Summary
Background
Speckle tracking can be used to measure left ventricular global longitudinal strain (GLS).
Aims
To study the effect of speckle tracking software product upgrades on GLS values and intervendor consistency.
Methods
Subjects (patients or healthy volunteers) underwent systematic echocardiography with equipment from Philips and GE, without a change in their position. Off-line post-processing for GLS assessment was performed with the former and most recent upgrades from these two vendors (Philips QLAB 9.0 and 10.2; GE EchoPAC 12.1 and 13.1.1). GLS was obtained in three myocardial layers with EchoPAC 13.1.1. Intersoftware and intervendor consistency was assessed. Interobserver variability was tested in a subset of patients.
Results
Among 73 subjects (65 patients and 8 healthy volunteers), absolute values of GLS were higher with QLAB 10.2 compared with 9.0 (intraclass correlation coefficient [ICC]: 0.88; bias: 2.2%). Agreement between EchoPAC 13.1.1 and 12.1 varied by myocardial layer (13.1.1 only): midwall (ICC: 0.95; bias: –1.1%), endocardium (ICC: 0.93; bias: 1.6%) and epicardial (ICC: 0.80; bias: –3.3%). Although GLS was comparable for QLAB 9.0 versus EchoPAC 12.1 (ICC: 0.95; bias: 0.5%), the agreement was lower between QLAB 10.2 and EchoPAC 13.1.1 endocardial (ICC: 0.91; bias: 1.1%), midwall (ICC: 0.73; bias: 3.9%) and epicardial (ICC: 0.54; bias: 6.0%). Interobserver variability of all software products in a subset of 20 patients was excellent (ICC: 0.97–0.99; bias: –0.8 to 1.0%).
Conclusion
Upgrades of speckle tracking software may be associated with significant changes in GLS values, which could affect intersoftware and intervendor consistency. This finding has important clinical implications for the longitudinal follow-up of patients with speckle tracking echocardiography.
Résumé
Contexte
Les possibles modifications des valeurs de strain longitudinal global (SLG), induites par les améliorations de version des logiciels, restent inconnues.
Objectifs
Cette étude avait pour but d’étudier l’impact des upgrades de ces logiciels sur les valeurs de SLG et donc sur la concordance entre constructeurs.
Méthodes
Des sujets (patients ou sujets sains) ont bénéficié d’une échocardiographie avec acquisition systématique avec des machines de deux constructeurs (Philips et GE), sans modification de position du patient. Le SLG était obtenu en déporté avec les versions anciennes et récentes des logiciels de ces deux constructeurs (Philips QLAB 9.0 et 10.2 ; GE EchoPAC 12.1 et 13.1.1). Le SLG était obtenu sur trois couches myocardiques avec EchoPAC 13.1.1. La concordance entre constructeurs était évaluée avec les deux versions des logiciels de chaque constructeur. La variabilité interobservateur de chaque logiciel était testée sur un échantillon de patients de l’étude.
Résultats
Parmi 73 sujets (68 patients et 8 sujets sains), le SLG était plus élevé en valeur absolue avec QLAB 10.2 qu’avec 9.0 (coefficient de corrélation intraclasse [CCI] : 0,88 ; biais : 2,2 %). Une bonne concordance était observée entre le SLG à mi-paroi (mid) obtenu à l’aide de l’EchoPAC 13.1.1 et le SLG obtenu à l’aide de l’EchoPAC 12.1 (CCI : 0,95 ; biais : –1,1 %). Le SLG à l’endocarde (endo) obtenu avec EchoPAC 13.1.1 était légèrement plus élevé en valeur absolue que le SLG obtenu avec EchoPAC 12.1 (CCI : 0,93 ; biais : 1,6 %), tandis que le SLG à l’épicarde (épi) obtenu avec EchoPAC 13.1.1 était plus bas en valeur absolue que le SLG obtenu avec EchoPAC 12.1 (CCI : 0,80 ; biais : –3,3 %). Bien que le SLG fût comparable entre QLAB 9.0 and EchoPAC 12.1 (CCI : 0,95 ; biais : 0,5 %), la concordance était plus faible entre QLAB 10.2 et EchoPAC 13.1.1 endo (CCI : 0,91 ; biais : 1,1 %), mid (CCI : 0,73 ; biais : 3,9 %) et epi (CCI : 0,54 ; biais : 6,0 %). La variabilité interobservateur de tous ces logiciels était excellente, sur un échantillon de 20 patients de l’étude avec des CCI situés entre 0,97 et 0,99.
