- Email: [email protected]
Reproducibility of Combined Acquisition and Measurement of Left Ventricular Longitudinal Peak Segmental Strain in Relation to the Severity of Left Ventricular Dysfunction
Reproducibility of Combined Acquisition and Measurement of Left Ventricular Longitudinal Peak Segmental Strain in Relation to the Severity of Left Ventricular Dysfunction Andrei D. Margulescu, MD, PhD, Maria-Claudia-Berenice S¸uran, MD, and Dragos¸ Vinereanu, MD, PhD, FESC, Bucharest, Romania; and Swansea, United Kingdom
Background: Whether left ventricular (LV) longitudinal peak systolic segmental strain (LPSS) has sufficient reproducibility to be used in clinical practice (e.g., in patient follow-up) remains unclear. The aim of this study was to assess the reproducibility of combined acquisition and measurements of LPSS across the spectrum of LV ejection fraction (LVEF). Methods: In this prospective study, 72 subjects (mean age, 63 6 14 years; 65% men) were included in four equal groups: group 1, LVEF $ 50%, healthy; group 2, LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3, LVEF 30%–49%; and group 4, LVEF # 29%. Two observers performed four sets of image acquisitions and measurements (three during the same session, one after a median of 1 day) to account for intraobserver, interobserver, and test-retest reproducibility of combined acquisition and measurements. LPSS was measured in each of the 17 LV segments. Results: On average, the intraobserver and test-retest intraclass correlation coefficients and mean absolute differences of repeated acquisition and measurement of LPSS were similar across groups. However, interobserver intraclass correlation coefficients and mean absolute differences decreased in group 4 compared with groups 1 to 3. The intraobserver, test-retest, and interobserver coefficients of variation of all LV segments became worse as LVEF decreased, especially in group 4, in which LPSS was not reproducible in most segments. Reproducibility of LPSS in basal LV segments was worse compared with apical segments. The average measurement uncertainty (defined as the 95% limits of agreement of repeated acquisition and measurements) of LPSS in a test-retest scenario was 68.9%, 611.8%, 610.7%, and 69.0% in groups 1, 2, 3, and 4, respectively. Conclusions: The clinical applicability of LPSS is hindered by suboptimal reproducibility, even if a single observer repeats both acquisition and measurements. Changes in LPSS during patient follow-up should be interpreted with caution. (J Am Soc Echocardiogr 2019;32:1451-61.) Keywords: Segmental strain, Echocardiography, Reproducibility, Left ventricular dysfunction
Two-dimensional (2D) speckle-tracking echocardiographic (STE) imaging is especially suited to assess myocardial deformation1 and has been proposed as a robust imaging modality for the detection and quantification of severity of cardiac dysfunction.2 From the University of Medicine and Pharmacy ‘‘Carol Davila’’ Bucharest (A.D.M., M.-C.-B.S¸., D.V.); Department of Cardiology, University and Emergency Hospital of Bucharest (A.D.M., M.-C.-B.S¸., D.V.), Bucharest, Romania; Department of Cardiology, Morriston Hospital NHS Trust (A.D.M.), Swansea, United Kingdom rgulescu, MD, PhD, Department of Cardiology, Reprint requests: Andrei D. Ma Morriston Hospital NHS Trust, Swansea, SA6 6NL, United Kingdom (E-mail: [email protected]). 0894-7317/$36.00 Copyright 2019 by the American Society of Echocardiography. https://doi.org/10.1016/j.echo.2019.07.007
Two-dimensional STE indices of global left ventricular (LV) function (e.g., global longitudinal LV strain [GLS]) and indices of regional (segmental) LV function (e.g., longitudinal peak systolic segmental strain [LPSS]) have been suggested to allow early diagnosis of pathologic processes that may affect the left ventricle globally (e.g., anthracycline cardiomyopathy) or regionally (e.g., myocardial infarction, acute myocarditis).3,4 Some 2D STE indices of global LV function (e.g., GLS) have good reproducibility that may allow their use for patient follow-up. However, we have recently shown that the reproducibility of most 2D STE indices of global LV function decreases as the LVejection fraction (LVEF) becomes depressed, thus limiting the utility of 2D STE indices in patients with overt LV systolic dysfunction.5 Regarding regional LV function, recent data from the European Association of Cardiovascular Imaging (EACVI)/American Society of 1451
1452 Margulescu et al
Abbreviations
2D = Two-dimensional CV = Coefficient of variation GLS = Global longitudinal left ventricular strain ICC = Intraclass correlation coefficient LOA = Limits of agreement LPSS = Longitudinal peak systolic segmental strain
LV = Left ventricular LVEF = Left ventricular ejection fraction
MD = Mean absolute difference between repeated measurements
STE = Speckle-tracking echocardiographic
Journal of the American Society of Echocardiography November 2019
Echocardiography (ASE) Strain Standardization Task Force suggested that LPSS is less reproducible compared with GLS and that single LPSS values should be interpreted with caution in clinical practice.6,7 In these studies however, only the intraobserver reproducibility of combined acquisition and measurements was assessed, and even if the included subjects had a wide range of regional wall motion abnormalities, the average LVEF was within normal limits, thus limiting the number of patients with LV systolic dysfunction. Thus, the aim of this study was to assess the intraobserver, test-retest and interobserver reproducibility of combined acquisition and measurement of LPSS in relation to the degree of LV systolic dysfunction as
defined by LVEF.
METHODS Patients A brief summary of the study design has been published elsewhere.5 Here, we provide the protocol relating to the assessment of LPSS. Subjects were included prospectively and consecutively into four groups according to LVEF determined during the index echocardiographic evaluation and the presence of cardiovascular risk factors and/or overt cardiac pathology. Subjects were recruited from patients admitted electively to our hospital. Normal subjects were recruited from patients admitted for ablation of symptomatic supraventricular tachycardia who were otherwise healthy and from healthy colleagues who volunteered for the study. Those four groups covered the range from normal cardiac function, through subclinical LV dysfunction and heart failure with preserved LVEF, to moderate and severe systolic LV dysfunction. Inclusion criteria were as follows. Group 1: LVEF $ 50%, no structural abnormality on echocardiography, normal results on electrocardiography, no cardiovascular symptoms (except regular palpitations in patients admitted for ablation), and no history of hypertension, diabetes, or cardiovascular disease. Group 2: LVEF $ 50% and at least one of the following conditions: 1. documented hypertension and diabetes; 2. significant abnormality detected on the index echocardiographic evaluation: LV hypertrophy (defined as LV mass > 115 g/m2 in men and >95 g/m2 in women, calculated using the cube formula),8 concentric remodeling (defined as relative wall thickness > 0.42),8 more than mild aortic and/or mitral valve disease, or
LV regional wall motion abnormalities; and 3. established cardiovascular disease: documented coronary artery disease, previous myocardial infarction, history of coronary revascularization, peripheral or cerebrovascular vascular disease, hypertrophic cardiomyopathy, known significant (more than mild) valvular disease. Group 3: LVEF between 30% and 49%, regardless of medical history. Group 4: LVEF # 29%, regardless of medical history. Exclusion criteria were as follows: 1. any change in cardiovascular symptoms and/or treatment within the prior month before the index echocardiographic evaluation; 2. admission for acute cardiovascular reasons within the prior month before the index echocardiographic evaluation (e.g., acute myocardial infarction, acute heart failure); 3. irregular heart rhythm (e.g., atrial fibrillation conducted to the ventricles, frequent ventricular or supraventricular extra beats; patients with paced rhythms [atrial, ventricular, cardiac resynchronization devices] were allowed into the study if the rhythm was regular); 4. isolated right ventricular dilatation and/or dysfunction; and 5. poor echocardiographic images, including poor 2D speckletracking. We initially recruited 20 subjects in group 2 and 19 subjects each in the other groups. One subject was subsequently excluded from group 4 because of poor 2D speckle-tracking. To maintain an equal number of subjects in all groups, four additional subjects were randomly excluded from the other three groups. Thus, the final study population consisted of 18 subjects per group (72 subjects in total; mean age, 63 6 14 years; 65% men). The study was approved by the University and Emergency Hospital of Bucharest local ethics committee, and subjects gave written informed consent. Echocardiography All echocardiographic studies were acquired using a single machine (Vivid E9; GE Vingmed Ultrasound, Horten, Norway) equipped with a 2.5-MHz probe. Echocardiography was performed with the patient in the left lateral decubitus position. Chamber diameters were measured from the parasternal long-axis view, as recommended.8 For this study, 2D grayscale images from the apical four-chamber, two-chamber, and long-axis views were stored for offline analysis. Images covered at least three complete cardiac cycles and were optimized to obtain frame rates between 50 and 80 frames/sec, as recommended.9 Every effort was made to minimize apical foreshortening in the apical views.8,9 Analysis was performed offline using a single software version for all studies (EchoPAC version 113; GE Vingmed Ultrasound). LV volumes (end-diastolic and end-systolic) and LVEF were measured using the AutoEF option in EchoPAC.10 Full-thickness LPSS was calculated using the Automated Function Imaging option in EchoPAC.8 Zero strain was set at end-diastole (defined by the R wave on the electrocardiogram). End-systole was defined by the timing of the aortic valve closure time from the apical long-axis view. Then, a region of interest was selected to contain only the LV
Margulescu et al 1453
Journal of the American Society of Echocardiography Volume 32 Number 11
HIGHLIGHTS LPSS has suboptimal reproducibility. Reproducibility of LPSS in basal LV segments was worse compared with apical segments. The reproducibility of LPSS became worse as the ejection fraction decreased. Changes in LPSS during patients’ follow-up should be interpreted with caution. myocardium (without the pericardium or the LV cavity). The automatic tracking software tracked the myocardial speckles throughout the cardiac cycle. The results were visually inspected, and if necessary, the region of interest was adjusted manually until the tracking was judged to be optimal (i.e., following closely the movement of the underlying myocardium). We excluded segments from analysis if the tracking was not optimal, and we completely excluded the patient from the study if more than two segments in one view were not tracked optimally. Then the LPSS (defined as the maximal strain value, either positive or negative, that occurs before the aortic valve closure) was measured in each LV myocardial segment, as given by Automated Function Imaging, using the 17-segment model (corresponding to six basal and six midventricular segments [septal anterior, anterior, lateral, posterior, inferior, septal posterior], four apical segments [septal, anterior, lateral, inferior], and one apical cap segment). For easier understanding, we use the absolute value for strain in our discussion (i.e., higher strain indicates higher shortening of the myocardial segment). Reproducibility Two observers, each with >5 years of experience in performing and analyzing 2D STE data, performed four sets of image acquisitions and five sets of measurements, in all subjects. Joint sessions of image acquisitions and analysis of 2D STE indices were performed by the two observers before the beginning of the study to minimize variability. In particular, we paid special attention to use similar gain settings (set at medium), sector widths, and frame rates when acquiring images. The first three sets of image acquisitions were performed in sequence during the same session, with a 5-min break between them. The first two acquisitions were performed by the first observer, and the third acquisition was performed by the second observer. The first echocardiogram of the sequence was used to define LVEF and presence or absence of cardiac abnormalities to include or exclude the subject in or from one of the study groups. The patient maintained the same lateral decubitus position between sessions. The observers were blinded to each other’s acquisition. The first observer repeated image acquisition after 1 day in all but three patients (in whom repeated acquisition was performed after 3 days; test-retest). Images were stored on a secure offline server. The observers performed offline measurements on their own acquisitions, on separate days and in random order, blinded to each other’s data. Thus, we assessed the intraobserver, test-retest, and interobserver reproducibility of combined acquisition and measurements of 2D STE indices. In addition, we also evaluated the interobserver reproducibility of 2D STE measurements only, where observer 1 repeated measurements on the acquisition performed by observer 2.
