Author + information
- Received October 18, 2012
- Revision received December 18, 2012
- Accepted January 23, 2013
- Published online May 1, 2013.
- Sebastian I. Sarvari, MD⁎,†,‡,
- Kristina H. Haugaa, MD, PhD⁎,†,‡,
- Wasim Zahid, MD⁎,†,‡,
- Bjørn Bendz, MD, PhD⁎,
- Svend Aakhus, MD, PhD⁎,
- Lars Aaberge, MD, PhD⁎ and
- Thor Edvardsen, MD, PhD⁎,†,‡,⁎ ()
- ↵⁎Reprint requests and correspondence:
Prof. Thor Edvardsen, Department of Cardiology, Oslo University Hospital, Rikshospitalet, N-0027 Oslo, Norway
Objectives Our objective was to assess whether patients with significant coronary artery disease (CAD) had reduced endocardial function assessed by layer-specific strain compared with patients without significant CAD.
Background The left ventricular (LV) wall of the heart comprises 3 myocardial layers. The endocardial layer is most susceptible to ischemic injury.
Methods Seventy-seven patients referred to coronary angiography due to suspected non–ST-segment elevation-acute coronary syndromes (NSTE-ACS) were prospectively included. Coronary occlusion was found in 28, significant stenosis in 21, and no stenosis in 28 patients. Echocardiography was performed 1 to 2 h before angiography. Layer-specific longitudinal and circumferential strains were assessed from endocardium, mid-myocardium, and epicardium by 2-dimensional (2D) speckle-tracking echocardiography (STE). Territorial longitudinal strain (TLS) was calculated based on the perfusion territories of the 3 major coronary arteries in a 16-segment LV model, whereas global circumferential strain (GCS) was averaged from 6 circumferential LV segments in all 3 layers.
Results Patients with significant CAD had worse function in all 3 myocardial layers assessed by TLS and GCS compared with patients without significant CAD. Endocardial TLS (mean –14.0 ± 3.3% vs. –19.2 ± 2.2%; p < 0.001) and GCS (mean –19.3 ± 4.0% vs. –24.3 ± 3.4%; p < 0.001) were most affected. The absolute differences between endocardial and epicardial TLS and GCS were lower in patients with significant CAD (Δ2.4 ± 3.6% and Δ6.7 ± 3.8%, respectively) than in those without significant CAD (Δ5.3 ± 2.1% and Δ10.4 ± 3.0%; p < 0.001). This reflects a pronounced decrease in endocardial function in patients with significant CAD. A receiver-operating characteristic curve analysis showed that endocardial and mid-myocardial TLS were superior to identify significant CAD compared with epicardial TLS (p < 0.05), wall motion score index (p < 0.01), and ejection fraction (EF) (p < 0.001).
Conclusions Assessment of layer-specific strain by 2D-STE might identify NSTE-ACS patients with significant CAD. Endocardial function was more affected in patients with significant CAD compared with epicardial function and EF.
The clinical presentation of coronary artery disease (CAD) varies from silent ischemia, stable angina pectoris to acute coronary syndromes (ACS) and death. ACSs comprise non-ST–segment elevation-acute coronary syndrome (NSTE-ACS) and ST-segment elevation myocardial infarction (STEMI). The presence of ST-segment elevation typically represents coronary occlusion requiring acute reperfusion therapy. On the other hand, patients with suspected NSTE-ACS are a more heterogeneous group. Coronary occlusion and/or significant stenosis may or may not be present, and coronary angiography and revascularization therapy might be unnecessary in as many as one-third of these patients (1). Therefore, a better selection of patients with real need for coronary angiography and revascularization therapy could reduce both complication rates and healthcare costs associated with this procedure.
The left ventricular (LV) wall of the heart comprises an endocardial, a mid-myocardial, and an epicardial layer. Of these 3 layers, the endocardium is most susceptible to ischemic injury (2–5). Careful evaluation of this layer might increase the diagnostic accuracy of CAD.