Conclusion
Les améliorations des logiciels de speckle tracking peuvent être associées à des modifications significatives des valeurs de SLG qui influencent la concordance entre constructeurs. Ces données peuvent avoir des implications cliniques importantes pour le suivi longitudinal des patients à l’aide du speckle tracking .
Introduction
Longitudinal strain describes myocardial deformation, the fractional change in length of a myocardial segment. Speckle tracking is a recent, largely angle-independent routine technique that has been used for the evaluation of myocardial longitudinal strain and has been validated against sonomicrometry . Global longitudinal strain (GLS) is defined by the average of peak systolic longitudinal strain values from all left ventricular (LV) segments.
Clinical studies have demonstrated major additional diagnostic and/or prognostic values of GLS compared with conventional indices of LV systolic function in various settings, such as heart failure, valvular heart disease and cardiomyopathies . Previous reports have demonstrated significant differences between longitudinal strain values obtained with speckle tracking using the first generation of software products released by various manufacturers . Owing to different post-processing algorithms , the American Society of Echocardiography (ASE) and the European Association of CardioVascular Imaging (EACVI) set up an expert group, combining researchers and industry members, to achieve a consensus document detailing speckle tracking measurements . As a result, reports have demonstrated an agreement between two software products from two major vendors (EchoPAC 12, GE Medical, Milwaukee, WI, USA and QLAB 9, Philips, Andover, MA, USA) . Changes in vendor and reader can be expected to influence GLS values by up to 5% . However, since the publication of these studies, upgrades of software products have been released. In addition, one software upgrade (GE EchoPAC 13.1.1) enables distinctive evaluation of endocardial, mid-myocardial and epicardial myocardial strain, hence complicating the comparison of GLS values between vendors . Whether changes in these software releases influence GLS assessment remains unknown. Hence, we studied the effect of software product upgrade releases on measurement and intervendor consistency of GLS.
Methods
Study population
Patients referred for echocardiography at the echocardiography laboratory of the Saint Philibert Hospital (Lille Catholic University) during a 2-week period were screened for inclusion in the study. Inclusion criteria were: good visualization of all LV segments (allowing speckle tracking and measurement of LV GLS), sinus rhythm and consent to participate. Patients with poor echogenicity, i.e. speckle tracking not possible in at least one LV segment, were excluded. In addition, control subjects from the hospital staff with normal ECG, echocardiogram and free of any cardiovascular disease were asked to enroll in the study population.
Standard echocardiography and workflow
Transthoracic echocardiograms were acquired by experienced echocardiographers with two commercially available ultrasound transducers and equipment (M5S-D probe, Vivid E9, GE Medical; X5-1 probe, iE33, Philips) located in the same echocardiography room. Each participant underwent comprehensive assessment of cardiac anatomy and function with one of the ultrasound systems. The order of examination with the two machines was randomized. Acquisitions with both systems were performed during the same echocardiographic examination and patients remained in the same position. Sector size and depth were adjusted to achieve optimal visualization of all LV segments at the highest possible frame rate. At least three video loops of one cardiac cycle were obtained for apical views.
