Development and validation of a diagnostic nomogram to predict significant stenosis of the left anterior descending branch of the coronary artery by stress echocardiography

Introduction

Coronary artery disease (CAD), with its complications, is one of the major causes of death in patients with cardiovascular disease around the world and causes a vast social and economic burden (1). Thus, accurate diagnosis and aggressive intervention are essential to prevent adverse cardiac events in patients with CAD. Although coronary angiography is the “gold standard” for the diagnosis of CAD, the latest international guidelines recommend stress imaging as the first choice for patients with suspected CAD (2). Currently, although stress imaging and noninvasive imaging are widely used to evaluate CAD, noninvasive quantitative assessment of coronary stenosis and myocardial ischemia in patients with CAD remains a major challenge.

Previous studies have used stress echocardiography (SE) combined with myocardial contrast and speckle tracking to diagnose myocardial ischemia in CAD (3,4). However, myocardial contrast is highly subjective, which affects the reproducibility of the results. Although SE combined strain improves the accuracy of quantitative evaluation of myocardial exercise capacity, the dependence of strain parameters on afterload may lead to incorrect assessment of myocardial function (5,6).

The myocardial work (MW) technique for noninvasive acquisition of MW parameters combines brachial artery blood pressure and longitudinal strain (LS) with speckle tracking. It overcomes the dependence of myocardial strain on afterload and can reflect global and regional myocardial mechanical changes more accurately and objectively (7). MW combined with SE detects global myocardial function under myocardial stimulation and may improve the detection of significant CAD (8). Additionally, using transthoracic Doppler flow imaging, coronary flow reserve (CFR) is measured noninvasively by determining the distal flow velocity of the left anterior descending branch (LAD) at rest and during stress. It can reflect the dilated dysfunction of the coronary artery and the degree of myocardial ischemia (9).

However, the diagnosis of coronary artery stenosis and myocardial ischemia based solely on myocardial mechanics or hemodynamics is relatively limited. The results of studies by Edwards et al. (10) and Wang et al. (11) show that global MW parameters are powerful predictors for screening CAD, but their effectiveness still needs to be improved [area under the curve (AUC) <0.800]. Therefore, it is necessary to develop a multi-parameter model that assesses myocardial ischemia from multiple perspectives. Lin et al. (8) demonstrated that a new model, including peak global work efficiency (WE) and recovery-phase global wasted work, has high diagnostic value for significant coronary artery stenosis but does not incorporate hemodynamic parameters. As a major branch of the coronary artery, the LAD provides nearly 50% of the blood supply to the left ventricle and is the most common location for coronary artery stenosis (12). However, there is a lack of research data on SE in detecting regional myocardial exercise capacity. Therefore, this study aimed to develop and validate the value of a multivariate logistic regression model (MLRM) in the preliminary exploration of the combined use of MW and CFR in patients with significant LAD stenosis regarding myocardial function and coronary reserve function. We present this article in accordance with the TRIPOD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1404/rc).

Methods

Study population

This study retrospectively collected 177 cases of patients suspected of having CAD from a single-center comprehensive hospital (the Fourth Affiliated Hospital of Harbin Medical University) between January 2023 and March 2025. Inclusion criteria included the presence of symptoms related to myocardial ischemia (e.g., chest pain, chest tightness, palpitations, and shortness of breath) and clear image quality on SE within 7 days after coronary angiography. The exclusion criteria were as follows: (I) congenital heart disease, myocardial infarction, heart failure, severe valvular disease, severe pulmonary hypertension, severe hypertension, and recent stroke; (II) abnormal left ventricular function at rest, with left ventricular ejection fraction (LVEF) <52% in men and LVEF <54% in women (13); (III) the stenosis rate of the left circumflex artery (LCx) and right coronary artery (RCA) was ≥50%; (IV) previous coronary intervention or coronary artery bypass grafting.

A total of 120 patients were finally enrolled and categorized into the LAD significant stenosis group and the LAD nonsignificant stenosis group based on the results of coronary angiography. The participants were randomized into the derivation cohort (n=84) and validation cohort (n=36) in a ratio of 7:3. All patients underwent conventional echocardiography, SE, and coronary angiography. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Ethics Committee of the Fourth Affiliated Hospital of Harbin Medical University (approval No. 2024-ethical review-83) and informed consent was provided by all individual participants. The patient screening flowchart is shown in Figure 1.

Figure 1 Patient screening flowchart. LAD, left anterior descending; LCx, left circumflex artery; LVEF, left ventricular ejection fraction; RCA, right coronary artery.

Conventional echocardiography

All patients underwent two-dimensional (2D) conventional echocardiographic assessments acquired by an experienced ultrasound physician using a M5Sc-D 1.5–4.6 MHz phased-array probe (Vivid E95, GE HealthCare, Chicago, IL, USA) in accordance with the American Society of Echocardiography guidelines (13). The internal diameter of each cardiac chamber was obtained using 2D ultrasound, mitral flow velocity was obtained using pulsed Doppler, and LVEF was calculated using a modified biplane Simpson method.

