Strain dyssynchrony index determined by three-dimensional speckle area tracking can predict response to cardiac resynchronization therapy

The study included 15 consecutive HF patients with New York Heart Association functional class III or IV, ejection fraction ≤ 35%, and QRS duration ≥ 120 msec, who underwent CRT. One of these patients was excluded from all subsequent analyses because of poor echocardiographic image quality, so that eventually 14 patients were enrolled in this study (Table 1). Their mean age was 70 ± 9 years, ejection fraction was 27 ± 7%, and QRS duration was 172 ± 30 msec. Five patients (36%) had ischemic cardiomyopathy, defined as the presence of ≥75% stenosis of at least one major epicardial coronary artery and/or prior coronary revascularization. Ten patients (71%) were diagnosed with sinus rhythm and there was no patient with atrial fibrillation. In addition, four of the patients (29%) had previously undergone implantation of a permanent right ventricular (RV) pacing device at least one year before enrollment and featured predominantly RV pacing, which was defined as ≥ 90% paced when the device was interrogated at the time of enrollment. Whenever tolerated, all patients were on optimal pharmacological therapy. Written informed consent was obtained from all patients.

All echocardiographic studies were performed with a 2.5-MHz 3-D matrix array transducer and 3-MHz sector transducer using a commercially available echocardiography system (Aplio Artida, Toshiba Medical Systems, Tochigi, Japan). Patients were studied before and 6 ± 1 months after CRT. We acquired digital LV 3-D volume data from the apical view with 4- or 6-beat electrocardiogram-gated acquisition. Routine digital grayscale 2-D and tissue Doppler cine loops were obtained, including mid-LV short axis views at the level of the papillary muscle and standard apical views (4-chamber, 2-chamber, and long-axis). Sector width was optimized to allow for complete myocardial visualization while maximizing the frame rate. Mean volume rates were 22 ± 3 volumes/s in the apical views for gray scale imaging used for 3-D speckle tracking analysis, 45 ± 4 frames/s in the short axis view for gray scale imaging used for 2-D speckle tracking analysis, and 44 ± 3 frames/s in the apical axis view for tissue Doppler imaging used for tissue Doppler velocity analysis. LV end-diastolic volume, end-systolic volume (ESV), and ejection fraction were obtained with the modified biplane Simpson's method[13]. A response to CRT was defined as a reverse remodeling detected by a relative reduction of ≥ 15% in ESV at the 6-month follow-up after CRT.

Our relatively simple version of the strain dyssynchrony index with 2-D speckle tracking strains represents the average of the wasted energy due to LV dyssynchrony as previously described in detail[10]. In the current study, the strain dyssynchrony index was calculated by means of area tracking. Briefly, ASDI analysis was performed with the aid of 3-D speckle area tracking strain using complete 3-D pyramidal datasets. First, a region of interest was traced on the endocardial cavity with a point and click approach. A second larger concentric circle was then automatically generated and manually adjusted near the epicardium (Figure 1). The 3-D area strain curve of the endocardium was calculated automatically (Figure 2 and 3). ASDI was then calculated as the average difference between peak and end-systolic area strain derived from 16-segment (Figure 2). The peak of the Q wave on the electrocardiogram was automatically used as the reference time point for end-diastole. The timing of minimum LV volume, determined by means of 3-D speckle tracking echocardiography, was used as the reference time point for end-systole. The wasted energy per segment of the endocardium as a result of dyssynchrony was expressed as the difference between peak strain (ε-peak) and end-systolic strain (ε-ES) (Figure 2). This difference increases with an increase in the degree of dyssynchrony, because this increase leads to a decrease in ε-ES. ASDI was then calculated as the average of the absolute difference between ε-peak and ε-ES derived from 16 segments. Three cardiac cycles were recorded and averaged for each measurement. If a segment showed a positive area strain during the entire cardiac cycle, the difference between ε-peak and ε-ES was assumed to be 0.

The interobserver and intraobserver reproducibility of the measurements of ASDI for randomly selected 10 patients was evaluated by two independent observers. To assess the interobserver reproducibility, measurements for all patients were analyzed by the second observer blinded to the values obtained by the first observer. For assessment of the intraobserver reproducibility, measurements of all patients were analyzed on 2 consecutive days by an observer blinded to the results of the previous measurements. The intraclass correlation coefficient was used to evaluate interobserver and intraobserver reproducibility.

