Author + information
- Received April 29, 2004
- Revision received October 31, 2004
- Accepted November 11, 2004
- Published online March 1, 2005.
- Cheuk-Man Yu, MD, FRCP*,* (, )
- Qing Zhang, BM, MM*,
- Jeffrey Wing-Hong Fung, MRCP, FHKAM*,
- Hamish Chi-Kin Chan, MRCP, FHKAM†,
- Yat-Sun Chan, MRCP, FHKAM*,
- Gabriel Wai-Kwok Yip, MRCP, FHKAM*,
- Shun-Ling Kong, BN, MN*,
- Hong Lin, BM, MM*,
- Yan Zhang, BM* and
- John E. Sanderson, MD, FACC*
- ↵*Reprint requests and correspondence:
Prof. Cheuk-Man Yu, Division of Cardiology, Department of Medicine and Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, N.T., Hong Kong, China.
Objectives This study was designed to investigate if tissue synchronization imaging (TSI) is useful to identify regional wall delay and predict left ventricular (LV) reverse remodeling after cardiac resynchronization therapy (CRT).
Background Echocardiographic assessment of systolic asynchrony is helpful to predict a positive response to CRT. Tissue synchronization imaging is a new imaging technique that allows quick evaluation of regional systolic delay.
Methods Tissue synchronization imaging was performed in 56 heart failure patients at baseline and three months after CRT. Regional wall delay was identified on TSI images and the time to regional peak systolic velocity (Ts) in LV was measured by the six-basal-six-mid-segmental model. Eight TSI parameters of systolic asynchrony were computed when Ts was measured in ejection phase or also included postsystolic shortening.
Results Severe lateral wall delay occurred in 17 patients, which predicted LV reverse remodeling (chi-square = 8.13, p = 0.004). Among the eight quantitative parameters of asynchrony, the predictive values were higher for parameters that measured Ts in ejection phase than in postsystolic shortening. The standard deviation of Ts of 12 LV segments in ejection phase (Ts-SD-12-ejection) was most powerful to predict reverse remodeling (r = −0.61, p < 0.001) and gain in ejection fraction (r = 0.53, p < 0.001). The area of the receiver-operating characteristic (ROC) curve was the largest for Ts-SD-12-ejection (0.90, p < 0.001), with a sensitivity of 87% and specificity of 81% at a cutoff of 34.4 ms. The combination of lateral wall delay with Ts-SD-12-ejection gave a sensitivity and specificity of 82% and 87%.
Conclusions Tissue synchronization imaging allows quick evaluation of regional wall delay, and combined with Ts-SD-12-ejection provides a reliable way of predicting reverse remodeling after CRT.
Cardiac resynchronization therapy (CRT) has been proven unequivocally beneficial for patients with advanced chronic heart failure (HF) with prolonged QRS complexes (1–3). Despite enthusiasm of giving this therapy to patients who fulfilled the current recommendation, nonresponse was observed in about one-third of patients who may not show clinical or left ventricular (LV) reverse remodeling response (2,4,5). It is appealing that electrocardiography is not an accurate marker of electromechanical delay, as electrical delay may not occur in patients with left bundle branch block (6), whereas significant mechanical asynchrony is absent in nearly 30% of patients with prolonged QRS duration (7). Recent studies have suggested the vital role of assessing systolic asynchrony by echocardiography to predict improvement of systolic function or LV reverse remodeling (4,5,8–11). Among various echocardiographic techniques, tissue Doppler imaging (TDI) has gained its acceptance by virtue of the ability to define regional timing and contractility, and is highly reproducible (3–5,8–11). Recently, TDI has evolved into another technical modality, tissue synchronization imaging (TSI). Tissue synchronization imaging portrays regional asynchrony on two-dimensional (2D) echocardiography by transforming the timing of regional peak velocity into color codes, which allows immediate visual identification of regional delay in systole by comparing the color mapping of orthogonal walls. In addition, quantitative measurement of regional delay is possible. However, the ability of TSI to assess systolic asynchrony and predict a positive response to CRT has not been explored. Therefore, the aims of the study were to assess if TSI is useful to predict LV reverse remodeling and improvement of systolic function after CRT, and to compare the predictive values of qualitative parameters that identified regional wall delay and eight different quantitative TSI parameters computed by different algorithms.