Stress echocardiography

Patients were instructed to hold their breath, and dynamic images of 3–5 cardiac cycles were acquired from apical four-chamber, two-chamber, and three-chamber cardiac sections, respectively. The optimal section showing distal LAD flow was explored in a modified left ventricular long-axis view, where distal LAD flow velocities were measured and averaged 3 times at rest. Brachial artery pressure and heart rate were also measured at rest. The LAD distal blood flow spectrum was continuously collected to record the LAD distal blood flow velocity at the peak flow rate by peripheral intravenous injection of Regadenoson (14) at the elbow. The CFR was calculated from the ratio of peak LAD blood flow velocity to resting blood flow velocity (14). Dynamic images of 3–5 cardiac cycles in apical four-chamber, two-chamber, and three-chamber cardiac sections were acquired at peak stress. Brachial artery pressure and heart rate were also measured under stress.

Image offline analysis

Acquired dynamic images of apical views were imported into a specialized Echo PAC 203 workstation (GE HealthCare) for offline analysis. Automated functional imaging (AFI) mode was selected and the region of interest boundaries were manually adjusted as prompted. The system automatically tracked and recognized the wall motion of the left ventricle, acquiring a bull’s-eye plot of LS in each segment. The MW module was and the brachial artery blood pressure was entered. The software calculates the MW parameters by multiplying the left ventricle LS by the instantaneous left ventricle pressure. (I) The global work index (WI) indicates the total work the left ventricle completes from mitral valve closure to opening; (II) global WE is the ratio of global construction work to the sum of global construction work and the global wasted work. Global construction work is a form of practical work that promotes left ventricular ejection and maintains normal left ventricular function. Global wasted work represents the energy loss during contraction and diastole of the left ventricle (15,16).

The WI bull’s-eye plot, WE bull’s-eye plot, and peak strain dispersion (PSD) were acquired. Regional LS_LAD, regional WI_LAD, and regional WE_LAD were calculated for LAD-supplying segments according to the ASE guideline-recommended 17-segment division of the left ventricle (13) (Figure 2).

Figure 2 Bull’s eye plot of MW parameters for nonsignificant stenosis (A,B) and significant stenosis (C,D) of the LAD under stress (the red solid line segments represent the main blood supply segments of the LAD). ANT, anterior; BP, blood pressure; GCW, global constructive work; GLS, global longitudinal strain; GWE, global work efficiency; GWI, global work index; GWW, global wasted work; INF, inferior; LAD, left anterior descending branch; LAT, lateral; LVP, left ventricular pressure; MW, myocardial work; POST, posterior; SEPT, septal.

Coronary angiography

Experienced cardiologists evaluated the coronary angiography images of all patients without prior knowledge of the echocardiographic findings. LAD significant stenosis was defined as ≥70% stenosis (10,11).

Statistical analysis

The software SPSS 25.0 (IBM Corp., Armonk, NY, USA) was used to perform the following statistical analyses: continuous variables were assessed for normality using the Shapiro-Wilk test. Normally distributed data were expressed as mean ± standard deviation (SD), and Student’s t-test was applied for comparison between groups. Non-normally distributed data were expressed as medians and interquartile ranges, and the Mann-Whitney U test was applied to compare groups. Categorical variables were expressed as frequency counts and percentages, and either the Chi-squared test or Fisher’s exact test was applied for comparison. Receiver operating characteristic (ROC) curves were plotted to compare the ability of SE parameters to discriminate significant stenosis in the LAD. Correlations between continuous variables were analyzed using Pearson’s or Spearman’s correlation coefficients.

R Software (R Foundation for Statistical Computing, Vienna, Austria) was applied to analyze the binary classification problem. The prediction model was developed in two steps: first step, to avoid overcomplexity of the model, we used a parameter-based least absolute shrinkage and selection operator (LASSO) model to automatically screen and identify potential information variables that detect LAD significant stenosis from the original data (17). LASSO regression is a statistical method widely used in biomedical research for feature selection. Its core advantage lies in automatically identifying and retaining the most predictive variables while compressing the contribution of irrelevant or redundant variables to zero. Second step, all participants were randomly divided into the derivation and validation cohorts in a 7:3 ratio. Variables were selected based on the λ values and a one standard error range of λ values from step 1. The MLRM for diagnosing LAD significant stenosis was constructed using the derivation cohort. A nomogram was developed to visualize the model prediction results, which provides clinicians with a more straightforward and quantitative tool for non-invasive diagnosis of significant stenosis of the LAD (18).

Afterward, ROC curves were applied to test the diagnostic capabilities of the model, and calibration curves were used to evaluate the model’s calibration (19). External data were applied to verify the robustness of the model. Decision curve analysis (DCA) was applied to evaluate the net benefit of the model. Lastly, the modeling conditions were evaluated based on the events per variable (EPV) principle (20), and external cohorts were applied to validate the robustness of the model. All tests were two-sided and were considered statistically significant at P<0.05.

