On the basis of the electromechanical coupling theory, an activation imaging system has been developed with three-dimensional speckle-tracking echocardiography. The aim of this study was to determine the association between left ventricular (LV) propagation patterns by activation imaging and response to cardiac resynchronization therapy (CRT).
This was a retrospective, single-center study. Eighty-one patients undergoing CRT, of whom 50 (61.7%) had left bundle branch block (LBBB), were enrolled. Activation imaging studies were performed with a three-dimensional speckle-tracking echocardiographic system, which allowed visualization of LV activation propagation and measurement of the time from the QRS complex to activation onset. A CRT volume responder was defined as a patient with ≥15% reduction of LV end-systolic volume at 6 months after CRT. Clinical outcomes were assessed with the composite end point of death due to cardiac causes or unplanned hospitalization for cardiac diseases.
In patients with LBBB, the main activation pattern (74%) was a U-shaped propagation pattern, which was characterized as propagation from the midseptum to the lateral or posterior wall through the apex. In patients without LBBB, various non-U-shaped propagation patterns were observed in the majority of patients (97%). Among the 41 CRT responders, almost all (87.8%) had the U-shaped propagation pattern. During follow-up (median, 20 months), 29 patients (35.8%) reached the clinical end points. In a multivariate Cox proportional hazards model, a U-shaped propagation pattern was associated with the end points independently of LBBB or LV end-diastolic volume.
The U-shaped propagation pattern on three-dimensional speckle-tracking echocardiography was significantly associated with a favorable CRT response. Activation pattern analysis may provide additional information to predict response to CRT.
Indications for cardiac resynchronization therapy (CRT) have been classified on the basis of QRS morphology and duration. In particular, patients with left bundle branch block (LBBB) clearly have been suitable candidates for CRT compared with those without LBBB. However, CRT responses are heterogeneous even in patients with LBBB. In contrast, assessment of mechanical dyssynchrony has been expected to predict CRT response effectively, but its reliability remains controversial. Given that previous studies with voltage-mapping systems demonstrated various propagation patterns even in patients with LBBB, the variability of electrical propagation may be a cause of the diversity in CRT responses. In addition, electrical propagation patterns in patients without LBBB would be more complex and heterogeneous and have not been well investigated. The use of voltage-mapping systems may be promising to identify electrical characteristics associated with CRT responses, but they cannot be a routine study modality in the clinical setting, because of their invasiveness and high cost.
Significant associations between mechanical and electrical propagation have been reported in experimental studies. On the basis of these pathophysiologic findings, we have developed an activation imaging system using three-dimensional (3D) speckle-tracking echocardiography (STE), which is modeled on 3D voltage-mapping systems and allows the visualization of contraction propagation. Our validation study against 3D voltage mapping systems revealed the reliability and feasibility of activation imaging by 3D STE.
For these reasons, we sought to better understand the relationship between LV propagation pattern and CRT response and hypothesized that activation imaging by 3D STE has the potential to identify the substrate of CRT responses on the basis of electromechanical coupling.
Therefore, we studied the association between LV propagation pattern by activation imaging with 3D STE and CRT response.
Study Design and Population
This was a retrospective, single-center study. We enrolled 81 patients (58 men and 23 women; mean age, 63.6 ± 14.2 years) who underwent CRT from March 2009 to December 2013 at the University of Tsukuba Hospital. All patient records were reviewed for collecting clinical information until June 2014 to ensure a follow-up period of ≥6 months in the last enrolled patient. As cardiac events after CRT, death due to cardiac causes and unplanned hospitalizations for heart failure or arrhythmic events were reviewed. We also enrolled 30 healthy, age-matched volunteers (20 men and 10 women; mean age, 61.5 ± 13.4 years) as the control group. The selection criteria for CRT included drug-refractory, symptomatic heart failure despite optimal pharmacologic therapy; heart failure of New York Heart Association class II, III, or IV; depressed LV ejection fraction (≤35%); and QRS duration ≥ 120 msec. The intrinsic rhythm in all patients was sinus rhythm. Patients with atrial fibrillation and those being upgraded from right ventricular pacing were excluded.
