We recruited 78 ANOCA patients referred for adenosine-based transthoracic Doppler echocardiography-assessed coronary flow reserve (CFR) measurement, adenosine stress echocardiography and conventional echocardiography. Diagnostic criteria of ANOCA: (1) the patient had symptoms of angina or angina equivalents; (2) coronary angiography or coronary CT angiography suggested coronary artery stenosis was <50% (13); (3) Objective evidence of myocardial ischemia. Exclusion criteria: (1) suboptimal image quality; (2) left ventricular ejection fraction (LVEF) <50%; (3) atrial fibrillation or other severe arrhythmias; (4) severe valvular disease; and (5) intracardiac shunt. All patients are requested to stop taking medications that may affect the test results the day before the test. According to the International Standardization of Diagnostic Criteria for Microvascular Angina issued by Coronary Vasomotion Disorders International Study Group, CMD was defined as CFR < 2.5 (13). The study was approved by the Ethics Committee of Scientific Research of Shandong University Qilu Hospital (KYLL202008019) and was conducted as per the Declaration of Helsinki. All patients were informed about the study.

A GE Vivid E95/E9 ultrasound diagnostic apparatus with an M5S probe (3.5 MHz) (GE Medical Systems, Milwaukee, WI, United States) was used. Brachial artery blood pressure was measured prior to image acquisition. All subjects were connected simultaneously to a thoracic-lead ECG. Parasternal left ventricular long-axis views, and apical four-chamber, three-chamber, and two-chamber views were acquired continuously for at least three cardiac cycles in the left lateral recumbent position at a frame rate of 50–80 frames/s. We measured conventional echocardiographic parameters, including: LV end-diastolic dimension (LVEDD), LV end-systolic dimension (LVESD), LV interventricular septal end-diastolic thickness (IVST), LV posterior wall thickness (LVPWT), LV mass index (LVMI), left atrium volume index (LAVI), rest LV end-diastolic volume index (rest-LVEDVI), rest LV end-systolic volume index (rest-LVESVI), and rest LV ejection fraction (rest-LVEF, Simpson's biplane method). Pulsed-wave Doppler (PW) and tissue Doppler imaging (TDI) of the mitral valve were also evaluated.

Patients underwent adenosine stress echocardiography according to the protocol recommended by the EACVI (European Association of Cardiovascular Imaging) (14). We used adenosine at a dose of 0.14 mg/kg/min over 6 min. The electrocardiogram was monitored continuously and blood pressure was monitored intermittently. Criteria for interrupting the test were severe chest pain, diagnostic ST-segment shift, excessive blood pressure increase (systolic blood pressure ≥240 mmHg, diastolic blood pressure ≥120 mmHg), dyspnea, hypotension (systolic blood pressure ≤90 mmHg, diastolic blood pressure ≤60 mmHg), maximal predicted heart rate, or significant arrhythmias. When the adenosine stress was maximal, that is, when the coronary blood flow velocity was maximal, the three apical views (the apical four-chamber, two-chamber, and long axis) were repeatedly recorded and the cardiac systolic function related indices were measured, including peak-LVEDVI, peak-LVESVI and peak LVEF.

Coronary flow reserve testing was performed using a previously published and validated protocol (15). The mid-distal segment of the left anterior descending (LAD) branch was identified in the interventricular sulcus under a modified left ventricular double-chamber view, and the flow spectrum was recorded by color Doppler echocardiography to detect the mid-distal coronary artery flow in the left anterior descending branch. CFR was defined as the ratio of the maximum diastolic flow velocity in the hyperemic state to the maximum diastolic flow velocity in the basal state.

Speckle-tracking echocardiography was performed using an offline workstation (EchoPAC version 204; GE Vingmed Ultrasound AS, GE Medical Systems) to calculate 2D-STE parameters. Based on three apical views, the software automatically identifies myocardial activity in the region of interest to enable automatic tracking of myocardial motion. If necessary, the region of interest was adjusted by correcting the edge of the endocardium or the width of the myocardium. According to the standardized 17-segment heart model (16), GLS was calculated from the mean of the longitudinal peak systolic strain of all the LV segments. From the time to reach the peak strain in each segment, the software automatically calculated the PSD.

To calculate MW-related indices, mitral and aortic valve opening and closing times were first determined in the apical three-chamber cardiac section, followed by inputting brachial artery systolic pressure to replace the peak LV pressure to obtain the noninvasive LV pressure-strain loop (LV-PSL). Based on LV-PSL, the following data were obtained: global work index (GWI), global contractive work (GCW), global waste work (GWW) and global work efficiency (GWE).