Two-dimensional (2D) speckle-tracking echocardiography (STE) is a semi-automated quantitative technique for assessment of strain, a measure of cardiac function based on grayscale images. Strain is a measure of deformation, an intrinsic mechanical property, and measures myocardial systolic function more directly compared with conventional cavity-based echocardiographic measures (6–8). Strain echocardiography has been introduced as an accurate tool for the assessment of global and regional (9) LV myocardial function and has been demonstrated to be more sensitive and accurate in the identification of coronary artery occlusion in patients with NSTE-ACS (10,11) compared with conventional echocardiographic measures of systolic LV function. Recent software allows separate evaluation of endocardial, mid-myocardial, and epicardial myocardial deformation.
The aim of this study was to evaluate myocardial deformation in 3 myocardial layers in patients with suspected NSTE-ACS.
This study was conducted in a single tertiary coronary care center. Seventy-seven patients with suspected NSTE-ACS (12,13) referred to our hospital for coronary angiography were prospectively included in the study. Exclusion criteria were: age <18 years and/or a history of previous myocardial infarction, percutaneous coronary intervention and open chest surgery, left bundle branch block, severe valvular dysfunction, atrial fibrillation with heart rate >100 beats/min, sustained severe arrhythmia, and/or any condition that might interfere with the patient's ability to comply. Electrocardiograms (ECGs) were evaluated by experienced cardiologists at admission. ECGs were described as ischemic if ST depression or T-wave changes were present. All patients received medical treatment according to current guidelines. Echocardiography was performed 1 to 2 h prior to coronary angiography and within 48 h after the last episode of chest pain. The echocardiographic data were analyzed blinded to all clinical information.
Written informed consent was given by all study participants. The study complied with the Declaration of Helsinki and was approved by the Regional Committees for Medical and Health Research Ethics.
Echocardiographic studies were performed with commercially available system (Artida, Toshiba Medical Systems Corporation, Tokyo, Japan). Routine grayscale 2D cine loops from 3 consecutive beats were obtained at end-expiratory apnea from standard parasternal short-axis view of the LV at the level of the papillary muscle and from the 3 apical views (4-chamber, 2-chamber, and long-axis). Frame rates were 47 ± 5 Hz for grayscale imaging. LV volume and ejection fraction (EF) were assessed by biplane Simpson method using manual tracing of digital images.
Wall motion was visually assessed in a 16-segment model, according to the American Society of Echocardiography (14), by an experienced observer. The observer evaluated image quality, and segments were discarded if the quality was found insufficient for analysis. Wall motion score index (WMSI) was calculated for each patient as the average of the analyzed segmental values.
Grayscale images were analyzed. Myocardial function by strain was evaluated on a frame-by-frame basis by automatic tracking of acoustic markers (speckles) throughout the cardiac cycle. The endocardial borders were traced in the end-systolic frame of the 2D images from the 3 apical views for analyses of longitudinal endocardial, mid-myocardial, and epicardial strains. Analyses of layer-specific circumferential strains were obtained from the parasternal short-axis view. Peak negative systolic longitudinal and circumferential strains from 3 layers were assessed using off-line software (Toshiba Medical Systems Corporation, Tokyo, Japan) in 16 longitudinal and 6 circumferential LV segments. All segmental values were averaged to global longitudinal strain (GLS) and global circumferential strain (GCS) for each myocardial layer (Fig. 1). Segments that failed to track properly were manually adjusted by the operator. Any segments that subsequently failed to track were excluded.
Territorial longitudinal strain (TLS) was calculated based on the perfusion territories of the 3 major coronary arteries in a 16-segment LV model (14) by averaging all segmental peak systolic strain values within each territory.
All patients underwent coronary angiography. The assessment of CAD was performed by visual estimate in each single stenosis with multiple projections, avoiding overlap of side branches and foreshortening of relevant coronary stenoses. Coronary occlusion was defined as TIMI (Thrombolysis in Myocardial Infarction) flow grade 0 or 1, while significant coronary artery stenosis was considered as a ≥50% reduction of vessel diameter in at least 1 major coronary artery (15). Fractional flow reserve measurement was performed only if the operator was in doubt about the hemodynamic significance of a stenosis. The lesion was considered functionally significant when fractional flow reserve was ≤0.8.
