Author + information
- Received August 11, 2005
- Revision received November 7, 2005
- Accepted November 21, 2005
- Published online April 18, 2006.
- Helge Skulstad, MDa,
- Stig Urheim, MD,
- Thor Edvardsen, MD, PhD,
- Kai Andersen, MD, PhD,
- Erik Lyseggen, MDa,
- Trond Vartdal, MDa,
- Halfdan Ihlen, MD, PhD and
- Otto A. Smiseth, MD, PhD⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Otto A. Smiseth, Department of Cardiology, Rikshospitalet University Hospital, N-0027 Oslo, Norway.
Objectives The aim of the study was to compare the ability of the tissue Doppler echocardiographic imaging (TDI) modalities velocity, strain, and displacement to quantify systolic myocardial function.
Background Several TDI modalities may be used to quantify regional myocardial function, but it is not clear how the different modalities should be applied.
Methods In 10 anesthetized dogs we measured left ventricular pressure, longitudinal myocardial velocity, strain, and displacement by TDI at baseline and during left anterior descending coronary artery (LAD) stenosis and occlusion. Reference methods were segmental shortening by sonomicrometry and segmental work. In 10 patients with acute anterior wall infarction (LAD occlusion) and 15 control subjects, velocity, strain, and displacement measurements were performed.
Results In the animal study, systolic strain correlated well with segmental shortening (r = 0.96, p < 0.01) and work (r = 0.90, p < 0.01), and differentiated well between non-ischemic (−13.5 ± 3.2% [mean ± SD]), moderately ischemic (−6.5 ± 2.8%), and severely ischemic myocardium (7.1 ± 13.2%). The ratio post-systolic strain/total strain also differentiated well between levels of ischemia. Displacement and ejection velocity had weaker correlations with segmental shortening (r = 0.92 and r = 0.74, respectively) and regional work (r = 0.85 and r = 0.69), and there was marked overlap between values at baseline and at different levels of ischemia. In the human study, systolic strain differentiated well between infarcted and normal myocardium (1.0 ± 5.0% vs. −17.8 ± 3.8%), whereas systolic displacement (−0.3 ± 1.3 mm vs. −2.3 ± 0.6 mm) and ejection velocity (0.9 ± 0.6 cm/s vs. 2.2 ± 0.6 cm/s) showed overlap. In the infarction group, strain was reduced in segments with infarcted tissue, while systolic velocity and displacement were reduced in all segments and did not reflect the extension of the infarct.
Conclusions Strain was superior to velocity and displacement for quantification of regional myocardial function. Provided technical limitations can be solved, strain Doppler is the preferred TDI modality for assessing function in ischemic myocardium.
Tissue Doppler imaging (TDI) was introduced several years ago as a method to quantify regional myocardial function, and was assumed to be useful in the evaluation of patients with suspected coronary artery disease (1,2). The most promising TDI measure of regional function has been peak systolic ejection velocity, which is typically reduced in ischemic regions (3,4). Although the TDI methodology has been extensively validated and is now installed in most echocardiographic recorders on the market, myocardial velocity imaging is still not widely used in clinical routine. The slow dissemination of TDI is most likely related to some of the limitations of the method, which include spatial non-uniformity of even normal velocities and the related problem that velocities in a particular region are, in part, caused by contractions in adjacent regions (5,6). Furthermore, peak ejection velocity appears to have limited ability to differentiate between different grades of ischemic dysfunction (7).
The present study investigates if grading of ischemic dysfunction by two other TDI modalities (i.e., strain Doppler echocardiography [SDE] and myocardial displacement imaging) may be superior to velocity imaging for quantification of regional myocardial dysfunction. Previous reports indicate that strain imaging in principle provides better quantification of regional function than velocity imaging (8,9). However, SDE is associated with methodological problems that limit its clinical application. This includes strong sensitivity of Doppler strain to misalignment between the cardiac axis and the ultrasound beam, and substantial drifting of the signal (6). Therefore, as an alternative modality, we evaluate displacement imaging, which is essentially calculation of the regional velocity-time integrals. This modality provides more comprehensive information than peak systolic ejection velocity, because displacement can be set to incorporate velocity data from isovolumic as well as ejection phases. Other likely advantages relative to SDE are less dependency on the angle between the cardiac axis and the ultrasound beam and virtually no problem with signal drift.
