’s method. We assessed LV diastolic function by measuring the mitral valve peak E velocity in early diastole, peak A velocity in atrial contraction, the E/A ratio, and the averaged e′ velocity at the lateral and medial mitral annulus via tissue Doppler, calculating the E/e′ ratio thereafter [18].

Regadenoson stress echocardiography and measurement of CFR

After the FMD examination was completed, a regadenoson stress echocardiography was performed to measure the CFR. Before conducting the CFR examination, beta-adrenergic blockers, cardiac glycosides, calcium channel blockers, and adenosine-enhancing agents such as dipyridamole and ticagrelor should be discontinued for 24 h. Standard preparations for stress echocardiography were made, including ready access to emergency facilities, and continuous monitoring of blood pressure, heart rate, and electrocardiography throughout the procedure. Blood flow spectra for the left anterior descending coronary artery in the apical region was obtained, peak and mean coronary blood flow velocities were measured at the end of exhalation. If the coronary arteries were not clearly displayed, a high-frequency probe such as S9-2 could be used for exploration. The median cubital vein access was established, regadenoson, serving as a vasodilator, was injected in a 5 ml solution at a dose of 0.4 mg in 10–20 s, immediately flushed by 5 ml saline. Two minutes later, the coronary arteries were in their maximum filling state and remained so for approximately 8 min. During this period, the coronary artery blood flow spectra could be measured multiple times, and the average value of the three heartbeats with the highest flow velocities were taken the peak and mean coronary artery blood flow velocities were measured at maximal hyperemia state. CFR was determined as the ratio of the coronary blood flow velocity at maximal hyperemia to that under resting conditions.

Measurement of EFT

EFT linear measurements were obtained from the parasternal long-axis and midventricular parasternal short-axis according to standard criteria. Using the aortic annulus as an anatomic landmark, maximum EFT linear measurements was measured from parasternal long-axis images at the point on the free wall of the right ventricle along the midline of the ultrasound beam, perpendicular to the aortic annulus. For midventricular parasternal short-axis assessment, maximum EFT is measured from the right ventricular free wall along the midline of the ultrasound beam perpendicular to the interventricular septum [19, 20]. The average value from three cardiac cycles for each echocardiographic view was calculated. To control intra-observer and inter-observer variability, the measurements of EFT were performed using a standardized protocol: a senior sonographer measured the EFT of 66 study participants, three independent measurements were taken in each of the EFT long-axis view and parasternal short-axis view, and the mean value was calculated. During the second measurement, the technician was not allowed to refer to the first measurement data to eliminate interference from previous results on the measurement judgment. Additionally, an independent senior sonographer with equivalent professional qualifications performed a secondary EFT measurement on the same 66 participants without knowledge of the initial measurement data. The consistency and variability of the measurements were evaluated between the two assessments by the same observer and between the two observers using Bland-Altman plots, descriptive statistics, and intraclass correlation coefficients (ICC).

An example is as follow (Fig. 2).

Fig. 2.

EFT Linear Measurement

Measurement of FMD

The endothelial function assessed using FMD measurement of the brachial artery conducted with a 7.0-MHz linear array transducer on the UNEXEF VG Ultrasound System. Before the FMD examination, patients should fast, avoid physical activity, avoid caffeine, vitamin C, polyphenols, alcohol and refrain from smoking [21]. Before the test, the patient lay supine in a quiet room with a suitable temperature (20–25℃) for 10–15 min. Place a blood pressure cuff near the elbow fossa on the forearm. Perform a longitudinal scan of the brachial artery 5–10 centimeters above the elbow. After optimizing the image, the baseline internal diameter of the brachial artery was measured. Subsequently, a blood pressure cuff was inflated to 50 mmHg above the systolic blood pressure and maintained for 5 min. The peak internal diameter following cuff deflation was measured within two minutes. FMD percentage was calculated using the formula: [(Hyperemia diameter - Baseline diameter) / Baseline diameter × 100%] [12].

