Low value of simple echocardiographic indices of ventricular dyssynchrony in predicting the response to cardiac resynchronization therapy
Abstract
Aims
A recent study suggested that no single echocardiographic index of cardiac dyssynchrony can reliably identify candidates for cardiac resynchronization therapy (CRT). We examined the value of three simple echocardiographic indices for predicting the 6-month clinical and echocardiographic responses to CRT.
Methods and results
We analysed data from 75 CRT-D system recipients. Standard echocardiography was used to measure aortic pre-ejection delay (APED), interventricular mechanical delay (IVMD), and delayed activation of the left ventricular (LV) infero-lateral wall (OVERLAP). Clinical responders were defined as patients who had an improved status, based on rehospitalization for heart failure, NYHA class, and peak oxygen consumption. Echocardiographic responders had a ≥10% decrease in LV end-systolic volume. During the study, one patient died and five were lost to follow-up. Of the remaining 69 analysable patients, 50 (72.5%) were classed as clinical responders and 41 (59.4%) as echocardiographic responders to CRT. Before CRT implantation, APED, IVMD, and OVERLAP were similar in responders and non-responders. The value of these indices of dyssynchrony as single or combined predictors of the clinical or echocardiographic response to CRT was low, with sensitivities ranging between 4 and 63%, and specificities between 37 and 100%.
Conclusion
Simple echocardiographic indices of dyssynchrony were poor predictors of response to CRT.
Introduction
In the large randomized trials which confirmed the safety and efficacy of cardiac resynchronization therapy (CRT), the patients were selected on the basis of a New York Heart Association (NYHA) functional class III or IV, a left ventricular (LV) ejection fraction (EF) < 35%, and a QRS duration > 120 ms.1–4 Despite compelling evidence that CRT improves clinical status, cardiac function, and prognosis, approximately one-third of patients do not respond to therapy.5,6 It has been suggested that, by measuring mechanical dyssynchrony, echocardiography could reliably identify patients most likely to respond to treatment.7,8 In small, single-centre, non-randomized studies, several indices of ventricular dyssynchrony, derived from standard echocardiography, have been shown to be predictors of a potential response to CRT.9–13 The large, randomized Cardiac Resynchronization-Heart Failure (CARE-HF) trial was the first to include the presence of LV echocardiographic dyssynchrony in its inclusion criteria.4 Patients with a QRS interval between 120 and 149 ms were required to fulfil two of the three following criteria of dyssynchrony: (i) an aortic pre-ejection delay (APED) > 140 ms, (ii) an inter ventricular mechanical delay (IVMD) > 40 ms, and (iii) a delayed activation of the LV infero-lateral wall (OVERLAP). However, a very low percentage of patients included in this trial had a QRS duration between 120 and 150 ms, precluding a precise evaluation of these echocardiographic indices. A recent prospective, multicentre study found that some indices of dyssynchrony were associated with a high inter- and intra-observer variability, and that no echocardiographic measurement alone reliably improved the process of patient selection for CRT.14
We hypothesized that a combination of measurements might improve the predictive power of echocardiography, and we therefore designed the present study to assess the reliability of the three simple echocardiographic criteria used in the CARE-HF study, for predicting the 6-month clinical and echocardiographic responses to CRT.
Methods
IDEF 13 (clinical evaluation of CRT with implantable cardioverter defibrillator therapy) was a multicentre, international, randomized, double-blind, parallel-control trial, which examined the clinical and echocardiographic outcomes in recipients of CRT-D systems. All patients presented with LV systolic dysfunction, cardiac dyssynchrony, and symptomatic heart failure (HF), despite optimal drug treatment. The study, which conforms to the principles outlined in the Declaration of Helsinki, was approved by the Ethics Committee of each participating institution and by the appropriate National Ethics Committees. All patients gave written informed consent to participate in the trial. The study was conducted in the following six countries: France, UK, Germany, Italy, Poland, and USA.
Between December 2003 and March 2006, eligible patients underwent implantation of Alto™ MSP 617, Alto™ 2 MSP 627, or Ovatio™ 6750 CRT-D devices (Sorin Group CRM, Milan, Italy/ELA Medical, Inc., Le Plessis-Robinson, France). Patients were then randomly assigned in a 2:1 ratio to either optimal pharmacological therapy plus CRT ON, or to optimal pharmacological therapy plus CRT OFF. ICD therapy was activated in all patients. This analysis focuses on the CRT ON group only.