Fifty patients (61.7%) had LBBB, which was defined on the basis of the following criteria: QRS duration ≥ 140 msec (men) or 130 msec (women), QS or rS in leads V 1 and V 2 , and mid-QRS notching or slurring in two or more of leads V 1 , V 2 , V 5 , V 6 , I, and aVL. Among the remaining 31 patients, 28 had nonspecific intraventricular conduction delays, and three had right bundle branch block. The study was approved by the local research ethics committee of Tsukuba University Hospital, and all patients gave their written informed consent.
Cardiac Resynchronization Therapy Procedure
CRT devices were implanted transvenously in all patients. After performing retrograde coronary venography, we selected a lateral or posterolateral vein as the target branch of the coronary sinus to stimulate the latest activation site. If attempts to access these veins failed because of unusual anatomy preventing access to the coronary sinus or resulting in poor sensing, phrenic nerve stimulation, or pacing failure, the middle cardiac vein was considered as an alternative branch.
Clinical outcomes were assessed with the composite end point of death due to cardiac causes or unplanned hospitalization for cardiac diseases including heart failure and arrhythmic events.
All patients were examined in the left lateral position. Baseline Doppler echocardiographic examinations were performed with an Artida ultrasound system (Toshiba Medical Systems, Tochigi, Japan). Standard echocardiographic examinations were performed using a multifrequency transducer. LV end-diastolic volume (LVEDV), LV end-systolic volume, and LV ejection fraction were measured using a modified Simpson’s method. In patients with CRT, echocardiographic examinations were performed within 48 hours before the procedures and were repeated at 7 days and 6 months after CRT. CRT volume responders were defined as patients with ≥15% reductions of LV end-systolic volume at 6 months after CRT.
For 3D STE, full-volume R waves of electrocardiographically gated 3D data sets were acquired from apical positions using a matrix-array 2.5-MHz transducer. To obtain these data sets, four or six sectors were scanned and automatically integrated into a wide-angle (maximum, 90° × 90°) pyramidal data image covering the entire left ventricle. The volume rate of each image was set at >30 Hz. The data were stored and transferred to a computer for offline analysis. The images were analyzed with 3D Wall Motion Tracking software (Toshiba Medical Systems) in a manner reported previously.
Activation Imaging by 3D STE
The activation imaging system aimed to quantify a time from the QRS complex to the onset of regional deformation in the 16 LV segments, but not the timing of maximum deformation corresponding to the maximum value of deformation parameter. Therefore, activation imaging allows estimation of the approximate timing of the onset of regional contraction. The regional contraction was determined by the area change ratio (ACR), which is a deformation parameter obtained from the area-tracking method by 3D STE. After 3D data sets were obtained, the initial frame for ACR calculation was manually moved to the frame including QRS onset, which was determined from the electrocardiographic waveform on the echocardiographic system. To visualize the timing of regional activation, if a regional ACR value exceeded a threshold value, which is defined as 25% of the maximum ACR value, in each regional area, the activated segments were colored by specific colors corresponding to different time from QRS onset to activation onset (TACT) on a polar map and 3D activation imaging of the LV endocardial surface ( Figure 1 ). The loop imaging of regional colorized images can produce a propagation image of LV activation, which is assessed with a video during one cardiac cycle ( Video 1 ; available at www.onlinejase.com ).
Dyssynchrony Assessed by Activation Imaging
First, we evaluated the propagation patterns on the activation imaging system. An independent investigator identified the patterns carefully. Second, intraventricular dyssynchrony was quantified as the intraventricular propagation time (IVPT), which was calculated as the difference between the minimum and maximum TACT in 16 LV segments, and the SD of TACT (TACT-SD) was also calculated.
Lead Position and the Latest Propagation Area
An independent investigator, who was blinded to information about the echocardiographic data, assessed LV lead positions. On postimplantation chest radiography and biplane fluoroscopy, the lead placement segments were categorized as basal, middle, or apical in the right anterior oblique projections and as anterior, lateral, posterior, or inferior in the left anterior oblique projections. The lead positions were defined with the following scores: 1 = LV lead position located at a segment more than two segments apart from the latest site in the activation image; 2 = LV lead position located at a segment adjacent to the latest site; and 3 = LV lead position located at the latest site.
Two observers independently assessed propagation patterns, TACT, IVPT, and TACT-SD in 20 patients. To test intraobserver variability, a single observer analyzed the data twice on occasions separated by a 1-month interval. To test interobserver variability, a second observer analyzed the data without knowledge of the first observer’s measurements. Reproducibility was assessed as the mean percentage error (absolute difference divided by the mean of the two observations).