Statistical Analysis Statistical analysis was performed using SPSS version 20 (SPSS, Chicago, IL). Quantitative data are presented as mean 6 SD. Multiple parametric values were compared using one-way analysis of variance, applying the Bonferroni correction. P values # .05 were considered to indicate statistical significance. Reproducibility is expressed as intraclass correlation coefficient (ICC; two-way mixed effect, absolute agreement model), mean absolute difference between repeated measurements (MD), coefficient of variation (CV), and measurement uncertainty. The MD was calculated as follows: Pn MD ¼
i¼1 ðjmeasurement 1
n
measurement 2jÞ
;
where n is the number of patients. Because LPSS is expressed as a percentage, the MD of LPSS is also expressed as a percentage. The CV was calculated as follows: CV ¼ SD=ðarithmetic mean of measurementsÞ 100; where SD is the standard deviation of the measurement error associated with a single measurement, calculated as the SD of residuals divided by O2. Measurement uncertainty was defined as the 95% limits of agreement (LOA) of repeated acquisition and measurements of LPSS.11 Bland-Altman analysis was used to estimate the bias and 95% LOA between paired measurements.
RESULTS Baseline and echocardiographic characteristics of the study groups are displayed in Tables 1 and 2, respectively. Subjects in group 2 were characterized by the presence of significant concentric LV remodeling compared with group 1 (smaller LV volumes, higher relative wall thickness). Frame rates of the acquired images were similar between observers (57.0 6 3.3 vs 56.2 6 2.9 frames/sec, P = .15) and between repeated acquisitions performed on different days by the first observers (57.1 6 3.4 vs 56.9 6 3.3 frames/sec, P = .75). Heart rate and blood pressure also remained stable between repeated evaluations performed on different days (heart rate: 71.7 6 13.6 vs 71.8 6 13.7 beats/min [P = .87]; systolic blood pressure: 127 6 16 vs 124 6 12 mm Hg [P = .20]; diastolic blood pressure: 74 6 10 vs 75 6 9 mm Hg [P = .48]). LPSS values for each LV segment in all groups are displayed in Figure 1. The average intraobserver, test-retest, and interobserver ICCs and 95% LOA in all four groups are displayed in Figure 2, while the average intraobserver, test-retest, and interobserver MDs and CVs in all four groups are displayed in Table 3. The ICCs, MDs, and CVs in individual LV segments are displayed in Supplemental Figures 1 to 3. When all segments were considered together, the intraobserver and test-retest ICCs, MDs, and 95% LOA of repeated acquisition and measurement of LPSS remained relatively unchanged throughout the spectrum of LVEF. However, interobserver ICCs, MDs, and 95% LOA were worse in patients with severely depressed LVEFs (group 4) compared with groups 1 to 3 (Table 3, Figure 2).
1454 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Table 1 Baseline characteristics of the study groups Variable
Group 1 (n = 18)
Group 2 (n = 18)
Group 3 (n = 18)
Group 4 (n = 18)
P
Demographic data Age, y
48.6 6 13.8
67.2 6 12.8
67.7 6 7.8
66.7 6 11.6
Sex, male, %
9 (50)
10 (56)
13 (72)
15 (83)
1.80 6 0.22
1.84 6 0.20
1.80 6 0.24
<.001* .13†
BSA, m
1.96 6 0.22
Hypertension
—
16 (89)
16 (89)
7 (39)
<.001†
Diabetes
—
3 (17)
5 (28)
4 (22)
.13†
Ischemic heart disease
—
7 (39)
9 (50)
7 (39)
.026†
Previous myocardial infarction
—
3 (17)
8 (44)
7 (39)
.007†
Peripheral or cerebrovascular vascular disease
—
1 (6)
2 (11)
0 (0)
.38†
Moderate or severe MR
—
0 (0)
4 (22)
9 (50)
.005†
Moderate or severe AS
—
0 (0)
0 (0)
1 (6)
.39†
Aspirin
—
7 (39)
12 (67)
13 (72)
.09†
Oral anticoagulants
—
9 (50)
5 (28)
5 (28)
.27†
b-blockers
—
14 (78)
13 (72)
17 (94)
.20†
ACE inhibitors/ARBs
—
13 (72)
16 (89)
8 (44)
.015†
Statin
—
16 (89)
13 (72)
10 (56)
.08†
Diuretics
—
8 (44)
13 (72)
18 (100)
.001†
2
.11*
Treatment
ACE, Angiotensin-converting enzyme; ARB, angiotensin receptor blocker; AS, aortic stenosis; BSA, body surface area; MR, mitral regurgitation. Data are expressed as mean 6 SD or as number (percentage). *One-way analysis of variance. † Chi-square test.
The ICCs, MDs, and CVs in individual LV segments in each group are displayed in Supplemental Figures 1 to 3. On average, LPSS decreased from group 1 to 4, while the MDs remained either
constant (for intraobserver and test-retest) or decreased in group 4 (for interobserver). As a result, the intraobserver, test-retest, and interobserver CVs (for combined acquisition and
Table 2 Echocardiographic characteristics of the study groups P value (one-way ANOVA with Bonferroni correction) Variable
Group 1 (n = 18)
Group 2 (n = 18)
Group 3 (n = 18)
Group 4 (n = 18)
Group 1 Group 1 Group 1 Group 2 Group 2 Group 3 vs group 2 vs group 3 vs group 4 vs group 3 vs group 4 vs group 4
General echocardiographic data Frame rate, frames/sec
58.4 6 3.5
57.6 6 3.3
57.2 6 3.3
54.6 6 1.8
$.999
$.999
.002
$.999
.022
.06
Heart rate, beats/min
71.6 6 12.5
64.7 6 7.9
71.3 6 16.2
79.4 6 13.0
.64
$.999
.43
.75
.006
.37
IVS, mm
8.8 6 1.2
12.9 6 1.8
12.8 6 1.5
11.0 6 2.0
<.001
<.001
.001
$.999
.004
.011
LVPW, mm
8.3 6 1.1
11.0 6 1.1
11.1 6 1.3
10.4 6 1.5
<.001
<.001
<.001
$.999
.92
.56
LVEDD, mm
49.5 6 5.1
42.9 6 5.9
52.8 6 6.8
66.1 6 9.7
.040
$.999
<.001
.001
<.001
<.001
LV indexed mass, g/m2
74.7 6 15.6 103.4 6 24.3 138.4 6 27.2 176.2 6 31.6
.007
<.001
<.001
.001
<.001
<.001
RWT
0.35 6 0.05
0.57 6 0.12
0.46 6 0.10
0.33 6 0.08
<.001
.002
$.999
.004
<.001
<.001
LA diameter at end-systole, mm
38.6 6 5.3
40.0 6 6.2
41.5 6 5.0
44.1 6 7.3
$.999
.93
.05
.23
$.999
$.999
LVEDV, mL
115.3 6 32.0
88.9 6 29.9 146.7 6 32.7 275.8 6 108.9
$.999
.76
<.001
.035
<.001
<.001
LVESV, mL
48.6 6 15.7
36.9 6 15.3
88.6 6 27.5 219.9 6 89.6
$.999
.09
<.001
.012
<.001
<.001
LVEF, %
58.5 6 4.3
59.2 6 6.3
37.8 6 6.5
$.999
<.001
<.001
<.001
<.001
<.001
20.6 6 4.3
ANOVA, Analysis of variance; IVS, interventricular septal thickness at end-diastole; LA, left atrial; LVEDD, LV end-diastolic diameter; LVEDV, LV end-diastolic volume; LVESV, LV end-systolic volume; LVPW, posterior wall thickness at end-diastole; RWT, relative wall thickness. Data are expressed as mean 6 SD.
Journal of the American Society of Echocardiography Volume 32 Number 11
Margulescu et al 1455
Figure 1 Polar map (bull’s-eye view) of LV longitudinal peak systolic strain for each segment of the LV wall (using a 17-segment model) in the four study groups. Values represent strain (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%. LV walls are displayed in the first bull’seye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
measurements) and the interobserver CVs for measurements only of all LV segments became worse as LVEF decreased. This was especially notable in group 4 (LVEF # 29%), in which LPSS was not reproducible in most segments (Table 3, Supplemental Figure 3). Overall, the intraobserver and test-retest ICCs and MDs of LPSS were marginally worse in basal versus apical segments; also the CVs decreased from basal to apical segments. This tendency was much more evident for interobserver ICCs, MDs, and CVs, for which the reproducibility of LPSS clearly decreased from apical to basal segments. Similar tendencies were observed when analyzing groups separately (Table 4).