This study included 56 HF patients (mean age 66 ± 12 years, 72% men) who received CRT and were followed up for at least three months. Inclusion criteria included New York Heart Association functional class III (n = 44) or IV (n = 12) HF despite optimal pharmacologic therapy, evidence of LV systolic dysfunction with ejection fraction <40%, and electrocardiographic evidence of prolonged QRS of >120 ms, either in the form of bundle branch block or intraventricular conduction delay. The etiologies of HF were ischemic in 28 patients (50%) and nonischemic in 28 patients (50%). Atrial fibrillation was present in three patients. Medications included diuretics in all patients, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers in 95% of patients, beta-blockers in 72%, spironolactone in 34%, and digoxin in 18%. Serial echocardiography with TDI and strain rate imaging was performed before and three months after CRT. The study was approved and conducted in compliance with the regulations of the ethics committee of the institution. Written informed consent was obtained from all patients.
Biventricular device implantation
Biventricular devices were implanted as previously described (1–3). The LV pacing lead was inserted by a transvenous approach through the coronary sinus into a cardiac vein of the free wall. The LV lead was placed at the lateral vein in 37 patients, posterolateral vein in 10 patients, posterior vein in 2 patients, anterolateral branch in 3 patients, and anterior cardiac vein in 4 patients. Thirty-eight patients received an Attain system (Medtronic Inc, Minneapolis, Minnesota) and 18 patients received the Easytrak over-the-wire lead (Model 4512, Guidant Inc., St. Paul, Minnesota). Apart from four patients who received biventricular cardiac defibrillators, all the others received biventricular pacemakers (InSync, InSync III, Contak TR, or Contak TR2). The atrioventricular interval was optimized by Doppler echocardiography for maximal transmitral diastolic filling.
Standard echocardiography, including Doppler studies, was performed (Vivid 5 or Vivid 7, GE Vingmed Ultrasound, Horten, Norway). The left ventricular end-diastolic volume (LVEDV) and left ventricular end-systolic volume (LVESV) and ejection fraction were assessed by biplane Simpson's equation using the apical four- and two-chamber views. Cardiac output was assessed by pulse-wave Doppler echocardiography over the LV outflow tract by continuity equation. The +dp/dt was estimated from the continuous-wave Doppler for mitral regurgitation velocity curve. Myocardial performance index was calculated as the sum of isovolumic contraction and relaxation time divided by LV ejection time (5,9). Sphericity index was calculated by dividing the maximum long-axis by the maximum short-axis dimension (5,9). The severity of mid-systolic mitral regurgitation was assessed by the percentage jet area relative to the left atrial area in the apical four-chamber view. At least three consecutive beats of sinus rhythm were measured and the average value taken.
Tissue Doppler imaging was performed using apical four-chamber, apical two-chamber, and apical long-axis views for the long-axis motion of the ventricles (3,5,12). Two-dimensional echocardiography with TDI-color imaging views was optimized for pulse repetition frequency, color saturation, and sector size and depth and allowed a highest possible frame rate. At least three consecutive beats were stored and the images were analyzed offline for TSI by a customized software package (EchoPac for PC, GE Vingmed Ultrasound).
Tissue synchronization imaging is a parametric imaging tool derived from 2D TDI images. It automatically calculates and color-codes the time to peak tissue velocity (Ts) in every position in the image with reference to the QRS signal (3,5,7). The TSI algorithm detects positive velocity peaks within a specified time interval, and the color coding ranges from green (earliest), yellow, orange, to red (latest) within this interval. The algorithm uses the automatically detected QRS onset as a reference. With the event timing tool, the time from the onset of the QRS to the aortic valve opening or closure was first measured in a separately recorded Doppler spectrum or M-mode through the valve. Then in TSI mode, the event timing tool allows the start and end times of TSI to be adjusted manually to align with the corresponding aortic valve opening and closure markers on electrocardiogram. In this way, only the peak systolic velocities in the ejection phase will be measured. Subsequently the end times of TSI were extended into early diastole to measure for possibly a higher velocity peak after aortic valve closure, i.e., post-systolic shortening (PSS). For qualitative assessment, the wall with most severe delay was identified on the basis of TSI at the three apical views. This was defined as the presence of red (occasionally orange) color-coding in at least one segment of that wall.
