Summary
Background
Transthoracic echocardiography (TTE) is the most commonly used method for measuring left ventricular ejection fraction (LVEF), but its reproducibility remains a matter of controversy. Speckle tracking echocardiography assesses myocardial deformation and left ventricular systolic function by measuring global longitudinal strain (GLS), which is more reproducible, but is not used routinely in hospital practice.
Aim
To investigate the feasibility of on-line two-dimensional GLS in predicting LVEF during routine echocardiographic practice.
Methods
The analysis involved 507 unselected consecutive patients undergoing TTE between August 2012 and November 2013. Echocardiograms were performed by a single sonographer. Echogenicity was noted as good, moderate or poor. Simple linear regression was used to assess the relationship between LVEF and GLS, overall and according to quality of echogenicity. Receiver operating curve (ROC) analysis was used to identify the threshold GLS that predicts LVEF ≤ 40%.
Results
Mean LVEF was 64 ± 11% and GLS was –18.0 ± 4.0%. A reasonable correlation was found between LVEF and GLS ( r = –0.53; P < 0.001), which was improved when echogenicity was good ( r = –0.60; P < 0.001). GLS explained 28.1% of the variation in LVEF, and for one unit decrease in GLS, a 1.45 unit increase in LVEF was expected. Correlations between LVEF and GLS were –0.51 for patients in sinus rhythm ( n = 490) and –0.86 in atrial fibrillation ( n = 17). Based on ROC analysis, the area under the curve was 0.97 for GLS ≥ –14%, allowing detection of LVEF ≤ 40% with a sensitivity of 95% and specificity of 86%.
Conclusion
Two-dimensional GLS is easy to obtain and accurately detects LVEF ≤ 40% in unselected patients. GLS may be especially helpful when a suboptimal acoustic window makes LVEF measurement by Simpson’s biplane method difficult and in atrial fibrillation patients with low heart rate variability.
Résumé
Objectifs
L’échocardiographie transthoracique (ETT) est l’outil le plus couramment utilisé pour l’évaluation de la fraction d’éjection ventriculaire gauche (FEVG), un paramètre pronostique majeur dans la majorité des maladies cardiovasculaires. Cependant, sa reproductibilité demeure controversée. Les fonctionnalités du speckle tracking permettent l’évaluation de la déformation myocardique et de la fonction systolique du ventricule gauche, par la mesure du Strain Longitudinal Global (GLS), qui est plus reproductible. Cependant, le GLS n’est pas encore entré dans la routine échocardiographique. Par conséquent, son rôle dans la prédiction de la FEVG dans la vraie vie de notre pratique échocardiographique n’a pas été pleinement évalué.
Méthodes
Nous avons étudié comment le GLS bi-dimensionnel calculé en direct pendant l’examen échocardiographique pouvait prédire la FEVG (calculée par la méthode de Simpson biplan chez les patients en rythme sinusal) dans le cadre d’une activité échocardiographique de routine d’un laboratoire. Les examens ont été réalisés par un unique échographiste dans sa pratique clinique. L’ échogénicité des patients a été notée (bonne, moyenne, insuffisante). L’ analyse a porté sur 507 ETT consécutives et non sélectionnées, réalisées entre août 2012 et novembre 2013 (53 examens exclus). La régression linéaire simple a été utilisée pour évaluer la relation entre la FEVG et le GLS. Les tests ont été répétés pour chaque classe d’échogénicité. L’analyse de ROC a été utilisée pour identifier le seuil de GLS qui prédit une FEVG ≤ 40 %.
