Inter‐ and intra‐observer variability in the echocardiographic evaluation of wall motion abnormality in patients with ST‐elevation myocardial infarction or takotsubo syndrome – A novel approach

Using existing transthoracic echocardiographic indices to quantify left ventricular wall motion abnormalities (WMAs) can be difficult due to the variations in the location of the abnormalities within the left ventricle, the quality of examinations, and the inter‐/intra‐observer variability of available indices. This study aimed to evaluate a new approach for measuring the extent of WMA by calculating the percentage of abnormal wall motion and comparing it to the wall motion score index (WMSI). The study also sought to assess inter‐ and intra‐observer variability.


INTRODUCTION
Non-invasive assessment of regional myocardial function is important to risk-stratify and treat patients with myocardial infarction and other cardiac disorders, and to compare treatment strategies in clinical trials. Transthoracic echocardiography (TTE) is a cost-effective and well-established tool to evaluate global and regional cardiac function.
Regional wall motion abnormality is defined as hypokinesia, akinesia, or as dyskinesia.
The most widely used TTE tool for quantifying wall motion abnormalities is the semi-quantitative wall motion score index (WMSI). 1 The WMSI is derived by dividing the left ventricle (LV) into 17 segments. Each segment is analyzed individually and given a score in a segmentation scheme that is supposed to reflect the coronary perfusion territories. The division of the LV into standardized segments has been shown to reduce inter-observer variability 2 and allows for comparison of WMSI across imaging modalities, including cardiac magnetic resonance, photon-emission computed tomography, and positron emission computed tomography. 3 However, because it relies on dividing the LV into segments and assigning the same score to the entire segment, WMSI could be influenced by the location of the wall motion abnormalities and could be less efficient in detecting more subtle differences in wall motion abnormalities. To allow for more detailed monitoring of myocardial wall motion abnormalities over time, we developed a novel echocardiographic approach for measuring the percentage of akinesia, proportion akinesia (PrA), and proportion akinesia/hypokinesia (PrAH) in the left ventricle. This method requires no extra equipment or collection of more images compared to other methods and can be part of normal routine evaluation.
Patients diagnosed with acute ST-elevation myocardial infarction (STEMI) or takotsubo syndrome (TS) may develop wall motion abnormalities in the acute phase due to acute occlusion of a coronary artery (STEMI), or triggered by stress (TS). In both conditions wall motion abnormality potentially resolve over time.
This study aimed to evaluate the inter-and intra-observer variability of PrA, PrAH, and WMSI in patients with STEMI or TS.

Patient selection
Patients without prior myocardial infarction or known pre-existing wall motion abnormalities who presented with STEMI or TS, accord- study, but no analysis was made between timepoints. All participants were examined in the left lateral recumbent position by a well-trained cardiac sonographer or physicians according to a standard protocol File (S1). All study-specific measurements were performed offline using images digitally stored on a custom workstation by a primary cardiac sonographer and a cardiologist.
The WMSI was defined as the mean of all the individual segments. 4

Specific training for assessment of proportion akinesia and proportion akinesia/hypokinesia
Two expert echocardiographers, termed Observer A and Observer B throughout the paper, underwent the following preparations. First, the two observers evaluated 20 echocardiograms in a joint session.
The observers then independently evaluated 20 randomly selected echocardiograms. After the calculation of the difference in the recorded PrA, PrAH, and WMSI for each of the 20 echocardiograms across observers, the two observers re-evaluated all 20 echocardiograms. This process was repeated twice before the study was started.
These 20 TTEs were only used for practice and were not included in the analysis.
To evaluate inter-and intra-observer variability, measurements of wall motion abnormality were initially made by the first observer.
Next, the second observer performed the same measurements in the same images blinded to the first observer's result. Finally, the first observer repeated the measurements again. All measurements were independently recorded and stored.

Statistical analysis
Cohen's kappa coefficient was used to compare agreement between the observers regarding the presence versus absence of akinesia.
Echocardiograms that were scored by both observers as having no akinesia or hypokinesia were excluded from the analysis that quantitively compared inter-and intra-observer variability for the extent of akinesia and/or akinesia or hypokinesia. To evaluate inter-and intra- PrA, PrAH, and WMSI between the two observers were determined using Pearson's correlation coefficient. The coefficient of variation (defined as the ratio between the standard deviation of the mean of measurements and the mean between observers in percent) was also used for the assessment of inter-and intra-observer variability. Results in the text are presented as mean ± SD.