Speckle tracking strain echocardiography
QLAB software versions 9.0 and 10.2 were used for images obtained from Philips iE33, while EchoPAC software products 12.1 and 13.1.1 were used to analyse data obtained with GE Vivid E9. Segmental longitudinal strain values (on a 17-segment ASE model) were calculated from the three apical views obtained on each ultrasound machine and with each software package to obtain GLS (%). All acquired apical views were available for off-line quantification. GLS values were computed after having determined the onset of aortic valve closure using Doppler recordings or visual inspection of the kinetics of the aortic valve in long-axis views.
The automatic tracking of the endocardial contour was performed in end-systole with EchoPAC and in end-diastole with QLAB. Tracking was carefully verified and the region of interest was manually corrected to ensure optimal tracking and to cover the entire thickness of the LV myocardium. The multilayer two-dimensional (2D) strain speckle tracking (EchoPAC 13.1.1) starts, similarly to EchoPAC 12.1, by delineating the endocardial border; however, instead of a single chain of nodes, the myocardial wall is automatically defined with multiple chains of nodes, allowing investigation of the three myocardial layers: endocardial (endo), midwall (mid) and epicardial (epi) . Longitudinal 2D speckle tracking strain values were analysed off-line by a blinded investigator (A.L.C.). The same investigator performed image analyses with both the upgrade and the former release of each software (GE or Philips). A gap of more than 2 weeks was required between reading sessions. A second investigator (S.M.) independently repeated the strain analyses to test the interobserver variability of each software in a randomly selected set of 20 echocardiograms.
Statistical analyses
Baseline data are presented as means ± standard deviations (SDs) or numbers and frequencies. For the sake of clarity, GLS has been converted into absolute values to depict comparisons as recommended by the European Association of Cardiovascular Imaging (EACVI)/ASE/Industry Task Force to standardize deformation imaging . Student’s t -test was used to compare mean GLS values. Intraclass correlation coefficients (ICCs) were obtained to compare GLS values between software versions, between vendors and between observers. Bland–Altman plots of differences between GLS by QLAB 9.0 versus 10.2, EchoPAC 12.1 versus 13.1.1, QLAB 9.0 versus EchoPAC 12.1 and QLAB 10.0 versus EchoPAC 13.1.1 and mean values of GLS were produced to study potential bias and to obtain 95% limits of agreement (LOA) . Variations in the range of GLS values for intersoftware comparison, intervendor comparison and interobserver comparison were derived from these LOAs. Both ICCs and Bland–Altman results are reported as these methods can provide inconsistent results in agreement studies . Interobserver variabilities of the software products were also evaluated using coefficients of variation (CV [%] = 100 × SD of the difference between observers/mean of the difference between observers). Two-sided P values < 0.05 were considered statistically significant. Statistical analyses were performed using PASW 18.0 (IBM, Inc., Bois-Colombes, France) and MedCalc for Windows version 12.5.0 (MedCalc Software, Mariakerke, Belgium).
Results
Of 96 subjects who were screened for inclusion in the study, 23 were excluded for poor echogenicity. Hence, the final study cohort consisted of 73 subjects (65 patients and 8 healthy volunteers) ( Fig. 1 ). Baseline characteristics of the study population are summarized in Table 1 .
| All subjects ( n = 73) | |
|---|---|
| Demographics | |
| Age (years) | 61 ± 16 |
| Men | 43 (59) |
| Body mass index (kg/m 2 ) | 26.0 ± 0.6 |
| Diabetes | 18 (25) |
| Hypertension | 31 (42) |
| Smoking | 15 (21) |
| Dyslipidaemia | 28 (38) |
| Main echocardiography findings (patients with structural heart disease) | |
| Heart failure with preserved or depressed LV ejection fraction | 18 (25) |
| Ischaemic heart disease | 3 (4) |
| Hypertrophic cardiomyopathy | 3 (4) |
| Significant valvular heart disease | 12 (16) |
| Reason for echocardiography (patients without structural heart disease) | |
| Chest pain | 8 (11) |
| Other a | 21 (29) |
| Healthy volunteers | 8 (11) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