Résultats
Les indications les plus fréquentes d’ETT étaient l’accident vasculaire cérébral quel qu’il soit ( n = 235) et la sclérose en plaques ( n = 44). La FEVG médiane (l’intervalle interquartile, IQR) était de 65 % (59, 70) et le GLS médian (IQR) était de –19 % (–21,–16). La FEVG moyenne était 64 ± 11 %, et la moyenne du GLS était de –18,0 ± 4,0 %. Il existait une corrélation entre la FEVG et le GLS dans la totalité de la série ( n = 507), avec r = –0,53 ( p < 0,001) et l’équation de régression FEVG = –1,45 GLS + 38,04. Dans la classe des échogénicités insuffisantes ( n = 76), le coefficient de corrélation était r = –0,53 et l’équation de régression FEVG = –1,48 GLS + 41,65, où GLS expliquait 26,6 % du modèle donc de la variation de la FEVG, et pour une diminution d’une unité du GLS, on s’attend à une augmentation de 1,48 unités de la FEVG. Lorsque l’échogénicité était moyenne ( n = 187) ou bonne ( n = 244), la corrélation était encore meilleure (valeurs de r respectivement à –0,54 et –0,60). Tous les coefficients de corrélation étaient significatifs ( p < 0,001). Plus surprenant, les coefficients de corrélation de Pearson pour la relation entre FEVG et GLS chez les patients en rythme sinusal (RS, n = 490) et ceux en fibrillation auriculaire (FA, n = 17) étaient respectivement de –0,51 et –0,86 ( p < 0,001 pour les deux). Les équations de régression étaient : FEVG = –1,41 GLS + 38,66 pour les patients en RS et FEVG = –2,42 GLS + 26,82 pour les patients en FA. Le GLS expliquait 25,8 % et 73,8 % des variations de la FEVG en RS et en FA respectivement. Sur la base de l’analyse de ROC pour l’ensemble de la série, l’aire sous la courbe était de 0,97 et un GLS ≥ –14 % permettait la détection d’une FEVG ≤ 40 % avec une sensibilité de 95 % et une spécificité de 86 %.
Conclusion
Le GLS bi-dimentionnel semi-automatique est facile et rapide à obtenir pendant l’examen échographique. Les valeurs de GLS obtenues en direct permettent la détection fiable d’une FEVG ≤ 40 % chez des patients consécutifs, non sélectionnés et représentant notre pratique de routine en laboratoire d’échocardiographie. Le GLS peut notamment être utile quand une fenêtre acoustique sub-optimale rend la mesure de la FEVG par la méthode de Simpson biplan difficile et chez les patients en fibrillation auriculaire, à condition que la variabilité de la fréquence cardiaque soit faible.
Introduction
Echocardiography is the most commonly used diagnostic tool for left ventricular (LV) systolic dysfunction. Given the poor prognosis associated with this condition, early detection is warranted . Simpson’s biplane method is the reference approach for calculating LV ejection fraction (LVEF), but it is time consuming to perform and is prone to intra- and interobserver variability.
New echocardiographic techniques are now available for assessing LV systolic function, one of which is speckle tracking. This technique can be used to study global myocardial deformation using global longitudinal strain (GLS) . Measurement of this index is now possible on-line during the ultrasound examination, with no requirement for post-processing. The aim of this study was to determine the value of GLS in measuring LVEF and identifying LV systolic dysfunction during routine in-hospital echocardiographic practice in consecutive patients, regardless of the indication for transthoracic echocardiography (TTE).
Methods
The study population comprised consecutive patients undergoing TTE at Fondation Rothschild Hospital, Paris, France, between August 2012 and November 2013.
Echocardiographic analysis
TTEs were performed using a commercial ultrasound system (Vivid 7; GE Health Medical, Horten, Norway) with a 4 MHz transducer. All echocardiograms were performed by one experienced cardiologist (NB).
Standard TTE included LV analysis, LVEF calculation (using Simpson’s biplane method in patients in sinus rhythm and by visual evaluation in patients in atrial fibrillation), Doppler tissue imaging and calculation of LV GLS. First, a comprehensive analysis of the cardiac anatomy was performed. Importantly, the patient’s echogenicity was noted as good, moderate or poor, to allow further analysis by “acoustic window” subgroups. The quality of the acoustic window had to be sufficient to allow the calculation of LVEF and GLS, otherwise the TTE was excluded.