2D-STE and MW-related indices were measured before adenosine stress and at maximum adenosine stress, where pre-adenosine stress was defined as rest and post-adenosine stress was defined as peak. The difference before and after adenosine stress was calculated as Δ.

Two experienced sonographers remeasured 20 randomly selected participants to assess the repeatability. The sonographers were blinded to the clinical data as well as to each other's results. A month later, the images were analyzed again by the same sonographers to assess the intraobserver variability and to assess the interobserver variability.

All data were collected, statistically analyzed, and tabulated using SPSS 26 software (SPSS Inc., Chicago, IL, United States), Med Calc 19.04 (Med Calc Software BVBA, Ostend, Belgium) and GraphPad Prism 9.0.0 (GraphPad Software, San Diego, CA, United States). Baseline variable statistical significance was assessed with the Wilcoxon test, analysis of variance, and the χ² test for continuous nonnormally distributed, continuous normal distributed, and categorical variables, respectively. Normally distributed continuous variables are presented as mean ± SD. Nonnormally distributed continuous variables are presented as median [Q1, Q3]. Categorical variables are presented as number (%). Pearson's correlation method was used to explore the variables influencing MW indices. Univariable and multivariable logistic regression analyses were performed to find the variables associated with CMD. The receiver operating characteristic (ROC) curve was drawn to analyze the predictive ability of 2D-STE and MW indices for CMD and to calculate the area under the curve (AUC). A value of p < 0.05 was considered significant.

The study population comprised 78 subjects with ANOCA. The average age of the entire population was 54.3 ± 10.3 years old; 28% of the subjects were female. The prevalence of CFR < 2.5, consistent with CMD, was 41% (32/78 subjects), and 59% (46/78 subjects) had CFR ≥ 2.5, consistent with non-CMD (Figure 1). Baseline demographics, cardiac risk factors, and relevant medications are compared by the presence of CMD in Table 1. The level of HDL-C was significantly higher in the non-CMD group than in the CMD groups. All other baseline clinical variables were not significantly different between the two groups.

Parameters by 2D echocardiography, PW, and TDI, including LVEDD, LVESD, E velocity and coronary flow velocity were compared between the CMD group and non-CMD group (Table 1). Except for the higher E/e′ in the CMD group, there were no significant differences in the 2D, PW or TDI indicators between the two groups. In addition, the resting flow velocity of the CMD group was faster, and the peak coronary flow velocity was slower.

After adenosine stress, the absolute values of GLS and LVEF were significantly increased in all groups. In the CMD group, GWW increased significantly and GWE decreased significantly, but no significant changes were observed in the remaining indicators. In the non-CMD group, there was a significant decrease in PSD and GWW along with a significant increase in GWI, GCW and GWE. Regarding hemodynamic parameters, heart rate increased significantly after adenosine stress in both groups. While, SBP and DBP decreased significantly after adenosine stress in the non-CMD group, but not in the CMD group (Table 2).

In the comparison of cardiac contractility metrics between the two groups, the differences appeared mainly after adenosine stress. Of these, the absolute values of peak GLS and ΔGLS were significantly lower in the CMD group than in the non-CMD group. The peak PSD and ΔPSD in the CMD group were significantly higher than those in the non-CMD group. The CMD group had lower peak GWI, peak GCW, and peak GWE, but higher peak GWW and ΔGWW (Figures 2, 3 and Table 2).

GWI was significantly correlated with SBP, DBP, the product of heart rate and blood pressure (RPP), GLS and coronary flow velocity at rest and under stress and significantly correlated with PSD, LVEF and LVESVI only under stress. The factors affecting GCW, both at rest and under stress, were the same as those affecting GWI, except that GCW at rest was correlated with LVEF. GWW was significantly associated with PSD in both the rest and stress states, while GWE was significantly associated with PSD and GLS in both the rest and stress states (Table 3).

Supplementary Table S1 shows the factors affecting ΔMW. ΔGWI was significantly correlated with ΔSBP, ΔDBP, ΔHR, ΔRPP and ΔGLS. GCW was significantly correlated with ΔSBP, ΔDBP, ΔHR, ΔGLS and ΔLVESVI. Both ΔGWW and ΔGWE were associated with ΔPSD, but ΔGWE was also associated with ΔSBP and ΔGLS.