Data are presented as mean ± SD, numbers (percentages), and median (interquartile range), respectively. Chi-square test (categorical variables) or Student t test (continuous variables) was used to determine differences between 2 groups. Mann-Whitney U test was used for non-normally distributed continuous variables (biomarkers) (SPSS version 18, SPSS Inc., Chicago, Illinois). Comparisons of means between 3 groups were performed by analysis of variance (ANOVA) with the Bonferroni post-hoc correction for multiple comparisons. The areas under the receiver-operating characteristic (ROC) curves (AUC) were calculated for endocardial, mid-myocardial, epicardial TLS, GLS (not shown) and GCS, WMSI, EF, and troponin T, and all variables were compared by using a nonparametric U test (Analyse-it, Analyse-it Software, Ltd., Leeds, United Kingdom). The value closest to the upper left corner of the ROC curves determined optimal sensitivity and specificity for the ability of endocardial, mid-myocardial, and epicardial TLS, GLS and GCS, WMSI, EF, and troponin T to predict the presence of significant CAD. Logistic regression analysis was performed for factors that could potentially influence myocardial function and to determine the diagnostic value of endocardial TLS for predicting significant CAD. Parameters influencing myocardial contraction, such as age, body mass index, systolic and diastolic blood pressure, plus the categorical variables hypertension and diabetes mellitus, in addition to endocardial TLS, were included in a multivariate logistic regression analysis. To avoid overfitting the model, in each analysis only 3 variables were included with endocardial TLS as a constant factor in the model. All possible combinations with 3 variables were tested. Due to co-linearity of strain parameters, only endocardial TLS was selected for inclusion in the multivariate analysis, and blood pressure parameters were included separately in the model. Correlation of strain parameters with troponin T was tested by the Spearman correlation coefficient. Reproducibility was expressed as intraclass correlation coefficient. p Values were 2-tailed; values <0.05 were considered significant.
We included 77 patients with suspected NSTE-ACS (Table 1). By coronary angiography, 49 (64%) had significant CAD. Patients with significant CAD were more frequently male compared with patients without (Table 1). No differences in age, comorbidity, or medication were observed between those with and without significant CAD at admission. Of the patients with significant CAD, 21 (28%) had significant coronary stenosis, and 28 (36%) had coronary artery occlusion in 1 or more coronary arteries. Five fractional flow reserve measurements were performed in 4 patients. In these 5 arteries, hemodynamically significant stenosis was found in 4 and no significant stenosis in 1 coronary artery. Detailed angiographic findings and revascularization results are shown in Table 2.
Patients with significant CAD had worse function in all 3 myocardial layers assessed by TLS, GLS, and GCS compared with patients without (Table 3). However, endocardial TLS (–14.0 ± 3.3% vs. –19.2 ± 2.2%; p < 0.001), GLS (–15.3 ± 2.2% vs. –19.2 ± 2.2%; p < 0.001), and GCS (–19.3 ± 4.0% vs. –24.3 ± 3.4%; p < 0.001) were most affected (Fig. 2). In general, there was a worsening of myocardial function assessed by TLS, GLS, and GCS in all 3 layers in individuals without significant CAD versus patients with significant coronary artery stenosis and patients with occlusion. However, this method could not differentiate between patients with significant coronary artery stenosis and patients with occlusion (Table 3).
Patients with coronary artery stenosis or occlusion showed generally lower TLS magnitude than GLS (Table 3); however, the difference between these parameters generally did not reach statistical significance. The absence of statistically significant differences can probably be explained by the relatively low sample size.
Endocardial TLS, GLS, and GCS (–15.4 ± 4.0%, –16.7 ± 2.9%, and –21.1 ± 4.5%, respectively) were significantly greater in magnitude than were epicardial TLS, GLS, and GCS (–11.8 ± 2.9%, –12.7 ± 2.0%, and –13.0 ± 2.1%, respectively; all, p < 0.001) in the total study population (N = 77).