As the aim was to evaluate GLS in detecting LV systolic dysfunction in everyday clinical practice, no rereading by a second observer was done. All measurements were performed on-line, without post-processing. Final results of LVEF were obtained by averaging two to three measures. Simpson’s biplane LVEF was calculated using the manual tracing. All of the images were stored digitally on EchoPAC imaging analysis software (EchoPAC; GE Health Medical).
Speckle tracking global longitudinal strain analysis
The three apical views (four-, two- and three-chamber) were recorded, with a frame rate of between 70 Hz and 80 Hz. Careful manual tracking of the endocardial contour was performed on-line and the entire thickness of the myocardium had to be covered correctly to ensure good tracking. GLS values were calculated during the standard TTE as the mean of the global longitudinal strains of each apical view. Using this method, myocardial deformation was assessed in a semiautomatic manner, based on grey-scale images. Segmental strain values were also recorded, according to a 19-segment model.
Statistical analysis
Values are presented as means ± standard deviations, medians (interquartile ranges) or absolute counts and frequencies. Pearson’s test was used to seek correlations, and simple linear regression was used to assess the relationship between LVEF and GLS. The tests were repeated for each class of echogenicity and according to cardiac rhythm (sinus or atrial fibrillation). Logistic regression was used to determine the threshold of GLS that predicts abnormal LVEF in this population. Thus, receiver operating characteristic curve analysis was used to identify the optimal cut-off value and the sensitivity and specificity of GLS in detecting an LVEF ≤ 40%. A P value of < 0.05 was considered to be statistically significant. Statistical analyses were performed using STATA ® software, version 13 (StataCorp LP, College Station, TX, USA).
Results
A total of 558 TTEs were performed, 51 (9.1%) of which were excluded from the analysis for the following reasons: 19 patients were in atrial fibrillation and had important variability in heart rhythm; and in 32 patients the image quality was insufficient to use the speckle tracking technique and/or to assess LVEF by Simpson’s biplane method. The study population therefore comprised 507 consecutive TTEs.
The indications for TTE were as follows: ischaemic stroke or transient ischaemic attack and haemorrhagic stroke ( n = 235), multiple sclerosis ( n = 44), dyspnoea ( n = 24), evaluation of various treatments ( n = 19), heart murmur exploration and control of valvular disease, endocarditis and suspicion of endocarditis ( n =17), LVEF evaluation and/or heart failure ( n = 17), preoperative assessment ( n = 15), hypertension ( n = 12), diabetes mellitus ( n = 11), assessment of the pericardium ( n = 5), occlusion of the central retinal artery ( n = 4), ischaemic heart disease ( n = 4), dissection of supra-aortic vessels ( n = 4), subarachnoid haemorrhage ( n = 3), assessment of pulmonary pressures ( n = 3) and other indications ( n = 90).
Table 1 displays the main clinical and echocardiographic characteristics of the population. Of note, the study population was relatively young, with a mean age of 59 ± 18 years; the median age was 60 years. The two youngest patients were aged 8 years and 14 years, and the oldest patient was aged 95 years.
| Characteristic | |
|---|---|
| Age (years) | 59 ± 18 |
| Men | 272 (53.7) |
| Atrial fibrillation | 17 (3.4) |
| LVM indexed to body surface (g) | 96.7 ± 33.3 |
| Wall-motion abnormality | 103 (20.3) |
| LVEF (%) | |
| Overall, mean a | 64 ± 11 |
| Overall, median a | 65 (59, 70) |
| According to Teichholz | 68 ± 10 |
| Without LVWMA a | 67 ± 8 |
| With LVWMA a | 54 ± 15 |
| GLS (%) | |
| Overall, mean | –18.0 ± 4.0 |
| Overall, median | –19 (–21,–16) |
| A3C | –17.9 ± 4.8 |
| A4C | –17.6 ± 4.1 |
| A2C | –18.4 ± 4.5 |
| Without LVWMA | –19.1 ± 3.0 |
| With LVWMA | –13.6 ± 4.7 |
| Echogenicity | |
| Good | 244 (48.1) |
| Moderate | 187 (36.9) |
| Poor | 76 (15.0) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