The aim of the present study was to determine how the different TDI modalities could be applied to obtain the most accurate quantification of myocardial function during acute ischemia. In a dog model, we used segmental shortening by implanted ultrasonic crystals and segmental work as reference methods for regional function. Furthermore, we compared the TDI modalities in a clinical study that included patients with acute myocardial ischemia.
Ten dogs with body weight 23 ± 1.6 kg (mean ± SEM) were prepared as previously described (6,10). The National Animal Experimentation Board approved the study.
Instrumentation and hemodynamic measurements
Pressures were measured by micromanometers (10). Coronary flow was reduced by a pneumatic constrictor around the proximal left anterior descending coronary artery (LAD), and was measured by an ultrasonic transit-time flow probe (Transonic Systems Inc., Ithaca, New York). Another constrictor was positioned around the ascending aorta for increment of left ventricle (LV) afterload.
One pair of ultrasonic crystals was implanted in the anterior LV wall (LAD region) and another pair in the posterior wall (circumflex artery region). Both pairs were positioned parallel to the LV long axis in the inner third of the myocardium and connected to a sonomicrometer (Sonometrics, London, Ontario, Canada).
Digital recordings were performed by a System FiVe ultrasonograph (GE Vingmed Ultrasound, Horten, Norway) with a combined tissue imaging (3.5 MHz) and Doppler (2.75 MHz) transducer. The mean frame rate was 87 s−1. To minimize noise, the pulse repetition frequency was set to 0.5 to 1.0 kHz. Recordings were obtained from the apical position with the image plane oriented through the regions in which crystals were positioned, and were digitally stored and analyzed off-line (Echopac, GE Vingmed Ultrasound, Horten, Norway). The echocardiographic measurements were performed from the same position as the ultrasonic crystals were located.
Measurements and calculations
Maximal systolic and end-diastolic LV pressures were measured along with maximum and minimum time derivatives of LV pressure (dP/dt) (Table 1).Left atrial (LA) pressure was measured at first diastolic crossover with LV pressure. Left anterior descending coronary artery flow was calculated as mean flow. Isovolumic contraction phase (IVC) was defined from peak R on electrocardiogram (ECG) to LV and aortic pressure crossover. End-systole was defined as time of LV dP/dtmin. Isovolumic relaxation (IVR) was defined from LV dP/dtminto first diastolic LA and LV pressure crossover. All values represent the mean of measurements from three consecutive heart cycles. The area of LV pressure-segment length loops was used as an index of segmental work (Fig. 1),as previously described and validated by regional wall stress-segment length relations (10).
Tissue velocitiesof a given region were obtained from a velocity trace generated throughout the cardiac cycle (Fig. 2).
Displacementof a given region was calculated by temporal integration of myocardial velocity, displayed in millimeters versus time. By convention, displacement from base towards apex was given negative values and displacement towards the base positive values. This is contrary to some previous reports, which assign positive values to displacement towards the apex (11). We felt that using negative signs is more appropriate and intuitive because systolic displacement in the LV long axis is due to shortening. This is also consistent with the convention of reporting systolic strains in the LV long axis as shortening strains (6,10). Velocity-time integration was started at end-diastole, and displacement was measured at end of IVC (Divc) and as the maximum value in systole (Dsys). Maximum post-systolic displacement (Dps) was calculated as maximal displacement after end-systole.