Statistical analysis

Statistical analyses were performed using SPSS version 25.0 (SPSS Inc., Chicago, IL, USA). We tested the normality distribution of all variables using the Shapiro-Wilk test. Measurement data adhering to normal distribution were expressed as mean ± standard deviation, and those not conforming were shown as medians and interquartile ranges [M (P25, P75)]. Categorical variables were reported as counts and percentages [n (%)]. Continuous variables were compared using the Student’s t-test or the Mann-Whitney U test, and categorical data were analyzed with the chi-square test. Correlations were examined using Pearson or Spearman correlation coefficients, as appropriate. To validate the stability and reliability of the regression model as well as the robustness of key predictors, the Bootstrap resampling method was employed with 1,000 iterations of resampling with replacement, accompanied by 95% confidence intervals (95% CI). Coefficient bias across all resamples was calculated to further corroborate the robustness of the observed associations. To control the intra-observer and inter-observer variability of EFT measurements, we constructed Bland-Altman plots, and conducted descriptive statistics as well as ICC analyse. Additionally, the Hosmer-Lemeshow test was performed to assess the goodness-of-fit of the logistic regression model, where a non-significant P value (P > 0.05) indicated that the model exhibited no significant bias and fitted the data adequately. The diagnostic thresholds for EFT and FMD in predicting CMVD were estimated via receiver operating characteristic (ROC) curve analysis, providing sensitivity and specificity. All variables with P < 0.05 in the univariate analysis were included in the multivariate binary logistic regression analysis model to examine the relationship between each factor. P < 0.05 was considered statistically significant.

Results

In the baseline data comparison, we compared age, gender, body mass index (BMI), total cholesterol, triglycerides, hypertension status, diabetes mellitus status, hyperlipidemia status, smoking status, EFT, and FMD, as well as the echocardiographic parameters: left ventricular ejection fraction (LVEF), left atrial diameter (LAD), left ventricular internal diameter at end-diastole (LVIDd), mitral valve peak E wave velocity (MV Peak E), mitral valve peak A wave velocity (MV Peak A), mitral annular early diastolic velocity (e’), and E/e’ ratio. And the following associations were identified between the two groups.

Patients were divided into the CFR decreased group (CFR < 2.0, n = 29, 43.94%) and CFR normal group (CFR ≥ 2.0, n = 37, 56%) based on the CFR results. Compared to those in the CFR normal group, patients in the CFR decreased group were older (P = 0.026), had lower FMD (P = 0.011), a higher prevalence of hypertension and thicker EFT (P = 0.003). No significant differences were noted in gender, BMI, total cholesterol, triglyceride levels, diabetes, hyperlipidemia, and smoking history between the two groups (P > 0.05) (Table 1).

Table 1.

Baseline clinical characteristics

	CFR decreased group (n= 29)	CFR normal group (n= 37)	P
Age (years)	67 (61.25, 71.25)	55 (49.25, 67.00)	0.026
Male gender (n%)	12 (41.4%)	17 (45.9%)	0.507
BMI (Kg/m2)	26.87 ± 3.29	26.8 ± 4.14	0.278
Total cholesterol (mmol/L)	4.30 ± 0.94	4.43 ± 0.85	0.652
Triglycerides (mmol/L)	1.67 ± 0.66	2.07 ± 1.39	0.467
Hypertension (n%)	19 (65.5%)	13 (35.1%)	0.005
Diabetes (n%)	18 (62.1%)	14 (45.9%)	0.284
Hyperlipidemia (n%)	19 (65.5%)	12 (32.4%)	0.261
Smoking (n%)	10 (34.4%)	24 (64.9%)	0.179
FMD (%)	4.05 (3.10, 5.40)	5.3 (4.85, 7.30)	0.011
EFT (mm)	6.063 ± 1.732	5.95 ± 1.718	0.003
LVEF (%)	68.00 (60.00, 70.00)	64.00 (60.00, 70.00)	0.884
LAD (mm)	36.50 (33.00, 40.00)	34 (31.25, 37.75)	0.093
LVIDd (mm)	45.00 (43.00, 47.75)	47 (43.25, 48.75)	0.205
MV Peak E (m/s)	0.63 ± 0.19	0.70 ± 0.16	0.141
MV Peak A (m/s)	0.83 ± 0.16	0.76 ± 0.15	0.132
e’ (cm/s)	5.95 (4.98, 7.65)	6.95 (5.70, 9.90)	0.065
E/e’	10.77 ± 4.55	10.45 ± 4.30	0.729

