Prediction of Significant Coronary Artery Disease Through Advanced Echocardiography: Role of Non-invasive Myocardial Work

Background: Non-invasive prediction of critical coronary artery stenosis (CAST) in patients with coronary artery disease (CAD) is challenging. Strain parameters can often capture an impairment of regional longitudinal function; however, they are load dependent. A novel non-invasive method to estimate Myocardial Work (MW) has been recently proposed, showing a strong correlation with invasive work measurements. Our aim was to investigate the ability of non-invasive MW to predict the ischaemic risk area underlying a CAST. Methods and Results: The study population comprises 80 individuals: 50 patients with CAST and 30 controls (CTRL). Echocardiography recordings were obtained before coronary angiography to measure global longitudinal strain (GLS), Myocardial Work Index (MWI), Myocardial Constructive Work (MCW), Myocardial Wasted work (MWW), Myocardial Work Efficiency (MWE). Global MWI (p = 0.048), MWE (p < 0.001), and MCW (p = 0.048) at baseline were significantly reduced in patients with CAST compared to controls (p < 0.05). Regional MWE within the myocardial segments underlying the CAST, but not LS, was significantly reduced compared to non-target segments (p < 0.001). At ROC analysis, the diagnostic performance to predict CAST for regional MWE (AUC = 0.920, p < 0.001) was higher compared to both regional post-systolic shortening index (PSI) (AUC = 0.600, p = 0.129) and regional LS (AUC = 0.546, p = 0.469). Conclusions: Non-invasive estimation of MW work indices is able to predict a CAST before invasive angiography.


INTRODUCTION
Reliable identification of inducible ischemia in patients with suspected coronary artery disease (CAD) is clinically crucial to their management (1,2). However, non-invasive prediction of critical coronary artery stenosis (CAST) is challenging, leading to unnecessary invasive angiographies (3,4).
Despite most patients with significant coronary stenosis have preserved left ventricular (LV) ejection fraction (EF) at rest, regional wall motion abnormalities (WMAs) may be found, but they are often too subtle to be captured.
Current non-invasive risk stratification algorithms are time consuming (2), not sufficiently accurate to prevent redundant invasive angiographies, and mostly not indicated in the emergency setting (5,6).
Boe et al. (21) demonstrated that regional MWI is a sensitive marker of acute coronary occlusion (ACO). However, their study was conducted on patients with acute myocardial infarction (AMI), and could not assess the ability of regional MWI to recognize CAST in patients with stable CAD. Another study demonstrated in a non-acute setting that non-invasive MW parameters were sensitive marker of myocardial ischemia during induced transient ACO (22).
Aim of this study was to test whether non-invasive global and regional myocardial work indices might be able to identify the myocardial area underlying a CAST (functional ischemic risk area, FRA) to reduce the number of unnecessary coronary angiographies.

Study Population
All patients undergoing coronary angiography at a single center during the study period were screened for this study. Among them, 50 consecutive patients with one critical coronary artery stenosis (CAST) at coronary angiography were included into the CAST group and 30 consecutive patients with no critical coronary stenoses were included into the control group. The following criteria have been used as inclusion criteria: • ≥18 years of age; • Planned coronary angiography; • Presence of a critical coronary stenosis, diagnosed during coronary angiography; • Patients gave their consent.
An acute myocardial infarction (AMI) within 30 days, QRS-width of ≥120 ms, severe valvular disease, previous cardiac surgery, extensive comorbidity, or the presence of atrial fibrillation at ECG were exclusion criteria. All study patients underwent echocardiography and invasive coronary angiography. All patients were stable and provided informed consent. The study was approved by the regional ethics committee.

Study Timeline, Procedures, and Analysis Plan
A complete echocardiography examination along with measurement of non-invasive automatic oscillometric blood pressure (NINV) were obtained before performing coronary angiography.
Aim of the study was the evaluation of the diagnostic performance of STE and non-invasive echocardiography-derived Myocardial Work parameters to predict CAST (≥70%) before the invasive study. To this purpose, baseline measurements of all study patients were assessed and compared to CTRLs. Regional LV performance indices, calculated in the functional ischemic risk area (FRA), downward of CAST, were compared to the remaining (non-ischemic) myocardial segments.