Strainwas calculated by TDI (6,12) with measurements 6 mm apart (10 mm in the clinical study). A strain curve of a given region was generated throughout the cardiac cycle, enabling measurement of strain at end of IVC (Sivc) and of maximum systolic (Ssys) and maximum post-systolic strain (Sps). Corresponding measurements were made by sonomicrometry, using end-diastolic segment length as reference.
As an index of effective segmental shortening, we calculated the ratio between systolic shortening and the sum of systolic and post-systolic shortening (Ssys/[Ssys+ Sps]), and the same ratio was calculated for displacement (Dsys/[Dsys+ Dps]). This effective shortening ratio expresses the fraction of segmental shortening that contributes to LV stroke volume.
To avoid interference between sonomicrometry and Doppler ultrasound, we first recorded pressures, ECG, and echocardiographic data during 10 s. Then, during the subsequent 10 s, pressures, flow, ECG, and sonomicrometry data were recorded. Data were recorded with the respirator off.
Moderate and severe ischemia was induced by LAD stenosis and occlusion, respectively. Moderate ischemia was obtained by gradually reducing LAD flow (by 47 ± 6%) until sonomicrometry demonstrated marked hypokinesis. The hemodynamic variables were allowed to stabilize for 1 to 2 min before recordings were made. Then the LAD constrictor was deflated. After a recovery period, another set of baseline recordings were performed. During re-established LAD stenosis, LV afterload was instantly increased by constriction of the ascending aorta. The constrictor was deflated after 10 s. After 30 min of reperfusion, which resulted in complete functional recovery, the LAD was completely occluded, and recordings were performed after 1 to 2 min. Aortic constriction was performed in six experiments.
We studied 15 (12 men) healthy volunteers, age 57 ± 2 years, and 10 patients (9 men), age 56 ± 3 years, with acute myocardial infarction in the anterior wall. Angiograms verified occlusion of the distal part of the proximal LAD (three patients) or mid-LAD (seven patients). Digital echocardiographic recordings were performed by a Vivid 7 ultrasonograph (GE Vingmed Ultrasound). Apical two-chamber views were used to achieve similarity to the animal study. Measurements were done in the basal and apical parts of the basal, mid-, and apical segments of the anterior wall. Mean frame rate was 108 ± 1 s−1and 150 ± 4 s−1(p < 0.01), and heart rate was 75 ± 8 and 61 ± 3 beats/min (p < 0.05), respectively.
Magnetic resonance imaging (MRI) obtained after six months demonstrated transmural anterior wall infarction in five of six patients.
Angle dependency of Doppler measurements
In five healthy volunteers, we compared the different TDI modalities with regard to angle dependency. The ultrasound insonation angle was modified by dislocating the echocardiographic probe from a strict apical position, aiming a deviation of approximately 30° from the LV longitudinal axis.
Values are expressed as mean ± SEM if not stated otherwise. In the animal study, non-parametric tests were used. Normal distribution was not assumed due to a small number of observations. Repeated measurements were analyzed by Friedman’s two-way analysis of variance by ranks (13). Paired data were analyzed with Wilcoxon matched pair test. In the clinical study, analysis of variance was performed with the six ventricular segments as factors. Comparisons between segments in patients and control subjects were performed with Student ttest. Bonferonni corrections for multiple comparisons were used in the non-parametric and parametric analysis. Regression analyses were performed with the least-squares method. A value of p < 0.05 was considered significant.
Strain by sonomicrometry, which served as reference method, showed systolic shortening that started during IVC and continued until aortic valve closure (Fig. 2). In some cases, there was slight post-systolic shortening. Strain by Doppler showed traces very similar to sonomicrometry.
The displacement traces were qualitatively similar to the strain traces. Velocity traces were dominated by a large, but brief, positive spike during IVC and a more protracted velocity during LV ejection. There were only minor velocity components during IVR. During early LV filling, all modalities indicated rapid segmental lengthening. At baseline, the findings were essentially identical in the posterior and anterior walls.