Intra- and inter-observer variability

A total of 15 patients were randomly selected to have CFR and MW-related parameters measured by two ultrasound physicians who were unaware of the patient’s clinical data and each other’s results. Intra-observer consistency was assessed by spacing the same observer two weeks apart. A second independent observer assessed inter-observer variability. The interclass correlation coefficient (ICC) was used to evaluate the variability between data.

Results

Patient characteristics

The clinical baseline characteristics of all study populations are shown in Table 1. Smoking rate, hypertension, and low-density lipoprotein abnormalities were significantly increased in the LAD-significant stenosis group compared with the nonsignificant stenosis group (P<0.05). In both groups, no significant differences were observed in terms of gender, age, body mass index (BMI), blood pressure, rest and peak heart rates, diabetes mellitus prevalence, and abnormalities in total cholesterol, triglycerides, and high-density lipoproteins (P>0.05).

Comparison of conventional echocardiographic parameters

Conventional echocardiographic parameters are shown in Table 2. All conventional echocardiographic parameters did not differ between the nonsignificant stenosis and significant stenosis groups (P>0.05).

Comparison of SE parameters

At rest, compared with the nonsignificant stenosis group, WI_LAD and WE_LAD were significantly lower in the significant stenosis group (P<0.05), and there was no significant difference between LS_LAD and PSD (P>0.05). At stress, LS_LAD, PSD, WI_LAD, and WE_LAD were different between the two groups (P<0.05). Compared with the nonsignificant stenosis group, the significant stenosis group had higher LS and PSD and lower WI_LAD and WE_LAD (Table 3).

According to the ROC analysis, CFR had the most significant AUC (0.860), with an optimal threshold of 2.04, a sensitivity of 0.789, and a specificity of 0.825. Among the MW-related parameters, Peak WE_LAD had the highest capacity to detect significant stenosis (AUC =0.806), with an optimal threshold of 88.50%, a sensitivity of 0.632, and a specificity of 0.810 (Figure 3).

Figure 3 Evaluation of diagnostic efficacy of stress echocardiography parameters. (A) ROC curves of stress echocardiography parameters; (B) the histogram of AUC for stress echocardiography parameter ROC. AUC, area under the curve; CFR, coronary flow reserve; LS, longitudinal strain; PSD, peak strain dispersion; ROC, receiver operating characteristic; WE, work efficiency; WI, work index.

Correlation analysis between MW-related parameters and CFR

Rest PSD, Rest WI_LAD, Peak LS_LAD, Peak PSD, and Peak WI_LAD correlate well with Rest LS_LAD (r=0.677, −0.732, 0.795, 0.734, −0.700). Rest PSD, Rest WI_LAD, and Peak LS_LAD correlated well with Peak PSD (r=0.858, −0.702, 0.696). Rest WI_LAD, Rest WE_LAD, Peak LS_LAD, and Peak PSD correlated well with Peak WI_LAD (r=0.869, 0.665, −0.718, 0.694). Rest WE_LAD, Peak WI_LAD, and Peak WE_LAD correlated well with CFR (r=0.716, 0.636, 0.720). Rest PSD and Rest WI_LAD correlated well with Peak LS_LAD (r=0.629, −0.696). Additionally, Rest PSD correlated with Rest WI_LAD (r=−0.657). Rest WE_LAD correlated well with Peak WE_LAD (r=0.754) (Table 4).

Variable screening and diagnostic predictive model construction

In step 1 of the diagnostic predictive model development, the derivation cohort selected five variables based on LASSO regression when the cross-validation error was minimal to construct the best-fitting model. The constructed diagnostic equations for MLRM were as follows: logitP1 =11.301 + 0.473 × hypertension + 0.058 × Peak DBP + 0.197 × LAd − 0.183 × Peak WE_LAD − 3.232 × CFR. In addition, LASSO regression identified the two most important variables within one standard error of the minimum error in cross-validation error to construct a more concise and generalizable model. Its MLRM diagnostic equation is as follows: logitP2 =19.683 − 0.160 × Peak WE_LAD − 2.644 × CFR (Figure 4).

Figure 4 LASSO regression feature selection. (A) Path diagram of LASSO regression coefficients; (B) LASSO regression cross-validation curve. A, peak late diastolic mitral flow velocity; AAo, ascending aorta; BMI, body mass index; CFR, coronary flow reserve; DBP, diastolic blood pressure; E, peak early diastolic mitral flow velocity; e', early diastolic velocity of the mitral annulus; HDL, high-density lipoprotein; HR, heart rate; IVSd, interventricular septal thickness at end-diastole; LAd, left atrial diameter; LASSO, least absolute shrinkage and selection operator; LDL, low-density lipoprotein; LS, longitudinal strain; LVEF, left ventricular ejection fraction; LVIDd, left ventricular internal dimension at end-diastole; LVPWd, left ventricular posterior wall thickness at end-diastole; PSD, peak strain dispersion; RV, right ventricular; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides; WE, work efficiency; WI, work index.

