| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1Division of Cardiology, Department of Medicine, 2Division of Cardiothoracic Surgery, Department of Surgery, and 3Carlyle Fraser Heart Center, Emory University School of Medicine, Emory University, Atlanta, Georgia
Submitted 9 February 2007 ; accepted in final form 16 April 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
tissue Doppler; strain
Recent developments in speckle tracking (13) allow assessment of radial and circumferential strain on B-mode echocardiography similar to tagged magnetic resonance imaging (1, 7). Whether conventional echocardiography, TDI, or novel speckle tracking accurately detect mechanical dyssynchrony remains undetermined; the various clinically utilized echocardiography-Doppler techniques have not undergone extensive testing in an experimental model of dyssynchrony. The present study tests the accuracy of detecting dyssynchrony of multiple echocardiography parameters in an experimental canine model of dyssynchrony with or without heart failure, with specific attention to a comparison between novel speckle tracking-derived radial and circumferential strain and widely adopted TDI-derived longitudinal velocity methods.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
The experimental procedures complied with Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publication No. 85-23, revised 1996) and were approved by The Institutional Animal Care and Use Committee of Emory University. (Supplemental data for this article is available online at The American Journal of Physiology-Heart and Circulatory Physiology website).
Adult mongrel dogs underwent atrioventricular junction ablation followed by implantation of a dual-chamber pacing system. Atrioventricular pacing (atrioventricular interval of 150 ms) at 110 bpm induced dyssynchrony and wide QRS without heart failure (D group, n = 12), and right-ventricular (RV) pacing at 170 bpm induced dyssynchrony and wide QRS with heart failure (DHF group, n = 9). A third group underwent implantation of a dual-chamber pacing system without junctional ablation and received atrial pacing at 190 bpm for 4 wk to induce heart failure with narrow QRS (HF group, n = 8). Intracardiac recordings confirmed the absence of intrinsic atrioventricular conduction in the animals that underwent junctional ablation. We measured QRS duration on surface lead II.
Echocardiography
We performed tissue Doppler and two-dimensional echocardiography (Vivid 7; General Electric, Vingmed, Norway) at baseline and 4 wk after pacing with animals under general anesthesia (closed chest). All the echocardiographic images were obtained at intrinsic normal sinus rhythm both at baseline and 4 wk later. One cardiologist (D. Sorescu) obtained the images, and a second investigator (T. Arita) analyzed them offline. We derived the ventricular volumes and ejection fractions by using the Simpson biplane method. Color-coded tissue Doppler images were obtained on apical two-, three-, and four-chamber views with a frame exceeding 150 frames/s (see supplemental video online). Pulse-wave tissue Doppler images were obtained on apical two-, three-, and four- chamber views. We excluded the foreshortened apical image view for measurement. TDI involved sampling a region of interest in the basal six and mid-six segments. The speckle-tracking analysis used two-dimensional images obtained in the mid-level short axis with a frame rate of 70–110 frames/s. We generated the mid-level short-axis images at the level of both papillary muscles perpendicular to the long axis of the ventricle. We recorded and stored at least three clips of three to five consecutive cardiac cycles of each apical and short-axis image for offline digital analysis. All the measured parameters except speckle-tracking analysis were calculated as the average of three consecutive beats.
Assessment of Dyssynchrony
Color-coded tissue Doppler image. We measured longitudinal dyssynchrony as the standard deviation (SD) of time from onset of QRS to peak systolic velocity (TPVs) within the ejection phase across 12 (6 basal and 6 mid) segments according to a previously described method (28) (Fig. 1A).
|
M mode. We defined the septal-to-posterior wall motion delay (SPWMD) as the time difference between the peak inward excursions of the interventricular septum and left posterior wall on M-mode images obtained from the mid-level short-axis view (17) (Fig. 1C).