The inclusion criteria were: (i) NYHA functional class III or IV despite optimal pharmacological therapy, due to ischaemic or non-ischaemic cardiomyopathy, (ii) a LVEF ≤ 35%, (iii) a QRS duration ≥ 130 ms on the surface electrocardiogram, and (iv) presence of sinus rhythm. Patients were excluded if they (i) had undergone prior implantation of a permanent pacemaker or implantable cardioverter defibrillator, (ii) had suffered from a cardiac ischaemic event within the previous month or cerebral ischaemic event within the previous 3 months, or (iii) had presented with an atrial arrhythmia within the previous month.
Study design
Patients who fulfilled the enrolment criteria underwent a baseline evaluation, which included (i) an assessment of the NYHA functional class, (ii) a maximal treadmill exercise test, (iii) an assessment of quality of life by means of the Minnesota Living with Heart Failure® questionnaire, and (iv) a two-dimensional transthoracic echocardiogram. Devices were programmed to DDD mode at a backup rate of 30 bpm. The atrioventricular delay, which after implantation of the device was echocardiographically optimized to obtain the longest filling time without truncation of the A wave, remained unchanged throughout the follow-up.
When re-evaluated after 6 months of CRT, patients with a ≥10% decrease in LV end-systolic volume (LVESV) index were classified as echocardiographic responders.15 A clinical responder was defined as: (i) the patient had not been re-hospitalized for management of decompensated HF and the NYHA functional class had decreased by ≥1 point or, (ii) NYHA was unchanged and peak oxygen consumption (VO2) had increased by >10%.
Echocardiographic studies
Standard two-dimensional and Doppler echocardiograms were obtained as previously described.16 Measurements of ≥3 cardiac cycles were made and averaged. Left ventricular volumes and LVEF were calculated by Simpson's biplane equation in the apical four- and two-chamber views. The standard indices of ventricular dyssynchrony measured were (i) APED, (ii) IVMD, and (iii) OVERLAP. Aortic pre-ejection delay was measured as the time interval between QRS onset and onset of aortic ejection. Interventricular mechanical delay was defined as the difference between left and right ventricular pre-ejection delays. The third measurement was the OVERLAP of (a) infero-lateral LV wall contraction, using M-mode echocardiography, and (b) onset of diastolic filling, using the transmitral Doppler flow. The measurement feasibility was 100% for IVMD, 100% for APED, and 65% for OVERLAP. The cut-off values used to define the presence of dyssynchrony were the same as in the CARE-HF trial: (i) an APED >140 ms, (ii) an IVMD >40 ms, or (iii) the presence of an OVERLAP.
The echocardiographic data gathered by the investigators were analysed by a core laboratory (Ultrasound Consulting, Pessac, France).
The intra-observer (5.5% for APED, 4.9% for IVMD, and 7.1% for OVERLAP) and the inter-observer coefficient of variability (6.1% for APED, 6.2% for IVMD, and 7.3% for OVERLAP) of the core laboratory have been calculated and reported previously.10
Statistical analyses
Unless specified otherwise, the results are presented as means ± standard deviation. The efficacy of CRT was examined by comparing the measurements made at baseline with those made after 6 months of CRT, using Student's t-test for paired samples. The sensitivity, specificity, and positive and negative predictive value (P/NPV) of each echocardiographic index of cardiac mechanical dyssynchrony with respect to the 6-month clinical and echocardiographic responses to CRT were calculated. A P-value < 0.05 was considered significant.
Results
A total of 89 patients were considered for this analysis, of whom 14 (16%) were not included because their echo recordings were either in an inadequate file format or lost (14%), or of poor image quality (2%). The clinical and echocardiographic characteristics of the remaining 75 patients are shown in Table Table 1.
| Baseline characteristics | |
|---|---|
| Age, year | 64 ± 9 |
| Men/women, n (%) | 59 (79)/16 (21) |
| New York Heart Association functional class | 3.1 ± 0.3 |
| Quality of life score | 54 ± 22 |
| QRS duration, ms | 163 ± 17 |
| Underlying heart disease, n (%) | |
| Ischaemic | 49 (65) |
| Non-ischaemic | 26 (35) |
| Peak oxygen consumption, l/min/kg | 11.7 ± 3.73 |
| Left ventricular | |
| Ejection fraction, % | 27.9 ± 8.8 |
| End-diastolic volume, mL | 231 ± 88 |
| End-systolic volume, mL | 170 ± 79 |
- a Unless specified otherwise, the values are means ± SD.