The variables that conform to the normal distribution are expressed as (mean ± standard) deviation. The variables that do not conform to the normal distribution are expressed as quartiles [ M (P25, P75)]. The categorical variables are expressed as numbers and percentages [ n (%) ]

Incorporating factors affecting coronary heart disease, EFT, and FMD into a univariate logistic regression analysis, EFT (odds ratio (OR) : 2.665, 95% confidence interval (CI): 1.282–5.540, P = 0.009), FMD (OR: 0.597, 95% CI: 0.399–0.896, P = 0.013), age (OR: 1.071, 95% CI: 1.006–1.140, P = 0.032), and hypertension (OR: 6.667, 95% CI: 1.690–26.298, P = 0.007) showed statistically significant differences between the two groups and were identified as influencing factors of CMVD (Table 2).

Table 2.

Univariate binary logistic regression analysis for CMVD influence factors

	B	SE	Wald	P	OR	95% CI
Age	0.068	0.032	4.618	0.032	1.071	1.006–1.140
Sex	-0.405	0.612	0.438	0.508	0.667	0.201–2.214
BMI	0.096	0.088	1.201	0.273	1.101	0.927–1.308
Diabetes	0.847	0.805	1.109	0.292	2.333	0.482–11.297
Hypertension	1.897	0.7	7.341	0.007	6.667	1.690-26.298
Hyperlipidemia	1.012	0.926	1.193	0.275	2.75	0.448–16.895
Smoking	-0.847	0.636	1.774	0.183	0.429	0.123–1.491
Triglycerides	-0.359	0.31	1.34	0.247	0.698	0.380–1.282
Total cholesterol	-0.163	0.351	0.215	0.643	0.85	0.427–1.691
EFT	0.98	0.373	6.894	0.009	2.665	1.282–5.540
FMD	-0.515	0.207	6.216	0.013	0.597	0.399–0.896

Additionally, no significant differences in baseline echocardiography parameters, including LVEF, LAD, LVIDd, peak E, peak A, e’, E/e’ were observed between the two groups (P > 0.05) (Table 1).

Through establishing a multifactorial binary logistic regression model incorporating factors such as age, hypertension, FMD, and EFT, we found a statistically significant association. Notably, a thicker EFT was associated with an increased likelihood of CFR < 2.0 and a higher risk of CMVD, with an OR of 1.626, (95% CI: 1.282–5.540, P = 0.009). Thus, EFT was identified as an independent predictor for CMVD. In contrast, the adjusted influence of FMD on CMVD was not statistically significant (P = 0.092), indicating that FMD does not independently affect CMVD outcomes (Table 3).

Table 3.

Multivariate binary logistic regression analysis for CMVD influence factors

	B	SE	Wald	P	OR	Bootstrap P	Bootstrap 95% CI
Age	0.017	0.055	0.098	0.754	1.017	0.351	-0.120-0.405
Sex	0.357	1.468	0.059	0.808	1.43	0.932	-4.710-3.296
BMI	0.079	0.182	0.19	0.663	1.082	0.658	0.813–1.232
Diabetes	0.316	1.751	0.033	0.857	1.371	0.302	0.967–4.834
Hyperlipidemia	0.296	1.325	0.05	0.823	1.345	0.678	0.765–2.745
Hypertension	2.41	1.105	4.76	0.029	11.135	0.015	1.587–21.077
Smoking	-2.291	1.326	2.984	0.084	0.101	0.09	-0.690-0.533
EFT	1.626	0.623	6.805	0.009	5.084	0.013	1.131–5.835
FMD	-0.752	0.446	2.836	0.092	0.472	0.241	-2.509-0.617