Echocardiographic Analyses
Two-dimensional (2D) 4-chambers, 3-chambers, and 2chambers apical views were acquired with a frame rate ≥60 fps, as described elsewhere (23), and transferred to a dedicated workstation (EchoPAC, GE Healthcare). The echocardiographic recordings were processed using a semi-automatic speckle-based strain software analysis (EchoPAC version 112.99, Research Release, GE Healthcare). LV peak systolic, longitudinal strain (LS), and post-systolic Shortening Index (PSI) were calculated by tracing the LV endocardium's inner border in each view. The EchoPAC software automatically identifies the region of interest (ROI), then approved by the operator supervising the whole process in order to ensure the proper delimitation of the LV wall and the correct assessment of the wall thickness into the analysis. Finally, data of segmental and global LV peak systolic LS and PSI, segmental and global, were processed and analyzed by the software through an 18-segment model. PSI was calculated as previously described by Kukulski et al. [PSI = (peak systolic strain -end-systolic strain)/peak systolic strain] (7). Reproducibility was assessed on a subset of randomly selected echocardiograms (n = 15). Two operators independently and blindly performed the assessment and one of the two additionally analyzed the same series of echocardiographic exams a second time blindly. Inter-and intra-rater reliability were then assessed using the intraclass correlation coefficient (ICC).

Evaluation of Regional Function in Patients With Critical Coronary Stenosis
The functional ischemic risk area (FRA) and the non-risk areas (NRA) were selected according to the vascular distribution of the coronary vessels with or without critical coronary stenosis (CAST), as previously described (24). Briefly, the inferior midand basal segments imaged in the apical two-chamber view were considered FRA segments in presence of a CCS of the right coronary artery (RCA). In presence of a CAST of the left circumflex coronary artery (LCx), the basal-and mid-lateral segments in the apical four chamber view were considered "atrisk" (25). Both septal mid-and apical segments from the apical four-chamber view, were considered "at-risk" in presence of a CAST of the left anterior descending coronary artery (LAD), as previously described (21,24,25).
Empirical cut-off values were set to identify segmental systolic dysfunction for MWI and according to Boe et al. published data (21). The presence of ≥4 adjacent dysfunctional segments assessed by MWI were used to assess the FRA area in order to detect the presence of CAST.

Data Analysis and Statistics
Continuous values are expressed as mean ± SD. Mean differences between groups were analyzed with ANOVA and the unpaired Student's t-test. The Mann Whitney U-test was used to compare two groups for non-normally distributed variables. Intra-and inter-rater agreement was assessed using the intraclass correlation coefficient (ICC).
Discrete variables are reported as absolute numbers and percentages of the group. Comparisons between discrete variables were performed using the Chi-squared test.
Receiver operating characteristic (ROC) curve analysis was used to identify optimal cut-off values, using the Youden method. Comparisons between the AUC of different ROC Curves were performed using the DeLong method, as already described (26,27).
Net reclassification improvement (NRI) index was calculated to quantify the proportion of tests being correctly reclassified with a test with a higher AUC compared to another, as already elsewhere described (28)(29)(30).
The statistical analyses were performed using IBM SPSS v.21.1 (IBM corp., Armonk, NY, USA). A two-tailed P-value of 0.05 was interpreted as significant.

Study Population
Eighty patients undergoing coronary angiography for clinical indication have been enrolled in this study, including fifty patients with coronary artery disease (CAD) and critical coronary

Baseline LV Performance Indices in Patients With Critical Coronary Stenoses
Left ventricle Ejection Fraction (LVEF) was within normal range at baseline, without significant difference between the groups (p = 0.089). No regional wall motion abnormalities were present at rest. Baseline global MWE (92.3 ± 5.1%) was significantly reduced in patients with CAST compared to controls (96.6 ± 1.7%, p < 0.001; Figure 1). Similarly, baseline global MWI (1,843 ± 540 mmHg% vs. 2,070 ± 246 mmHg%, p = 0.048), MCW  (Figure 1). Exemplificative pressure-strain curves (left panels) and bullseye views of MWI (right panels) of two patients: one with no coronary stenosis (upper panel) and one with CAST of the Left Anterior Descending (LAD) (lower panel) are shown in Figure 2.
Regional MWE, measured within the FRA (i.e. underlying the target vessel with CAST) was significantly decreased by 21% (p < 0.001) compared to NRA of the same patients, while no significant change was found for regional LS (p = 0.476). Regional PSI measured within the FRA was not significantly higher than in NRA, although it was numerically higher (5.9 ± 6% vs. 3.5 ± 4%; p = 0.129; Figure 3). At ROC analysis, the diagnostic performance to predict CAST for regional MWE (AUC = 0.920, p < 0.001) was higher compared to both regional PSI (AUC = 0.600, p = 0.129) and regional LS (AUC = 0.546, p = 0.469; Figure 4). Accordingly, direct comparison of ROC curves using the method developed by DeLong confirmed a significantly higher AUC for regional MWE compared to both regional PSI (p < 0.001) and regional LS (p < 0.001; Table 3).
Finally, we compared diagnostic classification by regional indices using Reclassification Statistics, revealing that regional MWE was able to correctly reclassify myocardial segment underlying a CAST in a substantially higher number of segments, as demonstrated by a strong NRI of 0.74 (Figure 5). The amount (segments/total) of FRA and NRA segments is shown in Table 4.