Speckle tracking of short-axis B-mode image. Analysis of mid-level short-axis images used dedicated software (EchoPac PC, BTO5.0.1; General Electric) to determine radial and circumferential dyssynchrony (Fig. 1, D and E). After tracing the endocardial border at the end systole, we set the region of interest from the endocardium to the epicardial edge. Automatic speckle tracking analyzed each short-axis image to track motion of speckles contained within the region of interest on a frame-to-frame basis. The software automatically computed distances between adjacent speckles to obtain the radial and circumferential strain curves as a function of time. Assessing radial and circumferential dyssynchrony involved exporting the strain data for automatic processing on a customized program (Matlab; MathWorks, Natick, MA) to calculate the SD of time from onset of QRS to peak radial and circumferential strain (Err and Ecc, respectively) across the six mid-level segments.
Reproducibility of Echocardiographic Analysis
We used the Bland-Altman method and standard error of the mean to assess inter- and intraobserver variability.
Intraobserver variability. Bland-Altman analysis computed the limits of agreement for TPVs (5.2 ± 20.9 ms), pkS (–4.3 ± 14.4 ms), SPWMD (–2.6 ± 28.9 ms), Err (–1.4 ± 15.8 ms), and Ecc (11.6 ± 76.4 ms). The variability for each parameter measured 7.0, 6.8, 14.4, 2.3, and 10.4%, respectively.
Interobserver variability. To determine the interobserver variability, two independent observers (T. Arita and G. P. Sorescu) performed independent analysis of the same echocardiographic images. Average differences were as follows: TPVs, –1.3 ± 19.4 ms; pkS, 0.25 ± 24.5 ms; SPWMD, –6.2 ± 23.0 ms; Err, –6.2 ± 33.4 ms; and Ecc, –4.7 ± 81.0 ms. The variability for each parameter measured 9.8, 7.9, 17.3, 6.3, and 15.4%, respectively.
Statistical Analysis
All data appear as means ± SD. For statistical analysis, we used the Wilcoxon t-test or Mann-Whitney U-test to analyze differences in the means. Receiver-operating curve (ROC) analysis was performed to obtain the cutoff point for determining intraventricular mechanical dyssynchrony with or without heart failure. Area under the curve (AUC) was computed for each significant parameter. In ROC analysis, the DHF and D groups served as separate positive controls, with the combined baseline group serving as a negative control. The relationship between dyssynchrony parameters and basic echocardiographic parameters was analyzed by using linear regression and is expressed as a Pearson's correlation coefficient followed by stepwise multivariate regression analysis. Statistical analysis was performed with commercial software (SPSS 13.0; SPSS, Chicago, IL). We defined the two-sided P value of <0.05 as statistically significant.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Figure 2 and Table 1 contain the results of the ablation and pacing intervention in the three groups of animals. RV pacing after His-bundle ablation increased the QRS duration from 91.2 ± 12.9 to 166.7 ± 14.6 ms but did not affect ejection fraction (D group). Rapid atrial pacing caused systolic dysfunction (left-ventricular ejection fraction reduced from 58.7 ± 6.0 to 22.2 ± 4.7%) but produced no significant change in QRS duration (HF group). Rapid RV pacing and His-bundle ablation worsened left-ventricular ejection fraction (58.4 ± 9.4 to 21.4 ± 7.6%) and prolonged QRS duration (93.7 ± 6.0 to 166.1 ± 17.5 ms ) to produce dyssynchrony and heart failure (group DHF).
|
|
Color-coded and pulse-wave tissue Doppler analysis of longitudinal dyssynchrony. Junctional ablation or rapid pacing alone did not change the color-coded tissue Doppler-derived TPVs in the D and HF groups. The combination of ablation and rapid RV pacing increased TPVs in the DHF group (Table 2 and Fig. 3A). Rapid atrial pacing-induced heart failure (HF with narrow QRS) and the ablation/rapid RV pacing combination (DHF, heart failure with wide QRS) increased the pulse-wave tissue Doppler-derived pkS; dyssynchrony alone (D group) did not affect the pkS (Table 2 and Fig. 3B). Applying multiple published color-coded and pulse-wave tissue Doppler methods found no statistical significant difference in time to peak longitudinal velocity in the D group; ablation and rapid RV pacing (DHF group) did affect some parameters (see Supplemental Table 1 online).