A quantitative measurement tool allowed numerical calculation of the median time to peak velocity within a 6-mm diameter circular region of interest manually positioned within the 2D TSI image. The six-basal-six-mid-segmental model was used (3,5). The myocardial velocity curves were constructed with the TSI images simultaneously when necessary to confirm the pattern of myocardial motion. Reproducibility of TSI and TDI was calculated in 120 consecutive measurements. The intra- and interobserver variability was 4.2% and 5.9%, respectively, for TSI, and 3.3% and 4.9%, respectively, for TDI. Eight parameters of systolic asynchrony were computed by the software. Depending on the setting of the TSI interval, these parameters included the assessment of segmental Ts only in ejection phase or also included PSS when it was present. Parameters that assessed only ejection phase included:
• Standard deviation of Ts of the 12 LV segments (Ts-SD-12-ejection)
• Standard deviation of Ts of the 6 basal LV segments (Ts-SD-6-ejection)
• Maximal difference in Ts between any 2 of the 12 LV segments (Ts-12-ejection)
• Maximal difference in Ts between any 2 of the 6 basal LV segments (Ts-6-ejection)
Similarly, the four corresponding parameters that also assessed Ts in PSS were called Ts-SD-12-PSS, Ts-SD-6-PSS, Ts-12-PSS, and Ts-6-PSS, respectively. All these indices were automatically calculated by the dedicated software (EchoPac for PC) with the use of preset equations. These values were computed when all the 12 LV segments were sampled on the 2D TSI images.
For the comparison of parametric variables before and after CRT, paired sample ttest was employed. The comparison of echocardiographic parameters between patient groups was performed by unpaired ttest or chi-square test where appropriate. Correlation analysis was used to compare the relationship between TSI parameters of systolic asynchrony and the change of LV end-systolic volume or ejection fraction after CRT by comparing the Pearson correlation coefficients, and receiver-operating characteristic (ROC) curves were also analyzed. All data were expressed as mean ± SD. A p value <0.05 was considered statistically significant.
LV reverse remodeling and cardiac function
There was significant improvement in cardiac function after CRT for three months. This was evidenced by the gain in ejection fraction (26.1 ± 9.3% vs. 34.2 ± 10.8%, p < 0.001), +dp/dt (644 ± 193 vs. 893 ± 307 mm Hg/s, p = 0.001), and decrease in myocardial performance index (1.02 ± 0.36 vs. 0.87 ± 0.26, p = 0.004). Left ventricular reverse remodeling was evident with reduction of LVESV (134 ± 68 vs. 107 ± 61 cm3, p < 0.001) and end-diastolic volume (178 ± 75 vs. 155 ± 71 cm3, p < 0.001) and increase in sphericity index at end-systole (1.77 ± 0.25 vs. 1.94 ± 0.39, p = 0.001) and end-diastole (1.61 ± 0.20 vs. 1.73 ± 0.28, p = 0.001). Mitral regurgitation was also reduced (30 ± 20% vs. 22 ± 19%, p = 0.001).
Patterns of regional wall delay by TSI and its correlation with reverse remodeling
Two-dimensional TSI that evaluated regional wall delay qualitatively in ejection phase showed a heterogeneous pattern of delay before CRT. Majority of patients had more severe delay in one LV wall (40 of 56, 71%). Significant delay in 2 walls was evident in 14 patients (25%), whereas severe delay in 3 walls was seen in 2 patients (4%). The commonest site of most severe delay was the inferior wall (25 of 56 patients, 45%). This was followed by lateral wall (17 patients, 30%), posterior wall (14 patients, 25%), septal wall (9 patients, 16%), anterior wall (6 patients, 11%), and anteroseptal wall (3 patients, 5%) (Fig. 1).For inferior wall delay, 9 out of 25 patients had accompanying delay of other walls. Only the presence of lateral wall delay at baseline was associated with a positive reverse remodeling response (defined as a reduction of LVESV >15% 3 months after CRT). In the responders, 14 of 30 patients (47%) had a lateral wall delay, whereas this occurred only in 3 of 26 (12%) nonresponders (defined as a reduction of LVESV ≤15%) (chi-square = 8.13, p = 0.004). Therefore, the presence of most severe delay in lateral wall gives a sensitivity of 47% and specificity of 89% to predict reverse remodeling. Lateral wall delay was present in eight patients with ischemic and nine with nonischemic etiology of HF.