Clinical outcomes
At the end of the 6-month follow-up period, six patients had to be excluded as follows: one due to a clinical cause (patient died from multiple organ failure), and five due to CRT device failure (device replacement or loss of capture). Therefore 69 patients completed the 6-month follow-up. The mean absolute changes in NYHA functional class, and mean percent changes in quality of life scores, peak VO2 during treadmill exercise test, and in LVEF and LV volumes at 6 months, are shown in Table Table 2.
| Outcome measure | Changes units (relative %) |
|---|---|
| New York Heart Association functional class | −1 ± 0.7 (−32 ± 21) |
| Quality of life score | −21 ± 21 (−34 ± 42) |
| Peak oxygen consumption | 1.3 ± 3.2 (18 ± 39) |
| Left ventricular | |
| Ejection fraction, % | 7.1 ± 10.3 (34 ± 52) |
| End-diastolic volume, mL | −21.3 ± 55.7 (−8 ± 24) |
| End-systolic volume, mL | −28.9 ± 51.3 (−16 ± 27) |
- a Values are means ± SD.
Clinical responders vs. non-responders to cardiac resynchronization therapy
After 6 months of CRT, 50 of the 69 analysable patients (72.5%) were classed as clinical responders to therapy. It is, however, noteworthy that the APED, IVMD, and presence of OVERLAP measured before CRT system implantation were similar in clinical responders (150 ± 37 ms, 51 ± 28 ms, and 23% of patients, respectively) and non-responders (149 ± 50 ms, 41 ± 30 ms, and 13% of patients, respectively). The value of each measurement, alone or in combination, as a predictor of the clinical response to CRT is shown in Table Table 3. An APED >140 ms was observed in 39 patients, corresponding to a 55% sensitivity and a 37% specificity. Similarly, the sensitivity and specificity of a >40 ms IVMD, present in 37 patients, were 63 and 56%, respectively, and those of an OVERLAP present in 12 patients were 23 and 88%, respectively. The presence, in 30 patients, of ≥2 variables was associated with a 53% sensitivity and 63% specificity. The sensitivity and specificity of the three indices of dyssynchrony combined (observed in two patients) in predicting a clinical response to CRT were 4 and 100%, respectively.
| Index | I: APED > 140 (n = 39) | II: IVMD > 40 (n = 37) | III: OVERLAP (n = 12) | ≥2 indices (n = 30) | I + II (n = 26) | I + III (n = 4) | II + III (n = 4) | I + II + III (n = 2) |
|---|---|---|---|---|---|---|---|---|
| Clinical response | ||||||||
| Sensitivity, % | 55 | 63 | 23 | 53 | 42 | 8 | 6 | 4 |
| Specificity, % | 37 | 56 | 88 | 63 | 74 | 100 | 95 | 100 |
| Predictive value, % | ||||||||
| Positive | 69 | 81 | 83 | 80 | 81 | 100 | 75 | 100 |
| Negative | 24 | 33 | 30 | 32 | 33 | 29 | 28 | 28 |
| Echocardiographic response | ||||||||
| Sensitivity, % | 54 | 57 | 24 | 45 | 37 | 5 | 10 | 5 |
| Specificity, % | 39 | 41 | 86 | 46 | 61 | 93 | 100 | 100 |
| Predictive value, % | ||||||||
| Positive | 56 | 57 | 75 | 57 | 58 | 50 | 100 | 100 |
| Negative | 37 | 41 | 40 | 34 | 40 | 40 | 43 | 42 |
Echocardiographic responders vs. non-responders to cardiac resynchronization therapy
After 6 months of reliable CRT, a ≥10% decrease in LVESV was observed in 41 patients (59.4%). The mean pre-implant APED, IVMD, and presence of OVERLAP in the echocardiographic responders (149 ± 41 ms, 46 ± 29 ms, and 24% of patients, respectively) and non-responders (150 ± 41 ms, 52 ± 30 ms, and 14% of patients, respectively) were similar. The values of each measurement, alone or in combination, as a predictor of a ≥10% decrease in LVESV are shown in Table Table 3. An APED >140 ms was associated with a 54% sensitivity and a 39% specificity. The sensitivity and specificity of a >40 ms IVMD were 57 and 41%, and those of an OVERLAP were 24 and 86%, respectively. The presence of at least two variables was associated with a 45% sensitivity and 46% specificity. The sensitivity and specificity of the three indices of dyssynchrony combined in predicting an echocardiographic response to CRT were 5 and 100%, respectively.