Reproducibility
The assessment of non-invasive myocardial work had an excellent reproducibility. In fact, the intraclass correlation coefficient for intra-rater assessment was excellent for all

DISCUSSION
Our results demonstrate that: (1) baseline global MWI, MCW, MWE are significantly worse in patients with CAST compared to CTRLs; (2) regional MWE at baseline is significantly worse in myocardial segments underlying a CAST and might therefore be exploited to identify the myocardial region downward of a CAST before invasive coronary angiography. Regional MWE outperforms both regional PSI and regional LS for this application.

Myocardial Work and Ventricular Function
Regional LV function is highly dependent on subendocardial blood flow (34). Parameters of longitudinal deformation are dependent by the loading conditions and this could limit their accuracy. Invasively measured Myocardial Work has been historically used as marker of ventricular contractility (35)(36)(37)(38). It was later demonstrated that it is feasible to provide similar physiological information as pressure/strain loops (33).
Recently, Russell et al. (18) validated a novel method for noninvasive MW estimation, using of speckle tracking analysis and an estimation of LV pressure from brachial artery cuff pressure. The NORRE sub-study demonstrated a good reproducibility, reporting reference ranges for non-invasive MW (39).
MW indices have been also evaluated by Chan et al. (40) in three different cardiovascular scenarios, such as hypertension, ischaemic, and not-ischaemic dilated cardiomyopathy. Indeed, this group demonstrated a significant increase of MWI in hypertensive patients, despite normal global longitudinal strain values.
All these evidences support the hypothesis that, since myocardial work indices take into account multiple hemodynamic factors, they could represent a valid implementation to the sole strain measurements.
In this context, our results confirm and further extend previous results by Edwards et al. (31). In line with their findings, we found that global MW indices, GLS and PSI at baseline were significantly different in CAD patients as compared to controls. However, unlike them, at ROC analysis we found a similar diagnostic performance of global MW indices and On the other hand, the study population included by Edwards et al. presents a lower degree of interindividual heterogeneity which advantages comparisons between the groups, while our population represents a real-World sample and is therefore nearer to actual clinical practice. In addition, Edwards et al. had a large prevalence of multivessel CAD (62.5% among CAD patients), which would suggest larger ischemic areas. The smaller ischemic areas in our all-comers population might have diluted the impact of regional impairment of LV performance on global indices. In this regard, a key novelty of our study compared to the previous by Edwards et al. is the assessment of regionalized LV performance indices. In particular, regional MWE might solve one of the most relevant limitations of the approach proposed by Edwards et al., namely the very low specificity. In fact, in the present study regional MWE had a 100% specificity using the cut off previously proposed by the NORRE Study (32,39,41).

Myocardial Function in Patients With Critical Coronary Stenoses
Regional and global myocardial work non-invasive assessment by LV pressure-segment length loops is a feasible method of evaluate myocardial performance during ischaemia in an experimental setting as recently demonstrated by Boe et al. (21).
CAST are associated with chronic myocardial blood flow reduction, which may cause impairment in regional function within the ischaemic risk area. However, the echocardiographic documentation of the resulting impairment in systolic function is frequently challenging. In fact, its manifestation is not only related to the myocardium contractile state, but it is also sensitive to variations in preload and afterload. Indeed, previous studies have already demonstrated that an elevation in afterload may cause a further decrease in systolic shortening of an ischaemic segment (42).
Boe et al. (21) exemplified how regional contraction may be influenced by load conditions, showing that segments can falsely be interpreted as dysfunctional using strain values alone in case of high systolic blood pressures. In fact, they (21) elegantly demonstrated that MWI reclassified several of those segments as normal, correcting in this way the falsely interpreted impairment in systolic function derived by the strain measurement in patients with a high afterload and no coronary occlusion. In line with those findings, our study complements results by Boe et al. (21) demonstrating for the first time that use of regional MWE is able to identify at baseline CAD patients with CAST before invasive angiography with excellent performance (AUC = 0.92), providing a substantial improvement of specificity compared to global LV performance indices in this setting. The use of regional MWE was superior to both global LV performance parameters than and other regional LV deformation indices. In particular, use of MWE was able to correctly reclassify most patients with a strong NRI of 0.74 compared to regional PSI. Further differences between MWI and strain are related to the hypothesis that peak strain values have restricted ability to reveal myocardial performance. As a matter of fact, peak strain is determined by the afterload concurrently with contraction, while numerically equal peak segmental strain values are not associated to the same level of segmental systolic function in early vs. mid-systole, reflecting the impact exerted by a higher afterload in mid-systole. Our results extend and complement previous findings (7,8,43). In our study, global PSI had a direct correlation with global MWW but an inverse correlation with global MWE. This is compatible with the hypothesis that the adjunctive post-systolic LV deformation measured as an increased PSI under myocardial ischemia mostly generates wasted work in the FRA, reducing global LV mechanics efficiency.

