Conventional Echocardiography

All participants underwent routine echocardiography (Philips iE33; Philips Medical Systems, Eindhoven, The Netherlands), and the images were postprocessed using Xcelera software (Philips Medical Systems). Standard 2D and Doppler echocardiographic measurements were determined in accordance with current American Society of Echocardiography guidelines. For the assessment of LA volume, orthogonal apical views, most commonly apical four-chamber and two-chamber views, were obtained to determine LA area and length (from the middle of the plane of the mitral annulus to the posterior wall). The apical long-axis view was then used instead of the two-chamber view if the left atrium in the latter view appeared foreshortened. Specifically, the maximal LA chamber area and length were measured at end ventricular systole, excluding the LA appendage and pulmonary veins. LA volume was calculated on the basis of the algorithm [(0.85 × A 1 × A 2)/ L ], where A 1 is the four-chamber LA area, A 2 is the two-chamber or apical long-axis LA area, and L is the average of the two lengths obtained from the orthogonal views and indexed to body surface area. Indexed LA diameter, four-chamber LA area, and indexed LA volume were categorized according to current American Society of Echocardiography guidelines. The LA emptying fraction was calculated from maximum and minimum volumes ([maximum volume − minimum volume]/maximum volume). Pulmonary artery systolic pressure was obtained using tricuspid regurgitation velocity, inferior vena cava diameter and its respiratory variation, and hepatic venous pulsed-wave Doppler. Left ventricular (LV) systolic function was assessed by atrioventricular plane excursion, and LV ejection fraction was assessed by a modified biplane Simpson’s method. Right ventricular mechanical function was measured by tricuspid annular plane systolic excursion. LV ejection fraction and LV filling pressure (by the ratio of mitral peak velocity of early filling to early diastolic mitral annular velocity, determined using the average early diastolic mitral annular velocities of the septal and lateral LV annulus obtained by pulsed-wave Doppler) were also estimated.

Speckle-Tracking Echocardiography

All 77 patients from the current database were eligible for speckle-tracking echocardiographic analysis. Longitudinal LA strain was computed by speckle-tracking echocardiography (2D Cardiac Performance Analysis; TomTec Imaging Systems, Munich, Germany). This software, unlike previous software, performs border tracking that is better suited to a region of interest commensurate with the thin LA wall. The relative velocity is estimated, for each border point, by extracting a space-time representation along a line that passes through the tracked border and has direction orthogonal to the direction used for border tracking. The velocity vector is then obtained by the vector summation of the Lagrangian (border) and Eulerian (relative) velocity contributions. The entire process makes use of the time periodicity to ensure a periodic result and avoid the drift effect. Strain is defined as the instantaneous local border lengthening or shortening, St ( t ) = [ L ( t ) − L ₀ ]/ L ₀ , with respect to one arbitrary instant, t ₀ , when L ( t ₀ ) = L ₀ , taken at the end of diastole (the electrocardiographic R wave). Strain rate is defined as the rate of lengthening/shortening as SR ( t ) = L ⁻¹ ( t )d L /d t .

Global LA strain was estimated by taking the average of longitudinal strain data obtained from the apical four-chamber and two-chamber projections. Data from a total of 12 LA segments (annular, mid, and superior segments along the septal, lateral, anterior, and inferior LA walls using apical four-chamber and two-chamber images) were averaged to determine global longitudinal strain at the end of LV ejection (LA reservoir phase; Figures 1 and 2 ). Regional LA strain was computed from the annular, mid, and superior segments from both four-chamber and two-chamber projections. The peak strain rate was measured during LV ejection (LA reservoir phase), and the early diastolic systolic strain rate was measured during LV early diastole (LA conduit phase). Assessment of LV strain was regarded as suboptimal when either (1) speckle tracking could not be obtained for at least four of the six myocardial segments in each view or (2) a theoretically unacceptable value or values were obtained.

Characterization of LA motion in (A) a control subject and (B) a patient with AF.

Longitudinal LA strain in (A) a control subject and (B) a patient with AF.

Follow-Up Data

Of the original 35 patients, 26 patients consented to participate in additional follow-up data analysis; the remaining patients either received follow-up care elsewhere or did not consent to additional follow-up data collection. Rehospitalization for cardiac events and death was recorded for each of the 26 individuals. Hospitalizations, cardiac events, and all-cause mortality were recorded from the date of echocardiography through May 2010, resulting in a median length of follow-up of 394 days. Cardiac events were defined as episodes of heart failure, AF with rapid ventricular rate, other arrhythmias, acute coronary syndromes, stroke, or a cardioembolic event that required hospitalization.

Statistical Analysis

Data are presented as mean ± SD. Unpaired t tests were performed to compare echocardiographic data between patients and controls. Comparison of the prevalence of comorbid conditions (diabetes, hypertension, and other medical illnesses) was made using the χ ² test. P values < .05 were considered statistically significant. Linear regression analysis was performed in all patients with AF and control subjects for identifying correlates of global longitudinal LA strain. We also performed logistic regression analyses in patients with AF to determine the predictors of a CHADS ₂ score ≥ 2. The associations of CHADS ₂ score, indexed LA volume, and longitudinal LA strain with outcomes were assessed in a series of Cox models. The incremental value of longitudinal LA strain over indexed LA volume and CHADS ₂ score was assessed in a series of Cox models in which the first model consisted of CHADS ₂ score ≥ 2 entered as a test group. Each echocardiographic variable was then entered into separate models in combination with CHADS ₂ score. Reliability of the LA speckle-tracking parameters was assessed by interobserver and intraobserver variability. Ten patients from each group were studied in duplicate by two observers (P.L.A. and G.C.) for LA emptying fraction, global longitudinal LA strain, and LA peak systolic and end-systolic strain. Bland-Altman plots and linear regression methods were used to assess the reproducibility.