To verify the stability of the core conclusions, the Bootstrap method was employed for validation of the regression model and predictors, involving 1,000 bootstrap resamples with replacement and 95% CI. The results showed that EFT, as an independent predictor of CMVD, had a regression coefficient of 1.626 (Bootstrap P = 0.013) and an OR of 5.084 (Bootstrap 95% CI: 1.131–5.835). The coefficient bias across the 1,000 resamples was only 4.431, indicating that the association between EFT and CMVD is stable and independent of other confounding factors. In contrast, FMD had a regression coefficient of -0.752 (Bootstrap P = 0.241) and an OR of 0.472 (Bootstrap 95% CI: 0.463–1.249), further confirming that FMD is not an independent predictor of CMVD but only provides supplementary diagnostic value. Additionally, the Hosmer-Lemeshow test yielded χ² = 6.074 (P = 0.639), demonstrating good fit of the regression model with no significant bias.

In all patients, ROC curve analysis demonstrated that when EFT > 5.51 mm, it significantly predicted a CFR < 2.0, suggesting a higher likelihood of CMVD. This association was characterized by a sensitivity of 90% and a specificity of 71% (AUC = 0.745, 95% CI: 0.593–0.896, P < 0.006) (Fig. 3a). In male patients, EFT > 5.51 mm significantly predicted a CFR < 2.0, with a sensitivity of 88% and specificity of 90% (AUC = 0.825, 95% CI: 0.595–1.000, P < 0.021) (Fig. 3b). For female patients, an EFT > 5.55 mm was found to significantly predict a CFR < 2.0, with a sensitivity of 92% and a specificity of 67% (AUC = 0.767, 95% CI: 0.570–0.965, P < 0.026) (Fig. 3c).

Fig. 3.

(a-c). ROC curve analysis of the EFT levels predicting CMVD in all patients, in male patients, and in female patients

Our findings indicated that the FMD values varied between 2.3% and 8.9%. ROC curve analysis demonstrated that an FMD value of less than 4.65% significantly predicted a CFR value below 2.0, thereby correlating with an elevated likelihood of CMVD. This association was characterized by a sensitivity of 60% and a specificity of 83% (AUC = 0.725, 95% CI = 0.571–0.879, P < 0.011) (Fig. 4).

Fig. 4.

ROC curve analysis of the FMD levels predicting CMVD

No significant correlations were detected between EFT and FMD in all patients (r = -0.196, P = 0.116), in the CFR decreased group (r = -0.134, P = 0.487), or in the CFR normal group (r = -0.034, P = 0.844) (Fig. 5a-c).

Fig. 5.

(a-c). Correlation between EFT and FMD in all patients, in the CFR decreased group, and in the CFR normal group

FMD significantly correlated with cardiovascular risk factors such as age (r = -0.337, P = 0.025) and total cholesterol (r = 0.388, P = 0.009), but showed no correlated with gender, BMI, hypertension, diabetes, smoking history, or triglycerides. There were no significant correlations between EFT and the above parameters (Table 4).

Table 4.

The correlation between FMD, EFT and cardiovascular risk factors

	FMD		EFT
	r	P	r	P
Sex	-0.133	0.389	-0.074	0.634
Age	-0.337	0.025	0.076	0.625
BMI	0.063	0.684	0.168	0.275
Total cholesterol	0.388	0.009	-0.13	0.401
Triglyceride	0.188	0.221	0.177	0.25
Diabetes	0.028	0.857	0.197	0.199
Hypertension	-0.118	0.447	0.087	0.576
Smoking	-0.113	0.465	0.137	0.377

The reliability of transthoracic echocardiographic EFT measurements was validated via intra- and inter-observer consistency assessments in all 66 study participants, with consistency evaluated using descriptive statistics, Cronbach’s alpha coefficient, ICC, and Bland-Altman plots.