|
|
Radial and circumferential strain analysis from short axis. The Err increased significantly in all three groups (Table 2 and Fig. 3D). The DHF group demonstrated the greatest change in the Err (greater than fivefold increase in mean SD; P = 0.001), with less change in the HF and D groups [twofold (P = 0.019) and greater than threefold (P = 0.018) increase over baseline, respectively]. The DHF group showed significantly greater Err than the D and HF groups (both P < 0.001). The results suggest that Err detects dyssynchrony irrespective of heart failure and may also reflect the degree of dyssynchrony.
The circumferential short-axis strain increased in all three groups (Table 2 and Fig. 3E). Circumferential dyssynchrony increased in the D and DHF groups (P = 0.03 and P < 0.001, respectively). However, the lack of difference in the changes detected in the HF and DHF groups (P = 0.20) suggests that, compared with radial strain analysis, circumferential strain analysis has sensitivity but lacks specificity in detecting dyssynchrony.
ROCs for Echocardiographic Parameters of Dyssynchrony
Using the D group as the positive control for dyssynchrony shows that Err (AUC = 0.819) and Ecc (AUC = 0.874) effectively detect dyssynchrony without heart failure (Table 3). Using the DHF group as the positive control shows that radial and circumferential dyssynchrony detect dyssynchrony in heart failure better (AUC = 1) than tissue Doppler- and M-mode-derived parameters (Table 4). Tissue Doppler and M-mode parameters appear insensitive, whereas radial and circumferential dyssynchrony appear more sensitive and specific for detecting dyssynchrony independent of QRS duration. All parameters detect dyssynchrony with adequate sensitivity and specificity in heart failure associated with prolonged QRS, but radial dyssynchrony had the best overall sensitivity and specificity.
|
|
Adopting cutoff values from the ROC analysis (using DHF as positive control) yields the following results. With cutoff point for Err > 63.94 ms, 37.5% (3 of 8) of heart failure animals demonstrated radial dyssynchrony; with a cutoff value for Ecc > 107.69 ms, 50% (4 of 8) showed circumferential dyssynchrony, but results were less specific.
Relationship Between Echocardiographic Dyssynchrony Parameters and Global Cardiac Function
The longitudinal dyssynchrony parameters (i.e., TPVs and pkS) correlated with ejection fraction, end-diastolic volume, and end-systolic volume but not with QRS duration during univariate analysis. The radial strain-derived dyssynchrony parameter correlated with ejection fraction, end-diastolic volume, end-systolic volume, and QRS. The circumferential strain-derived dyssynchrony parameter and SPWMD correlated with ejection fraction, end-systolic volume, and QRS but not end-diastolic volume (Table 5). Multivariate analysis showed that TPVs and pkS independently correlated with ejection fraction (P < 0.001) but not with QRS duration. In contrast, SPWMD and Err correlated both with ejection fraction and QRS duration (P < 0.01) and Ecc correlated with ejection fraction, end-diastolic volume, and QRS duration (P < 0.01; Table 6).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
Preliminary clinical reports demonstrate the value of echocardiography for assessing cardiac dyssynchrony and predicting response to CRT (2, 17, 28). However, these clinical parameters to measure mechanical dyssynchrony lack testing in an animal model for cardiac dyssynchrony. Because of large variability of clinical variables in humans, to better test the sensitivity and specificity of widely used echocardiography measures of dyssynchrony, we designed an experimental model of dyssynchrony and heart failure using the same echocardiographic technology used in humans. We used three canine groups: HF with narrow QRS (fast atrial pacing with intact atrioventricular node), dyssynchrony without HF (wide QRS, RV pacing only without heart failure), and the combination of both (RV pacing with heart failure and wide QRS and heart failure). We chose a model of RV apical pacing after His-bundle ablation for its similarity with dyssynchrony encountered in humans with spontaneous left bundle branch block (3). Although this may not reflect the whole spectrum of human disease, it nonetheless induces dyssynchronous mechanical activation of the heart and is therefore useful as positive control for cardiac dyssynchrony. RV pacing induced mechanical dyssynchrony independent of ventricular size and function (20). Rapid RV pacing after atrioventricular block induced both dyssynchrony and heart failure. Rapid atrial pacing produced cardiomyopathy without echocardiographic criteria of dyssynchrony. We used the model to test various conventional echocardiographic and Doppler parameters and a novel speckle-tracking method to determine their accuracy in detecting dyssynchrony.
