Background
The aim of this study was to investigate the incremental value of global longitudinal strain (GLS) by automated function imaging in respect to wall motion (WM) for the detection of coronary artery disease (CAD) during dipyridamole stress echocardiography.
Methods
Fifty-two patients (mean age, 65.3 ± 8.7 years; 22 men) underwent dipyridamole stress echocardiography followed by coronary angiography within 1 week. Diagnostic accuracy for the identification of single-vessel CAD was evaluated for WM and GLS. The study population was divided into two groups according to coronary angiographic findings: those with CAD ( n = 38; mean age, 67.2 ± 5.9 years; 19 men) and those without CAD ( n = 14; mean age, 63.3 ± 6.4 years; three men).
Results
A trend toward lower resting GLS values was found in patients with CAD than in those without (−18.7 ± 2.2% vs −20 ± 2.8%, P = .061). In patients without CAD, GLS progressively increased up to peak dose (from −20 ± 2.8% at rest to −20.7 ± 1.9% at low dose, P = .045; from −20.7 ± 1.9% at low dose to −21.5 ± 3.1% at peak dose, P = .032), whereas in patients with CAD, an increase of GLS from rest to low dose (from −18.7 ± 2.2% to −19.2 ± 3.9%, P = .046) followed by a decrease from low to peak dose (from −19.2 ± 3.9% to −17.5 ± 2.4%, P = .007) was observed. In addition, with regard to diagnostic accuracy in detecting CAD, WM yielded sensitivity of 44%, specificity of 55%, positive predictive value of 73%, and negative predictive value of 26%, whereas GLS, alternatively evaluated as the difference between peak dose and resting values or between peak and low-dose values, provided sensitivity of 61%, specificity of 90%, positive predictive value of 94%, and negative predictive value of 47% (respectively, P = .020, P = .001, P = .023, and P = .031, all vs WM) and sensitivity of 84%, specificity of 92%, positive predictive value of 96%, and negative predictive value of 68% (respectively, P < .001, P < .001, P = .001, P < .001, all vs WM).
Conclusions
GLS analysis, particularly performed by comparing peak-dose with low-dose values, improves the accuracy of dipyridamole stress echocardiography in the detection of single-vessel CAD compared with the sole assessment of WM changes.
Stress echocardiography, performed through the combination of echocardiography and physical, electrical, or pharmacologic stress, is a widely used diagnostic technique for detecting coronary artery disease (CAD). Among the currently available ischemic stressors, dipyridamole is safe but limited by lower sensitivity in the diagnosis of single-vessel CAD, particularly in cases of single-vessel disease. Various adjustments to the dipyridamole protocol, such as the addition of atropine and the measurement of coronary flow reserve in the descending anterior artery, have been introduced to increase its diagnostic accuracy.
Independent of the ischemic stressor used, one of the main limitations of stress echocardiography is the subjective nature of visual wall motion (WM) interpretation, which requires specific technical skills and experience. In addition, assessment of WM, which is determined primarily by radial thickening of myocardial walls, does not take into account longitudinal function, which is derived from deformation of subendocardial fibers, the most vulnerable to myocardial ischemia. New technologies and techniques, such as tissue Doppler imaging and two-dimensional (2D) strain analysis, allow accurate measurement of myocardial function and could possibly provide a quantitative estimation of ischemic response during stress echocardiography, which would overcome the limitation of subjective WM interpretation. The aim of our study was to investigate the role of longitudinal strain (LS) analysis in addition to the assessment of WM in patients undergoing dipyridamole stress echocardiography (DipSE) with single-vessel CAD.
Methods
Study Population
We prospectively enrolled 52 patients (mean age, 65.3 ± 8.7 years; 22 men) who were scheduled for coronary angiography to undergo DipSE; coronary catheterization was performed within 1 week. According to the findings on coronary angiography (i.e., the presence of a coronary stenosis of >50% affecting one of the epicardial vessels), the overall study population was divided into two groups: those with CAD (38 patients; mean age, 67.2 ± 5.9 years; 19 men; group A) and those without CAD (14 patients; mean age, 63.3 ± 6.4 years; three men; group B). Exclusion criteria were atrial fibrillation, concomitant valvular heart disease beyond mild, CAD involving more than one vessel on coronary angiography, and inadequate acoustic window for speckle-tracking analysis. All patients gave written informed consent to our study protocol, which was supported by our hospital’s ethics committee.