Model evaluation and validation

The derivation cohort ROC results showed AUC sizes as follows: MLRM1 > MLRM2 > univariate logistic regression model (ULRM)-CFR > ULRM-Peak WE_LAD (Table 5). Moreover, the results of the calibration curve for the best MLRM1 indicated a high agreement between its predicted probability of significant stenosis of the LAD and the actual observed probability (Figure 5). The results of DCA showed that the model provided high clinical net benefit within a reasonable threshold probability range (Figure 5). The EPV of this cohort was 11.4 (57 positive events/5 variables), exceeding the methodological threshold (EPV >10) and meeting the requirements for robust modeling. Applying the predictive model to an independent validation cohort showed that MLRM1 had good stability (AUC of 0.889 for the derivation cohort and 0.888 for the validation cohort) (Figure 6). The best predicted MLRM1 results are shown as nomograms in Figure 7.

Figure 5 Calibration curves for MLRM1 (A) and DCA (B). DCA, decision curve analysis; MLRM, multivariate logistic regression model.

Figure 6 ROC curves for MLRM1 derivation cohort (A) and validation cohort (B). MLRM, multivariate logistic regression model; ROC, receiver operating characteristic.

Figure 7 Nomogram of MLRM1. Instructions: locate the CFR value obtained by the SE on the CFR axis. Draw a line vertically upward from that point on the axis to determine how many points represent the probability of significant stenosis in the LAD. Repeat this process for each variable. Add up the scores obtained for each predictor. Find the final sum on the total score axis. Draw a straight line downward to calculate the probability of significant stenosis of the patient’s LAD. CFR, coronary flow reserve; DBP, diastolic blood pressure; LAd, left atrial diameter; LAD, left anterior descending branch; MLRM, multivariate logistic regression model; SE, stress echocardiography; WE, work efficiency.

Repeatability test

All MW-related parameters and CFR showed good reliability and stability. The same observer measured the parameter results at intervals with higher ICC values. In contrast, inter-observer ICC was slightly lower but still well correlated (Table 6).

Discussion

The main findings of this study are summarized as follows: (I) WI_LAD and WE_LAD were significantly reduced in the significant stenosis group compared with the nonsignificant stenosis group in patients with preserved LVEF, and MW parameters have potential diagnostic value in identifying significant stenosis in the LAD; (II) CFR is the best single factor for diagnosing significant stenosis in the LAD after drug stress, followed by Peak WE_LAD; (III) the MLRM1 constructed by multivariate logistic regression had the best efficacy in predicting significant stenosis of the LAD, and the model had good stability.

Comparison of the detection capabilities of MW, LS, and LVEF for significant stenosis in the LAD

The pressure-volume loop, first proposed by Otto Frank, describes the relationship between left ventricular volume and pressure at various stages of the cardiac cycle, objectively and quantitatively reflecting myocardial mechanical functions (21). However, its invasiveness and complexity of operation limit clinical application. In recent years, noninvasive pressure-strain loops obtained by combining LS with brachial blood pressure rather than left ventricular pressure have been used to estimate left ventricular MW parameters. It has been shown that noninvasive pressure-strain loops correlate well with invasive methods (5,22). Furthermore, it has been demonstrated that MW assessed by noninvasive stress–strain loops was in good agreement with ¹⁸F-deoxyglucose positron emission tomography (¹⁸F-FDG PET) for the detection of myocardial metabolism and oxygen consumption (23). In a hypoxic environment, myocardial dependence on fatty acid energy supply is blocked, favoring anaerobic glycolysis. Ischemic myocardium uptake of ¹⁸F-FDG is increased, but perfusion is decreased, and ¹⁸F-FDG PET shows a perfusion–metabolism mismatch in the ischemic region (24). MW parameters in the group with significant LAD stenosis at rest in our study were significantly reduced before significant changes in LVEF and LS, which reflect early ischemic hypoxia and metabolic changes in cardiomyocytes.

Findings suggest that longitudinal motion abnormalities may allow for earlier detection of ischemia-induced myocardial dysfunction (25,26). The LVEF of all participants in our study was within the normal range. In contrast, the LS_LAD at stress was statistically different between the two groups, and the MW parameters in the significant stenosis group were significantly reduced both at rest and during stress. The reasons why LS and MW can identify early ischemic injury in the myocardium more sensitively than can LVEF may be as follows: The myocardial wall of the left ventricle consists of inner longitudinal, middle circular, and outer oblique muscle fibers, respectively (27). The vessels supplying the subendocardial myocardium have small diameters and fewer branches, and the inner layer of the myocardium is directly exposed to blood, making it more sensitive to early ischemia. In addition, LS and LS-derived MW parameters are based on longitudinal myocardial motion, whereas LVEF mainly reflects radial wall motion of the left ventricle (8).

Patients with CAD often have hypertension and increased left ventricular afterload. One of the main limitations of LS is load dependence. Increased afterload decreases the degree of myocardial deformity, leading to false positives in the assessment of myocardial ischemia by LS (28). The integration of blood pressure information using the MW parameter overcomes the load dependence of strain and can improve the accuracy of assessing myocardial function. In addition, LS reflects only the peak systolic strain, and the MW parameter acquired by the pressure-strain loop includes the entire cardiac cycle. Therefore, MW parameters can more comprehensively and accurately reflect myocardial function. Ran et al. (29) categorized patients with CAD into three classes according to their coronary angiographic Gensini scores. In this cohort, global MW was superior to LVEF and global LS in predicting early myocardial dysfunction.