Previous studies have shown that SBP is associated with GWI only in hypertensive patients, but in our study, SBP was associated with GWI not only in ANOCA patients with hypertension (rest: r = 0.695, p < 0.001, peak: r = 0.489, p = 0.002) but also in ANOCA patients without hypertension (rest: r = 0.727, p < 0.001, peak: r = 0.517, p < 0.001). And this correlation was also seen in ΔSBP and ΔGWI (hypertension group: r = 0.577, p < 0.001, non-hypertension group: r = 0.586, p < 0.001).

The significant predictors of CMD on univariable regression analysis were E/e′, peak GLS, ΔGLS, peak PSD, ΔPSD, peak GWI, peak GCW, peak GWW, ΔGWW, peak GWE and ΔGWE. The significant predictors of CMD on multivariable regression analysis were ΔPSD and ΔGWW (Table 4). We also performed multivariable modeling that adjusted for age, sex, diabetes mellitus, hyperlipidemia, obesity, and hypertension. ΔPSD and ΔGWW remained significantly correlated with CMD (p = 0.024, OR = 1.165; p = 0.020, OR = 1.037, respectively) after controlling for these factors. To eliminate the effect of blood pressure, we performed multivariable modeling that adjusted for stress-SBP and stress-DBP. ΔPSD and ΔGWW remained significantly correlated with CMD (p = 0.043, OR = 1.124; p = 0.015, OR = 1.031, respectively).

According to ROC analysis, 2D-STE and MW indicators have diagnostic value for CMD, with GWW and PSD having the larger AUCs (Figure 4 and Table 5). To improve the diagnostic efficacy for CMD, we combined ΔPSD and ΔGWW into a new joint predictive index, which showed stronger predictive performance (AUC: 0.913, sensitivity: 75%, specificity: 95.65%).

Excellent intra-observer and inter-observer variabilities were observed while measuring the MW parameters (Figure 5). For the intra-observer variability, the interclass correlations coefficients (ICC) of rest GLS, rest PSD, rest GWI, rest GWW, peak GLS, peak PSD, peak GWI and peak GWW were found to be 0.912, 0.941, 0.950, 0.937, 0.907, 0.957, 0.946 and 0.977 respectively. For the inter-observer variability, the ICC of rest GLS, rest PSD, rest GWI, rest GWW, peak GLS, peak PSD, peak GWI and peak GWW were found to be 0.954, 0.974, 0.976, 0.973, 0.956, 0.982, 0.952 and 0.991, respectively.

Most cardiovascular risk factors, such as advanced age, obesity, diabetes, and hyperlipidemia, promote the development of a systemic proinflammatory state and the accumulation of reactive oxygen species in the vascular endothelium, thereby causing an inflammatory response in the coronary microvascular endothelium and the extracellular interstitium of the myocardium, which are considered the pathophysiological basis of CMD. The inflammatory response in the myocardial extracellular interstitium promotes myocardial fibrosis and remodeling of the extracellular matrix, which reduces the elasticity of the heart and causes changes in GLS, an indicator of cardiac deformation (3, 25). In addition, myocardial interstitial fibrosis induces myocardial heterogeneous activation, and slows conduction in fibrotic areas, thereby impairing the synchronization of electromechanical conduction and causing deterioration of indicators of cardiac contraction synchronization such as PSD (7). As our findings demonstrated, the absolute value of GLS was lower in the CMD group than in the non-CMD group after adenosine stress, which is consistent with the findings of Hugo Rodriguez et al. and Tagliamonte et al. (26, 27).The CMD group also had impaired synchronization of cardiac contractions.

However, as adenosine is a vasodilator, it has potential effects on arterial blood pressure, and the load-dependent properties of 2D-STE-related indices dictate that they are not well interpreted under conditions of blood pressure variability.

Therefore, our study also explored the responsiveness of ANOCA patients with and without CMD to adenosine, which is a good complement to previous studies. In terms of hemodynamic indices, the CMD group showed no significant change in blood pressure after adenosine stress and a significant increase in RPP, but the non-CMD group showed a significant decrease in blood pressure. As for 2D-STE related indices, there was no significant change in PSD from before to after adenosine stress in the CMD group, but peak PSD was significantly lower than resting PSD in the non-CMD, indicating that adenosine improved cardiac systolic synchronization in the non-CMD group but not in the CMD group. Recent studies have confirmed that, under similar pathophysiological mechanisms, ANOCA patients with CMD often have combined peripheral vascular endothelial cell dysfunction, resulting in peripheral vasodilator dysfunction similar to that of coronary microvascular dysfunction (28, 29). Our results corroborate this idea. Recent studies have confirmed that many diseases that cause systemic inflammation can lead to CMD (30). In fact, in recent years, some scholars have viewed CMD as a cardiac manifestation of systemic disease. Cardiac fibrosis caused by CMD (31), especially interstitial fibrosis, leads to a reduced sensitivity to altered cardiac synchrony, which is why PSD in the CMD group did not change significantly before and after stress.