The absolute differences between endocardial and epicardial TLS, GLS, and GCS were lower in patients with significant CAD (patients with significant stenosis and/or occlusion; n = 49) (Δ2.4 ± 3.6%, Δ3.4 ± 1.3%, and Δ6.7 ± 3.8%, respectively) than in those without significant CAD (n = 28; values are presented in Table 4) (p < 0.001). This reflects a pronounced decrease in endocardial function in patients with significant CAD. The absolute differences between epicardial and endocardial deformation were generally lower in territorial analysis than in the global analyses. The difference between these parameters, however, did not reach statistical significance (Table 4). The transmural gradient of myocardial strain was less pronounced in individuals with coronary artery occlusion (n = 28) and significant stenosis (n = 21) than in patients without significant CAD (n = 28) (Table 4).
Of 1,232 segments, 1,087 (88%) were analyzable by wall motion scoring. Of these segments, 124 (11%) had wall motion abnormalities: hypokinesis in 119 and akinesis in 5 segments. In hypokinetic and akinetic segments, the absolute differences between endocardial and epicardial longitudinal strains were lower than in segments with normal function (Δ3.1 ± 6.3 vs. Δ5.2 ± 2.3; p = 0.01).
Figures 3 and 4⇓ demonstrate ROC analyses of the ability of different strain parameters, WMSI, EF, and troponin T, to identify patients with significant CAD. Comparison of ROC curves showed that endocardial and mid-myocardial TLS and endocardial GLS were significantly better for identifying significant CAD than were epicardial TLS and GLS (all, p < 0.05), WMSI (all, p < 0.01), and EF (all, p < 0.001). In addition, endocardial GCS was better for identifying significant CAD than were epicardial GCS (p = 0.005) and EF (p = 0.01). Troponin T and mid-myocardial GLS were better only than EF (p < 0.05). TLS parameters had greater AUCs than did GLS parameters; however, the differences were not significant. The other apparent differences between AUCs also were not significant.
Multivariate regression analyses, including parameters influencing myocardial function, showed that reduced myocardial function by endocardial TLS (per 1% change) was the only predictor of the presence of significant CAD (Table 5), independently of which variables were included in the model in addition to endocardial TLS.
There was a significant difference in myocardial function by mid-myocardial TLS between those with 1-vessel occlusion and those with occlusion and 3-vessel disease (p < 0.02), confirming a relationship of function and apparent ischemic burden. There was a similar significant difference in epicardial TLS (p = 0.05). There were, however, no differences in myocardial function assessed by GLS or GCS between patients with 1-, 2-, or 3-vessel disease.
All longitudinal and circumferential strain parameters showed significant correlations with troponin-T values. Endocardial GLS and TLS and endocardial GCS demonstrated greatest correlation, with correlation coefficients of 0.64, 0.58, and 0.47, respectively (all, p < 0.01).
Assessment of longitudinal strain could be performed in 1,303 (94%) endocardial, 1,149 (83%) mid-myocardial, and 1,142 (82%) epicardial LV segments. The corresponding circumferential analyses could be done in 361 (78%) endocardial, 361 (78%) mid-myocardial, and 343 (74%) epicardial segments.
Intraobserver and interobserver intraclass correlations were performed in 10 patients and were 0.96 and 0.96, respectively, for endocardial GLS; 0.87 and 0.92 for mid-myocardial GLS; 0.94 and 0.93 for epicardial GLS; 0.81 and 0.84 for endocardial GCS; 0.84 and 0.81 for mid-myocardial GCS; and 0.82 and 0.76 for epicardial GCS measurements.
This study demonstrates impaired LV function in all 3 myocardial layers in patients with NSTE-ACS and significant CAD compared with patients without significant CAD. We introduce LV layer-specific longitudinal and circumferential strain assessed by 2D-STE as a novel method for the identification of significant CAD in patients with suspected NSTE-ACS. ECG alone frequently failed to identify patients with significant CAD in our study. These results confirm those from earlier studies and suggest that assessment of layer-specific strain can be of valuable help as an additional clinical tool in the diagnostic workup of these patients.