In our study, both MW parameters, WE_LAD and WI_LAD, were significantly reduced in the LAD significant stenosis group. ROC curve results showed that WE_LAD exhibited better predictive value for significant stenosis. The global WE is the ratio of the global construction work to the sum of the global construction work and the global ineffective work, which reflects the efficiency with which energy is converted into effective output work (15). Myocardial ischemia results in reduced adenosine triphosphate (ATP) production and diminished myocardial contractility, as well as delayed calcium reuptake and impaired active relaxation of the myocardium, leading to a reduction in global effective work. In addition, myocardial segmental asynchronous contraction and post-systolic contraction increased global ineffective work (1). This may explain the detection of myocardial ischemia due to significant coronary stenosis in the early stages of CAD by WE_LAD.

Value of drug burden on LS and MW parameters compared to resting state

Regadenoson is a highly selective A_2A adenosine receptor agonist that rapidly dilates coronary arteries and reduces side effects in other organs (30). Patients with CAD have compensatory dilation of stenotic vessels at rest to fulfill the needs of myocardial oxygen consumption. When drug stress is applied in this case, the dilatation of vessels with stenosis is not significant, whereas the branches of vessels without stenosis are significantly dilated, which leads to the phenomenon of blood stealing and aggravates ischemia in the lesion area. Therefore, our results showed that the absolute value of LS_LAD and MW parameters in the ischemic region was reduced in the significant stenosis group after drug stress compared with the resting state. The differences in LS_LAD and MW between the groups after stress were more significant than they were in the resting state.

Lin et al. (8) showed that there was no difference in global LS and MW parameters in the significant coronary stenosis group compared to the nonsignificant stenosis group at rest. There was a difference between groups in the global LS and MW parameters at the peak of treadmill exercise. The reason for the inconsistency with the results of the present study could be that exercise stress is more consistent with physiological alterations, which more significantly increase cardiac workload by accelerating heart rate and increasing cardiac output to increase myocardial oxygen consumption, inducing myocardial ischemia. Therefore, the difference in global LS and MW of the left ventricle between exercise stress and the resting state was more significant. In addition, Lin et al. (8) showed an increase in the global WI after stress compared with rest in all groups. In contrast, the regional WI_LAD after drug stress was reduced in both groups of our study compared to that at rest, which may be explained by the different mechanisms of myocardial ischemia induced by drug stress and exercise stress, with the degree of coronary steal being more pronounced in the lesions induced by Regadenoson. Apart from that, our study focused on regional cardiac function in LAD-supplied myocardium, whereas Lin et al. (8) investigated the global cardiac function changes. Borrie et al. (31) showed an increase in global WI and no significant change in global WE in patients with normal exercise stress. In contrast, stress-positive patients had decreased WI in ischemic-affected segments, no increase in global WI, and a significant decrease in global WE. Both Borrie et al. (31) and our study showed the feasibility of identifying myocardial ischemia by stress MW, especially ischemic region MW.

Functional evaluation of CFR on coronary circulation

The updated the European Society of Echocardiography guidelines propose that chronic coronary syndromes are syndromes caused by structural or functional changes associated with chronic disease of the coronary arteries and microcirculation. The definition transitions from a simple model of fixed stenosis in the large and middle arteries to a more complex and dynamic model (32). Thus, the decisive factor in determining the outcome of CAD should be the existence or absence of myocardial ischemia due to coronary stenosis. Flow reserve fraction (FFR) is the gold standard for coronary artery functional assessment. However, its invasive nature and high price limit clinical accessibility. CFR noninvasively reflects the potential reserve capacity of coronary circulation and myocardial perfusion by changes in coronary blood flow velocity before and after stress (33,34). It has been demonstrated that CFR is significantly correlated with FFR. Meimoun et al. (35) compared the diagnostic value of CFR, FFR, and instantaneous wave-free ratio for moderate stenosis of the LAD. The results showed the best correlation between CFR and FFR, and CFR demonstrated good diagnostic performance. In the univariate analysis of this study, CFR demonstrated the best predictive value for significant stenosis of the LAD, suggesting its potential advantage as a noninvasive coronary blood flow assessment index, which may assist in identifying myocardial ischemia associated with coronary anatomical stenosis. This finding provides preliminary evidence for future prospective studies exploring the use of CFR to guide clinical decision-making.

Machine learning model performance

Machine learning-based algorithms demonstrated potential for identifying significant stenosis of the LAD in this study. The results of the ULRM showed that CFR was the most effective noninvasive parameter for predicting significant stenosis of the LAD in this cohort. van de Hoef et al. (36) investigated the 5-year target vessel failure rate in patients with chronic coronary syndromes based on different combinations of CFR and FFR groupings; their results suggest that the combination of CFR and FFR can optimize functional assessment of coronary stenosis and clinical decision-making. Among our research, Peak WE_LAD was the most sensitive predictor of the degree of coronary stenosis differentiated by the MW parameter. The results of Borrie et al. (31) demonstrated the feasibility of global WE reduction in identifying patients with stress-induced ischemia in heterogeneous populations. In contrast, changes in pre- and post-stress WI more accurately identified myocardial ischemia. Borrie et al.’s (31) research differs from ours, however, in using visual assessment of SE positivity as a grouping criterion and in applying exercise stress to explore the recognition of myocardial ischemia by changes in MW parameters.

