Longitudinal strain (LS) is a quantitative parameter that adds incremental value to wall motion analysis. The aim of this study was to compare the reproducibility of LS derived from Doppler tissue imaging and speckle-tracking between an expert and a novice strain reader during dobutamine stress echocardiography (DSE).
Forty-one patients (mean age, 65 ± 15 years; mean ejection fraction, 58 ± 11%) underwent DSE per clinical protocol. Global LS derived from speckle-tracking and regional LS derived from both speckle-tracking and Doppler tissue imaging were measured twice by an expert strain reader and also measured twice by a novice strain reader. Intraobserver and interobserver analyses were performed using intraclass correlation coefficients (ICC), Bland-Altman analysis, and absolute difference values (mean ± SD).
Global LS measured by the expert strain reader demonstrated high intraobserver measurement reproducibility (rest: ICC = 0.95, absolute difference = 5.5 ± 4.9%; low dose: ICC = 0.96, absolute difference = 5.7 ± 3.7%; peak dose: ICC = 0.87, absolute difference = 11.4 ± 8.4%). Global LS measured by the novice strain reader also demonstrated high intraobserver reproducibility (rest: ICC = 0.97, absolute difference = 4.1 ± 3.4%; low dose: ICC = 0.94, absolute difference = 5.4 ± 5.9%; peak dose: ICC = 0.94, absolute difference = 6.1 ± 4.8%). Global LS also showed high interobserver agreement between the expert and novice readers at all stages of DSE (rest: ICC = 0.90, absolute difference = 8.5 ± 7.5%; low dose: ICC = 0.90, absolute difference = 8.9 ± 7.1%; peak dose: ICC = 0.87, absolute difference = 10.8 ± 8.4%). Of all parameters studied, LS derived from Doppler tissue imaging had relatively low interobserver and intraobserver agreement.
Global LS is highly reproducible during all stages of DSE. This variable is a potentially reliable and reproducible measure of myocardial deformation.
Dobutamine stress echocardiography (DSE) is a widely used noninvasive and functional imaging modality to assess for ischemic heart disease and myocardial viability. Conventional DSE relies on the detection of wall motion abnormalities by qualitative visual analysis. However, this technique is subjective and limited by low reproducibility and variability, even among expert observers. Both the American Society of Echocardiography and the European Association of Echocardiography recommend supervised reporting of ≥100 stress echocardiograms to achieve minimum competence. There is a need for substantial training, and a learning curve exists to improve visual wall motion analysis during DSE.
Wall motion can be quantified by strain analysis, which allows more objective evaluation of regional myocardial deformation. Regional longitudinal strain (LS) and strain rate by Doppler tissue imaging (DTI) and speckle-tracking (ST) can identify myocardial ischemia and viability during DSE with correlation to wall motion analysis and coronary angiography. LS is predominately found in the subendocardial region and is the most sensitive marker of myocardial ischemia and nontransmural infarction. ST-derived global LS is a relatively new prognostic marker for quantifying left ventricular (LV) systolic function, particularly in patients with extensive wall motion abnormalities and congestive heart failure. After acute myocardial infarction, LV global LS can predict myocardial viability during low-dose DSE. Global LS combined with wall motion analysis during DSE provides incremental diagnostic accuracy for myocardial ischemia detection. Despite the accumulation of published data in the medical literature confirming the incremental value of strain over conventional echocardiography, the technique has not yet been incorporated into widespread routine clinical application. The primary concerns include limitations of expertise, need for meticulous acquisition, time-consuming postprocessing analysis, signal-to-noise variability and lack of reproducibility, particularly during high-dose DSE. We evaluated the feasibility and reproducibility of regional and global LS between expert and novice strain readers during all stages of DSE.
We prospectively enrolled 50 patients who were referred with clinical indications for DSE from August 2010 to May 2011. Seven patients with suboptimal images, in whom administration of an echocardiographic contrast agent would have been justified on the basis of American Society of Echocardiography criteria for contrast administration, were excluded from the study. We also excluded two more patients who had arrhythmias during DSE. Forty-one patients (mean age, 65 ± 15; 41% men) were eligible for this study.