On the basis of the 95% LOA between paired acquisitions and measurements, the average measurement uncertainty of LPSS in a scenario in which a single observer repeats both echocardiographic acquisition and measurements during patient follow-up was 68.9%, 611.8%, 610.7%, and 69.0% in groups 1, 2, 3, and 4, respectively. An estimation of the measurement uncertainty of LPSS assessment in each LV segment and in each study group in a similar scenario is given in Figure 3. An estimation of the measurement uncertainty of LPSS in basal, mid, and apical LV segments for each group is given in Table 5. We hypothesized that the worse reproducibility of basal versus apical segments was related to slight differences in the acquisition of
1456 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Figure 2 Correlation (top) and Bland-Altman plots (bottom) of repeated measurements in the four study groups. ICCs with 95% CIs and bias with 95% LOA of repeated measurements are also displayed.
apical images by the two observers, so we expected that this should have been reflected by a bias in measuring LV end-diastolic and end-systolic volumes. Indeed, there was no significant bias for intraob-
server and test-retest acquisition and measurements of LV volumes (end-diastolic, 1.3 and 3.2 ml; end-systolic, 1.0 and 1.3 ml), but there was a significant bias in inter-observer assessment of LV volumes
Margulescu et al 1457
Journal of the American Society of Echocardiography Volume 32 Number 11
Table 3 Mean and SD of MD and CV of LV LPSS in each of the 17 LV segments in all study groups P value (one-way ANOVA with Bonferroni correction) Group 1 (n = 18)
Group 2 (n = 18)
Group 3 (n = 18)
Group 4 (n = 18)
Group 1 vs group 2
Group 1 vs group 3
Group 1 vs group 4
Group 2 vs group 3
Group 2 vs group 4
Group 3 vs group 4
MD, % Intraobserver
2.9 6 1.0 2.7 6 0.7 3.0 6 1.3
3.1 6 0.9
$.999
$.999
$.999
$.999
$.999
$.999
Test-retest
2.8 6 0.7 3.7 6 1.2 3.7 6 1.4
3.6 6 1.4
.26
.29
.41
$.999
$.999
$.999
Interobserver
3.4 6 0.8 3.9 6 1.0 4.0 6 0.8
4.1 6 1.6
.94
.60
.38
$.999
$.999
$.999
Interobserver 2.8 6 0.8 3.3 6 0.9 3.8 6 1.3 (measurements only)
4.3 6 2.2
$.999
.25
.017
$.999
.31
$.999
CV, % Intraobserver
14 6 9
17 6 8
27 6 17
263 6 645
$.999
$.999
.17
$.999
.18
.22
Test-retest
14 6 5
24 6 12
33 6 20
289 6 649
$.999
$.999
.10
$.999
.12
.15
Interobserver
16 6 7
21 6 8
33 6 11
113 6 82
$.999
$.999
<.001
$.999
<.001
<.001
Interobserver (measurements only)
12 6 4
17 6 5
35 6 15
213 6 341
$.999
$.999
.006
$.999
.008
.021
Data are expressed as mean 6 SD.
(acquisition and measurement and measurement only; end-diastolic, 16.7 and 13.3 ml; end-systolic, 25.9 and 21.5 ml). An illustrative example of the difference between LPSS values from repeated acquisition and measurements in a single patient is given in Figure 4.
DISCUSSION In this study, we assessed the intraobserver, test-retest and interobserver reproducibility of combined acquisition and measurement of LPSS in four groups ranging from normal subjects to patients with severe LV systolic dysfunction. The main results of this study suggest that the clinical applicability of LPSS is hindered by suboptimal reproducibility, even if a single observer repeats both acquisition and measurements. This is especially noticeable in basal LV segments and in patients with LV systolic dysfunction. The lower interobserver reproducibility of basal and mid segments compared with intraobserver and test-retest reproducibility may be related to slight differences in the window and angle of insonation when acquiring apical images. Thus, changes in LPSS during patient follow-up should be interpreted with caution. Unlike GLS, for which most of the studies revealed good reproducibility across the spectrum of LVEF,5 data on the clinical utility and reproducibility of LPSS have been more conflicting. Initial studies suggested that LPSS has good reproducibility.12,13 For example, Shiino et al.12 assessed the reproducibility of LPSS in 55 patients, using the same equipment as in our study, but analyzed LPSS reproducibility according to presumed coronary artery myocardial distribution and did not include the acquisition process in their analysis (measurements only), making their data less relevant to clinical practice. In a study of 290 subjects (two thirds of whom had various cardiac conditions), Barbier et al.13 reported that LPSS has good or moderate testretest reproducibility depending on segment (CV ranging from 8% to 23%); they concluded that LPSS is ‘‘suitable for diagnosis and follow-up of LV regional systolic function.’’ However, the mean reported LVEF was near normal (52 6 16%).
More recently, the EACVI/ASE Strain Standardization Task Force published a series of articles assessing the variability and reproducibility of LPSS acquisition and measurements, for all current major echocardiographic machine manufacturers, including GE.6,7,14 The authors showed that the ICC was 0.90 (95% CI, 0.89–0.91), and the MD was 3.0 6 2.9% for the pooled analysis of LPSS in all segments. The average MD for individual segments ranged from 1.7 6 1.3% (basal inferior septum) to 4.5 6 3.9% (anterior apex).6 Also, the basal LV segments had the lowest reproducibility, while the apical LV segments had the highest reproducibility of LPSS. The results reported in the articles published by the EACVI/ASE task force are comparable with the results presented here if one takes into account some important differences between these studies. The EACVI/ASE task force defined ‘‘test-retest’’ reproducibility on the basis of repeated acquisitions and measurements performed sequentially, during the same echocardiographic session, by a single observer. This is similar to what we defined herein as ‘‘intraobserver reproducibility of combined acquisition and measurements’’; in our article, we define ‘‘test-retest’’ reproducibility on the basis of repeated acquisition and measurements on different days. This distinction is critical for the correct interpretation of the data; indeed, our results on ‘‘intraobserver’’ reproducibility are similar with the ‘‘test-retest’’ reproducibility previously published by Mirea et al.6 However, we think that the protocol and definition used in our study are more relevant for clinical practice. For example, if LPSS, or any other echocardiographic parameter, is to be used in a patient’s follow-up, one is more interested in how stable and reliable measurements are when images are reacquired after a period of time (under reasonably similar physiologic conditions). A challenge in such situations is minimizing the potential bias that would result from differences in the acquisitions. In this more clinically relevant scenario, the reproducibility of LPSS was worse compared with that previously reported.6 In addition, even if the subjects analyzed by the EACVI/ASE task force had a wide range of segmental and functional abnormalities (on average, there were 5 6 3 transmural scarred segments per patient, as defined by cardiac magnetic resonance imaging) the average LVEF in the study was 52 6 10%. Thus, similar to Barbier et al.,12 the number
1458 Margulescu et al
Table 4 Reliability and reproducibility of LV peak segmental strain in basal, mid, and apical segments in all patients combined and in each study group separately All patients
Group 1
Group 2
Group 3
Group 4
Basal
Mid
Apical
Basal
Mid
Apical
Basal
Mid
Apical
Basal
Mid
Apical
Basal
Mid
Apical
0.90 (0.88– 0.92)
0.94 (0.93– 0.95)
0.94 (0.93– 0.95)
0.78 (0.68– 0.85)
0.73 (0.61– 0.82)
0.73 (0.58– 0.82)
0.68 (0.54– 0.78)
0.79 (0.69– 0.85)
0.81 (0.72– 0.88)
0.85 (0.79– 0.90)
0.86 (0.80– 0.91)
0.93 (0.90– 0.96)
0.85 (0.78– 0.90)
0.85 (0.78– 0.90)
0.63 (0.42– 0.76)
Intraobserver ICC (95% CI)
MD, %
3.2
2.6
3.0
2.8
2.5
3.3
3.0
2.5
2.9
3.6
2.7
2.2
3.5
2.7
3.4
CV, %
30.5
20.6
17.9
19.8
12.6
12.8
22.2
13.7
13.2
39.9
26.4
15.7
87.3
90.5
62.5
0.83 (0.79– 0.86)
0.92 (0.90– 0.93)
0.93 (0.91– 0.94)
0.82 (0.73– 0.87)
0.56 (0.12– 0.75)
0.63 (0.42– 0.76)
0.32 (0.01– 0.53)
0.42 (0.15– 0.60)
0.77 (0.66– 0.85)
0.70 (0.56– 0.80)
0.86 (0.80– 0.91)
0.79 (0.68– 0.86)
0.76 (0.65– 0.84)
0.73 (0.60– 0.82)
0.71 (0.54– 0.81)
Test-retest ICC (95% CI)
MD, %
4.1
3.2
3.1
2.9
2.4
3.3
4.6
3.6
3.1
4.5
2.9
2.9
4.2
3.6
2.9
CV, %
38.5
24.9
18.7
16.9
12.2
14.0
34.1
20.5
13.4
51.5
25.3
22.7
105.2
136.2
52.1
0.77 (0.73– 0.81)
0.88 (0.86– 0.90)
0.93 (0.90– 0.95)
0.48 (0.24– 0.65)
0.25 (0.06– 47)
0.77 (0.64– 0.85)
0.14 (–0.25– 0.42)
0.48 (0.24– 0.65)
0.76 (0.62– 0.85)
0.63 (0.46– 0.75)
0.68 (0.53– 0.78)
0.76 (0.60– 0.85)
0.55 (0.35– 0.69)
0.58 (0.39– 0.72)
0.70 (0.52– 0.81)
Interobserver ICC (95% CI)
4.8
4.1
3.3
4.0
3.7
3.1
4.7
3.8
3.9
4.7
4.1
3.7
5.6
4.6
2.5
CV, %
38.9
27.3
17.9
23.0
16.8
11.1
30.8
19.1
14.5
46.7
30.8
25.6
106.2
121.3
46.2
0.80 (0.75– 0.83)
0.88 (0.85– 0.90)
0.96 (0.93– 0.97)
0.56 (0.36– 0.70)
0.33 (0.04– 0.54)
0.88 (0.80– 0.92)
0.54 (0.34– 0.69)
0.52 (0.30– 0.67)
0.83 (0.67– 0.91)
0.54 (0.33– 0.69)
0.61 (0.44– 0.74)
0.88 (0.76– 0.94)
0.46 (0.20– 0.63)
0.46 (0.21– 0.63)
0.88 (0.81– 0.93)
Interobserver measurements only ICC (95% CI)
MD, %
4.4
4.0
2.5
3.2
3.0
2.3
3.4
3.7
3.1
4.7
4.1
2.5
6.3
5.2
1.7
CV, %
35.1
27.5
13.6
16.6
13.9
8.6
19.8
18.2
12.2
48.5
33.4
17.8
109.3
126.4
26.5
Journal of the American Society of Echocardiography November 2019
MD, %
Margulescu et al 1459
Journal of the American Society of Echocardiography Volume 32 Number 11
Figure 3 Estimation of measurement uncertainty of LV LPSS in the four study groups. The polar map (bull’s-eye view) displays the 95% LOA of repeated acquisitions and measurements on different days, by the same observer, of LPSS in each segment of the LV wall (using a 17-segment model), equivalent to a best-case clinical scenario in which a single echocardiographer follows patients. Values represent strain (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%. LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
Table 5 Estimation of measurement uncertainty (95% LOA between repeated acquisitions and measurements) of LV peak systolic strain in basal, mid, and apical LV segments All patients Basal
Mid
Group 1 Apical
Basal
Mid
Group 2 Apical
Basal
Mid
Group 3 Apical
Basal
69.9 63.6 68.2
69.5
Mid
Group 4 Apical
Basal
Apical
69.7
67.4 68.0
69.4 66.8 68.1
Test-retest, %
612.1
68.8 68.4
67.1 66.5 68.6 614.5 65.2 68.5 613.1
67.8 68.1 611.9 610.1 67.8
Interobserver, %
612.8 610.1 68.3 611.1 69.3 67.3 613.6 65.0 69.3 612.0
69.7 69.4 614.2 611.6 66.9
Intraobserver, %
Interobserver 617.9 610.3 66.4 (measurements only), %
67.9 65.5 610.0
Mid
67.8 69.6
68.9 67.7 65.7 611.3 64.8 67.8 621.5 610.6 66.6 612.3 612.8 64.2
1460 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Figure 4 An example of repeated acquisition and measurements of LV LPSS in a single patient from group 1. Individual LPSS curves taken from the three apical views (four chamber [A4c], two chamber [A2c], and long-axis [APLAX]) are displayed in the top three rows, while the 17-segment polar maps (bull’s-eye view) are displayed in the bottom row. Each column represents a sequence of acquisition and measurements (three performed by observer 1, one performed by observer 2). One can appreciate that between evaluations, LPSS in basal segments displays larger intraobserver, test-retest, and interobserver variability, compared with apical segments because the measured LPSS in the basal segments is more sensitive compared with apical segments to slight changes in the angle of insonation.