Results are expressed as number (percentage) or as mean ± SD. Comparisons between two groups were performed using the Student t test for continuous variables and the χ 2 test for categorical variables. One-way analysis of variance with the post hoc Tukey-Kramer test was used to compare variables between three or more groups. The risk for clinical end points was determined with Cox proportional hazard models. The univariate factors with P values < .05 were entered into the multivariate model to assess the effects of the parameters on the end points. Kaplan-Meier analysis was done to determine the influence of dyssynchrony parameters on the end points. P values < .05 were considered to indicate statistical significance. All calculations were performed with SPSS version 22 (SPSS, Inc, Chicago, IL).
Feasibility and Reproducibility of Activation Imaging
Agreement on propagation patterns was perfect between the two investigators. Among a total of 1,296 segments in 81 patients, optimal 3D speckle-tracking echocardiographic analyses could be done on 1,239 segments (95.6%), whereas all segments were available in the control subjects. Among the 57 excluded segments, 20 were LV basal anterior segments, followed by 14 midanterior, 12 basal lateral, and 11 apical anterior wall segments. Individually, three patients had three inadequate segments, 15 patients had two inadequate segments, and 18 patients had one inadequate segment. Intra- and interobserver variability of TACT, IVPT, and TACT-SD measurements were as follows: TACT, 5.4 ± 4.1% and 7.1 ± 4.3%; IVPT, 7.1 ± 4.0% and 8.0 ± 4.4%; and TACT-SD, 5.3 ± 3.9% and 7.7 ± 3.3%, respectively.
Activation Imaging Analysis at Baseline
A featured activation pattern was U-shaped propagation ( Figure 1 ), which was found mainly in patients with LBBB. In contrast, other various non-U-shaped propagation patterns were observed, which were more frequent in patients without LBBB. The comparisons of characteristics between the U-shaped propagation pattern and non-U-shaped propagation patterns are summarized in Table 1 .
|Variable||Control ( n = 30)||U-shaped ( n = 38)||Non-U-shaped ( n = 43)||P value (U-shaped vs non-U-shaped)|
|Age (y)||61.5 ± 13.4 ∗||67.3 ± 13.1||59.3 ± 13.6||.01|
|Men||20 (66.7%)||20 (52.6%)||38 (88.4%)||<.001|
|Ischemic etiology||—||3 (7.9%)||11 (25.6%)||.03|
|NYHA class II/III/IV||—||15 (39.5%)/19 (50%)/4 (10.5%)||11 (25.6%)/27 (62.8%)/5 (11.6%)||.41|
|SBP (mm Hg)||120.4 ± 11.2 †||115.9 ± 28.9||112.2.2 ± 15.6||.81|
|DBP (mm Hg)||68.5 ± 12.3||69.6 ± 15.3||61.0 ± 10.5||.01|
|LVEDV (mL)||99.1 ± 14.2 ‡||184.5 ± 73.6||228.5 ± 97.5||.04|
|LVESV (mL)||26.9 ± 5.6 ‡||129.1 ± 57.1||166.7 ± 88.3||.04|
|LVEF (%)||73.1 ± 3.4 ‡||30.3 ± 7.6||29.9 ± 10.7||.85|
|QRS duration (msec)||91.5 ± 8.1 ‡||167.2 ± 21.5||152.9 ± 38.9 ∗||.04|
|LBBB||—||37 (97.4%)||13 (26.0%)||<.001|
|First activated segment||<.001|
|Basal septum||13 (43.3%)||—||15 (34.9%)|
|Middle septum||13 (43.3%)||26 (68.5%)||17 (39.5%)|
|Apex||4 (13.3%)||12 (31.5%)||10 (23.3%)|
|Latest activated segment||.01|
|Anterior wall||3 (10%)||—||4 (9.3%)|
|Lateral wall||9 (30%)||28 (73.7%)||8 (18.6%)|
|Posterior wall||18 (60%)||10 (26.3%)||22 (51.2%)|
|Inferior wall||—||7 (16.3%)|
|IVPT baseline (msec)||69.5 ± 22.6 ‡||192.9 ± 47.0||159.8 ± 58.7||.01|
|TACT-SD baseline (msec)||19.3 ± 5.3 ‡||57.8 ± 14.1||52.2 ± 16.9||.19|
Among 50 patients with LBBB, a U-shaped propagation pattern was found in 37 (74%). Thirteen patients (26%) with LBBB had non-U-shaped propagation patterns. In contrast, only one patient without LBBB had a U-shaped propagation pattern, and the remaining 30 patients without LBBB showed various non-U-shaped propagation patterns. In particular, 11 patients with non-U-shaped propagation patterns had no apparent intraventricular conduction delays, with IVPTs of <100 msec, which was defined because the 95th percentile of IVPT was 97.9 msec in control subjects.
