This study was retrospective in fashion, approved by the ethics committee in Osaka Medical College with notification for guaranteed withdrawal of participants on the website providing means of “opt-out” (No. 2194-01).
Study population
We examined 73 patients with paroxysmal AF (PAF), 39 MRFs patients without history of AF (disease controls), and 61 healthy subjects (healthy controls), all of whom underwent transthoracic echocardiography from February 2012 through December 2018. Patients with previous cardiac surgery including pacemaker implantation, known coronary artery disease, left ventricular (LV) ejection fraction < 30%, and those with dialysis treatment were excluded (Fig. 1). Patients who had echocardiographic images inadequate for assessing indispensable measures, described later, were also excluded.
PAF patients: The patients were scheduled for pulmonary vein isolation, with their cardiac rhythm being “sinus” during echocardiographic examinations.
Disease controls: This group of patients had been hospitalized in the department of neurology or neurosurgery in our institution under a diagnosis of non-lacunar ischemic stroke, transient ischemic attack, or peripheral artery occlusion. They were found neither to have PAF detected on an automated cardiac rhythm monitor [14], nor to have apparent embolic sources evidenced by carotid ultrasound and transesophageal echocardiography.
Healthy controls: These subjects were screened for any cardiovascular disease in our outpatient department.
Standard echocardiography
Transthoracic echocardiography was performed by experienced sonographers using commercially available ultrasound apparatus (Vivid 7 Dimensions or Vivid E9; GE Vingmed Ultrasound, Horten, Norway). During each examination, one-lead electrocardiogram, usually the limb-lead II was recorded continuously. Under 2-dimensional guidance in the parasternal view, LA diameter, LV end-diastolic dimension, and LV wall thickness were measured. LA volume was calculated by the disc method in the apical 2- and 4-chamber views and indexed by the body-surface area leading to LA volume index. LV ejection fraction was measured by the modified Simpson’s rule. LV mass was calculated using the Devereux formula, and indexed by the body surface area (LV mass index). LV mass index ≥ 115 g/m2 in men and ≥ 95 g/m2 in women were considered as the presence of LV hypertrophy [15].
For assessing LV diastolic function, pulsed Doppler LV inflow indices of early (E) and late filling (A) wave velocities, their ratio (E/A), and E-wave deceleration time were obtained. In the apical 4-chamber view, using the spectral type of TDI, early (E’) and late (A’) diastolic velocities were measured with the sample volume placed at the septal and lateral sides of the mitral annulus. The ratio of E to E’ (E/E’) was used as a surrogate of LV filling pressure [16]. In the present study, A’, meaning “velocity”, averaged for both mitral annuli, was considered as LA systolic function.
Measurement of AEMD
AEMD was measured from the beginning of the electrocardiogram P-wave to the initial point of the spectral TDI-derived A’ as described previously [9–12]. In this study, AEMD was obtained for the septal (septal EMD) and lateral (lateral EMD) sides of the mitral annulus. The time difference of the lateral to septal EMD was defined as intra-LA EMD as reported previously [12]. All AEMD measurements were performed by independent observers without knowledge of patients’ background.
To assess interobserver variability of AEMD, 40 individuals were randomly selected and Bland–Altman plot analysis was performed (KA and TI). It was found that measurements were similar and statistically comparable with each other (Fig. 2). The mean difference was 1.9 ms (3.2%) and the coefficient of variation was 4.9.
Statistical analysis
Continuous variables were expressed as mean ± SD and categorical variables as percentages. Comparisons of categorical variables were made using the chi-square test or Fisher’s exact test. Continuous variables across the 3 groups were compared using one-way analysis of variance or Kruskal Wallis test according to whether normally distributed or not, as tested by Welch test. Tukey's HSD test or Steel–Dwass test was applied for intergroup comparisons as appropriate. Univariate and multivariate logistic regression analysis were performed to predict significant variables for identifying PAF patients. The sensitivity and specificity of AEMD and other echocardiographic variables for identifying PAF patients were calculated by receiver operating characteristic (ROC) analysis. Comparisons of area under curves (AUCs) between models of ROC analysis were also performed. All analyses, except for ROC analysis, were performed using SPSS for Windows ver. 24.0 (IBM, Chicago, IL). For ROC analysis, JMP Pro ver. 13.0 (SAS Institute, Cary, NC) was used. P < 0.05 was considered significant.