During LAD stenosis sonomicrometry demonstrated moderate reductions in systolic shortening and a marked increase in post-systolic shortening (Fig. 2). Similar changes were observed by strain Doppler and displacement analysis. During LAD occlusion sonomicrometry showed systolic lengthening and more marked post-systolic shortening, indicating further deterioration of regional function. Similar changes were observed using strain-Doppler and displacement imaging. As shown in Figure 3,both strain-Doppler and displacement imaging differentiated between LAD stenosis and occlusion (i.e., between moderate and severe ischemia). When comparing baseline and LAD stenosis, however, there was overlap between individual values of peak systolic strain and displacement.
Peak systolic ejection velocity decreased during LAD stenosis, but individual values showed substantial overlap with baseline measurements. During LAD occlusion, there was further decrease in peak systolic ejection velocity but with considerable overlap with LAD stenosis. Therefore, peak ejection velocity did not differentiate between moderate and severe ischemia. The negative component of IVC velocity and the positive component of IVR velocity were larger (p < 0.05 and p < 0.01, respectively) in severe than in moderate ischemia, but there was overlap between individual values.
There were no significant changes in the Doppler or sonomicrometry parameters in the posterior wall during LAD constriction.
We also investigated if incorporation of post-systolic motion could enhance the diagnostic power of strain and displacement imaging. The relative contribution of post-systolic strain was quantified as the ratio between systolic strain and total strain, and a similar ratio was calculated for displacement. The ratio of systolic strain/total strain discriminated completely between the various degrees of ischemia. There was, however, some overlap between ratios for displacement during baseline and LAD stenosis (Fig. 3, Table 2).
Increased LV afterload during ischemia
During moderate ischemia, the aorta was abruptly constricted, and this caused a marked rise in LV systolic pressure. Regional function by sonomicrometry showed a shift from systolic shortening to systolic lengthening (Fig. 2). This change in regional function was clearly demonstrated by strain and displacement imaging, and both modalities differentiated well between the two conditions (Fig. 3). Peak systolic ejection velocity decreased significantly (p < 0.05), but there was overlap of individual values before and during aortic constriction.
Comparison to reference methods
Figure 4summarizes in scatter plots how each of the TDI modalities performed as compared to the two reference methods for grading myocardial function.
Velocity and displacement were maximum at the base and decreased gradually towards the apex in healthy control subjects and in patients with acute anterior wall myocardial infarctions as well (Fig. 5).In all segments, mean values of velocities and displacements in the infarction group were reduced relative to control subjects (p < 0.05), but there was substantial overlap between individual values in the two groups. Importantly, positive displacement values, which were observed in some of the infarcted segments, were not seen in normal hearts. Inclusion of post-systolic velocity and post-systolic displacement did not provide better differentiation between groups or individuals with normal and infarcted ventricles (Table 3).Because velocity and displacement values decreased gradually towards the apex, these measures did not define the anatomical extension of dysfunctional myocardium.
In healthy individuals, systolic strain was relatively uniform between base and apex, except for a reduction in the most distal part of the apical segment (Table 3). In the infarction group, however, strain decreased sharply in the mid-segment, consistent with severe dysfunction in the distal portion of the anterior LV wall (Figs. 5 and 6).⇓Figure 6illustrates that strain reverses from systolic shortening (negative strain) to systolic lengthening (positive strain) in the distal segments. Systolic ejection velocity, however, approached zero in the mid-segment and was essentially similar at more distal measurement sites. Therefore, systolic ejection velocity did not reflect more severe dysfunction in the apical segments, as indicated by the strain measurements.
Angle dependency of Doppler parameters in a clinical setting
Figure 7illustrates misalignment between insonation angle and the LV long axis. An angle deviation of 30° reduced apical velocity, strain, and displacement by approximately 50%. For the mid- and basal segments, there was less effect of angle deviation. Importantly, strain values did not show stronger angle dependency than velocity and displacement. This was explained by the geometrical principles that are illustrated in Figure 7.