Both MLRM1 and MLRM2, which we created based on variables screened by LASSO regression, demonstrated superior composite diagnostic performance over individual factors. MLRM1, which includes hypertension, Peak DBP, LAd, Peak WE_LAD, and CFR, has the highest diagnostic value, and this model can provide high clinical net benefit within a reasonable threshold range. Meanwhile, MLRM2, which has a slightly lower AUC, is more concise and generalizable. However, both new models show great diagnostic potential for identifying patients with significant stenosis of the LAD. However, before clinical application, they should also be applied to larger populations to validate their usefulness and to select the appropriate model concerning specific clinical situations. In conclusion, machine learning models are expected to provide complementary value and new perspectives for the noninvasive diagnosis of significant coronary stenosis.

Limitations

There are some potential limitations to the current study: (I) this is a single-center retrospective study with a small sample size, and expansion of multicenter and prospective studies is needed to validate our preliminary findings; (II) there are multiple variants of LAD myocardial supply segments. The calculation of myocardial segments involved in the regional MW and LS parameters in this study was statistically analyzed using the eight myocardial segments supplied by the LAD, as referred to in the references (13), which may differ from the actual blood-supplying segments to some extent; (III) CFR is defined by blood flow at the epicardial and microvascular levels. Microcirculatory disorders were not excluded from the patients enrolled in this study, which may affect the accuracy of CFR in assessing the degree of coronary stenosis. Subsequent studies should refine the testing of microcirculation in this patient population for a comprehensive assessment; (IV) lack of CFR compared with FFR, an objective measure of functional ischemia; (V) this study only explored the value of MW and CFR on the degree of coronary LAD stenosis. It is necessary to expand the study of the LCx and the RCA. We hope that the above shortcomings and limitations can be remedied and improved in future work.

Conclusions

MW parameters following drug stress and CFR effectively identify patients with significant stenosis of the LAD and impaired left ventricular function in this cohort. Among these, peak WE_LAD and CFR demonstrate predictive potential for distinguishing the degree of LAD stenosis. MLRM enhanced diagnostic performance, suggesting that it may serve as a candidate non-invasive screening strategy for future exploration to reduce unnecessary invasive examinations. This model offers potential incremental value for the early identification of myocardial ischemia caused by significant coronary artery stenosis. However, its clinical applicability requires validation through prospective studies.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1404/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1404/dss