Stress Echocardiography
An ultrasound system (Vivid 7; GE Vingmed Ultrasound AS, Horten, Norway) equipped with a cardiac M4S transducer was used. Dipyridamole stress echocardiography was performed in each subject using a standard protocol (0.84 mg/kg over 10 min divided into two doses, 0.56 mg/kg over 4 min followed by 0.28 mg/kg over 2 min after a 4-min delay) with atropine injection (doses of 0.25 mg, repeated at 1-min intervals, up to a maximum of 1 mg). Four-, two-, and three-chamber views were acquired at rest, low dose, peak dose, and recovery. Regional WM was visually assessed according to the 16-segment model of the American Society of Echocardiography by an experienced reader blinded to all patient data, and WM score index (WMSI) was calculated. A 12-lead electrocardiogram was recorded during each stress test. We considered stress test results as positive for WM abnormalities when the kinetic of two or more adjacent myocardial segments worsened. Main criteria for test interruption were occurrence of anginal chest pain, worsening of WM in at least two adjacent myocardial segments, and electrocardiographic ischemic changes. With regard to cardiac medications, therapy with β-blockers was gradually suspended 1 week before stress echocardiography. Other drugs were continued.
Two-Dimensional Strain Analysis
Analysis of 2D strain was performed offline using dedicated software (EchoPAC Workstation version 8.0; GE Vingmed Ultrasound AS). At each step of the stress test, LS was measured using automated function imaging, a new method for LS assessment that allows measurement of regional LS and global LS (GLS) through the positioning of three endocardial markers (two markers at the mitral annulus and one at the apex) in each apical view. Subsequently, the obtained segmental values of LS were visualized as a bull’s-eye map in a quick and feasible manner.
We had optimal left ventricular endocardial tracking in the overall population, with manual adjustments for endocardial tracking in 6% of patients ( n = 3) at rest, 8% ( n = 4) at low dose, and 4% ( n = 2) at peak dose, thus confirming the feasibility of 2D strain during DipSE. In each patient, we used a frame rate ≥ 70 frames/sec for adequate 2D strain analysis.
We analyzed GLS at each step of DipSE and compared peak-dose values with rest or low dose; we considered test results positive for abnormal myocardial deformation when GLS was found to be lower at peak dose compared with rest or low dose. We also analyzed GLS during recovery.
Statistical Analysis
Data are expressed as mean ± SD. Statistical analysis was performed using SPSS version 17 for Windows (SPSS, Inc, Chicago, IL). Independent t and χ 2 tests were used for comparison between groups of continuous and categorical variables, respectively. One-way analysis of variance for repeated measures was performed when comparing variables through different steps; analysis of variance for independent samples was performed for the comparison of variables between more than two groups. To estimate the ability to detect CAD, we calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV); McNemar’s χ 2 test was performed to compare these parameters between groups. Intraclass correlation coefficients were used to assess intra- and interobserver variability of measurements of myocardial strains. P values ≤ .05 were considered significant.
Results
Overall Study Population
Main clinical and echocardiographic features of the overall study population are shown in Table 1 . Among patients with single-vessel CAD documented on coronary angiography, there was a higher prevalence of disease involving the left anterior descending coronary artery (18 patients) than the right coronary and left circumflex coronary arteries (11 and nine patients, respectively).
| Variable | Value |
|---|---|
| Clinical parameters | |
| Age (y) | 65.3 ± 8.7 |
| Men | 22 (42%) |
| Arterial hypertension | 23 (44%) |
| Diabetes | 16 (30%) |
| Dyslipidemia | 24 (46%) |
| Smokers | n = 9 (17%) |
| Heart rate at rest (beats/min) | 69 ± 2 |
| Heart rate at low dose (beats/min) | 73 ± 5 |
| Heart rate at peak dose (beats/min) | 80 ± 6 |
| Mean systolic blood pressure at rest (mm Hg) | 140 ± 9 |
| Mean systolic blood pressure at low dose (mm Hg) | 137 ± 8 |
| Mean systolic blood pressure at peak dose (mm Hg) | 133 ± 9 |
| Mean systolic blood pressure at recovery (mm Hg) | 139 ± 8 |
| Mean diastolic blood pressure at rest (mm Hg) | 86 ± 5 |
| Mean diastolic blood pressure at low dose (mm Hg) | 83 ± 2 |
| Mean diastolic blood pressure at peak dose (mm Hg) | 82 ± 1 |
| Mean diastolic blood pressure at recovery (mm Hg) | 85 ± 6 |
| Echocardiographic parameters | |
| EDV (mL) | 103.4 ± 28.6 |
| ESV (mL) | 46.7 ± 19.9 |
| Ejection fraction (%) | 55.2 ± 7.7 |
| WMSI | 1.15 ± 0.22 |
| GLS (%) | −18.8 ± 2.6 |
| Coronary angiography | |
| Nonsignificant stenosis | 14 (27%) |
| LAD stenosis | 18 (35%) |
| RCA stenosis | 11 (21%) |
| LCX stenosis | 9 (17%) |
No significant differences with regard to clinical and echocardiographic parameters were found between patients with or without CAD, with the only exception being gender prevalence ( P = .023) ( Table 2 ).