A recent report showed that about two-thirds of coronary angiographies lead to an intervention in patients with NSTE-ACS (1). Although there are several different noninvasive methods for the purpose of identifying significant CAD (16), many of these are expensive and unavailable in the acute setting, and the number of false-positive and -negative test results remain significant. A combination of troponin and strain measurements might have the potential to increase diagnostic accuracy; however, this should be tested in a larger study including outcomes data, before strong conclusions can be made.
The LV wall of the heart comprises 3 myocardial layers: the inner oblique, middle circular, and outer oblique myocardial layers. Differences of myocardial deformation and function across the ventricular wall are well known from previous observations in the normal myocardium (17–19). The endocardium undergoes greater dimensional changes (both thickening and shortening) during systole than does the epicardium in healthy myocardium (17,18).
In patients with CAD, the inner oblique myocardium is most susceptible to ischemic injury (2). Myocardial infarction models as well as reperfusion studies of myocardial infarction have indicated that the endocardial layer is first affected by ischemia (4), causing morphologic (3) and functional alterations predominant in this layer. With increasing severity, ischemia and necrosis propagate in a transmural wave front extending from the endocardium to the epicardium (5). Due to the fact that greatest shortening in a healthy myocardium occurs in the endocardial layer and that endocardium is first affected by ischemia, one will expect that significant CAD causes greatest reduction of function in this particular layer. Reduced endocardial function in patients with significant CAD can be explained by either coronary occlusion and direct myocardial damage or by reversible myocardial dysfunction caused by myocardial stunning or hibernation (20).
Most imaging techniques focus on the evaluation of global and regional LV function. Traditionally, the entire myocardial wall thickness is considered in the analysis of myocardial function without taking into account the differences in the layers of the myocardium. Imaging techniques allowing layer-specific myocardial function analysis are likely to increase the morphologic and pathophysiologic understanding of myocardial ischemia and may help to improve characterization of patients with CAD.
In our study, endocardial longitudinal and circumferential strains were better than epicardial strains for identifying patients with significant CAD. Furthermore, our results indicate that all 3 myocardial layers were affected by both coronary artery occlusion and significant coronary artery stenosis, although to different extents. Reduction of function in all 3 layers could be explained by several factors. First, deformations of the myocardial layers are not independent. The contraction of viable myocardium may result in the deformation of the neighboring nonviable myocardium through passive translational or tethering movements. On the contrary, nonviable myocardium may negatively influence contraction in the adjacent viable myocardium. The deformation of each layer is a sum of an active contraction within the layer and the passive influence of the adjacent layer. Second, nonocclusive lesions and occlusions may represent different degrees of transmural ischemia. Coronary artery occlusions, in the absence of collaterals, may cause transmural ischemia, affecting all 3 myocardial layers with both longitudinal and circumferential dysfunction. Third, lateral resolution might be a factor limiting the delineation of delicate structures without apparent boundaries such as the endocardium, mid-myocardium, and epicardium. These problems are further exaggerated in the basal LV segments, where the density of echo beams is lower.
Both global and territorial strains were assessed in this study for a number of reasons. One is the variability of coronary anatomy (21). Myocardial function assigned to a specific vascular territory by Cerqueira et al. (22) may not necessarily reflect the real coronary distribution. Assessment of global strain is not exposed to these anatomic variations and has been shown to detect even minor reductions in myocardial function (9). Furthermore, networks formed by microvascular communications between coronary arteries can give rise to zones of dual arterial perfusion (23), making strict regional analysis somewhat inaccurate. However, CAD results primarily in segmental wall motion abnormalities; therefore, TLS was assessed as well. Conversely, we chose to omit territorial circumferential strain results because we had short-axis acquisition from only 1 LV slice. The results from territorial circumferential strain would therefore consist of only 2 LV segments each.
All 3 myocardial layers are affected by significant CAD. The greatest decrease in myocardial function, however, occurs in the endocardial layer because shortening normally is most prominent in this layer. Therefore, layer-specific strain analyses might increase diagnostic accuracy in these patients.