Funding: This study was supported by the Project of Heilongjiang Provincial Health Commission (No. 20230909020093), Special funded project of the Fourth Affiliated Hospital of Harbin Medical University (No. HYDSYTB202226), Natural Science Foundation of Heilongjiang Province (No. PL2024H156), and Natural Science Foundation of Heilongjiang Province (No. LH2023H049).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1404/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Ethics Committee of the Fourth Affiliated Hospital of Harbin Medical University (approval No. 2024-ethical review-83) and informed consent was obtained from all individual participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Zhu H, Guo Y, Wang X, Yang C, Li Y, Meng X, Pei Z, Zhang R, Zhong Y, Wang F. Myocardial Work by Speckle Tracking Echocardiography Accurately Assesses Left Ventricular Function of Coronary Artery Disease Patients. Front Cardiovasc Med 2021;8:727389. [Crossref] [PubMed]
2. Gulati M, Levy PD, Mukherjee D, Amsterdam E, Bhatt DL, Birtcher KK, et al. 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR Guideline for the Evaluation and Diagnosis of Chest Pain: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation 2021;144:e368-454. [Crossref] [PubMed]
3. Qian L, Xie F, Xu D, Porter TR. Long-term prognostic value of stress myocardial perfusion echocardiography in patients with coronary artery disease: a meta-analysis. Eur Heart J Cardiovasc Imaging 2021;22:553-62. [Crossref] [PubMed]
4. Nishi T, Funabashi N, Ozawa K, Nishi T, Kamata T, Fujimoto Y, Kobayashi Y. Regional layer-specific longitudinal peak systolic strain using exercise stress two-dimensional speckle-tracking echocardiography for the detection of functionally significant coronary artery disease. Heart Vessels 2019;34:1394-403. [Crossref] [PubMed]
5. Hubert A, Le Rolle V, Leclercq C, Galli E, Samset E, Casset C, Mabo P, Hernandez A, Donal E. Estimation of myocardial work from pressure-strain loops analysis: an experimental evaluation. Eur Heart J Cardiovasc Imaging 2018;19:1372-9. [Crossref] [PubMed]
6. Mor-Avi V, Patel MB, Maffessanti F, Singh A, Medvedofsky D, Zaidi SJ, Mediratta A, Narang A, Nazir N, Kachenoura N, Lang RM, Patel AR. Fusion of Three-Dimensional Echocardiographic Regional Myocardial Strain with Cardiac Computed Tomography for Noninvasive Evaluation of the Hemodynamic Impact of Coronary Stenosis in Patients with Chest Pain. J Am Soc Echocardiogr 2018;31:664-73. [Crossref] [PubMed]
7. Ivanov SI, Alekhin MN. Myocardial work in assessment of left ventricular systolic function. Kardiologiia 2020;60:80-8. [Crossref] [PubMed]
8. Lin J, Wu W, Gao L, He J, Zhu Z, Pang K, Wang J, Liu M, Wang H. Global Myocardial Work Combined with Treadmill Exercise Stress to Detect Significant Coronary Artery Disease. J Am Soc Echocardiogr 2022;35:247-57. [Crossref] [PubMed]
9. Taqueti VR. Coronary flow reserve: a versatile tool for interrogating pathophysiology, and a reliable marker of cardiovascular outcomes and mortality. Eur Heart J 2022;43:1594-6. [Crossref] [PubMed]
10. Edwards NFA, Scalia GM, Shiino K, Sabapathy S, Anderson B, Chamberlain R, Khandheria BK, Chan J. Global Myocardial Work Is Superior to Global Longitudinal Strain to Predict Significant Coronary Artery Disease in Patients With Normal Left Ventricular Function and Wall Motion. J Am Soc Echocardiogr 2019;32:947-57. [Crossref] [PubMed]
11. Wang RR, Tian T, Li SQ, Leng XP, Tian JW. Assessment of Left Ventricular Global Myocardial Work in Patients With Different Degrees of Coronary Artery Stenosis by Pressure-Strain Loops Analysis. Ultrasound Med Biol 2021;47:33-42. [Crossref] [PubMed]
12. Han X, Cao Y, Ju Z, Liu J, Li N, Li Y, Liu T, Shi H, Gu J. Assessment of regional left ventricular myocardial strain in patients with left anterior descending coronary stenosis using computed tomography feature tracking. BMC Cardiovasc Disord 2020;20:362. [Crossref] [PubMed]
13. Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, Flachskampf FA, Foster E, Goldstein SA, Kuznetsova T, Lancellotti P, Muraru D, Picard MH, Rietzschel ER, Rudski L, Spencer KT, Tsang W, Voigt JU. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. Eur Heart J Cardiovasc Imaging 2015;16:233-70. [Crossref] [PubMed]
14. Matsuda K, Hoshino M, Usui E, Hanyu Y, Sugiyama T, Kanaji Y, Hada M, Nagamine T, Nogami K, Ueno H, Sayama K, Sakamoto T, Yonetsu T, Sasano T, Kakuta T. Noninvasive transthoracic doppler flow velocity and invasive thermodilution to assess coronary flow reserve. Quant Imaging Med Surg 2024;14:421-31. [Crossref] [PubMed]
15. Ilardi F, D'Andrea A, D'Ascenzi F, Bandera F, Benfari G, Esposito R, Malagoli A, Mandoli GE, Santoro C, Russo V, Crisci M, Esposito G, Cameli MOn Behalf Of The Working Group Of Echocardiography Of The Italian Society Of Cardiology Sic. Myocardial Work by Echocardiography: Principles and Applications in Clinical Practice. J Clin Med 2021;10:4521. [Crossref] [PubMed]
16. Lan J, Wang Y, Zhang R, Li J, Yu T, Yin L, Shao T, Lu H, Wang C, Xue L. The value of speckle-tracking stratified strain combined with myocardial work measurement in evaluating left ventricular function in patients with heart failure with preserved ejection fraction. Quant Imaging Med Surg 2024;14:2514-27. [Crossref] [PubMed]