of included patients with severe systolic dysfunction and the number of segments showing very low or positive strain (dyskinetic segments) were relatively small. Also, the EACVI/ASE task force evaluated the reproducibility of endocardial and midmyocardial LPSS, while we measured full-thickness LPSS. However, we believe that this distinction is unlikely to be relevant for reproducibility assessment, as previous data demonstrated that the test-retest variability of layer-specific GLS is similar at different myocardial levels.14 In our study, we provide data on the reproducibility of LPSS throughout the spectrum of LVEF, which is important for clinical practice and especially for patient follow-up. We recently showed that GLS might have acceptable reproducibility (test-retest CV < 15%) even in patients with severe LV systolic dysfunction.5 GLS is calculated by averaging all values of LPSS, but in this study we have shown that LPSS has suboptimal reproducibility, especially in patients with poor LVEFs. Moreover, the 95% LOA for LPSS are too wide to allow interpretation of changes between evaluations even if both acquisitions and measurements are performed by the same echocardiographer, regardless of LVEF. It becomes apparent that large differences of individual LPSS are attenuated by averaging all values when GLS is calculated. Indeed, Barbier et al.13 showed that LPSS variability correlates with minor changes in scan angulation, which is inevitable between scans. Changes in scan angulation between eval-
uations also help explain why LPSS is more reproducible at apical segments (which are relatively fixed) compared with basal segments. Thus, changes in LPSS should be interpreted with caution, in the context of changes in GLS and the pattern of change in the polar map (bull’s-eye) displayed by the Automated Function Imaging software. Clinical Perspective Several articles have suggested that LPSS may allow early diagnosis of global and regional LV pathological processes (e.g., anthracycline cardiomyopathy, myocardial infarction, acute myocarditis).3,4 This has also been initially supported by reported good reproducibility of LPSS. However, recent data suggest that LPSS cannot detect isolated segmental scarred segments (surrounded by normal myocardium, as defined by cardiac magnetic resonance imaging).7 Similarly, recent studies showed that the reproducibility of LPSS may not be adequate for clinical application.6 We confirm these latest results, and we also provide useful estimates of the expected measurement uncertainty for repeated assessments of LPSS in scenarios resembling ideal clinical practice (i.e., a single observer performing both acquisition and measurements during patient follow-up). These results confirm that assessment of LPSS cannot be reliably used in clinical practice.
Margulescu et al 1461
Journal of the American Society of Echocardiography Volume 32 Number 11
In the future, the shortcoming of assessing 2D LPSS may be overcome by full-volume three dimensional LV strain analysis. It would be expected that full-volume three-dimensional strain would allow assessment of LPSS with less dependency on the angle of insonation and slight changes in apical window acquisition, which may improve variability. Until then, current efforts to standardize deformation imaging among vendors are welcome to homogenize the reported range of values of deformation parameters but are unlikely to result in an improvement of clinical applicability of LPSS. Limitations of the Study The number of patients in each group was relatively small. This was a consequence of the design of the study, in which a single echocardiographic machine, probe, and software were used to eliminate the well-described variability induced by different machines, probes, and software versions.6,7,15 However, our results are in line those of with other recent studies,6,7 so we believe that our study provides useful insights regarding the reproducibility of LPSS and its impact on clinical practice when image acquisition is also taken into account and when other potential sources of variation are minimized. We used a single vendor. However, for LPSS, this vendor has shown higher fidelity of the 2D speckle-tracking algorithm in following LV myocardial segmental deformation and lower variability of LPSS measurements compared with other vendors. Thus, the results are unlikely to be improved by using other current vendors and software packages.6 We did not assess the reproducibility of detecting transmural scar by LPSS. However, previous studies have already shown that LPSS may detect transmural scar only when more than two neighboring segments are affected.7
2.
3.
4.
5.
6.
7.
8.
9.
CONCLUSION 10.
In our study, the clinical applicability of LPSS was hindered by suboptimal reproducibility, especially in basal LV segments, and in patients with LV systolic dysfunction, even if a single observer repeated both acquisition and measurements. The lower interobserver reproducibility of basal and mid segments compared with intraobserver and test-retest reproducibility may have been related to slight differences in the window and angle of insonation when apical images were acquired. Thus, changes in LPSS during patient follow-up should be interpreted with caution.
11.
12.
13.
SUPPLEMENTARY DATA Supplementary data to this article can be found online at https://10. 1016/j.echo.2019.07.007.
14.
REFERENCES 15. 1. Mor-Avi V, Lang RM, Badano LP, Belohlavek M, Cardim NM, Derumeaux G, et al. Current and evolving echocardiographic techniques for the quantitative evaluation of cardiac mechanics: ASE/EAE consensus
statement on methodology and indications endorsed by the Japanese Society of Echocardiography. Eur J Echocardiogr 2011;12:167-205. Potter E, Marwick TH. Assessment of left ventricular function by echocardiography: the case for routinely adding global longitudinal strain to ejection fraction. JACC Cardiovasc Imaging 2018;11:260-74. Sjøli B, Ørn S, Grenne B, Ihlen H, Edvardsen T, Brunvand H. Diagnostic capability and reproducibility of strain by Doppler and by speckle tracking in patients with acute myocardial infarction. JACC Cardiovasc Imaging 2009;2:24-33. Kostakou PM, Kostopoulos VS, Tryfou ES, Giannaris VD, Rodis IE, Olympios CD, et al. Subclinical left ventricular dysfunction and correlation with regional strain analysis in myocarditis with normal ejection fraction. A new diagnostic criterion. Int J Cardiol 2018;259:116-21. Margulescu AD, Suran MCB, Vinereanu D. Do 2-dimensional speckle tracking indexes have sufficient reproducibility to be used in clinical practice when image acquisition and severity of LV dysfunction are taken into account? JACC Cardiovasc Imaging 2019;12:756-9. Mirea O, Pagourelias ED, Duchenne J, Bogaert J, Thomas JD, Badano LP, et al., EACVI-ASE-Industry Standardization Task Force. Variability and reproducibility of segmental longitudinal strain measurement: a report from the EACVI-ASE Strain Standardization task force. JACC Cardiovasc Imaging 2018;11:15-24. Mirea O, Pagourelias ED, Duchenne J, Bogaert J, Thomas JD, Badano LP, et al., EACVI-ASE-Industry Standardization Task Force. Intervendor differences in the accuracy of detecting regional functional abnormalities: a report from the EACVI-ASE Strain Standardization task force. JACC Cardiovasc Imaging 2018;11:25-34. Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr 2015;28:1-39.e14. Voigt JU, Pedrizzetti G, Lysyansky P, Marwick TH, Houle H, Baumann R, et al. Definitions for a common standard for 2D speckle tracking echocardiography: consensus document of the EACVI/ASE/Industry task force to standardize deformation imaging. J Am Soc Echocardiogr 2015;28: 183-93. Szulik M, Pappas CJ, Jurcut R, Magro M, Peeters E, Goetschalckx K, et al. Clinical validation of a novel speckle-tracking-based ejection fraction assessment method. J Am Soc Echocardiogr 2011;24:1092-100. Working Group 1 of the Joint Committee for Guides in Metrology. Evaluation of measurement data—guide to the expression of uncertainty in measurement. Available at: https://www.bipm.org/utils/common/ documents/jcgm/JCGM_100_2008_E.pdf. Accessed July 22, 2019. Shiino K, Yamada A, Ischenko M, Khandheria BK, Hudaverdi M, Speranza V, et al. Intervendor consistency and reproducibility of left ventricular 2D global and regional strain with two different high-end ultrasound systems. Eur Heart J Cardiovasc Imaging 2017;18:707-16. Barbier P, Mirea O, Cefal u C, Maltagliati A, Savioli G, Guglielmo M. Reliability and feasibility of longitudinal AFI global and segmental strain compared with 2D left ventricular volumes and ejection fraction: intraand inter-operator, test-retest, and inter-cycle reproducibility. Eur Heart J Cardiovasc Imaging 2015;16:642-52. € u S, Mirea O, Duchenne J, Pagourelias ED, Bezy S, Thomas JD, et al. Unl€ Comparison of feasibility, accuracy, and reproducibility of layer-specific global longitudinal strain measurements among five different vendors: a report from the EACVI-ASE Strain Standardization task force. J Am Soc Echocardiogr 2018;31:374-80.e1. Costa SP, Beaver TA, Rollor JL, Vanichakarn P, Magnus PC, Palac RT. Quantification of the variability associated with repeat measurements of left ventricular two-dimensional global longitudinal strain in a real-world setting. J Am Soc Echocardiogr 2014;27:50-4.