Discussion
Assessment of cardiac dyssynchrony in this study vs. other studies
QRS duration is the only criterion for the selection of candidates for CRT that has been widely validated.15 However, approximately 30% of patients do not respond clinically to CRT, and 45% have no evidence of reverse remodelling.17 Studies of cardiac mechanical dyssynchrony, assessed by means of non-invasive imaging techniques, have recently been conducted with a view to improving the selection of CRT responders.9–13 In clinical practice, echocardiography is readily available, and enables the non-invasive evaluation of regional and global ventricular function at the bedside.
Several techniques have been described to evaluate LV dyssynchrony and predict the response to CRT, including M-mode, two- and three-dimensional echocardiography, and tissue Doppler imaging (TDI). However, no consensus has been reached with respect to the baseline characteristics that reliably identify a candidate for CRT. The measurements of dyssynchrony among studies have been inconsistent, and direct comparisons between predictors of response to CRT have been few. Tissue Doppler imaging, using 2- 4- 6- or 12-segment models of the left ventricle, has been the most widely tested technique.8–11 However, the optimal number of LV segments that need to be included remains to be determined. Furthermore, despite the promising results of small, non-randomized studies from single centres, concerns related to the quality and reproducibility of TDI have intensified the quest for clinically more practical, less time-consuming, and less operator-dependent echocardiographic indices of dyssynchrony.
Aortic pre-ejection delay, measured as the time interval between the QRS onset and the onset of aortic ejection, reflects the interaction between ventricular electro-mechanical activation and ventricular pre- and afterload. Interventricular mechanical delay, expressed as the difference between left and right ventricular pre-ejection delays, is due to the delayed onset of LV contraction and relaxation, which has effects on the trans-septal pressure gradient, interventricular septal motion, and LV ejection. Aortic pre-ejection delay and IVMD are inter-dependent, since changes in the latter are, by definition, related to changes in the former. Left ventricular lateral wall contraction delay, i.e. the overlap of (i) infero-lateral wall contraction on M-mode echocardiogram, and (ii) onset of diastolic filling on transmitral Doppler flow, is a simple measurement, which requires no special instrumentation, though is limited to the analysis of a single LV segment.
Findings of our study in the context of previous observations
The present study showed that the simple echocardiographic indices of dyssynchrony applied in the CARE-HF trial,4 used either singly or in combination, were poor predictors of response to CRT. The PROSPECT study has suggested that the inter- and intra-observer reproducibility of dyssynchrony ascertained by conventional echocardiography may be higher than that based on TDI, and therefore may be clinically more applicable.14 Our analysis, however, suggests that these measurements are of limited value for selecting candidates for CRT. Used singly or in combination, they failed indeed to correlate with changes in clinical status or in echocardiographic measurements. The sensitivity and specificity of APED and IVMD as predictors of clinical or echocardiographic responses were low, and certainly not superior to what was achieved in studies using QRS duration. Recently published results from the CARE-HF trial18 showed that IVMD, along with aetiology of LV dysfunction, is an independent predictor of echocardiographic response at 18 months, defined as survival and end-systolic volume reduced by at least 40 mL. Our study could not evidence any prediction ability of IVMD towards clinical and echocardiographic responses, which may be related to the shorter follow-up (6 months) and a higher proportion of ischaemic patients (65 vs. 46% in CARE-HF). We found, however, a noteworthy specificity in the presence of OVERLAP alone for predicting clinical and echocardiographic responses to CRT, suggesting that the 20% of patients who, exhibited this index of dyssynchrony in this study, were suitable candidates for CRT. On the other hand, patients whose LV infero-lateral wall contraction is not delayed should not be excluded systematically, since the sensitivity of this index was very low. The combination of these three indices of dyssynchrony, which was present in only a few patients, was an insensitive though highly specific observation, which identified patients highly likely to respond to CRT. However, a strategy insisting on the coexistence of these three indices, while limiting the proportion of non-responders, would deny therapy to a large number of legitimate candidates for CRT, except if applied after a pre-selection based on current recommendation criteria.
Study limitations
The number of patients included in this study was too small, and the duration of follow-up too short to study infrequent endpoints, such as death or hospitalization. These limitations might also have been in the way of quantifying a fully developed LV remodelling process. Furthermore, we did not analyse patients separately to examine the possible effects of ischaemic heart disease, prior myocardial infarction, and scar burden on the long-term clinical and echocardiographic responses to CRT.
Conclusions
In the present study, the value of the CARE-HF echocardiographic indices of dyssynchrony as predictors of response to CRT was low, limiting their clinical use in the selection of candidates for CRT especially in a monoparametric approach.
Acknowledgement
We wish to thank Rodolphe Ruffy, MD, for reviewing the manuscript.
Conflict of interest: none declared.