For intra-observer consistency, the mean EFT values of the two views were 5.458 ± 1.166 mm (coefficient of variation [CV] = 21.37%) and 5.374 ± 1.165 mm (CV = 21.68%), respectively, with a negligible mean difference of 0.084 mm, indicating no significant systematic bias between the two views. The measurements exhibited extremely high internal consistency (Cronbach’s alpha = 0.997) and excellent reproducibility (ICC = 0.994, 95% CI: 0.990–0.996, P < 0.001). Bland-Altman analysis confirmed low intra-observer measurement bias (0.084 mm), with all data points distributed within the 95% limits of agreement and no outliers, verifying the stability of repeated measurements by the same operator. (Table 5).

Table 5.

Consistency of EFT measurement

		CV,%	ICC	ICC 95% IC	ICC P
Intra-observer consistency	Parasternal short axis	21.3697	0.994	0.990–0.996	<0.01
	Parasternal long axis	21.6506
Inter-observer consistency	Observer 1	21.6789	0.994	0.990–0.996	<0.01
	Observer 2	21.3965

For inter-observer consistency, the mean EFT values of Observer 1 and Observer 2 were 5.374 ± 1.165 mm (CV = 21.68%) and 5.401 ± 1.156 mm (CV = 21.40%), respectively, with a mean difference of only 0.027 mm and comparable CV values, ruling out inter-operator systematic deviation. Consistency metrics were comparable to intra-observer results (Cronbach’s alpha = 0.997, ICC = 0.994, 95% CI: 0.990–0.996, P < 0.001). Bland-Altman analysis further demonstrated minimal inter-observer bias (0.027 mm), with all data falling within the 95% agreement range, confirming low inter-operator variability.

Collectively, transthoracic echocardiographic EFT measurement showed high reliability and low variability, providing a robust methodological basis for its application as a non-invasive indicator in CMVD assessment. (Fig. 6).

Fig. 6.

(a,b). Bland-Altman Plots for Assessing Intra-observer and Inter-observer Consistency of EFT Measurements

Discussion

In this study, we found that increased EFT significantly correlates with decreased CFR, suggesting that EFT might serve as an independent predictor for the presence of CMVD in patients with characteristics consistent with the study group, and that elevated EFT increases the risk of CMVD. These results align with the findings of Eroglu on chest pain in women with normal coronary angiography [22], and Kaya’s research on the correlation between EFT and CFR in patients with CAD [23].

Under normal physiological conditions, EAT functions in secretion and metabolism, secreting anti-atherosclerotic cytokines, generates heat, and acting as a shock absorber during heart movements. Collectively, these activities help protect the heart against atherosclerosis and regulate coronary artery tone [24].However, when EAT accumulates excessively, the release of anti-atherosclerotic cytokines will be reduced [25]. Meanwhile, EAT may lead to adipose tissue hypoxia, initiating the release of inflammatory factors and atherogenic cytokines [26]. These factors are secreted to adjacent myocardium and directly enter the outer membrane of coronary vessels, inducing local inflammation and atherosclerosis, and destroying the ultrastructure of myocardium and coronary arteries, causing CFR damage [27]. Inflammation contributes to the pathogenesis of non-obstructive CAD by impairing CFR, and serves as an independent risk factor for this entity. This mechanism may account for the observed association between increased EFT and reduced CFR; furthermore, we hypothesize that elevated EFT is closely linked to the presence of CMVD in the study cohort.