Continuous variables were expressed as mean ± SD. The paired t test was used to compare group data obtained before and 6 months after CRT, and the unpaired Student's t test for comparison of group data for responders and non-responders. The diagnostic performance of LV dyssynchrony indexes for predicting response to CRT was evaluated by means of receiver operating characteristic curve analysis. Univariate linear correlation analysis was used for a comparison of baseline ASDI and reduction in ESV after CRT. For all tests, p value < 0.05 was considered statistically significant. All the analyses were performed with commercially available software (SPSS version 15.0, SPSS Inc., Chicago, IL, USA).

The baseline clinical and echocardiographic characteristics of the 14 patients are summarized in Table 1. Overall findings showed that CRT reduced LV end-diastolic volume and LVESV (from 175 ± 72 to 153 ± 68 ml, p < 0.005 and from 132 ± 67 to 98 ± 53 ml, p < 0.001, respectively) and increased LV ejection fraction (from 27 ± 7 to 38 ± 9%, p < 0.001). Response to CRT, defined as a relative decrease in ESV of ≥ 15%, was observed in nine patients (64%), and the remaining five patients (36%) were classified as non-responders. Compared with non-responders, responders were more likely to have higher ASDI (4.8 ± 1.1% vs. 2.8 ± 0.7%, p < 0.005) (Table 1). In contrast, there was no significant difference between responders and non-responders in the IVMD, Yu index, and 2-D radial dyssynchrony determined by speckle tracking strain (Table 1).

Of the individual parameters including ASDI and conventional dyssynchrony measures, ASDI ≥ 3.8% proved to be the best predictor of response to CRT with a sensitivity of 78%, specificity of 100% and area under the curve (AUC) of 0.93 (p < 0.001, Table 2). 2-D radial dyssynchrony was also found to be a successful predictors for response to CRT with an AUC of 0.82 (p < 0.005, Table 2). IVMD and Yu index, on the other hand, were not predictive of response to CRT.

Using pre-defined cut-off values for the conventional dyssynchrony parameters, sensitivity of IVMD, Yu index, and 2-D radial dyssynchrony was found to be 67%, 11%, and 100%, and specificity 60%, 100%, and 60%, respectively (Table 2).

Noteworthy was that, ASDI ≥ 3.8% was associated with the highest incidence of clinical improvement after CRT with a response rate of 100% (7/7). On the other hand, pre-defined cut-off values of the conventional dyssynchrony parameters showed that CRT response rates were 67% (6/9), 62% (8/13), and 82% (9/11), respectively (Figure 4). Furthermore, baseline ASDI correlated significantly with % reduction in ESV after CRT (r = 0.80, p < 0.001; Figure 5). In contrast, there was no correlation between the conventional dyssynchrony parameters and % reduction in ESV. Figure 3 shows representative findings of baseline ASDI for a post-CRT responder patient.

The intraclass correlation coefficient for interobserver reproducibility was 0.961 (95% CI, 0.842 to 0.990), and the intraclass correlation coefficient for intraobserver reproducibility was 0.966 (95% CI, 0.864 to 0.992).

CRT is an established therapeutic option for HF patients with severe symptoms and a wide QRS complex[1‐8]. Although initial results have been promising, roughly one-third of patients selected according to standard clinical criteria do not respond to CRT. Tissue Doppler imaging was useful to evaluate LV longitudinal function [24], and a number of publications to quantify LV dyssynchrony and predict response to CRT has focused on LV longitudinal shortening velocities on tissue Doppler imaging from the apical views. However, the ability of echocardiographic measures of dyssynchrony, in particular tissue Doppler imaging to predict response to CRT, has recently been criticized as a result of the findings by the predictors of responders to CRT (PROSPECT) trial[14]. On the contrary, speckle tracking strain has the advantage of differentiating active contraction from passive motion and is not affected by Doppler angle of incidence. Since response to CRT may be associated with LV mechanical dyssynchrony[5, 6, 17, 25], LV lead position[17, 26‐29], scar burden or myocardial viability[30‐34], irreversibly advanced HF[34], and AV and VV optimization, a comprehensive approach addressing these factors may well be required to minimize non-responders to CRT. The strain delay index reported by Lim et al[9] and our relatively simple version of the strain dyssynchrony index[10], both determined by 2-D speckle tracking strain, proved to be strong predictor of response to CRT. These two indexes combined thus constitute a marker of both LV mechanical dyssynchrony and residual myocardial contractility.