| Variable | With CAD ( n = 38) | Without CAD ( n = 14) | P |
|---|---|---|---|
| Clinical parameters | |||
| Age (y) | 67.2 ± 5.9 | 63.3 ± 6.4 | NS |
| Men | 19 (50%) | 3 (21%) | .023 |
| Arterial hypertension | 17 (44%) | 6 (42%) | NS |
| Diabetes | 11 (29%) | 5 (35%) | NS |
| Dyslipidemia | 17 (44%) | 7 (50%) | NS |
| Smokers | 6 (15%) | 3 (21%) | NS |
| Use of antianginal drugs | 8 (21%) | 4 (28%) | NS |
| Heart rate at rest (beats/min) | 70 ± 5 | 68 ± 4 | NS |
| Heart rate at low dose (beats/min) | 74 ± 4 | 73 ± 2 | NS |
| Heart rate at peak dose (beats/min) | 80 ± 6 | 79 ± 4 | NS |
| Heart rate at recovery (beats/min) | 75 ± 7 | 74 ± 5 | NS |
| SBP at rest (mm Hg) | 140 ± 17 | 138 ± 14 | NS |
| SBP at low dose (mm Hg) | 137 ± 12 | 136 ± 11 | NS |
| SBP at peak dose (mm Hg) | 134 ± 14 | 133 ± 10 | NS |
| SBP at recovery (mm Hg) | 140 ± 15 | 138 ± 11 | NS |
| DPB at rest (mm Hg) | 87 ± 7 | 86 ± 4 | NS |
| DPB at low dose (mm Hg) | 84 ± 5 | 83 ± 5 | NS |
| DPB at peak dose (mm Hg) | 83 ± 3 | 81 ± 4 | NS |
| DBP at recovery (mm Hg) | 86 ± 5 | 85 ± 2 | NS |
| Echocardiographic parameters | |||
| EDV (mL) | 103.8 ± 27.1 | 108.9 ± 39.5 | NS |
| ESV (mL) | 45.2 ± 16.8 | 49.3 ± 26.2 | NS |
| Ejection fraction (%) | 55.6 ± 5.5 | 56.3 ± 8.8 | NS |
| WMSI at rest | 1.16 ± 0.20 | 1.08 ± 0.15 | NS |
| WMSI at low dose | 1.14 ± 0.16 | 1.07 ± 0.13 | .041 |
| WMSI at peak dose | 1.22 ± 0.26 | 1.08 ± 0.16 | .011 |
| WMSI at recovery | 1.17 ± 0.13 | 1.07 ± 0.14 | .017 |
| GLS at rest (%) | −18.7 ± 2.2 | −20 ± 2.8 | NS |
| GLS at low dose (%) | −19.2 ± 3.9 | −20.7 ± 1.9 | NS |
| GLS at peak dose (%) | −17.5 ± 2.4 | −21.5 ± 3.1 | <.001 |
| Mean GLS at recovery (%) | −19.1 ± 2.7 | −20.8 ± 2.9 | NS |
With regard to myocardial deformation analysis, a trend toward lower GLS at rest found in patients with CAD than in those without CAD (−18.7 ± 2.2% vs −20 ± 2.8%, P = .061) was not significant, whereas WMSI was comparable between groups (1.16 ± 0.20 vs 1.08 ± 0.15, P = .23).
Wall Motion Versus 2D Strain Analysis in the Detection of CAD
No significant differences were found regarding WMSI from baseline to low and peak dose in patients without CAD (respectively, from 1.08 ± 0.15 to 1.07 ± 0.13, P = .158, and from 1.07 ± 0.13 to 1.08 ± 0.16, P = .133); similarly, WMSI remained substantially unchanged from rest to low dose in patients with CAD, showing only a nonsignificant trend toward a decrease (from 1.16 ± 0.20 to 1.14 ± 0.16, P = .061), whereas it significantly increased from low to peak dose (from 1.14 ± 0.16 to 1.22 ± 0.26, P = .003). When comparing WMSI between patients with and without CAD, values at low and peak dose as well as at recovery were significantly different between groups (respectively, P = .041, P = .011, and P = .017).