To our knowledge, this is the first study that directly compares the different TDI modalities SDE, tissue velocity, and displacement imaging. We demonstrate that strain is superior to velocity and displacement for grading of myocardial segmental dysfunction. During experimental ischemia and in patients with anterior wall infarctions, peak systolic velocity could not differentiate between hypokinetic and dyskinetic myocardium, while systolic strain by SDE was an excellent tool for quantification of function in non-ischemic as well as ischemic myocardium. Furthermore, strain was superior to velocity and displacement imaging for defining the anatomical extension of dysfunctional myocardium.
Best TDI modality for quantification of regional dysfunction
In the present study, the different TDI modalities were validated against two different reference methods (i.e., regional systolic shortening by sonomicrometry and regional work from pressure-segment length loops). The rationale for using systolic shortening as a reference method for function is that it reflects the contribution by a particular segment to LV stroke volume. However, systolic shortening, similar to stroke volume and global LV ejection fraction, is markedly load-dependent. As an alternative method, we used the area of the regional LV pressure-segment length loop, which is a measure of regional myocardial work and reflects energy consumption. Segments with loop areas <0 are passive, and the negative loop area reflects work performed on such segments by the rest of the ventricle. When loop areas are positive, it means that the segment performs active work, and the loop area reflects the magnitude of regional work. Figure 4illustrates how the different TDI modalities performed against the two “gold standards” for regional systolic function. The figure demonstrates that values for systolic velocities in entirely passive segments (negative loop areas) overlapped with velocities in segments that were actively contracting (positive loop areas). This overlap implies that peak systolic velocity could not identify aggravation or improvement of ischemic dysfunction. For systolic strain and displacement, however, there was no overlap between segments with positive and negative work. Therefore, strain by SDE and displacement imaging differentiated better between moderate and severe ischemic dysfunction. There was some overlap between values for non-ischemic and moderately ischemic myocardium for all three modalities.
In non-ischemic myocardium, virtually all contraction occurs during systole with very little post-systolic shortening. Therefore, post-systolic motion has been introduced as a potentially useful marker of ischemic dysfunction. In the present study, post-systolic strain differentiated well between non-ischemic and ischemic myocardium. Post-systolic strain, however, provided no added value in staging of ischemic dysfunction, as there was substantial overlap between post-systolic strain values in segments with moderate and severe dysfunction. Post-systolic velocity and displacement in ischemic myocardium exceeded values in non-ischemic myocardium, but there was more overlap than for strain measurements. Thus, post-systolic motion appears to be a marker of ischemia, but does not help in grading of ischemic dysfunction.
We also evaluated the ratio between systolic strain and combined systolic and post-systolic strain. In non-ischemic myocardium, virtually all contraction occurred in systole, and, therefore, the ratio was near one. In ischemic myocardium, the ratio was markedly reduced. The ratio differentiated better between different levels of ischemia than just measuring systolic or post-systolic strain. In contrast, calculating a similar ratio for displacement and velocity did not improve grading of ischemic dysfunction.
Best TDI modality for anatomical localization of dysfunction
Due to tethering between myocardial segments, impairment of function in one region causes apparent reduction of function in adjacent and even remote regions (14). Therefore, assessment of systolic velocity is not well suited for defining anatomical extension of dysfunctional myocardium. There was reduction in velocities and displacement in all segments between base and apex, and there was no obvious transition zone between intact and dysfunctional myocardium. This implies that velocity imaging and displacement imaging show apparent dysfunction outside the injured region. Systolic strain, however, is not limited by tethering to the same extent, and in patients there was a relatively sharp transitional zone between intact and dysfunctional myocardium. Therefore, systolic strain appears superior to velocity and displacement imaging as a method to define anatomical extension of dysfunctional myocardium.