The first activated segments were located at the middle or apical septum in patients with the U-shaped propagation pattern, whereas the basal septal segment was more prevalent in patients with non-U-shaped propagation patterns. In addition, almost all of the latest activated segments in patients with the U-shaped pattern were present in the lateral wall, whereas latest activation in other segments was more frequent in patients with non-U-shaped propagation patterns. IVPT and TACT-SD did not differ significantly between the U-shaped and non-U-shaped propagation patterns but were significantly larger than those in the control group.
Left Ventricular Reverse Remodeling after CRT
Among the 81 patients, 41 (50.6%) were identified as CRT responders ( Table 2 ). Almost all responders had LBBB and the U-shaped propagation pattern. In contrast, even among CRT nonresponders, 13 patients (32.5%) had LBBB, but the U-shaped propagation pattern was observed in only two nonresponders. In two nonresponders with LBBB, marked regional conduction delay was observed in the posterolateral wall, which was revealed to have delayed enhancement regions by cardiac magnetic resonance imaging. Because appropriate coronary veins for the implantation of LV leads were not present in other regions, the LV leads were positioned at the posterolateral wall. On the other hand, three patients with LBBB and similar characteristics of the posterolateral wall, in whom the LV leads were positioned at the anterolateral wall without delayed enhancement regions, became responders. In nonresponders, plasma creatinine level and LVEDV were significantly higher than in responders. At 7 days after CRT, both IVPT and TACT-SD in responders were significantly reduced compared with baseline data; in contrast, IVPT in nonresponders was significantly increased compared with baseline values.
|Variable||Responders ( n = 41)||Nonresponders ( n = 40)||P value|
|Age (y)||67.2 ± 14.4||60.2 ± 13.4||.03|
|Men||22 (53.6%)||36 (90%)||.001|
|Ischemic etiology||8 (19.5%)||6 (15%)||.45|
|NYHA class II/III/IV||16 (39.0%)/21 (51.2%)/4 (9.8%)||10 (25.0%)/25 (62.5%)/5 (12.5%)||.32|
|QRS duration (msec)||163.3 ± 24.5||156.3 ± 39.5||.33|
|LBBB||37 (90.2%)||13 (32.5%)||<.001|
|Hb (g/dL)||12.8 ± 1.6||13.4 ± 1.9||.10|
|Cre (mg/dL)||0.95 ± 0.24||1.43 ± 1.22||.02|
|BNP (pg/mL)||323.1 ± 312.3||478.5 ± 682.5||.28|
|Lead position score 1/2/3||3 (7.3%)/18 (43.9%)/20 (48.8%)||2 (5.0%)/18 (45%)/20 (50.0%)||.86|
|LVEDV (mL)||180.7 ± 65.1||234.1 ± 101.3||.006|
|LVEF (%)||31.1 ± 7.5||29.7 ± 10.8||.51|
|U-shaped propagation pattern||36 (87.8%)||2 (5.0%)||<.001|
|IVPT baseline (msec)||192.9 ± 48.8||157.7 ± 59.43||.005|
|IVPT 7 d after CRT (ms)||115.9 ± 45.6 ∗||178.7 ± 47.2 †||<.001|
|TACT-SD baseline (msec)||58.7 ± 15.0||51.2 ± 16.9||.04|
|TACT-SD 7 d after CRT (msec)||35.3 ± 12.9 ∗||53.4 ± 13.1||<.001|
Then, changes of LV dyssynchrony parameters after CRT were followed by the comparison between the U-shaped and non-U-shaped propagation patterns ( Figure 2 ). In patients with the U-shaped propagation pattern, both IVPT and TACT-SD were significantly reduced after CRT. In contrast, IVPT and TACT-SD did not differ significantly after CRT in patients with non-U-shaped propagation patterns. Representative examples are shown in Figure 3 and Videos 2 and 3 (available at www.onlinejase.com ).