A standard dobutamine stress echocardiographic protocol with incremental dobutamine infusion rates of 5, 10, 20, 30, 40, and 50 μg/kg/min for 3 min and the addition of atropine and/or isometric exercise if clinically indicated to achieve target heart rate was used. Standard end point criteria included achievement of target heart rate (85% of the maximal age-predicted heart rate), worsening angina, significant ST-segment change on continuous electrocardiographic monitoring (in patients without left bundle branch block), significant dysrhythmia, severe systemic hypertension or hypotension, and worsening wall motion abnormalities.
Patients were scanned in the left lateral recumbent position using a Vivid E9 ultrasound scanner (GE Vingmed Ultrasound AS, Horten, Norway) with a 3.5-MHz transthoracic phased-array matrix transducer. An experienced sonographer acquired standard apical and parasternal images at rest and at low-dose (5 or 10 μg/kg/min) and peak-dose (40 or 50 μg/kg/min) dobutamine infusions. Acquired images were electrocardiographically gated to three cardiac cycles in the apical four-, two-, and three-chamber views in both conventional two-dimensional grayscale and color DTI. Echocardiographic images were stored digitally and analyzed offline using a dedicated postprocessing software package (EchoPAC BT11; GE Vingmed Ultrasound AS).
Standard Echocardiographic Measurements
Ejection fraction, visual wall motion score index, and number of dysfunctional segments were determined by an experienced observer blinded to clinical and strain data. Ejection fraction was calculated using the modified Simpson’s biplane method with manual tracing of the endocardial borders at end-diastole and end-systole in the apical four- and two-chamber views. All segmental analyses were based on the conventional American Society of Echocardiography 16-segment LV model. Each segment was assigned a wall motion score of 1 to 4, where 1 = normal, 2 = hypokinetic, 3 = akinetic, and 4 = dyskinetic or aneurysmal. Wall motion score index was derived as the sum of individual segment scores divided by the number of segments visualized.
DTI-derived measurements were obtained using previously described standard techniques using color Doppler myocardial imaging at a frame rate of >100 Hz ( Figure 1 A). On a still frame in each apical view, the observer placed a standard sample volume (15 × 2 mm) in the middle of each segment at the onset of systole and then manually tracked frame by frame throughout the cardiac cycle to maintain a precise midwall position.
ST-derived strain was acquired from grayscale images at frame rates of 50 to 80 Hz, as previously described ( Figure 1 B). Regional and global LS was measured using the software-specific semiautomated function image tool. Endocardial borders were defined, and myocardial motion was tracked using a semiautomated algorithm in all three apical views using EchoPAC automated function imaging analysis software. The observer was able to override an unsatisfactory tracking result if deemed visually satisfactory, and vice versa. Once approved, global LS and peak regional LS values were displayed for all segments.
Inter- and Intraobserver Variability
Two independent, blinded observers performed myocardial strain analysis offline. The first observer was considered a novice strain reader, as the cardiology fellow-in-training specialized in echocardiography but had no previous experience in strain analysis. The second observer was considered an expert strain reader, as this individual was a cardiologist with extensive experience (>5 years) in the performance and interpretation of strain imaging. Intraobserver agreement was tested for each independent reader, and all analyses were repeated in all 41 patients >1 week apart. Blinded repeat analysis was performed using the same images from the same cardiac cycles. Interobserver agreement was evaluated between the novice and expert analyses.
Exclusion of Segments
Segments that were unsuitable for myocardial strain analysis because of limited image quality or poor tracking were excluded. Global LS for the left ventricle cannot be calculated if more than two segments have suboptimal tracking.
Summary data are expressed as numbers, mean ± SD, or percentages. Data were analyzed using standard statistical software (SPSS version 13; SPSS, Inc, Chicago, IL). For all strain measurements, interobserver and intraobserver variability was assessed using Bland-Altman analysis and intraclass correlation coefficients (ICCs) with absolute difference values between repeated measurements expressed as percentages of the mean of repeated measurements. P values < .05 was considered statistically significant.