Study population
A total of 140 TTEs from 54 patients were available for evaluation. Due to absence of any wall motion abnormality 30 TTEs were excluded before the analysis (Figure 3).
The Pearson correlation coefficient in PrA between the two observers was .986 ( Figure 4B). Intraclass coefficient for PrA was .993 with a 95% confidence interval from .989 to .995 (Table 3). When TTE performed in patients with STEMI or TS was analyzed for the presence of akinesia and/or hypokinesia (PrAH), PrAH between the two observers were 32.9 ± 14.3 and 27.5 ± 13.6, respectively. The mean difference in estimated PrAH was .47 ± 2.64 ( Table 3). The 95% limits of agreement for PrAH extended from −4.70 to 5.64 ( Figure 4C). The Pearson correlation coefficient between the two observers was .986 (Table 3, Figure 4D). The intraclass coefficient in PrAH was .972 with a confidence interval between .959 to .981. WMSI calculated by Observers A and B were 1.65 ± .31 and 1.62 ± .3, respectively ( Table 2). The mean difference in the estimated wall motion score index (WMSI) between the two observers was .03 ± .11. The 95% limits of agreement for WMSI extended from −.19 to .25 ( Figure 4E). The Pearson correlation coefficient for WMSI between the two observers was .934 (Table 3, Figure 4F). The intraclass coefficient was .965 and the 95% confidence interval extended between .949 and .976.

Intra-observer variability
In echocardiograms from patients with STEMI or TS the mean difference in estimated PrA between two separate evaluations by one   Figure 5A). The 95% limits of agreement for PrA extended between −4.96 and 6.31 ( Figure 5A). The Pearson correlation coefficient for PrA between two separate evaluations by one single observer was .972 (Table 3, Figure 5B). The intraclass coefficient in PrA was .986 with a 95% confidence interval between .976 and .992. The mean difference in estimated PrAH between two separate evaluations spaced at least 1 month apart was .30 ± 3.99 (Table 3, Figure 5C). The Pearson correlation coefficient for PrAH was .951 (Table 3, Figure 5D).
The 95% limits of agreement for PrAH extended between −7.52 and 8.12. The intraclass coefficient was .984 and the 95% confidence interval extended between .977 to .989 (Table 3). The mean difference in estimated WMSI between two separate evaluations spaced at least 1 month apart was −.06 ± .19. The 95% limits of agreement for WMSI extended from −.42 to .31 (Table 3, Figure 5E). The Pearson correlation coefficient for WMSI was .841 (Table 3, Figure 5F). The intraclass coefficient was .914 and the 95% confidence interval extended between .875 to .941 (Table 3).

DISCUSSION
The main finding in our study is that the myocardial wall motion abnormalities can be reproducibly quantified by PrA; a measurement that is available using standard TTE projections. This study shows that PrA can be acquired in a reproducible manner, with inter-and  PrA allows for a more detailed quantification of the extent of regional myocardial dysfunction. PrA is less dependent on excellent image quality than more advanced methods, such as speckle-tracking based methods, for assessing regional myocardial strain 8 ; and the agreement was greater in our study for PrA than what has typically been reported for strain in the literature, 7 especially when contrast agents are not used. 9 Whereas strain provides a more granular assessment of the severity of the wall motion abnormalities across regions, PrA and PrAH provides an easily interpretable measure for the extent of myocardial akinesia (or the extent of the myocardium with any WMA, i.e. affected myocardium). 8 The inter-and intra-observer variability for PrA in our study is similar to that reported for several of the most commonly used echocardiographic parameters in different patient groups 8

Study limitations
The measurements were made by experienced echocardiographers.
Echocardiography is user-dependent which could have implications for the quality of echocardiograms obtained and by extension it may have consequences for the reproducibility of analysed echocardiograms. All participants had good acoustic windows, resulting in acceptable image quality in all examinations. Examinations with poor image quality would be expected to result in a higher inter-and intra-observer variability. All participants had sinus-rhythm making it easier to identify wall motion abnormalities. In patients with different rhythm disorders any wall motion abnormality might be harder to identify and quantify. Most participants were examined in the left lateral decubitus position without the bedside approach. The bedside approach often complicates the ability to achieve eligible image quality and complicates wall motion abnormality identifications.

CONCLUSION
PrA obtained using standard TTE projections can quantify the extent of left ventricular wall motion abnormality with similar inter-and intraobserver agreement as WMSI.