Introduction
Left ventricular ejection fraction (LVEF) is the established method for evaluation of LV systolic function and can be measured by a number of imaging modalities. LVEF by echocardiography has been regarded as a cornerstone in the prediction of outcome and is the most widely available method for evaluation of LV function. It is a vital measurement when determining whether patients benefit from an implantable cardioverter-defibrillator (ICD) or cardiac resynchronization therapy (CRT) [
1]. In addition, LVEF is used to define systolic heart failure and has a great impact on the selection of medical treatment [
1]. Several echocardiographic methods have been used to measure LVEF but at present, the Simpson’s biplane method is most widely used [
2]. Determining LVEF by echocardiography is associated with a high level of inter-observer variability, which to a certain degree can be improved using contrast enhanced echocardiography and 3D echocardiography [
3]. Reliability of LVEF depends on image quality and in particular the ability to visualize the endocardial border. Studies have shown that LVEF measured by cardiovascular magnetic resonance imaging (CMR), radionuclide ventriculography and echocardiography is not easily interchangeable [
3].
Strain by speckle tracking echocardiography is a technique that utilizes 2-dimensional gray scale images to evaluate both global and regional function of the left ventricle. Peak global longitudinal strain (GLS) may be used to measure systolic function. Previous studies have shown that GLS may both diagnose and exclude acute coronary heart disease better than LVEF [
4‐
6]. In addition, GLS has better intra- and inter-observer reproducibility in post hoc analysis compared to LVEF [
4,
7,
8]. Furthermore, GLS may be analyzed in a majority of patients with good feasibility [
9] and may be measured as fast as LVEF [
4,
10]. Since several studies have shown advantages of GLS compared to LVEF in the evaluation of LV function especially for mild systolic dysfunction [
11], GLS is increasingly used in clinical practice. In ESC guidelines for management of acute coronary syndrome in patients presenting without persistent ST-segment elevation, echocardiography is recommended and strain is suggested as a tool to identify reduced regional function [
12]. GLS is also recommended used in early detection of cardiotoxicity during chemotherapy [
13]. However, it is not well studied how the level of echocardiographic training impact performance of GLS compared to LVEF. It is therefore of interest to study the effect of echocardiographic training on reproducibility of GLS and LVEF.
The aim of this study was to investigate reproducibility of LVEF by Simpson’s biplane and GLS by speckle tracking echocardiography when echocardiographers with different levels of expertise obtain images. Furthermore, we compared inter observer variability of GLS and LVEF between expert and trainee both in image acquisition, image analysis and cross analysis.
Discussion
To our knowledge, this is one of very few studies to demonstrate that GLS is a more reproducible method for evaluation of LV systolic function than LVEF regardless of echocardiographic training. These findings support the emerging clinical use of GLS as an additional and incremental diagnostic tool in specific myocardial diseases.
A number of previous studies have compared the ability of GLS and LVEF to detect small reductions in LV function, particularly in ischemic heart disease [
10,
14,
17‐
19]. In these studies, GLS and segmental strain had better ability than LVEF to predict infarct size and segmental viability in patients with myocardial infarction [
18,
20], diagnose coronary artery occlusion in patients with NSTEMI [
5,
21], exclude coronary artery disease in patients with chest pain [
6], predict risk of ventricular arrhythmias [
22] and predict mortality [
4,
11]. The use of GLS in stress echocardiography increases diagnostic precision compared to LVEF and wall motion scoring, even for novice readers [
23]. Earlier studies have described inter- and intra-observer variability in LVEF [
24] and GLS outperforming LVEF [
4,
7]. In these studies, image readers were regarded as experts. However, despite several advantages of GLS compared to LVEF in clinical practice, LVEF is still the most used method for evaluation of LV systolic function. In order to use GLS in clinical practice it is important to know to how echocardiographic training may affect the analysis of GLS. We found that measurement of GLS by echocardiography in clinical practice is a highly reproducible method independent of echocardiographic training and significantly better reproducible than LVEF. There may be several reasons for this finding.