Notably, CMVD leads to left ventricular diastolic dysfunction (LVDD) through mechanisms such as myocardial ischemic fibrosis, inflammatory factor release, and energy metabolism disorders. According to Hirocoshi et al.’s research, EAT drives the occurrence of LVDD by promoting myocardial fibrosis, and an increased EAT volume index is an independent predictor of LVDD in patients with chronic coronary syndrome (CCS) [28]. As a subset of EAT surrounding coronary arteries, pericoronary adipose tissue (PCAT) can release free fatty acids and monocyte chemoattractant protein-1 under inflammatory states, NO synthesis, and thereby promote the formation of coronary atherosclerotic plaques. Yamaura et al.’s research further indicates that high inflammatory levels in PCAT (characterized by a CT attenuation value of PCAT ARCA ≥ -76.6 HU) are an independent driver of coronary atherosclerotic plaque burden in CCS patients, with this effect not interfered by EAT volume [29]. Additionally, patients with increased EAT volume have a higher risk of coronary artery calcification and plaque burden, and this association is unrelated to PCAT inflammation. In terms of measurement methods, compared to PCAT and EAT volume, EFT measured by transthoracic echocardiography does not precisely target pericoronary adipose tissue. However, due to its simplicity, high reproducibility, and cost-effectiveness, EFT can still effectively indicate the increased risk of coronary atherosclerotic plaques and CMVD.

Systematic intra- and inter-observer consistency analyses confirmed the low variability and high reliability of transthoracic echocardiographic EFT measurements: both intra-observer repeated measurements across views and inter-observer independent assessments showed excellent consistency. Additionally, Bland-Altman analyses revealed minimal measurement bias.Clinically, the high reliability of EFT measurement is critical to validating its role as an independent predictor of CMVD. Low measurement variability ensures EFT value accuracy and comparability, eliminating measurement error-induced misdiagnoses and supporting EFT’s utility as a non-invasive CMVD assessment indicator. Notably, the high consistency of EFT measurements aligns with prior evidence that ultrasonic EFT assessment has good reproducibility [20].

Although there is no standard range for EFT, studies suggest it generally lies between 1.1 mm and 22.6 mm [30]. Hasan Kaya identified a cutoff value of 5.8 mm for predicting CAD [23], while Green et al. established that EFT greater than 5.6 mm optimally detects myocardial flow reserve impairment in non-obstructive CAD [31]. Our study supports these findings, with an EFT greater than 5.51 mm demonstrating similar sensitivity and specificity for predicting impaired CFR.

Moreover, we found differing EFT cutoff values for predicting CFR impairment between male and female patients. Previous studies indicate that cardiovascular factors, including age and BMI, influence EFT, with middle-aged women exhibiting higher levels of EFT [32, 33]. Our study did not observe the effects of age and BMI on EFT. However, the older age and lower BMI of female patients may explain their higher EFT cutoff values compared to male patients.

We also observed that patients with decreased CFR had significantly lower FMD. In the logistic regression analysis, under the influence of other variables, the influence of FMD on CMVD was no longer significant. Nevertheless, ROC curve analysis showed that when FMD < 4.65%, it was possible to predict CMVD. These results suggested that although FMD may not be an independent influencing factor for CMVD, it is still important to consider FMD levels when predicting CMVD.

Laboratory and clinical studies have consistently shown that ED is a key component in the pathogenesis of CMVD [34]. The gold standard for endothelial testing is an invasive infusion of acetylcholine into the coronary arteries [14]. Research from the Mayo Clinic revealed that approximately two-thirds of patients with chest pain in non-obstructive CAD showcased endothelium-dependent or endothelium-independent CMVD when tested with acetylcholine [35]. FMD is a non-invasive technique that assesses endothelium-dependent function in the peripheral brachial arteries, primarily through NO-mediated arterial dilation. Given that CMVD might represent the cardiac manifestation of systemic microcirculatory dysfunction, we measured FMD to explore any significant correlations with peripheral vascular endothelial function.

In this study, we found that CFR decreased patients also had lower FMD, suggesting that vascular endothelial damage is more susceptible to lead to CMVD. This is consistent with previous studies that NO-mediated vasodilation function is significantly impaired in patients with microvascular angina [36]. We can use the FMD technique to aid in the diagnosis of CMVD patients.