Prediction of reverse remodeling and improvement of systolic function by TSI
A comparison of various TSI parameters of systolic asynchrony between responders and nonresponders of LV reverse remodeling is shown in Table 1.In general, all the TSI parameters were significantly higher in the responders than nonresponders, illustrating more severe systolic asynchrony in the responders. However, the differences were more significant in those parameters where Ts were measured only in the ejection phase than those that included PSS. There was a very close correlation between Ts measured by myocardial velocity curves of TDI in ejection phase and Ts measured by TSI (r = 0.97, p < 0.001). In addition, the Ts-SD-ejection derived from myocardial velocity curves of TDI correlated very closely with those derived by TSI (r = 0.93, p < 0.001) (Fig. 2).A number of TSI-derived parameters of systolic asynchrony were tested on their predictive value of LV reverse remodeling (% reduction in LVESV) and improvement of systolic function (absolute gain in ejection fraction). In general, TSI parameters that measured Ts only in ejection phase were consistently better than those that also included the PSS phase for prediction of both reverse remodeling and gain in ejection fraction (Table 2).Among TSI parameters that measured Ts only in ejection phase, the Ts-SD-12-ejection and Ts-12-ejection were good predictors of LV reverse remodeling (r = −0.61 and −0.60, both p < 0.001), and were consistently better than Ts-SD-6-ejection and Ts-6-ejection (both r = −0.53, both p < 0.001) (Table 2, Fig. 3).For corresponding TSI parameters that also measured Ts in PSS, the predictive values were substantially reduced. Therefore, when the ROC curves were compared among all the TSI parameters, the areas were highest for Ts-SD-12-ejection and Ts-12-ejection (0.90 and 0.91, both p < 0.001), which were followed by Ts-SD-6-ejection and Ts-6-ejection, and were lowest for all TSI parameters that included PSS (Table 3,Fig. 4).
For prediction of gain in ejection fraction (Tables 2 and 3), as similar to reverse remodeling, the predictive values were consistently higher for TSI parameters that measured Ts in ejection phase than when PSS was included. In addition, those parameters that measured 12 LV segments were more powerful than those that measured 6 LV segments at the base.
The sensitivity, specificity, and cutoff values derived from ROC curves for each of the TSI parameters are listed in Table 4.The TSI parameters that measured Ts in ejection phase had reasonably good sensitivity and specificity for predicting reverse remodeling. On the other hand, the corresponding parameters that measured Ts, including the PSS, were suffering from low specificity for similar sensitivity. Combining Ts-SD-12-ejection with either Ts-12-ejection or Ts-6-PSS was attempted to examine if that would improve the predictive value (Table 4). The Ts-6-PSS was chosen as it was the best-performing parameter among those that included the PSS phase. It appeared that combining two parameters in the ejection phase did not significantly alter the predictive values. The Ts-SD-12-ejection combined with Ts-6-PSS, although resulting in an increase in sensitivity, resulted in undesirably low specificity (Table 4). On the other hand, combining Ts-SD-12-ejection and qualitative assessment of lateral wall delay was helpful (Fig. 5).If Ts-SD-12-ejection was computed only when lateral wall delay was absent, such algorithm gave a sensitivity of 82% and specificity of 87%.
Identifying nonresponders to CRT is a challenge to physicians (3–5). This study compared a number of parameters that assessed systolic synchrony by TSI to predict LV reverse remodeling and improvement of ejection fraction. For quantitative analysis, the predictive value for a positive response to CRT by TSI is higher when Ts in ejection phase only (rather than including PSS) were measured, in particular when 12 (rather than 6) LV segments were analyzed. Furthermore, qualitative assessment of lateral wall delay is helpful.
Left ventricular reverse remodeling is the structural premise to reveal the improvement of cardiac function and ventricular hemodynamics (13,14). Among various parameters of reverse remodeling, LVESV was demonstrated to be the most powerful predictor of clinical outcome (13,14). In patients receiving CRT, previous studies consistently reported a reverse remodeling response (3,15,16). Despite these positive results, lack of response to CRT occurs in more than one-third of patients receiving the therapy (3,15). The absence of systolic asynchrony despite prolonged QRS duration is the key factor if LV leads are placed appropriately over the free wall region (6,7). Therefore, a number of recent studies aiming to identify nonresponders of CRT were based on various echocardiographic techniques, in particular TDI (5,9,10). This study confirmed previous notions that the relative lack of systolic asynchrony is the main reason for the occurrence of volumetric nonresponders (9,11,15).
Previous studies observed that measuring Ts from myocardial velocity curves of TDI was very useful for quantitative assessment of systolic asynchrony (3,5,11). In particular, the asynchrony index (i.e., Ts-SD) was the single most powerful predictor of LV reverse remodeling after CRT (5,11). Tissue synchronization imaging is a technologic advancement in the assessment of systolic asynchrony by transforming the Ts into different color-coding. It has the advantage of providing a visual aid for quick identification of regional delay in the LV, even on 2D images.
