Echocardiography

All echocardiographic studies were performed using a commercially available echocardiographic system (Artida; Toshiba Medical Systems, Tochigi, Japan) and a dedicated software package (Ultra Extend; Toshiba Medical Systems) with a 2.5-MHz variable-frequency harmonic phased-array transducer.

Standard echocardiographic views, including apical four- and two-chamber views, with the patient in the left lateral recumbent position, were obtained in 2D and color tissue Doppler modes. Mitral inflow velocities were recorded by standard pulsed-wave Doppler at the tips of the mitral valve leaflets in an apical four-chamber view. In addition to that, we evaluated LV ejection fraction, derived by using the modified Simpson’s method.

All patients underwent 3D echocardiography from the apical approach to evaluate LA volume and strain analysis. After the entire left atrium was acquired, 3D data were sent to the dedicated software for subsequent offline analysis. Three-dimensional speckle-tracking analysis was done as reported previously. In brief, the endocardial contour of the left atrium was traced manually on an end-diastolic cavitary frame, and after defining the thickness of the region to be considered, the software automatically tracked the atrial wall on subsequent frames. Adequate tracking can be verified and corrected by adjusting the region of interest or the contour ( Figure 1 ). The software automatically divided the LA wall into the following 16 segments: four segments at the roof, six segments in the middle, and six segments in the basal part ( Figure 1 D). While retaining the entire left atrium, depth and sector width were decreased as much as possible to improve the temporal and spatial resolution of the image, resulting in a volume rate of 36.4 ± 6.2 volumes/sec in this study. LA dyssynchrony was quantified using the 3D strain speckle-tracking method. We analyzed area tracking, which is deformation data of the endocardial surface. For each segment, LA wall lengthening (positive strain value) is observed during LV systole ( Figure 2 ). A curve plotting the average of the 16 segments’ strain curves was also automatically generated, referred to as global strain, which is an index of the amount of LA deformation. Global strain was originally established and reported to refer to the average of the each strain value obtained from 16 LV segments throughout the entire cardiac cycle. Global strain is automatically calculated and displayed as a global strain curve by the software in the echocardiographic machine, and similar to global LV strain, LA global strain was derived from the continual averaging of all segmental strain values over time and was automatically calculated and displayed. On the other hand, the averaged peak strain is the average of each peak strain value in all segments. Whereas global strain is influenced by dyssynchrony because it is the average of strain values at the same time, peak strain is not influenced by the timing of their peaks, because peak strain values have nothing to do with the time at which they appear. Time to peak strain was defined as the time from end-diastole (the R wave on the electrocardiogram) to maximal positive deformation or peak strain ( Figure 2 ). As an index of LA dyssynchrony, TP-SD of the area tracking was computed and expressed as a percentage of the R-R′ interval, as reported previously. Higher grades of dyssynchrony were recognized as larger values of TP-SD.

Three-dimensional tracking approach. After the LA endocardial contour was traced ( yellow dotted line ) and the epicardial contour was set ( green dotted line ), the software automatically tracked the atrial wall, allowing real-time adjustments and in-motion verification. (A) Apical four-chamber view, (B) apical view orthogonal to plane A, (C1) the basal portion of the left atrium, (C2) the middle portion, (C3) the roof portion, and (D) the scheme of 16 segments.

Derived strain curves. Time to peak (TP) was defined as the time from end-diastole (the R wave on the electrocardiogram) to maximal time positive deformation marked with circle of the curve. TP-SD was calculated as the standard deviation of TP and expressed as a percentage of the R-R′ interval. Higher grades of dyssynchrony were identified by larger values of TP-SD.

All echocardiographic studies were performed by an experienced cardiologist who was blinded to the patients’ clinical information. Ten subjects, including patients and controls, were randomly selected to test intra- and interobserver variability for strain and TP-SD. Their data were analyzed again 2 to 4 weeks after the first analysis by the same investigator and by a second independent investigator.

Statistical Analysis

All analyses were performed using SPSS version 21 (SPSS, Inc, Chicago, IL). All continuous variables are expressed as mean ± SD. The Shapiro-Wilk test was used to assess the distribution of continuous variables. Comparisons between two sets of continuous variables were performed using either t tests or Mann-Whitney U tests, as appropriate. Comparisons between two sets of categorical variables were done using either χ ² tests or Fisher’s exact tests, as appropriate. Comparisons of continuous variables before and after PVI were performed using either paired t test or Wilcoxon signed rank sum tests, as appropriate. Medications before and after PVI were analyzed using the McNemar test.

Intra- and interobserver variability were assessed in 10 randomly selected subjects, as stated previously. Intraclass correlation coefficients and the absolute difference between two paired measurements divided by the mean of the repeated observation were calculated. P values < .05 were considered significant.