We further explored the effect of CMD on MW and the factors influencing MW. Since noninvasive MW has proven reliable, its application in ischemic heart disease is valid. For example, Natalie et al. demonstrated that GWI, GCW, and GWE were significantly reduced in patients with obstructive coronary artery disease (32), and Rodolfo et al. demonstrated that in STEMI patients treated with primary PCI, MW in the culprit vessel territory is independently associated with early adverse LV remodeling (33). However, an analysis of the effect of CMD as a type of ischemic heart disease on MW has been lacking relevant studies. Our study showed no significant difference in MW between ANOCA patients with and without CMD at rest. However, after adenosine administration, GWI, GCW, and GWE were lower in the CMD group than in the non-CMD group, while GWW was higher in the CMD group.

MW is an indirect measure of cardiac contractility and myocardial oxygen consumption, and our study confirms that this is mainly reflected in GWI and GCW. While ΔSBP, ΔDBP and ΔRPP were similar in the CMD and non-CMD groups, the difference between peak GWI and peak GCW should be attributed to the difference in contractility at stress between the two groups, which was corroborated by the differences in peak GLS and ΔGLS between the two groups. Peak GWW and ΔGWW showed a positive correlation with peak PSD and ΔPSD, so we considered that higher GWW in the CMD group at stress was associated with cardiac contractile dyssynchrony. Thus, we believe that 2D-STE related indices and MW have potential clinical value in the diagnosis of ANOCA with CMD.

However, MW is not independent of blood pressure. Tsai et al. (11) showed a correlation between MW and blood pressure in their study of untreated hypertensive patients. They found that compared to controls, hypertensive patients had significantly higher GWI and GCW, but lower GWE. In the correlation analysis, Tsai et al. found a significant positive correlation between GWI and GCW with blood pressure, which is identical to our results and further corroborates our conclusions. Our study showed that patients in the non-CMD group had a significant decrease in blood pressure at stress, while GWI and GCW increased significantly, whereas patients in the CMD group with no change in blood pressure had no significant increase in GWI or GCW. This suggests that ANOCA patients with CMD have lost cardiac contractile reserve capacity at stress, a manifestation of subclinical impairment of cardiac systolic function. The changes in MW after adenosine stress in ANOCA patients with CMD in the present study followed the same pattern of changes in MW in patients with positive exercise stress echocardiograms in the study by Andrew et al. (34). Our findings further reveal the essence of CMD as ischemic heart disease, which is also supported by the relationship between CMD and markers of myocardial injury (35, 36). As for GWW and GWE, since their correlation with blood pressure was not found in the correlation analysis, their changes could be considered to be related to changes in the synchrony of cardiac contraction.

PSD is a strain-based quantification of the time synchronization of electromechanical conduction and an important index for the study of cardiac systolic function (7). Abnormal electromechanical synchronization decreases the mechanical efficiency of left ventricular ejection, increases energy loss and ineffective work, and reduces global work and efficiency. Previous studies have confirmed the significant alteration and prognostic value of PSD in various cardiovascular diseases such as hypertrophic cardiomyopathy and aortic stenosis, and its correlation with patients' symptoms (7, 37). Our multivariate regression analysis indicated that ΔPSD and ΔGWW were independently associated with CMD, suggesting that altered cardiac systolic synchrony and increased wasted work under stress may be the main effects of CMD on cardiac systolic function. For predicting CMD, both ΔGWW and ΔPSD showed high predictive value, while the joint predictor consisting of both demonstrated superior predictive power (AUC: 0.923). Although GWW is thought to be associated with cardiac contraction asynchrony, there has been a lack of data from relevant clinical studies to support this hypothesis, but our findings support it.

In summary, our study shows that the changes in cardiac systolic function in ANOCA patients with CMD after adenosine stress are mainly manifested as reduced systolic synchronization and increased useless work, resulting in reduced cardiac contraction efficiency and increased energy depletion, which eventually lead to impaired cardiac function. These findings provide important targets and ideas for the early diagnosis and treatment of ANOCA patients with CMD.