1461.e1 Margulescu et al
Journal of the American Society of Echocardiography November 2019
APPENDIX
Supplemental Figure 1 Polar map (bull’s-eye view) of ICCs of LV LPSS for each segment of the LV wall (using a 17-segment model) when image acquisition and measurement were considered (intraobserver, test-retest, and interobserver) and compared with interobserver reproducibility of 2D STE measurements only on the same image data set. ICCs $ 0.90 are displayed in green, ICCs < 0.90 and $0.75 in yellow, ICCs < 0.75 and $0.50 in red, ICCs < 0.50 and $0.25 in blue, and ICCs < 0.25 in black. Values are ICCs. Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%). LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
Journal of the American Society of Echocardiography Volume 32 Number 11
Margulescu et al 1461.e2
Supplemental Figure 2 Polar map (bull’s-eye view) of MDs of LV LPSS for each segment of the LV wall (using a 17-segment model) when image acquisition and measurement were considered (intraobserver, test-retest, and interobserver) and compared with interobserver reproducibility of 2D STE measurements only on the same image data set. Values are MDs (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%. LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
1461.e3 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Supplemental Figure 3 Polar map (bull’s-eye view) of CVs of LV LPSS for each segment of the LV wall (using a 17-segment model) when image acquisition and measurement were considered (intraobserver, test-retest, and interobserver) and compared with interobserver reproducibility of 2D STE measurements only on the same image data set. CVs # 10 are displayed in green, CVs > 10 and #20 in yellow, CVs > 20 and #40 in red, CVs > 40 and #60 in blue, and CVs > 60 in black. Values represent CVs (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%). LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
Background: Whether left ventricular (LV) longitudinal peak systolic segmental strain (LPSS) has sufficient reproducibility to be used in clinical practice (e.g., in patient follow-up) remains unclear. The aim of this study was to assess the reproducibility of combined acquisition and measurements of LPSS across the spectrum of LV ejection fraction (LVEF). Methods: In this prospective study, 72 subjects (mean age, 63 6 14 years; 65% men) were included in four equal groups: group 1, LVEF $ 50%, healthy; group 2, LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3, LVEF 30%–49%; and group 4, LVEF # 29%. Two observers performed four sets of image acquisitions and measurements (three during the same session, one after a median of 1 day) to account for intraobserver, interobserver, and test-retest reproducibility of combined acquisition and measurements. LPSS was measured in each of the 17 LV segments. Results: On average, the intraobserver and test-retest intraclass correlation coefficients and mean absolute differences of repeated acquisition and measurement of LPSS were similar across groups. However, interobserver intraclass correlation coefficients and mean absolute differences decreased in group 4 compared with groups 1 to 3. The intraobserver, test-retest, and interobserver coefficients of variation of all LV segments became worse as LVEF decreased, especially in group 4, in which LPSS was not reproducible in most segments. Reproducibility of LPSS in basal LV segments was worse compared with apical segments. The average measurement uncertainty (defined as the 95% limits of agreement of repeated acquisition and measurements) of LPSS in a test-retest scenario was 68.9%, 611.8%, 610.7%, and 69.0% in groups 1, 2, 3, and 4, respectively. Conclusions: The clinical applicability of LPSS is hindered by suboptimal reproducibility, even if a single observer repeats both acquisition and measurements. Changes in LPSS during patient follow-up should be interpreted with caution. (J Am Soc Echocardiogr 2019;32:1451-61.) Keywords: Segmental strain, Echocardiography, Reproducibility, Left ventricular dysfunction
Two-dimensional (2D) speckle-tracking echocardiographic (STE) imaging is especially suited to assess myocardial deformation1 and has been proposed as a robust imaging modality for the detection and quantification of severity of cardiac dysfunction.2 From the University of Medicine and Pharmacy ‘‘Carol Davila’’ Bucharest (A.D.M., M.-C.-B.S¸., D.V.); Department of Cardiology, University and Emergency Hospital of Bucharest (A.D.M., M.-C.-B.S¸., D.V.), Bucharest, Romania; Department of Cardiology, Morriston Hospital NHS Trust (A.D.M.), Swansea, United Kingdom rgulescu, MD, PhD, Department of Cardiology, Reprint requests: Andrei D. Ma Morriston Hospital NHS Trust, Swansea, SA6 6NL, United Kingdom (E-mail: [email protected]). 0894-7317/$36.00 Copyright 2019 by the American Society of Echocardiography. https://doi.org/10.1016/j.echo.2019.07.007
Two-dimensional STE indices of global left ventricular (LV) function (e.g., global longitudinal LV strain [GLS]) and indices of regional (segmental) LV function (e.g., longitudinal peak systolic segmental strain [LPSS]) have been suggested to allow early diagnosis of pathologic processes that may affect the left ventricle globally (e.g., anthracycline cardiomyopathy) or regionally (e.g., myocardial infarction, acute myocarditis).3,4 Some 2D STE indices of global LV function (e.g., GLS) have good reproducibility that may allow their use for patient follow-up. However, we have recently shown that the reproducibility of most 2D STE indices of global LV function decreases as the LVejection fraction (LVEF) becomes depressed, thus limiting the utility of 2D STE indices in patients with overt LV systolic dysfunction.5 Regarding regional LV function, recent data from the European Association of Cardiovascular Imaging (EACVI)/American Society of 1451
1452 Margulescu et al
Abbreviations
2D = Two-dimensional CV = Coefficient of variation GLS = Global longitudinal left ventricular strain ICC = Intraclass correlation coefficient LOA = Limits of agreement LPSS = Longitudinal peak systolic segmental strain
LV = Left ventricular LVEF = Left ventricular ejection fraction
MD = Mean absolute difference between repeated measurements
STE = Speckle-tracking echocardiographic
Journal of the American Society of Echocardiography November 2019
Echocardiography (ASE) Strain Standardization Task Force suggested that LPSS is less reproducible compared with GLS and that single LPSS values should be interpreted with caution in clinical practice.6,7 In these studies however, only the intraobserver reproducibility of combined acquisition and measurements was assessed, and even if the included subjects had a wide range of regional wall motion abnormalities, the average LVEF was within normal limits, thus limiting the number of patients with LV systolic dysfunction. Thus, the aim of this study was to assess the intraobserver, test-retest and interobserver reproducibility of combined acquisition and measurement of LPSS in relation to the degree of LV systolic dysfunction as
defined by LVEF.
METHODS Patients A brief summary of the study design has been published elsewhere.5 Here, we provide the protocol relating to the assessment of LPSS. Subjects were included prospectively and consecutively into four groups according to LVEF determined during the index echocardiographic evaluation and the presence of cardiovascular risk factors and/or overt cardiac pathology. Subjects were recruited from patients admitted electively to our hospital. Normal subjects were recruited from patients admitted for ablation of symptomatic supraventricular tachycardia who were otherwise healthy and from healthy colleagues who volunteered for the study. Those four groups covered the range from normal cardiac function, through subclinical LV dysfunction and heart failure with preserved LVEF, to moderate and severe systolic LV dysfunction. Inclusion criteria were as follows. Group 1: LVEF $ 50%, no structural abnormality on echocardiography, normal results on electrocardiography, no cardiovascular symptoms (except regular palpitations in patients admitted for ablation), and no history of hypertension, diabetes, or cardiovascular disease. Group 2: LVEF $ 50% and at least one of the following conditions: 1. documented hypertension and diabetes; 2. significant abnormality detected on the index echocardiographic evaluation: LV hypertrophy (defined as LV mass > 115 g/m2 in men and >95 g/m2 in women, calculated using the cube formula),8 concentric remodeling (defined as relative wall thickness > 0.42),8 more than mild aortic and/or mitral valve disease, or
LV regional wall motion abnormalities; and 3. established cardiovascular disease: documented coronary artery disease, previous myocardial infarction, history of coronary revascularization, peripheral or cerebrovascular vascular disease, hypertrophic cardiomyopathy, known significant (more than mild) valvular disease. Group 3: LVEF between 30% and 49%, regardless of medical history. Group 4: LVEF # 29%, regardless of medical history. Exclusion criteria were as follows: 1. any change in cardiovascular symptoms and/or treatment within the prior month before the index echocardiographic evaluation; 2. admission for acute cardiovascular reasons within the prior month before the index echocardiographic evaluation (e.g., acute myocardial infarction, acute heart failure); 3. irregular heart rhythm (e.g., atrial fibrillation conducted to the ventricles, frequent ventricular or supraventricular extra beats; patients with paced rhythms [atrial, ventricular, cardiac resynchronization devices] were allowed into the study if the rhythm was regular); 4. isolated right ventricular dilatation and/or dysfunction; and 5. poor echocardiographic images, including poor 2D speckletracking. We initially recruited 20 subjects in group 2 and 19 subjects each in the other groups. One subject was subsequently excluded from group 4 because of poor 2D speckle-tracking. To maintain an equal number of subjects in all groups, four additional subjects were randomly excluded from the other three groups. Thus, the final study population consisted of 18 subjects per group (72 subjects in total; mean age, 63 6 14 years; 65% men). The study was approved by the University and Emergency Hospital of Bucharest local ethics committee, and subjects gave written informed consent. Echocardiography All echocardiographic studies were acquired using a single machine (Vivid E9; GE Vingmed Ultrasound, Horten, Norway) equipped with a 2.5-MHz probe. Echocardiography was performed with the patient in the left lateral decubitus position. Chamber diameters were measured from the parasternal long-axis view, as recommended.8 For this study, 2D grayscale images from the apical four-chamber, two-chamber, and long-axis views were stored for offline analysis. Images covered at least three complete cardiac cycles and were optimized to obtain frame rates between 50 and 80 frames/sec, as recommended.9 Every effort was made to minimize apical foreshortening in the apical views.8,9 Analysis was performed offline using a single software version for all studies (EchoPAC version 113; GE Vingmed Ultrasound). LV volumes (end-diastolic and end-systolic) and LVEF were measured using the AutoEF option in EchoPAC.10 Full-thickness LPSS was calculated using the Automated Function Imaging option in EchoPAC.8 Zero strain was set at end-diastole (defined by the R wave on the electrocardiogram). End-systole was defined by the timing of the aortic valve closure time from the apical long-axis view. Then, a region of interest was selected to contain only the LV
Margulescu et al 1453
Journal of the American Society of Echocardiography Volume 32 Number 11
HIGHLIGHTS LPSS has suboptimal reproducibility. Reproducibility of LPSS in basal LV segments was worse compared with apical segments. The reproducibility of LPSS became worse as the ejection fraction decreased. Changes in LPSS during patients’ follow-up should be interpreted with caution. myocardium (without the pericardium or the LV cavity). The automatic tracking software tracked the myocardial speckles throughout the cardiac cycle. The results were visually inspected, and if necessary, the region of interest was adjusted manually until the tracking was judged to be optimal (i.e., following closely the movement of the underlying myocardium). We excluded segments from analysis if the tracking was not optimal, and we completely excluded the patient from the study if more than two segments in one view were not tracked optimally. Then the LPSS (defined as the maximal strain value, either positive or negative, that occurs before the aortic valve closure) was measured in each LV myocardial segment, as given by Automated Function Imaging, using the 17-segment model (corresponding to six basal and six midventricular segments [septal anterior, anterior, lateral, posterior, inferior, septal posterior], four apical segments [septal, anterior, lateral, inferior], and one apical cap segment). For easier understanding, we use the absolute value for strain in our discussion (i.e., higher strain indicates higher shortening of the myocardial segment). Reproducibility Two observers, each with >5 years of experience in performing and analyzing 2D STE data, performed four sets of image acquisitions and five sets of measurements, in all subjects. Joint sessions of image acquisitions and analysis of 2D STE indices were performed by the two observers before the beginning of the study to minimize variability. In particular, we paid special attention to use similar gain settings (set at medium), sector widths, and frame rates when acquiring images. The first three sets of image acquisitions were performed in sequence during the same session, with a 5-min break between them. The first two acquisitions were performed by the first observer, and the third acquisition was performed by the second observer. The first echocardiogram of the sequence was used to define LVEF and presence or absence of cardiac abnormalities to include or exclude the subject in or from one of the study groups. The patient maintained the same lateral decubitus position between sessions. The observers were blinded to each other’s acquisition. The first observer repeated image acquisition after 1 day in all but three patients (in whom repeated acquisition was performed after 3 days; test-retest). Images were stored on a secure offline server. The observers performed offline measurements on their own acquisitions, on separate days and in random order, blinded to each other’s data. Thus, we assessed the intraobserver, test-retest, and interobserver reproducibility of combined acquisition and measurements of 2D STE indices. In addition, we also evaluated the interobserver reproducibility of 2D STE measurements only, where observer 1 repeated measurements on the acquisition performed by observer 2.