Application of Non-invasive Myocardial Work in Clinical Practice
The findings of this study might have a significant impact on clinical practice. In fact, approximately 60% of coronary angiograms performed for clinical indication find no critical coronary stenoses, according to the National Cardiovascular FIGURE 4 | The figure depicts the ROC curves for regional MWE (green), regional PSI (blue), and regional LS (red), reflecting their respective diagnostic performance to predict CAST. ROC, receiver operating characteristic; MWE, Myocardial Work Efficiency; PSI, post-systolic shortening index; LS, Longitudinal Strain; CAST, critical coronary stenosis.  (31). # Clinically validated cutoff for PSI (32). § Previously published cutoff for MWE (33). @ MWE cutoff found in the present study according to the Youden Index method.
Data Registry of the American College of Cardiology, exposing a large number of patients to the avoidable risks of an invasive diagnostic procedure (44). In this context, the use of a noninvasive ultrasound-based screening test could be useful to improve the clinical management process of patients with suspected CAD, a currently unmet need in clinical practice.
Several diagnostic methods have been tested already for this purpose, with unsatisfactory results. Among the others, the delay of onset of flow mediated dilation (FMD) seems to be the most promising, yet being able to predict the presence of CAST in only 50% of patients with evidence of coronary stenosis above 70% at invasive angiography (45   non-useful invasive coronary angiograms, although they expose patients to ionizing radiation and/or high magnetic fields. The results of the present study should be contextualized in the larger picture of clinical practice. In particular, it is important to take into account pre-test probability (PTP), namely the likelihood that the patient has coronary atherosclerosis and thus CAST. This depends in turn on the prevalence of CAD in the study population. Diagnostic testing is most useful when the likelihood is intermediate. The sample size of the present study did not allow stratification based on PTP. Nevertheless, most patients were in the intermediate PTP range. It is paramount to assess PTP before taking any clinical decision or performing a diagnostic test. I fact, a low PTP should suggest a conservative approach with no further diagnostic testing, as these patients have an excellent prognosis without clinical action. On the other hand, patients with high or very high PTP shouldn't undergo intermediate diagnostic tests, as a false negative result might eventually avoid the appropriate clinical pathway (46). Of course, PTP also impacts the results of the present study. Consequently, these results might not apply to populations with substantially different PTP.
Our finding that regional MWE was significantly impaired in the FRA has potential clinical implications, as it could be of help to guide percutaneous treatment in case of multivessel disease with multiple intermediate stenoses.

Limitations
Echocardiographic images in 7 patients (16%) were not analysable due to poor image quality. These results are in agreement with recent studies (33). However, this could limit the applicability of our findings. On the other hand, the contrast agent used during the study might improve feasibility. Although the results coherently point in the same direction, the limited sample size does not allow to draw a definite conclusion on this topic, neither they can be generalized to different cohorts. The present results should be regarded as preliminary and hypothesis-generating and need to be validated in larger studies before they can be applicable. CONCLUSIONS Global and regional assessment of non-invasive Myocardial Work Indices under basal conditions is potentially able to identify myocardial segments underlying CAST. Large prospective studies will be needed to validate our findings and to confirm whether they might be helpful to improve clinical stratification of patients with suspected coronary stenosis.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Magna Graecia University Catanzaro. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
JS and SD: conceptualization. SD and CI: methodology. AS, IL, AP, and SS: formal analysis. JS and AP: writing-original draft preparation. CI, GD, JS, and SD: writing-review and editing. CS, AP, and SS: visualization. CI: supervision. All authors contributed to the article and approved the submitted version.

FUNDING
This study was partly supported through an ESC Training Grant 2019, awarded to JS.