When measuring GLS and calculating LVEF there are numerous sources of error in both image acquisition and image analysis that may affect measurements and results. The sources of error include how each operator records and analyzes images offline. Level of echocardiographic training influence both image acquisition and image analysis and may potentially lead to high variability for both LVEF and GLS.
Obtaining images suitable for both LVEF and GLS analysis requires several technical considerations [
2]. In LVEF calculation, we need high-quality visualization of the endocardial border in both apical 4-chamber and 2-chamber views. The images should display LV cavity with minimum foreshortening. Timing of end systole and diastole is critical. End-diastole is defined as the first frame after mitral valve closure or the frame which LV dimension is the largest. [
25] End-systole is defined as the frame after aortic valve closure or the frame in which the cardiac dimension or volume is smallest. [
25] Error in these steps will lead to miscalculation of cavity volume and LVEF. As illustrated in Table
3, it seems that the expert echocardiographer generally was able to achieve better images for LVEF analysis. The differences in image quality and visibility of endocardial border may depend on factors as gain setting, focus depth and sector width. Small differences in cavity foreshortening and apical transducer rotation might introduce variability in volume calculation as well. Variability may increase with increasing LVEF due to larger variation in endocardial border between end diastole and end systole.
A major limitation of LVEF in patients with myocardial infarction is that the Simpson’s biplane method is based on an assumption of symmetric LV geometry. The presence of regional myocardial dysfunction as a result of myocardial infarction alters LV geometry. [
2] As a consequence, the Simpson’s biplane method by echocardiography may partly fail to measure LVEF with precision, and level of echocardiographic experience may affect how LVEF is measured. Correlation between GLS and LVEF is reported higher in healthy subjects than in patients with myocardial infarction and heart failure. [
26] Our study population is a heterogeneous group regarding ischemic myocardial pathology (Table
1) and degree of LV dysfunction (Table
4).
Table 4
Distribution of patients according to LVEF
Trainee echocardiographer | 11 (23.4) | 18 (38.3) | 18 (38.3) |
Expert echocardiographer | 8 (17.0) | 15 (31.9) | 24 (51.1) |
GLS does not rely on geometric assumptions but measures myocardial function with precision as we have demonstrated previously [
5,
10,
14,
17,
21]. Strain by speckle tracking measure directly segmental myocardial deformation of the LV in a 16-segment model. Average deformation of LV is expressed as GLS. LVEF describes LV systolic function indirectly on the basis of changes in calculated LV volume during the systole. In addition, GLS may be more sensitive than LVEF to changes in long-axis shortening, which makes GLS useful in evaluation of LV function where LVEF is preserved [
27,
28]. After the region of interest (ROI) is set in strain measurement, speckle tracking is performed automatically by the respective software.
Image acquisition for strain analysis by speckle tracking has different sources of error [
29]. Recognition and elimination of acoustic phenomena as reverberation and acoustic shadowing is important. Tracking of these phenomena will result in underestimation of true deformation [
29,
30]. Since strain by speckle tracking is essentially angle independent, this can to a certain degree be eliminated by adjusting probe position. The software performs automated speckle tracking frame by frame which imply that the frame rate needs to be optimized. A frame rate between 40 and 80 frames per second (FPS) is often recommended [
29,
30]. Low frame rate and tachycardia may result in undersampling where systolic events are missed, resulting in underestimation of true deformation [
30].
Images suitable for strain analysis may be easier to obtain than images for endocardial tracing. Manual tracing of the endocardium in two image planes may be performed with significant variability between observers due to differences in defining the endocardial border in both end diastole and end systole even in high quality images. Variability is influenced by differences in image acquisition. Since GLS on the other hand is a direct and objective measurement of myocardial deformation and function, this may reduce variability between echocardiographers.
Our findings demonstrate that GLS is a more reproducible parameter regardless of echocardiographic training and image quality compared to LVEF. Level of training is probably more important for LVEF calculation. Our results are similar to a study of Medvedkovsky et al. who addressed the same issue but with another vendor and software [
31] and of Negishi et al. [
8]. The findings in this study supports the use of GLS in clinical practice as an important supplement in describing LV function with low variability between observers even among echocardiographic trainees.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.