The LV myocardial function represents three different patterns of myocardial deformation, including radial direction (myocardial thickening), circumferential direction (myocardial shortening), and longitudinal direction (myocardial shortening). Because LV dyssynchrony is also a 3-D phenomenon, these three types of deformation do not provide the same information about the failing heart. STAR (the Speckle Tracking and Resynchronization) study, which is the recently conducted first prospective multi-centre study to assess the utility of radial, circumferential, transverse and longitudinal speckle tracking strain for predicting response to CRT and important long-term outcome events after CRT, verified the utility of multidirectional analysis for the quantification of LV dyssynchrony[25]. This study demonstrated that patients who lacked dyssynchrony before CRT, as determined by either the 2-D radial or transverse speckle tracking strain approach, suffered had serious unfavorable clinical events three times more frequently than those with significant baseline dyssynchrony. Furthermore, lack of dyssynchrony before CRT, as determined by the combined use of 2-D radial and transverse speckle tracking strains, was associated with implantation of a left ventricular assist device, heart transplant or death in approximately 50% of patients, in contrast to these unfavorable events occurring in 11-13% of patients if baseline 2-D radial or transverse speckle tracking dyssynchrony were present. Moreover, we previously reported that combining assessment of 2-D radial, circumferential, and longitudinal strain dyssynchrony indexes can enhance the prediction of CRT responders[10].

In this study, we found that baseline ASDI, as determined by means of novel 3-D speckle area tracking strain can predict response to CRT as well as predict LV reverse remodeling following CRT. 3-D area strain may provide new and more predictive endocardial information, because 3-D area strain coupled with both 3-D longitudinal and circumferential factors pertaining to the endocardium is considered to be the most sensitive to changes in myocardial function, especially in the failing heart or as a result of ischemia. 3-D area strain thus appears to be the ideal parameter to express LV dyssynchorny[35]. Furthermore, ASDI, which represents circumferential and longitudinal mechanical dyssynchrony and residual endomyocardial function, provide more comprehensive information for predicting response to CRT than do other methods.

3-D speckle tracking strain method was a novel developed technology and used in a few human studies. The potential advantages of 3-D speckle strain method were expression of myocardial/endocardial function of the whole heart, independency of tomographic imaging planes, and being able to analyze regional ventricular function using 3 different 3-D strains (radial/transverse, circumferential, and longitudinal) from the same heart beat acquisition. On the contrary, the potential disadvantages were relatively low volume rate of 25-30 volume/sec and relatively low spatial resolution. However, these disadvantages will be improved with technical development in the future.

3-D speckle tracking is a simple, feasible, and reproducible method for quantifying LV dyssynchrony[11, 12], and it is considered to be faster than 2-D speckle tracking strain analysis[36]. As previously mentioned, comprehensive assessment may be crucial for selecting HF patients who will benefit most from CRT. It therefore appears that ASDI, which represents circumferential and longitudinal mechanical dyssynchrony as well as residual endomyocardial function, could be an alternative method for the assessment of HF patients in order to minimize non-responders to CRT. In view of the present study, the clinical algorithm is that a patient, whose ASDI is >3.8%, should be considered undergoing CRT.

Because the area tracking by means of 3-D speckle tracking system was developed only recently, this study covered a very small number of patients in a single-center study. Future studies of larger patient populations are therefore needed to test the accuracy of ASDI for predicting response to CRT and LV reverse remodeling. Moreover, the problem of the small number of the patients may be enhanced by the heterogeneous study population including 3 patients with ischemic cardiomyopathy and 4 patients with previously undergone implantation of a permanent RV pacing device. However, previous investigators have demonstrated that HF patients with RV pacing had similar dyssynchronous patterns and similar EF response and long-term outcome compared to those with left bundle branch block[11]. A limitation of the image acquisition for 3-D speckle tracking is the relatively slow volume rate of 25 to 30 volumes/s. The much slower volume rates of 3-D speckle tracking system compared to 2-D may limit analysis of rapid events such as isovolumic contraction and relaxation phase. Nevertheless, the reproducibility of ASDI in this study was acceptable. Finally, this 3-D speckle tracking strain methodology has been validated against sonomicrometry in animals[37], but there is no true non-invasive "gold standard" technique that can be used in humans to validate regional ventricular function. Therefore, this 3-D speckle tracking strain method does not establish the accuracy in humans.