Our results suggest that radial or circumferential strain measured by the novel speckle tracking in short-axis images detects dyssynchrony with or without heart failure better than the tissue Doppler-derived echocardiographic parameters (TPVs). Radial and circumferential strain analysis may be superior to measures of longitudinal velocity for several reasons. First, the contraction of the ventricle occurs in a more circumferential rather than longitudinal direction due to the helical architecture of myocardial fibers (10, 14). Second, Waldman and Covell (25) showed that the amplitude of longitudinal strain did not change significantly during epicardial pacing, whereas circumferential and radial strain did change. Third, strain data reflect cardiac deformation, whereas velocity data reflect cardiac motion (8); tethering, translational motion, or rotation of the heart may affect time to peak velocity in a given myocardial segment. The peaks of the velocity and strain curves do not coincide. Peak systolic velocity occurs in the early-to-middle phase of contraction; however, peak strain for each myocardial segment occurs at the end of segmental contraction.
Pulse-wave TDI detected dyssynchrony, but not in the absence of heart failure. Pulse-wave TDI has excellent temporal resolution compared with color-coded TDI (16). The multivariate regression analysis demonstrated that both color-coded and pulse-wave tissue Doppler indices correlated better with ejection fraction (systolic dysfunction) and less with QRS duration. Yu et al. (27) reported that TPVs correlated with left-ventricular end-systolic volume in heart failure patients with narrow and wide QRS duration during multivariate regression analysis. Their results and ours suggest that global ventricular size and function, and not the timing of the altered sequence of contraction, affect longitudinal velocity.
The SPWMD has also been proposed to quantitatively assesses dyssynchrony in patients with heart failure (17). Because the acquisition measures the timing of excursion points, the angle of acquisition and translation can affect the measurement and distort displacement. Factors that independently affect septal movement such as RV pressure overload or prior cardiac surgery could impact radial septal motion and decrease the reliability of this measurement for the detection of dyssynchrony. Also, although relatively easy to obtain, the low reproducibility of the SPWMD, as shown in our study and also by Marcus et al. (15), limits its value in clinical use.
Radial or Circumferential Strain from Speckle-Tracking Analysis
Speckle-tracking analysis provides angle-independent information on motion and deformation in any direction of a cardiac coordinate system. Several studies demonstrated its validity against sonomicrometry (1) and tagged MRI (7). Speckle-tracking echocardiography offers higher temporal resolution than MRI, making it suitable for assessing the timing of cardiac motion. Speckle tracking offers easier access, costs less, and consumes less time compared with tagged MRI. In our study, the processing and analysis of echocardiographic data was automatically performed by the software and was therefore independent of the operator. The limitations of speckle tracking include its dependence on the angle of acquisition in the short axis. In the present study, this limitation had greater impact on the measurement of circumferential (especially posterior wall) rather than radial strain, and this might explain the observed large variability in circumferential strain determination even in the presumably normal ventricles (baseline dog). With less variability and similar accuracy in detecting dyssynchrony, the radial strain derived from short-axis views was superior to circumferential strain. Our observations agree with Suffoletto et al. (24), who reported that speckle tracking-derived radial dyssynchrony has high sensitivity and specificity in predicting response in CRT in humans. That analysis used the difference in Err between two segments (anteroseptal and posterior wall) to measure dyssynchrony. We used the SD in Err across six segments in the mid-short axis view to achieve greater accuracy. We observed that the difference in Err between the anteroseptal and posterior wall segments increased in the D group (P = 0.047, AUC = 0.776) and also in DHF (P = 0.0019, AUC = 0.977). However, differences were lower than those calculated using the SD in the six-segment model (Tables 3 and 4).