Statistical Analysis Statistical analysis was performed using SPSS version 20 (SPSS, Chicago, IL). Quantitative data are presented as mean 6 SD. Multiple parametric values were compared using one-way analysis of variance, applying the Bonferroni correction. P values # .05 were considered to indicate statistical significance. Reproducibility is expressed as intraclass correlation coefficient (ICC; two-way mixed effect, absolute agreement model), mean absolute difference between repeated measurements (MD), coefficient of variation (CV), and measurement uncertainty. The MD was calculated as follows: Pn MD ¼
i¼1 ðjmeasurement 1
n
measurement 2jÞ
;
where n is the number of patients. Because LPSS is expressed as a percentage, the MD of LPSS is also expressed as a percentage. The CV was calculated as follows: CV ¼ SD=ðarithmetic mean of measurementsÞ 100; where SD is the standard deviation of the measurement error associated with a single measurement, calculated as the SD of residuals divided by O2. Measurement uncertainty was defined as the 95% limits of agreement (LOA) of repeated acquisition and measurements of LPSS.11 Bland-Altman analysis was used to estimate the bias and 95% LOA between paired measurements.
RESULTS Baseline and echocardiographic characteristics of the study groups are displayed in Tables 1 and 2, respectively. Subjects in group 2 were characterized by the presence of significant concentric LV remodeling compared with group 1 (smaller LV volumes, higher relative wall thickness). Frame rates of the acquired images were similar between observers (57.0 6 3.3 vs 56.2 6 2.9 frames/sec, P = .15) and between repeated acquisitions performed on different days by the first observers (57.1 6 3.4 vs 56.9 6 3.3 frames/sec, P = .75). Heart rate and blood pressure also remained stable between repeated evaluations performed on different days (heart rate: 71.7 6 13.6 vs 71.8 6 13.7 beats/min [P = .87]; systolic blood pressure: 127 6 16 vs 124 6 12 mm Hg [P = .20]; diastolic blood pressure: 74 6 10 vs 75 6 9 mm Hg [P = .48]). LPSS values for each LV segment in all groups are displayed in Figure 1. The average intraobserver, test-retest, and interobserver ICCs and 95% LOA in all four groups are displayed in Figure 2, while the average intraobserver, test-retest, and interobserver MDs and CVs in all four groups are displayed in Table 3. The ICCs, MDs, and CVs in individual LV segments are displayed in Supplemental Figures 1 to 3. When all segments were considered together, the intraobserver and test-retest ICCs, MDs, and 95% LOA of repeated acquisition and measurement of LPSS remained relatively unchanged throughout the spectrum of LVEF. However, interobserver ICCs, MDs, and 95% LOA were worse in patients with severely depressed LVEFs (group 4) compared with groups 1 to 3 (Table 3, Figure 2).
1454 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Table 1 Baseline characteristics of the study groups Variable
Group 1 (n = 18)
Group 2 (n = 18)
Group 3 (n = 18)
Group 4 (n = 18)
P
Demographic data Age, y
48.6 6 13.8
67.2 6 12.8
67.7 6 7.8
66.7 6 11.6
Sex, male, %
9 (50)
10 (56)
13 (72)
15 (83)
1.80 6 0.22
1.84 6 0.20
1.80 6 0.24
<.001* .13†
BSA, m
1.96 6 0.22
Hypertension
—
16 (89)
16 (89)
7 (39)
<.001†
Diabetes
—
3 (17)
5 (28)
4 (22)
.13†
Ischemic heart disease
—
7 (39)
9 (50)
7 (39)
.026†
Previous myocardial infarction
—
3 (17)
8 (44)
7 (39)
.007†
Peripheral or cerebrovascular vascular disease
—
1 (6)
2 (11)
0 (0)
.38†
Moderate or severe MR
—
0 (0)
4 (22)
9 (50)
.005†
Moderate or severe AS
—
0 (0)
0 (0)
1 (6)
.39†
Aspirin
—
7 (39)
12 (67)
13 (72)
.09†
Oral anticoagulants
—
9 (50)
5 (28)
5 (28)
.27†
b-blockers
—
14 (78)
13 (72)
17 (94)
.20†
ACE inhibitors/ARBs
—
13 (72)
16 (89)
8 (44)
.015†
Statin
—
16 (89)
13 (72)
10 (56)
.08†
Diuretics
—
8 (44)
13 (72)
18 (100)
.001†
2
.11*
Treatment
ACE, Angiotensin-converting enzyme; ARB, angiotensin receptor blocker; AS, aortic stenosis; BSA, body surface area; MR, mitral regurgitation. Data are expressed as mean 6 SD or as number (percentage). *One-way analysis of variance. † Chi-square test.
The ICCs, MDs, and CVs in individual LV segments in each group are displayed in Supplemental Figures 1 to 3. On average, LPSS decreased from group 1 to 4, while the MDs remained either
constant (for intraobserver and test-retest) or decreased in group 4 (for interobserver). As a result, the intraobserver, test-retest, and interobserver CVs (for combined acquisition and
Table 2 Echocardiographic characteristics of the study groups P value (one-way ANOVA with Bonferroni correction) Variable
Group 1 (n = 18)
Group 2 (n = 18)
Group 3 (n = 18)
Group 4 (n = 18)
Group 1 Group 1 Group 1 Group 2 Group 2 Group 3 vs group 2 vs group 3 vs group 4 vs group 3 vs group 4 vs group 4
General echocardiographic data Frame rate, frames/sec
58.4 6 3.5
57.6 6 3.3
57.2 6 3.3
54.6 6 1.8
$.999
$.999
.002
$.999
.022
.06
Heart rate, beats/min
71.6 6 12.5
64.7 6 7.9
71.3 6 16.2
79.4 6 13.0
.64
$.999
.43
.75
.006
.37
IVS, mm
8.8 6 1.2
12.9 6 1.8
12.8 6 1.5
11.0 6 2.0
<.001
<.001
.001
$.999
.004
.011
LVPW, mm
8.3 6 1.1
11.0 6 1.1
11.1 6 1.3
10.4 6 1.5
<.001
<.001
<.001
$.999
.92
.56
LVEDD, mm
49.5 6 5.1
42.9 6 5.9
52.8 6 6.8
66.1 6 9.7
.040
$.999
<.001
.001
<.001
<.001
LV indexed mass, g/m2
74.7 6 15.6 103.4 6 24.3 138.4 6 27.2 176.2 6 31.6
.007
<.001
<.001
.001
<.001
<.001
RWT
0.35 6 0.05
0.57 6 0.12
0.46 6 0.10
0.33 6 0.08
<.001
.002
$.999
.004
<.001
<.001
LA diameter at end-systole, mm
38.6 6 5.3
40.0 6 6.2
41.5 6 5.0
44.1 6 7.3
$.999
.93
.05
.23
$.999
$.999
LVEDV, mL
115.3 6 32.0
88.9 6 29.9 146.7 6 32.7 275.8 6 108.9
$.999
.76
<.001
.035
<.001
<.001
LVESV, mL
48.6 6 15.7
36.9 6 15.3
88.6 6 27.5 219.9 6 89.6
$.999
.09
<.001
.012
<.001
<.001
LVEF, %
58.5 6 4.3
59.2 6 6.3
37.8 6 6.5
$.999
<.001
<.001
<.001
<.001
<.001
20.6 6 4.3
ANOVA, Analysis of variance; IVS, interventricular septal thickness at end-diastole; LA, left atrial; LVEDD, LV end-diastolic diameter; LVEDV, LV end-diastolic volume; LVESV, LV end-systolic volume; LVPW, posterior wall thickness at end-diastole; RWT, relative wall thickness. Data are expressed as mean 6 SD.
Journal of the American Society of Echocardiography Volume 32 Number 11
Margulescu et al 1455
Figure 1 Polar map (bull’s-eye view) of LV longitudinal peak systolic strain for each segment of the LV wall (using a 17-segment model) in the four study groups. Values represent strain (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%. LV walls are displayed in the first bull’seye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
measurements) and the interobserver CVs for measurements only of all LV segments became worse as LVEF decreased. This was especially notable in group 4 (LVEF # 29%), in which LPSS was not reproducible in most segments (Table 3, Supplemental Figure 3). Overall, the intraobserver and test-retest ICCs and MDs of LPSS were marginally worse in basal versus apical segments; also the CVs decreased from basal to apical segments. This tendency was much more evident for interobserver ICCs, MDs, and CVs, for which the reproducibility of LPSS clearly decreased from apical to basal segments. Similar tendencies were observed when analyzing groups separately (Table 4).
On the basis of the 95% LOA between paired acquisitions and measurements, the average measurement uncertainty of LPSS in a scenario in which a single observer repeats both echocardiographic acquisition and measurements during patient follow-up was 68.9%, 611.8%, 610.7%, and 69.0% in groups 1, 2, 3, and 4, respectively. An estimation of the measurement uncertainty of LPSS assessment in each LV segment and in each study group in a similar scenario is given in Figure 3. An estimation of the measurement uncertainty of LPSS in basal, mid, and apical LV segments for each group is given in Table 5. We hypothesized that the worse reproducibility of basal versus apical segments was related to slight differences in the acquisition of
1456 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Figure 2 Correlation (top) and Bland-Altman plots (bottom) of repeated measurements in the four study groups. ICCs with 95% CIs and bias with 95% LOA of repeated measurements are also displayed.
apical images by the two observers, so we expected that this should have been reflected by a bias in measuring LV end-diastolic and end-systolic volumes. Indeed, there was no significant bias for intraob-
server and test-retest acquisition and measurements of LV volumes (end-diastolic, 1.3 and 3.2 ml; end-systolic, 1.0 and 1.3 ml), but there was a significant bias in inter-observer assessment of LV volumes
Margulescu et al 1457
Journal of the American Society of Echocardiography Volume 32 Number 11
Table 3 Mean and SD of MD and CV of LV LPSS in each of the 17 LV segments in all study groups P value (one-way ANOVA with Bonferroni correction) Group 1 (n = 18)
Group 2 (n = 18)
Group 3 (n = 18)
Group 4 (n = 18)
Group 1 vs group 2
Group 1 vs group 3
Group 1 vs group 4
Group 2 vs group 3
Group 2 vs group 4
Group 3 vs group 4
MD, % Intraobserver
2.9 6 1.0 2.7 6 0.7 3.0 6 1.3
3.1 6 0.9
$.999
$.999
$.999
$.999
$.999
$.999
Test-retest
2.8 6 0.7 3.7 6 1.2 3.7 6 1.4
3.6 6 1.4
.26
.29
.41
$.999
$.999
$.999
Interobserver
3.4 6 0.8 3.9 6 1.0 4.0 6 0.8
4.1 6 1.6
.94
.60
.38
$.999
$.999
$.999
Interobserver 2.8 6 0.8 3.3 6 0.9 3.8 6 1.3 (measurements only)
4.3 6 2.2
$.999
.25
.017
$.999
.31
$.999
CV, % Intraobserver
14 6 9
17 6 8
27 6 17
263 6 645
$.999
$.999
.17
$.999
.18
.22
Test-retest
14 6 5
24 6 12
33 6 20
289 6 649
$.999
$.999
.10
$.999
.12
.15
Interobserver
16 6 7
21 6 8
33 6 11
113 6 82
$.999
$.999
<.001
$.999
<.001
<.001
Interobserver (measurements only)
12 6 4
17 6 5
35 6 15
213 6 341
$.999
$.999
.006
$.999
.008
.021
Data are expressed as mean 6 SD.