The present study validated the good correlation between Ts or Ts-SD derived from myocardial velocity curve of TDI and TSI in ejection phase. Furthermore, the Ts-SD-ejection had a high predictive value for LV reverse remodeling and improvement of systolic function. Also, the predictive values were better when 12 rather than 6 segments were assessed: the more, the better. With the pre-installation of equations to compare various parameters of asynchrony and the advancement of computer hardware, it will take only a few minutes to measure Ts of 12 segments from three apical views to calculate the Ts-SD and Ts-12 automatically. If the time spent in these measurements is going to predict a positive response to CRT more accurately, it will be highly justified when balancing between the time and efficacy of the procedure. This is particularly valid for employing TDI as Ts is a highly reproducible signal in myocardial velocity curve (3,5,7). Whether there is a need to pursue “simpler” parameters of systolic asynchrony at the expense of reducing diagnostic accuracy becomes dubious, in particular when the time saved may be only a few minutes.
Another interesting finding is that parameters that measure PSS (if present) are inferior to Ts, which measures only ejection phase. A previous study observed that counting the number of basal LV segments with PSS correlated rather closely with LV reverse remodeling (10). However, such a method is semiquantitative and makes it hard to define the cutoff value of predicting a favorable response to CRT. Our modified method tried to examine whether quantitative assessment of timing by TSI that included PSS was a good predictor of volumetric response. To our disappointment, the correlation coefficient and areas of ROC curves were unfavorably reduced. Because of such limitations, combining any of the Ts parameters that included PSS did not increase the yield of Ts-SD-ejection.
Examination of wall delay in TSI-coded 2D images in ejection phase also provided new insight for the pattern of systolic asynchrony. In these patients, severe asynchrony could occur in single or multiple walls. Interestingly, although the LV lateral wall was not the commonest site of most severe delay, it was the only region that predicted a positive response to CRT with high specificity. This assessment is also a fast and easy one to perform. Therefore, we suggest the following algorithm of echocardiographic examination to quickly identify potential responders of CRT (Fig. 5). If TSI images in ejection phase show that delay is worst at the lateral wall, the patient has a high likelihood to show reverse remodeling response, and quantitative measurement may not be necessary. In the absence of severe lateral wall delay or presence of severe delay in sites other than the lateral wall, it is recommended to measure Ts-SD-12-ejection where a value above the cutoff of 34.4 ms is likely to predict a positive response.
In conclusion, TSI is a useful echocardiographic method to predict a reverse remodeling and gain in ejection fraction after CRT. Qualitative identification of lateral wall delay is a quick and specific guide to predict a favorable response. In the absence of lateral wall delay, quantitative computation of asynchrony index from 12 LV segments in ejection phase is beneficial.
We thank Andreas Heimdal, PhD, from GE-Vingmed Corp. for technical support of TSI software.
- Abbreviations and acronyms
- cardiac resynchronization therapy
- heart failure
- left ventricular/ventricle
- left ventricular end-systolic volume
- post-systolic shortening
- receiver-operating characteristic
- tissue Doppler imaging
- tissue synchronization imaging
- standard deviation of Ts of the 12 LV segments in ejection phase
- standard deviation of Ts of the six basal LV segments in ejection phase
- maximal difference in Ts between any of the 2 out of 12 LV segments in ejection phase
- maximal difference in Ts between any of the 2 out of 6 basal LV segments in ejection phase
- Received April 29, 2004.
- Revision received October 31, 2004.
- Accepted November 11, 2004.
- American College of Cardiology Foundation
- Yu C.M.,
- Chau E.,
- Sanderson J.E.,
- et al.
- Pitzalis M.V.,
- Iacoviello M.,
- Romito R.,
- et al.
- Fung J.W.,
- Yu C.M.,
- Yip G.,
- et al.
- Yu C.M.,
- Lin H.,
- Zhang Q.,
- Sanderson J.E.
- Sogaard P.,
- Egeblad H.,
- Kim W.Y.,
- et al.
- Yu C.M.,
- Fung J.W.,
- Zhang Q.,
- et al.
- Miyatake K.,
- Yamagishi M.,
- Tanaka N.,
- et al.
- Bristow M.R.,
- Gilbert E.M.,
- Abraham W.T.,
- et al.
- Stellbrink C.,
- Breithardt O.A.,
- Franke A.,
- et al.
- St. John Sutton M.G.,
- Plappert T.,
- Abraham W.T.,
- et al.