In our study, FMD < 4.65% was the cutoff value of CMVD under the combined influence of multiple coronary artery risk factors such as old age, hypertension, and hyperlipidemia. Previous studies have also confirmed that FMD shows greater variability in subjects who are older, hypertensive, or dyslipidemia [37]. In a large cohort study of 7277 participants, the threshold of FMD for normal endothelial function in Japanese subjects was 7.1% [38]. Junko Soga also found that the risk of adverse cardiovascular events would be increased in CAD patients with an FMD < 7.1% [12]. These studies all demonstrate that even if FMD is not an independent influencing factor of CMVD, but it also provides value for the imaging diagnosis of CMVD. Meanwhile, since EFT is easier to obtain and more accurately measurable than FMD, we can routinely perform ultrasonic EFT measurement in patients suspected of having CMVD. This may provide a novel direction for the clinical management of CMVD, namely reducing the increase in EFT. Regarding the peripheral vascular endothelial function reflected by FMD, we may need to compare it with coronary vascular endothelial function, identify the association between them, and thereby clarify whether FMD is related to CMVD.

Limitations and future work

This study has certain limitations. Firstly, the diagnosis of CMVD was based on regadenoson stress echocardiography-derived CFR rather than the gold standard of myocardial flow reserve measurement via PET; thus, the diagnostic accuracy of the cohort grouping may be limited. Although transthoracic echocardiography is a convenient tool for EFT measurement, it does not specifically target PCAT—the direct pathological contributor to coronary microvascular injury—leading to a mismatch between the measured indicator and the underlying pathological mechanism.Additionally, CT could provide more precise quantificant of total EAT volume compared with the linear EFT measurement used in this study. Secondly, the small sample size and single-center design may limit the generalizability of our findings. Thirdly, the brief duration of this study precluded the long-term follow-up of patients for prognosis and cardiovascular outcomes, falling it verify of EFT and FMD in CMVD patients. Future studies should aim to validate the diagnostic value of EFT and FMD in a multicenter cohort with CMVD confired by gold-standard diagnostic methods and conduct long-term follow-up to explore the prognostic significance of these non-invasive markers.

Conclusions

EFT and FMD may serve as potential evaluation methods for CMVD in patients with INOCA. Notably, EFT was independently associated with the presence of CMVD in this cohort, offering significant clinical diagnostic. EFT is a novel independent predictor of CMVD in this specific population, while FMD provides complementary diagnostic value for risk stratification. These non-invasive markers may help to improve risk stratification in patients with microvascular dysfunction characterized by reduced CFR.

Abbreviations

BMI

body mass index

CAD

coronary artery disease

CCS

chronic coronary syndrome

CFR

coronary flow reserve

CMVD

coronary microvascular disease

EAT

epicardial adipose tissue

ED

endothelial dysfunction

EFT

epicardial fat thickness

FMD

flow-mediated dilatation

INOCA

ischemia with non-obstructive coronary arteries

LAD

left atrium anteroposterior diameter

LVDD

left ventricular diastolic dysfunction

LVEF

left ventricular ejection fraction

LVIDd

left ventricular internal dimension diastole

NO

nitric oxide

PCAT

pericoronary adipose tissue

Authors’ contributions

Mengjiao Li: Conceptualization, Methodology, Data curation, Formal analysis, Investigation, Writing-Original draft preparation. Yingying Huang: Data curation, Formal analysis, Writing-Original draft preparation， Subsequent draft revision. Linghui Bai: Visualization, Formal analysis, Investigation. Wei Li: Visualization, Investigation. Chenbo Chu: Visualization, Investigation. Hui Liu: Conceptualization, Methodology, Writing - Review & Editing, Supervision, Project administration, Funding acquisition.

Funding

This work was supported by funding for scientific research from the Heilongjiang Provincial Department of Education (12541296) and Heilongjiang Renxin Bone Health Medical Assistance Foundation (2020HX016).

Data availability

The raw data underlying this article will be shared on reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

The study adhered to the ethical guidelines of the 1975 Declaration of Helsinki and received approval from the institution's human research committee (IRB-AF/SC-04/02.0, dated March 13, 2024).

Competing interests

The authors declare no competing interests.