(acquisition and measurement and measurement only; end-diastolic, 16.7 and 13.3 ml; end-systolic, 25.9 and 21.5 ml). An illustrative example of the difference between LPSS values from repeated acquisition and measurements in a single patient is given in Figure 4.
DISCUSSION In this study, we assessed the intraobserver, test-retest and interobserver reproducibility of combined acquisition and measurement of LPSS in four groups ranging from normal subjects to patients with severe LV systolic dysfunction. The main results of this study suggest that the clinical applicability of LPSS is hindered by suboptimal reproducibility, even if a single observer repeats both acquisition and measurements. This is especially noticeable in basal LV segments and in patients with LV systolic dysfunction. The lower interobserver reproducibility of basal and mid segments compared with intraobserver and test-retest reproducibility may be related to slight differences in the window and angle of insonation when acquiring apical images. Thus, changes in LPSS during patient follow-up should be interpreted with caution. Unlike GLS, for which most of the studies revealed good reproducibility across the spectrum of LVEF,5 data on the clinical utility and reproducibility of LPSS have been more conflicting. Initial studies suggested that LPSS has good reproducibility.12,13 For example, Shiino et al.12 assessed the reproducibility of LPSS in 55 patients, using the same equipment as in our study, but analyzed LPSS reproducibility according to presumed coronary artery myocardial distribution and did not include the acquisition process in their analysis (measurements only), making their data less relevant to clinical practice. In a study of 290 subjects (two thirds of whom had various cardiac conditions), Barbier et al.13 reported that LPSS has good or moderate testretest reproducibility depending on segment (CV ranging from 8% to 23%); they concluded that LPSS is ‘‘suitable for diagnosis and follow-up of LV regional systolic function.’’ However, the mean reported LVEF was near normal (52 6 16%).
More recently, the EACVI/ASE Strain Standardization Task Force published a series of articles assessing the variability and reproducibility of LPSS acquisition and measurements, for all current major echocardiographic machine manufacturers, including GE.6,7,14 The authors showed that the ICC was 0.90 (95% CI, 0.89–0.91), and the MD was 3.0 6 2.9% for the pooled analysis of LPSS in all segments. The average MD for individual segments ranged from 1.7 6 1.3% (basal inferior septum) to 4.5 6 3.9% (anterior apex).6 Also, the basal LV segments had the lowest reproducibility, while the apical LV segments had the highest reproducibility of LPSS. The results reported in the articles published by the EACVI/ASE task force are comparable with the results presented here if one takes into account some important differences between these studies. The EACVI/ASE task force defined ‘‘test-retest’’ reproducibility on the basis of repeated acquisitions and measurements performed sequentially, during the same echocardiographic session, by a single observer. This is similar to what we defined herein as ‘‘intraobserver reproducibility of combined acquisition and measurements’’; in our article, we define ‘‘test-retest’’ reproducibility on the basis of repeated acquisition and measurements on different days. This distinction is critical for the correct interpretation of the data; indeed, our results on ‘‘intraobserver’’ reproducibility are similar with the ‘‘test-retest’’ reproducibility previously published by Mirea et al.6 However, we think that the protocol and definition used in our study are more relevant for clinical practice. For example, if LPSS, or any other echocardiographic parameter, is to be used in a patient’s follow-up, one is more interested in how stable and reliable measurements are when images are reacquired after a period of time (under reasonably similar physiologic conditions). A challenge in such situations is minimizing the potential bias that would result from differences in the acquisitions. In this more clinically relevant scenario, the reproducibility of LPSS was worse compared with that previously reported.6 In addition, even if the subjects analyzed by the EACVI/ASE task force had a wide range of segmental and functional abnormalities (on average, there were 5 6 3 transmural scarred segments per patient, as defined by cardiac magnetic resonance imaging) the average LVEF in the study was 52 6 10%. Thus, similar to Barbier et al.,12 the number
1458 Margulescu et al
Table 4 Reliability and reproducibility of LV peak segmental strain in basal, mid, and apical segments in all patients combined and in each study group separately All patients
Group 1
Group 2
Group 3
Group 4
Basal
Mid
Apical
Basal
Mid
Apical
Basal
Mid
Apical
Basal
Mid
Apical
Basal
Mid
Apical
0.90 (0.88– 0.92)
0.94 (0.93– 0.95)
0.94 (0.93– 0.95)
0.78 (0.68– 0.85)
0.73 (0.61– 0.82)
0.73 (0.58– 0.82)
0.68 (0.54– 0.78)
0.79 (0.69– 0.85)
0.81 (0.72– 0.88)
0.85 (0.79– 0.90)
0.86 (0.80– 0.91)
0.93 (0.90– 0.96)
0.85 (0.78– 0.90)
0.85 (0.78– 0.90)
0.63 (0.42– 0.76)
Intraobserver ICC (95% CI)
MD, %
3.2
2.6
3.0
2.8
2.5
3.3
3.0
2.5
2.9
3.6
2.7
2.2
3.5
2.7
3.4
CV, %
30.5
20.6
17.9
19.8
12.6
12.8
22.2
13.7
13.2
39.9
26.4
15.7
87.3
90.5
62.5
0.83 (0.79– 0.86)
0.92 (0.90– 0.93)
0.93 (0.91– 0.94)
0.82 (0.73– 0.87)
0.56 (0.12– 0.75)
0.63 (0.42– 0.76)
0.32 (0.01– 0.53)
0.42 (0.15– 0.60)
0.77 (0.66– 0.85)
0.70 (0.56– 0.80)
0.86 (0.80– 0.91)
0.79 (0.68– 0.86)
0.76 (0.65– 0.84)
0.73 (0.60– 0.82)
0.71 (0.54– 0.81)
Test-retest ICC (95% CI)
MD, %
4.1
3.2
3.1
2.9
2.4
3.3
4.6
3.6
3.1
4.5
2.9
2.9
4.2
3.6
2.9
CV, %
38.5
24.9
18.7
16.9
12.2
14.0
34.1
20.5
13.4
51.5
25.3
22.7
105.2
136.2
52.1
0.77 (0.73– 0.81)
0.88 (0.86– 0.90)
0.93 (0.90– 0.95)
0.48 (0.24– 0.65)
0.25 (0.06– 47)
0.77 (0.64– 0.85)
0.14 (–0.25– 0.42)
0.48 (0.24– 0.65)
0.76 (0.62– 0.85)
0.63 (0.46– 0.75)
0.68 (0.53– 0.78)
0.76 (0.60– 0.85)
0.55 (0.35– 0.69)
0.58 (0.39– 0.72)
0.70 (0.52– 0.81)
Interobserver ICC (95% CI)
4.8
4.1
3.3
4.0
3.7
3.1
4.7
3.8
3.9
4.7
4.1
3.7
5.6
4.6
2.5
CV, %
38.9
27.3
17.9
23.0
16.8
11.1
30.8
19.1
14.5
46.7
30.8
25.6
106.2
121.3
46.2
0.80 (0.75– 0.83)
0.88 (0.85– 0.90)
0.96 (0.93– 0.97)
0.56 (0.36– 0.70)
0.33 (0.04– 0.54)
0.88 (0.80– 0.92)
0.54 (0.34– 0.69)
0.52 (0.30– 0.67)
0.83 (0.67– 0.91)
0.54 (0.33– 0.69)
0.61 (0.44– 0.74)
0.88 (0.76– 0.94)
0.46 (0.20– 0.63)
0.46 (0.21– 0.63)
0.88 (0.81– 0.93)
Interobserver measurements only ICC (95% CI)
MD, %
4.4
4.0
2.5
3.2
3.0
2.3
3.4
3.7
3.1
4.7
4.1
2.5
6.3
5.2
1.7
CV, %
35.1
27.5
13.6
16.6
13.9
8.6
19.8
18.2
12.2
48.5
33.4
17.8
109.3
126.4
26.5
Journal of the American Society of Echocardiography November 2019
MD, %
Margulescu et al 1459
Journal of the American Society of Echocardiography Volume 32 Number 11
Figure 3 Estimation of measurement uncertainty of LV LPSS in the four study groups. The polar map (bull’s-eye view) displays the 95% LOA of repeated acquisitions and measurements on different days, by the same observer, of LPSS in each segment of the LV wall (using a 17-segment model), equivalent to a best-case clinical scenario in which a single echocardiographer follows patients. Values represent strain (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%. LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
Table 5 Estimation of measurement uncertainty (95% LOA between repeated acquisitions and measurements) of LV peak systolic strain in basal, mid, and apical LV segments All patients Basal
Mid
Group 1 Apical
Basal
Mid
Group 2 Apical
Basal
Mid
Group 3 Apical
Basal
69.9 63.6 68.2
69.5
Mid
Group 4 Apical
Basal
Apical
69.7
67.4 68.0
69.4 66.8 68.1
Test-retest, %
612.1
68.8 68.4
67.1 66.5 68.6 614.5 65.2 68.5 613.1
67.8 68.1 611.9 610.1 67.8
Interobserver, %
612.8 610.1 68.3 611.1 69.3 67.3 613.6 65.0 69.3 612.0
69.7 69.4 614.2 611.6 66.9
Intraobserver, %
Interobserver 617.9 610.3 66.4 (measurements only), %
67.9 65.5 610.0
Mid
67.8 69.6
68.9 67.7 65.7 611.3 64.8 67.8 621.5 610.6 66.6 612.3 612.8 64.2
1460 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Figure 4 An example of repeated acquisition and measurements of LV LPSS in a single patient from group 1. Individual LPSS curves taken from the three apical views (four chamber [A4c], two chamber [A2c], and long-axis [APLAX]) are displayed in the top three rows, while the 17-segment polar maps (bull’s-eye view) are displayed in the bottom row. Each column represents a sequence of acquisition and measurements (three performed by observer 1, one performed by observer 2). One can appreciate that between evaluations, LPSS in basal segments displays larger intraobserver, test-retest, and interobserver variability, compared with apical segments because the measured LPSS in the basal segments is more sensitive compared with apical segments to slight changes in the angle of insonation.