Longitudinal and circumferential, but not radial, strain was assessed in the present study. We chose not to analyze radial strain because it has methodological limitations and has been shown to be inferior to longitudinal and circumferential strain in identifying ischemia and necrosis (24). In addition, longitudinal and circumferential strains have been well validated against EF in previous studies (11,25), are reproducible, and are easily obtained with only a minor increase of procedure duration. We did not have 3 short-axis projections available for the assessment of circumferential strain in a 16-segment model. In our study, 16 longitudinal segments were analyzed by strain compared with only 6 mid-circumferential segments.
Fractional flow reserve measurements were not performed in all patients; therefore, the true hemodynamic relevance of the stenoses is not known.
It is not known whether coronary artery occlusions represent transmural or nontransmural infarctions because no viability studies have been performed on our patients.
All patients were treated according to guidelines on admission. Obviously, this could have caused dissolution of thrombi, allowing recovery of the affected myocardium. However, one may also speculate that even in case of dissolution of thrombi, the myocardium might be still stunned or hibernated within the first 48 h.
Deformation of the entire wall thickness of the myocardium was not assessed because the software did not allow such analysis. Consequently, we could not evaluate whether layer-specific strain analysis has additional value compared with traditional global strain analysis in diagnosing CAD.
The gradient of strain across the myocardium is a nonlinear phenomenon, and the definition of the layers is arbitrary and is based on simple division into 3 parts. Because the spatial resolution of ultrasound is limited, there will always be a certain degree of overlap.
Assessment of endocardial and mid-myocardial TLS by layer-specific strain echocardiography provided higher accuracy than did epicardial strain, WMSI, and EF in the identification of patients with NSTE-ACS and significant CAD. Endocardial function was more affected in patients with significant CAD compared to those with epicardial function.
This work was supported by the South-Eastern Norway Regional Health Authority, the Norwegian Research Council, and the Inger and John Fredriksen Foundation. Dr. Sarvari has received honoraria from Toshiba for a lecture at one occasion. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- coronary artery disease
- ejection fraction
- global circumferential strain
- global longitudinal strain
- left ventricular
- non–ST-segment elevation-acute coronary syndrome(s)
- speckle-tracking echocardiography
- territorial longitudinal strain
- Received October 18, 2012.
- Revision received December 18, 2012.
- Accepted January 23, 2013.
- American College of Cardiology Foundation
- Damman P.,
- van G.N.,
- Wallentin L.,
- et al.
- Duncker D.J.,
- Traverse J.H.,
- Ishibashi Y.,
- Bache R.J.
- Ono S.,
- Waldman L.K.,
- Yamashita H.,
- Covell J.W.,
- Ross J. Jr.
- Haugaa K.H.,
- Smedsrud M.K.,
- Steen T.,
- et al.
- Sarvari S.I.,
- Haugaa K.H.,
- Anfinsen O.G.,
- et al.
- Sjoli B.,
- Orn S.,
- Grenne B.,
- Ihlen H.,
- Edvardsen T.,
- Brunvand H.
- Grenne B.,
- Eek C.,
- Sjoli B.,
- et al.
- Bassand J.P.,
- Hamm C.W.,
- Ardissino D.,
- et al.
- Alpert J.S.,
- Thygesen K.,
- Jaffe A.,
- White H.D.
- Lang R.M.,
- Bierig M.,
- Devereux R.B.,
- et al.
- Arnold J.R.,
- Karamitsos T.D.,
- Bhamra-Ariza P.,
- et al.
- Achenbach S.,
- Kramer C.M.,
- Zoghbi W.A.,
- Dilsizian V.
- Sabbah H.N.,
- Marzilli M.,
- Stein P.D.
- Shivalkar B.,
- Flameng W.,
- Szilard M.,
- Pislaru S.,
- Borgers M.,
- Vanhaecke J.
- Segall G.M.,
- Atwood J.E.,
- Botvinick E.H.,
- Dae M.W.,
- Lucas J.R.
- Cerqueira M.D.,
- Weissman N.J.,
- Dilsizian V.,
- et al.
- Gjesdal O.,
- Helle-Valle T.,
- Hopp E.,
- et al.
- Stanton T.,
- Leano R.,
- Marwick T.H.