On 2D strain analysis, opposite changes in GLS were observed from baseline through low and peak dose when comparing patients with and without CAD. Indeed, on one hand, GLS progressively increased up to peak dose in patients without CAD (from −20 ± 2.8% to −20.7 ± 1.9%, P = .045; from −20.7 ± 1.9% to −21.5 ± 3.1%, P = .032), whereas on the other hand, an increase of GLS from rest to low dose (from −18.7 ± 2.2% to −19.2 ± 3.9%, P = .046) followed by a decrease from low to peak dose (from −19.2 ± 3.9% to −17.5 ± 2.4%, P = .007) was observed in patients with CAD ( Figures 1 and 2 ; Videos 1–3 ; available at www.onlinejase.com ). When comparing GLS at each stage between groups, only peak dose values were significantly different between patients with and without CAD ( P < .001). At the recovery stage, as reported in Table 2 , both WMSI and GLS showed values comparable with those at rest in the two groups of patients, with and without CAD.
On the basis of these results, we compared WM and myocardial deformation analysis as to the ability to detect CAD. In particular, regarding 2D strain analysis, we tested two methods to identify CAD, evaluating (1) the differences between strain values at peak dose and rest and (2) the differences between strain values at peak and low dose.
By considering WM, we obtained the following results: sensitivity of 44%, specificity of 55%, PPV of 73%, and NPV of 26%. On the other hand, using myocardial deformation analysis, we observed sensitivity of 61%, specificity of 90%, PPV of 94%, and NPV of 47% (respectively, P = .020, P = .001, P = .023, and P = .031, all vs WM), when considering the differences between strain values at peak dose and rest and sensitivity of 84%, specificity of 92%, PPV of 96%, and NPV of 68% (respectively, P < .001, P < .001, P = .001, and P < .001, all vs WM), when comparing values at peak and low dose ( Figure 3 ).
In 10 patients (19% of the overall population), both a decrease in GLS from low to peak dose and a decrease in GLS from rest to peak dose were found. Interestingly, all these patients were affected by CAD, suggesting that this finding (i.e., the decrease of GLS values at peak dose from both rest and low dose) may be indicative of a high likelihood of CAD.
Subgroup Analysis: Study of Myocardial Viability in Patients with Resting WM Abnormalities
Among the overall population, 11 patients had resting WM abnormalities involving, on the whole, a maximum of 56 myocardial segments; of these, four segments were excluded from the analysis because of echocardiographic appearance of myocardial scar (i.e., hyperechoic and thinned walls).
By using visual assessment of WM, nine segments (17.3% of segments with resting WM abnormalities) were found to be likely viable: of these, seven segments showed a “biphasic response” (improvement of contractility at low dose followed by worsening at peak dose), indicative of ischemic viability, and two segments had progressive improvement of contractility through low and peak dose, suggesting nonischemic viability. On the other hand, by performing 2D strain analysis, 27 segments (51.9% of segments with resting WM abnormalities) could be considered likely viable: of these, 24 showed a “biphasic response” (increase of strain values at low dose followed by decrease of strain values at peak dose), and three segments had a progressive improvement of longitudinal deformation through low and peak dose. These differences between WM and 2D strain analysis in the ability to detect viable segments were statistically significant ( P = .023). Mean values of LS of the myocardial segments with resting WM abnormalities, at rest, low dose, and peak dose were 14.03 ± 5.81%, 16.05 ± 7.31%, and 11.94 ± 4.80%, respectively (rest vs low dose, P = .034; low dose vs peak dose, P = .027).
LS Analysis in Different Coronary Territories
According to current American Society of Echocardiography recommendations, we reviewed LS in three territories corresponding to the three epicardial coronary vessels: inferior to the right coronary artery (including basal and middle segments of the inferior wall), inferolateral to the left circumflex artery (including basal and middle segments of the inferolateral wall), and anterior to the left anterior descending coronary artery (including all apical segments, with the exception of the lateral segment, basal and middle segments of the anterior wall, and anterior septum). As shown in Table 3 , GLS was significantly lower at peak dose compared with baseline in each coronary territory with a vessel affected by CAD, whereas an increase in GLS values from baseline to peak dose was observed in the unaffected territories. Overall, by using myocardial deformation analysis, the correct identification of the affected coronary vessel territory was obtained in about 75%.