of included patients with severe systolic dysfunction and the number of segments showing very low or positive strain (dyskinetic segments) were relatively small. Also, the EACVI/ASE task force evaluated the reproducibility of endocardial and midmyocardial LPSS, while we measured full-thickness LPSS. However, we believe that this distinction is unlikely to be relevant for reproducibility assessment, as previous data demonstrated that the test-retest variability of layer-specific GLS is similar at different myocardial levels.14 In our study, we provide data on the reproducibility of LPSS throughout the spectrum of LVEF, which is important for clinical practice and especially for patient follow-up. We recently showed that GLS might have acceptable reproducibility (test-retest CV < 15%) even in patients with severe LV systolic dysfunction.5 GLS is calculated by averaging all values of LPSS, but in this study we have shown that LPSS has suboptimal reproducibility, especially in patients with poor LVEFs. Moreover, the 95% LOA for LPSS are too wide to allow interpretation of changes between evaluations even if both acquisitions and measurements are performed by the same echocardiographer, regardless of LVEF. It becomes apparent that large differences of individual LPSS are attenuated by averaging all values when GLS is calculated. Indeed, Barbier et al.13 showed that LPSS variability correlates with minor changes in scan angulation, which is inevitable between scans. Changes in scan angulation between eval-
uations also help explain why LPSS is more reproducible at apical segments (which are relatively fixed) compared with basal segments. Thus, changes in LPSS should be interpreted with caution, in the context of changes in GLS and the pattern of change in the polar map (bull’s-eye) displayed by the Automated Function Imaging software. Clinical Perspective Several articles have suggested that LPSS may allow early diagnosis of global and regional LV pathological processes (e.g., anthracycline cardiomyopathy, myocardial infarction, acute myocarditis).3,4 This has also been initially supported by reported good reproducibility of LPSS. However, recent data suggest that LPSS cannot detect isolated segmental scarred segments (surrounded by normal myocardium, as defined by cardiac magnetic resonance imaging).7 Similarly, recent studies showed that the reproducibility of LPSS may not be adequate for clinical application.6 We confirm these latest results, and we also provide useful estimates of the expected measurement uncertainty for repeated assessments of LPSS in scenarios resembling ideal clinical practice (i.e., a single observer performing both acquisition and measurements during patient follow-up). These results confirm that assessment of LPSS cannot be reliably used in clinical practice.
Margulescu et al 1461
Journal of the American Society of Echocardiography Volume 32 Number 11
In the future, the shortcoming of assessing 2D LPSS may be overcome by full-volume three dimensional LV strain analysis. It would be expected that full-volume three-dimensional strain would allow assessment of LPSS with less dependency on the angle of insonation and slight changes in apical window acquisition, which may improve variability. Until then, current efforts to standardize deformation imaging among vendors are welcome to homogenize the reported range of values of deformation parameters but are unlikely to result in an improvement of clinical applicability of LPSS. Limitations of the Study The number of patients in each group was relatively small. This was a consequence of the design of the study, in which a single echocardiographic machine, probe, and software were used to eliminate the well-described variability induced by different machines, probes, and software versions.6,7,15 However, our results are in line those of with other recent studies,6,7 so we believe that our study provides useful insights regarding the reproducibility of LPSS and its impact on clinical practice when image acquisition is also taken into account and when other potential sources of variation are minimized. We used a single vendor. However, for LPSS, this vendor has shown higher fidelity of the 2D speckle-tracking algorithm in following LV myocardial segmental deformation and lower variability of LPSS measurements compared with other vendors. Thus, the results are unlikely to be improved by using other current vendors and software packages.6 We did not assess the reproducibility of detecting transmural scar by LPSS. However, previous studies have already shown that LPSS may detect transmural scar only when more than two neighboring segments are affected.7
2.
3.
4.
5.
6.
7.
8.
9.
CONCLUSION 10.
In our study, the clinical applicability of LPSS was hindered by suboptimal reproducibility, especially in basal LV segments, and in patients with LV systolic dysfunction, even if a single observer repeated both acquisition and measurements. The lower interobserver reproducibility of basal and mid segments compared with intraobserver and test-retest reproducibility may have been related to slight differences in the window and angle of insonation when apical images were acquired. Thus, changes in LPSS during patient follow-up should be interpreted with caution.
11.
12.
13.
SUPPLEMENTARY DATA Supplementary data to this article can be found online at https://10. 1016/j.echo.2019.07.007.
14.
REFERENCES 15. 1. Mor-Avi V, Lang RM, Badano LP, Belohlavek M, Cardim NM, Derumeaux G, et al. Current and evolving echocardiographic techniques for the quantitative evaluation of cardiac mechanics: ASE/EAE consensus
statement on methodology and indications endorsed by the Japanese Society of Echocardiography. Eur J Echocardiogr 2011;12:167-205. Potter E, Marwick TH. Assessment of left ventricular function by echocardiography: the case for routinely adding global longitudinal strain to ejection fraction. JACC Cardiovasc Imaging 2018;11:260-74. Sjøli B, Ørn S, Grenne B, Ihlen H, Edvardsen T, Brunvand H. Diagnostic capability and reproducibility of strain by Doppler and by speckle tracking in patients with acute myocardial infarction. JACC Cardiovasc Imaging 2009;2:24-33. Kostakou PM, Kostopoulos VS, Tryfou ES, Giannaris VD, Rodis IE, Olympios CD, et al. Subclinical left ventricular dysfunction and correlation with regional strain analysis in myocarditis with normal ejection fraction. A new diagnostic criterion. Int J Cardiol 2018;259:116-21. Margulescu AD, Suran MCB, Vinereanu D. Do 2-dimensional speckle tracking indexes have sufficient reproducibility to be used in clinical practice when image acquisition and severity of LV dysfunction are taken into account? JACC Cardiovasc Imaging 2019;12:756-9. Mirea O, Pagourelias ED, Duchenne J, Bogaert J, Thomas JD, Badano LP, et al., EACVI-ASE-Industry Standardization Task Force. Variability and reproducibility of segmental longitudinal strain measurement: a report from the EACVI-ASE Strain Standardization task force. JACC Cardiovasc Imaging 2018;11:15-24. Mirea O, Pagourelias ED, Duchenne J, Bogaert J, Thomas JD, Badano LP, et al., EACVI-ASE-Industry Standardization Task Force. Intervendor differences in the accuracy of detecting regional functional abnormalities: a report from the EACVI-ASE Strain Standardization task force. JACC Cardiovasc Imaging 2018;11:25-34. Lang RM, Badano LP, Mor-Avi V, Afilalo J, Armstrong A, Ernande L, et al. Recommendations for cardiac chamber quantification by echocardiography in adults: an update from the American Society of Echocardiography and the European Association of Cardiovascular Imaging. J Am Soc Echocardiogr 2015;28:1-39.e14. Voigt JU, Pedrizzetti G, Lysyansky P, Marwick TH, Houle H, Baumann R, et al. Definitions for a common standard for 2D speckle tracking echocardiography: consensus document of the EACVI/ASE/Industry task force to standardize deformation imaging. J Am Soc Echocardiogr 2015;28: 183-93. Szulik M, Pappas CJ, Jurcut R, Magro M, Peeters E, Goetschalckx K, et al. Clinical validation of a novel speckle-tracking-based ejection fraction assessment method. J Am Soc Echocardiogr 2011;24:1092-100. Working Group 1 of the Joint Committee for Guides in Metrology. Evaluation of measurement data—guide to the expression of uncertainty in measurement. Available at: https://www.bipm.org/utils/common/ documents/jcgm/JCGM_100_2008_E.pdf. Accessed July 22, 2019. Shiino K, Yamada A, Ischenko M, Khandheria BK, Hudaverdi M, Speranza V, et al. Intervendor consistency and reproducibility of left ventricular 2D global and regional strain with two different high-end ultrasound systems. Eur Heart J Cardiovasc Imaging 2017;18:707-16. Barbier P, Mirea O, Cefal u C, Maltagliati A, Savioli G, Guglielmo M. Reliability and feasibility of longitudinal AFI global and segmental strain compared with 2D left ventricular volumes and ejection fraction: intraand inter-operator, test-retest, and inter-cycle reproducibility. Eur Heart J Cardiovasc Imaging 2015;16:642-52. € u S, Mirea O, Duchenne J, Pagourelias ED, Bezy S, Thomas JD, et al. Unl€ Comparison of feasibility, accuracy, and reproducibility of layer-specific global longitudinal strain measurements among five different vendors: a report from the EACVI-ASE Strain Standardization task force. J Am Soc Echocardiogr 2018;31:374-80.e1. Costa SP, Beaver TA, Rollor JL, Vanichakarn P, Magnus PC, Palac RT. Quantification of the variability associated with repeat measurements of left ventricular two-dimensional global longitudinal strain in a real-world setting. J Am Soc Echocardiogr 2014;27:50-4.
1461.e1 Margulescu et al
Journal of the American Society of Echocardiography November 2019
APPENDIX
Supplemental Figure 1 Polar map (bull’s-eye view) of ICCs of LV LPSS for each segment of the LV wall (using a 17-segment model) when image acquisition and measurement were considered (intraobserver, test-retest, and interobserver) and compared with interobserver reproducibility of 2D STE measurements only on the same image data set. ICCs $ 0.90 are displayed in green, ICCs < 0.90 and $0.75 in yellow, ICCs < 0.75 and $0.50 in red, ICCs < 0.50 and $0.25 in blue, and ICCs < 0.25 in black. Values are ICCs. Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%). LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
Journal of the American Society of Echocardiography Volume 32 Number 11
Margulescu et al 1461.e2
Supplemental Figure 2 Polar map (bull’s-eye view) of MDs of LV LPSS for each segment of the LV wall (using a 17-segment model) when image acquisition and measurement were considered (intraobserver, test-retest, and interobserver) and compared with interobserver reproducibility of 2D STE measurements only on the same image data set. Values are MDs (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%. LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.
1461.e3 Margulescu et al
Journal of the American Society of Echocardiography November 2019
Supplemental Figure 3 Polar map (bull’s-eye view) of CVs of LV LPSS for each segment of the LV wall (using a 17-segment model) when image acquisition and measurement were considered (intraobserver, test-retest, and interobserver) and compared with interobserver reproducibility of 2D STE measurements only on the same image data set. CVs # 10 are displayed in green, CVs > 10 and #20 in yellow, CVs > 20 and #40 in red, CVs > 40 and #60 in blue, and CVs > 60 in black. Values represent CVs (%). Group 1: LVEF $ 50%, healthy; group 2: LVEF $ 50%, presence of cardiovascular disease and/or risk factors; group 3: LVEF 30% to 49%; group 4: LVEF # 29%). LV walls are displayed in the first bull’s-eye graph of the series (left). The apical segments are septal, anterior, lateral, inferior, and apical cap. AW, Anterior wall; IVSa, anterior septum; IVSp, posterior septum; IW, inferior wall; LW, lateral wall; PW, posterior wall.