17. Dong YM, Sun J, Li YX, Chen Q, Liu QQ, Sun Z, Pang R, Chen F, Xu BY, Manyande A, Clark TG, Li JP, Orhan IE, Tian YK, Wang T, Wu W, Ye DW. Development and Validation of a Nomogram for Assessing Survival in Patients With COVID-19 Pneumonia. Clin Infect Dis 2021;72:652-60. [Crossref] [PubMed]
18. He Y, Zhu Z, Chen Y, Chen F, Wang Y, Ouyang C, Yang H, Huang M, Zhuang X, Mao R, Ben-Horin S, Wu X, Ouyang Q, Qian J, Lu N, Hu P, Chen M. Development and Validation of a Novel Diagnostic Nomogram to Differentiate Between Intestinal Tuberculosis and Crohn's Disease: A 6-year Prospective Multicenter Study. Am J Gastroenterol 2019;114:490-9. [Crossref] [PubMed]
19. Yanjun Z, Jin S, Qiuyue L, Tao L. Risk factors and the establishment of nomogram model for incisional-wound infection in patients with bladder cancer after radical resection. Asian J Surg 2023;46:1418-9. [Crossref] [PubMed]
20. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;49:1373-9. [Crossref] [PubMed]
21. Frișan AC, Mornoș C, Lazăr MA, Șoșdean R, Crișan S, Ionac I, Luca CT. Echocardiographic Myocardial Work: A Novel Method to Assess Left Ventricular Function in Patients with Coronary Artery Disease and Diabetes Mellitus. Medicina (Kaunas) 2024;60:199. [Crossref] [PubMed]
22. Gao L, Wang Y, Gao M, Chen L. Clinical research progress of myocardial work in assessment and prediction of coronary artery disease in noninvasive pressure-strain loop technique. J Clin Ultrasound 2023;51:38-45. [Crossref] [PubMed]
23. Russell K, Eriksen M, Aaberge L, Wilhelmsen N, Skulstad H, Remme EW, Haugaa KH, Opdahl A, Fjeld JG, Gjesdal O, Edvardsen T, Smiseth OA. A novel clinical method for quantification of regional left ventricular pressure-strain loop area: a non-invasive index of myocardial work. Eur Heart J 2012;33:724-33. [Crossref] [PubMed]
24. Aoyama R, Takano H, Kobayashi Y, Kitamura M, Asai K, Amano Y, Kumita SI, Shimizu W. Evaluation of myocardial glucose metabolism in hypertrophic cardiomyopathy using 18F-fluorodeoxyglucose positron emission tomography. PLoS One 2017;12:e0188479. [Crossref] [PubMed]
25. Liu C, Li J, Ren M, Wang ZZ, Li ZY, Gao F, Tian JW. Multilayer longitudinal strain at rest may help to predict significant stenosis of the left anterior descending coronary artery in patients with suspected non-ST-elevation acute coronary syndrome. Int J Cardiovasc Imaging 2016;32:1675-85. [Crossref] [PubMed]
26. Zhou F, Yuan H, Sun J, Ran H, Pan H, Wu P, Yang Q. Two-dimensional speckle tracking imaging cardiac motion-based quantitative evaluation of global longitudinal strain among patients with coronary Heart Disease and functions of left ventricular ischemic myocardial segment. Int J Cardiovasc Imaging 2024;40:351-9. [Crossref] [PubMed]
27. Vasan RS, Urbina EM, Jin L, Xanthakis V. Prognostic Significance of Echocardiographic Measures of Cardiac Remodeling in the Community. Curr Cardiol Rep 2021;23:86. [Crossref] [PubMed]
28. Li X, Zhang P, Li M, Zhang M. Myocardial work: The analytical methodology and clinical utilities. Hellenic J Cardiol 2022;68:46-59. [Crossref] [PubMed]
29. Ran H, Yao Y, Wan L, Ren J, Sheng Z, Zhang P, Schneider M. Characterizing stenosis severity of coronary heart disease by myocardial work measurement in patients with preserved ejection fraction. Quant Imaging Med Surg 2023;13:5022-33. [Crossref] [PubMed]
30. Elkholy KO, Hegazy O, Okunade A, Aktas S, Ajibawo T. Regadenoson Stress Testing: A Comprehensive Review With a Focused Update. Cureus 2021;13:e12940. [Crossref] [PubMed]
31. Borrie A, Goggin C, Ershad S, Robinson W, Sasse A. Noninvasive Myocardial Work Index: Characterizing the Normal and Ischemic Response to Exercise. J Am Soc Echocardiogr 2020;33:1191-200. [Crossref] [PubMed]
32. Vrints C, Andreotti F, Koskinas KC, Rossello X, Adamo M, Ainslie J, et al. 2024 ESC Guidelines for the management of chronic coronary syndromes. Eur Heart J 2024;45:3415-537. [Crossref] [PubMed]
33. Ziadi MC, Dekemp RA, Williams KA, Guo A, Chow BJ, Renaud JM, Ruddy TD, Sarveswaran N, Tee RE, Beanlands RS. Impaired myocardial flow reserve on rubidium-82 positron emission tomography imaging predicts adverse outcomes in patients assessed for myocardial ischemia. J Am Coll Cardiol 2011;58:740-8. [Crossref] [PubMed]
34. Kim J, Shin D, Lee JM, Lee SH, Hong D, Choi KH, et al. Differential Prognostic Value of Revascularization for Coronary Stenosis With Intermediate FFR by Coronary Flow Reserve. JACC Cardiovasc Interv 2022;15:1033-43. [Crossref] [PubMed]
35. Meimoun P, Clerc J, Ardourel D, Djou U, Martis S, Botoro T, Elmkies F, Zemir H, Luycx-Bore A, Boulanger J. Assessment of left anterior descending artery stenosis of intermediate severity by fractional flow reserve, instantaneous wave-free ratio, and non-invasive coronary flow reserve. Int J Cardiovasc Imaging 2017;33:999-1007. [Crossref] [PubMed]
36. van de Hoef TP, Lee JM, Boerhout CKM, de Waard GA, Jung JH, Lee SH, et al. Combined Assessment of FFR and CFR for Decision Making in Coronary Revascularization: From the Multicenter International ILIAS Registry. JACC Cardiovasc Interv 2022;15:1047-56. [Crossref] [PubMed]