Angle dependency of TDI modalities
The present study confirms that measurements in apical segments were highly dependent on optimal apical positioning of the echocardiographic probe. Measurements from mid- and basal anterior wall segments were less dependent on medial shifts of probe position. Importantly, strain showed only minor changes when probe position was shifted. This apparent angle-independency of strain is explained by geometrical factors (i.e., medial repositioning of the probe gave good alignment for the mid- and basal segments of the anterior wall). It does not mean that strain is angle independent. On the contrary, strain is very sensitive to angle problems (6).
The reference methods for systolic function that were used in the animal study represent best possible approaches for in situ hearts, but do not provide perfect measures of myocardial contractility. Segmental shortening, however, reflects the contribution of a particular segment to stroke volume, and may be considered a measure of effective systolic function. The pressure-segment length loop is an index of work and, therefore, is closely related to myocardial energy consumption. Each of these reference methods reflects the impairment in cardiac mechanics that is typical for ischemic dysfunction.
In clinical studies we lack reliable reference methods for measuring regional contractility. However, systolic strain provides a measure of regional myocardial shortening, which, in turn, is related to contractility. In a previous clinical study that used MRI tagging as a reference method, we showed that TDI could measure regional strain accurately (15). Therefore, in the present study, we used Doppler strain as reference method for regional function. Furthermore, we used MRI at follow-up to confirm scar tissue in infarcted segments. In contrast to echocardiography, MRI is not possible to perform in the acute phase of myocardial infarction.
Only six dogs underwent ischemia and aortic constriction, but similar responses were demonstrated in each experiment. The results were tested with non-parametric analyses of variance by ranks, which ensured validity of the findings despite the low number of experiments.
Due to the technical limitations with strain-rate imaging, most importantly artefacts and random noise, strain rate was not included in this study.
The present study demonstrates that strain is superior to systolic velocity and displacement for quantification of myocardial function. Strain had excellent ability to quantify function in intact as well as dysfunctional myocardium, while systolic ejection velocity could not differentiate between stages of ischemic dysfunction. Displacement performed somewhat better than velocity. Strain imaging should undergo further testing to determine how this modality can be implemented in clinical routine.
↵a Drs. Skulstad, Lyseggen, and Vartdal were recipients of clinical research fellowships from the Norwegian Council on Cardiovascular Diseases.
- Abbreviations and Acronyms
- time derivatives of left ventricular pressure
- isovolumic contraction
- isovolumic relaxation
- left atrial/atrium
- left anterior descending coronary artery
- left ventricle/ventricular
- magnetic resonance imaging
- strain Doppler echocardiography
- tissue Doppler imaging
- Received August 11, 2005.
- Revision received November 7, 2005.
- Accepted November 21, 2005.
- American College of Cardiology Foundation
- Fleming A.D.,
- Xia X.,
- Sutherland G.R.,
- Fenn L.
- Gorcsan J. 3rd.,
- Strum D.P.,
- Mandarino W.A.,
- et al.
- Derumeaux G.,
- Ovize M.,
- Loufoua J.,
- et al.
- Mädler C.F.,
- Payne N.,
- Wilkenshoff U.,
- et al.
- Urheim S.,
- Edvardsen T.,
- Torp H.,
- Angelsen B.,
- Smiseth O.A.
- Edvardsen T.,
- Urheim S.,
- Skulstad H.,
- Steine K.,
- Ihlen H.,
- Smiseth O.A.
- Edvardsen T.,
- Skulstad H.,
- Aakhus S.,
- Urheim S.,
- Ihlen H.
- Skulstad H.,
- Edvardsen T.,
- Urheim S.,
- et al.
- Altman D.G.
- Lima J.A.C.,
- Becker L.C.,
- Melin J.A.,
- et al.
- Edvardsen T.,
- Gerber B.L.,
- Garot J.,
- Bluemke D.A.,
- Lima J.A.C.,
- Smiseth O.A.