Thirty patients had known histories of ischemic heart disease before DSE or were newly diagnosed with ischemic heart disease after DSE. Clinical and baseline echocardiographic characteristics of the patients are shown in Table 1 . For all patients, the mean systolic and diastolic blood pressures and heart rate were 138 ± 25 mm Hg, 74 ± 11 mm Hg, and 64 ± 12 beats/min at rest, respectively. The mean heart rate increased to 137 ± 17 beats/min (88 ± 8% of predicted maximal heart rate) at peak DSE. Thirty-one patients achieved their target heart rate of 85% of the predicted maximal heart rate during peak DSE.
|Age (y)||65 ± 15|
|Ejection fraction (%)||58 ± 11 (range, 25–71)|
|LVEDV (mL/m 2 )||48 ± 21|
|Coronary artery disease||30 (73%)|
|Heart rate (beats/min)||64 ± 12|
|Systolic blood pressure (mm Hg)||138 ± 25|
|Diastolic blood pressure (mm Hg)||74 ± 11|
Feasibility and Reproducibility of Global LS Assessment
Global LS measurements were possible in all patients for all stages of DSE with 100% feasibility in those with good image quality included in the study ( Table 2 ). Global LS had excellent correlations for intra- and interobserver agreement throughout all stages of DSE (interobserver agreement at rest: ICC = 0.90 [95% confidence interval (CI), 0.81–0.94], absolute difference = 8.5 ± 7.5%, mean difference = 0.1 ± 4.0; low dose: ICC = 0.90 [95% CI, 0.82–0.95], absolute difference = 8.9 ± 7.1%, mean difference = 0.3 ± 4.4; peak dose: ICC = 0.87 [95% CI, 0.76–0.93], absolute difference = 10.8 ± 8.4%, mean difference = 0.8 ± 4.0; Table 2 ). Bland-Altman analysis in interobserver agreement is demonstrated in Figure 2 .
|Feasibility||Baseline||Low dose||Peak dose|
|Absolute difference (%) ∗||5.5 ± 4.9||5.7 ± 3.7||11.4 ± 8.4|
|Mean difference †||1.0 ± 1.8||0.8 ± 2.2||1.7 ± 2.7|
|Absolute difference (%) ∗||4.1 ± 3.4||5.4 ± 5.9||6.1 ± 4.8|
|Mean difference †||0.5 ± 1.8||0.0 ± 3.4||0.1 ± 2.9|
|Absolute difference (%) ∗||8.5 ± 7.5||8.9 ± 7.1||10.8 ± 8.4|
|Mean difference †||0.1 ± 4.0||0.3 ± 4.4||0.8 ± 4.0|
Feasibility and Reproducibility of ST- and DTI-Derived LS
ST-derived regional LS measurements were obtainable in 95% of segments at rest, 89% of segments at low dose, and 78% of segments at peak dose ( Table 3 ). DTI-derived regional LS measurements were obtainable in 90% of segments at rest, 87% of segments at low dose, and 72% of segments at peak dose ( Table 4 ). ST-derived regional LS was more feasible than DTI-derived regional LS for all stages of DSE, with less exclusion of segments from analysis.
|Feasibility||Baseline||Low dose||Peak dose|
|Absolute difference (%) ∗||9.6 ± 8.8||11.8 ± 10.6||15.2 ± 12.7|
|Mean difference †||0.6 ± 4.2||0.9 ± 6.1||1.4 ± 6.9|
|Absolute difference (%) ∗||11.0 ± 12.2||13.2 ± 17.1||13.0 ± 11.9|
|Mean difference †||0.3 ± 5.6||0.2 ± 7.4||0.2 ± 6.9|
|Absolute difference (%) ∗||14.4 ± 13.5||16.7 ± 16.4||21.7 ± 19.4|
|Mean difference †||0.1 ± 7.0||0.8 ± 9.8||0.9 ± 10.4|
|Feasibility||Baseline||Low dose||Peak dose|
|Absolute difference (%) ∗||14.6 ± 12.6||13.7 ± 14.6||14.7 ± 12.6|
|Mean difference †||0.4 ± 7.7||0.3 ± 7.5||0.2 ± 8.2|
|Absolute difference (%) ∗||16.2 ± 19.0||19.0 ± 18.3||19.9 ± 22.1|
|Mean difference †||1.0 ± 7.7||0.5 ± 10.5||0.3 ± 10.9|
|Absolute difference (%) ∗||28.7 ± 34.8||26.9 ± 24.8||29.9 ± 26.4|
|Mean difference †||0.1 ± 15.0||0.9 ± 14.2||1.2 ± 17.0|