Mechanical Dyssynchrony with Narrow QRS
The atrial pacing cardiomyopathy model in this study produced a valid representation of heart failure with normal QRS duration (fast atrial pacing with normal atrioventricular node conduction). In our study, 30% of the animals in the HF group displayed radial or circumferential dyssynchrony according to the criteria derived from analysis of the DHF group. However, although this method is very sensitive to detect mechanical dyssynchrony per se, it does not identify the threshold value, which defines the response to cardiac resynchronization therapy in subjects with heart failure and normal QRS. Whether echocardiography and TDI can detect dyssynchrony accurately enough to predict CRT response remains to be determined, and further studies are required to test this notion. Our results suggest that longitudinal velocity-derived parameters do not detect dyssynchrony well; therefore, the role of these methods in expanded clinical application of CRT in the narrow QRS population remains unproven.
Finally, our data suggest that systolic dysfunction may beget intraventricular mechanical dyssynchrony. Severe systolic dysfunction caused radial dyssynchrony of similar magnitude to that observed in D group, and this was further enhanced in DHF group. The pattern of mechanical dyssynchrony caused by systolic dysfunction likely differs from that associated with echocardiogram-defined dyssynchrony (such as in left bundle branch block) (9). Whether CRT corrects heart failure-related dyssynchrony with normal QRS duration as effectively as it improves dyssynchrony in heart failure associated with left bundle branch block needs further testing in prospective human clinical studies.
Limitations of Our Study
The current study suggests that speckle tracking-derived strain analysis is more sensitive and specific than velocity-based echocardiographic methods to detect cardiac dyssynchrony in a canine model of cardiomyopathy with dyssynchrony. The purpose of our study was to generate a relative homogenous model of dyssynchrony (electrical dyssynchrony with left-ventricular systolic dysfunction) so we could compare various echocardiographic methods that have been proposed in the human studies to detect dyssynchrony. Therefore, to induce dyssynchrony we chose a model of RV pacing with nonischemic cardiomyopathy. Obviously, our current results cannot be automatically extrapolated to all various subgroups of cardiac dyssynchrony encountered in human heart failure (for example, ischemic cardiomyopathy).
One limitation of this model is that we were unable to analyze speckle tracking-derived longitudinal strain because the quality of our echocardiographic images from apical views in dogs was inadequate to perform this analysis. In humans, with current technology this can be performed reliably and reproducibly. Furthermore, the specific cutoffs for various dyssynchrony parameters need to be validated against response to cardiac resynchronization therapy in humans. Nonetheless, our study suggests that speckle-tracking strain analysis is superior to longitudinal velocity data and provides the scientific basis to design further studies in which radial, circumferential, or longitudinal strain are compared with longitudinal velocity to test response from cardiac resynchronization therapy, especially in patients with narrow QRS and heart failure.
Conclusion
We demonstrated that radial and circumferential strains derived by speckle tracking in short-axis images accurately identify mechanical dyssynchrony in the presence or absence of left-ventricular dysfunction. The majority of the animals with heart failure and narrow QRS showed mechanical dyssynchrony during speckle-tracking analysis. Prospective randomized clinical studies will need to determine whether speckle tracking radial or circumferential strain detects dyssynchrony in humans with heart failure accurately enough to identify appropriate candidates for CRT.
|
GRANTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONGRANTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  S. Gupta, F. Khan, M. Shapiro, S. G. Weeks, S. E. Litwin, and A. D. Michaels The associations between tricuspid annular plane systolic excursion (TAPSE), ventricular dyssynchrony, and ventricular interaction in heart failure patients Eur J Echocardiogr, April 19, 2008; (2008) jen147v1. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
