Recurring patterns of atrial fibrillation in surface ECG predict restoration of sinus rhythm by catheter ablation

Abstract

Background: Non-invasive tools to help identify patients likely to benefit from catheter ablation (CA) of atrial fibrillation (AF) would facilitate personalised treatment planning.

Aim: To investigate atrial waveform organisation through recurrence plot indices (RPI) and their ability to predict CA outcome.

Methods: One minute 12-lead ECG was recorded before CA from 62 patients with AF (32 paroxysmal AF; 45 men; age 57±10 years). Organisation of atrial waveforms from i) TQ intervals in V1 and ii) QRST suppressed continuous AF waveforms (CAFW), were quantified using RPI: percentage recurrence (PR), percentage determinism (PD), entropy of recurrence (ER). Ability to predict acute (terminating vs. non-terminating AF), 3-month and 6-month postoperative outcome (AF vs. AF free) were assessed.

Results: RPI either by TQ or CAFW analysis did not change significantly with acute outcome. Patients arrhythmia-free at 6-month follow-up had higher organisation in TQ intervals by PD (p<0.05) and ER (p<0.005) and both were significant predictors of 6-month outcome (PD (AUC=0.67, p<0.05) and ER (AUC=0.72, p<0.005)). For paroxysmal AF cases, all RPI predicted 3-month (AUC(ER)=0.78, p<0.05; AUC(PD)=0.79, p<0.05; AUC(PR)=0.80, p<0.01) and 6-month (AUC(ER)=0.81, p<0.005; AUC(PD)=0.75, p<0.05; AUC(PR)=0.71, p<0.05) outcome. CAFW-derived RPIs did not predict acute or postoperative outcomes. Higher values of any RPI from TQ (values greater than 25th percentile of preoperative distribution) were associated with decreased risk of AF relapse at follow-up (hazard ratio ≤0.52, all p<0.05).

Conclusions: Recurring patterns from preprocedural 1-minute recordings of ECG TQ intervals were significant predictors of CA 6-month outcome.

Keywords: Atrial fibrillation; Electrocardiogram segmentation; Principal component analysis; Recurrence quantification analysis; TQ interval.