Association of left atrial low-voltage area burden and reservoir strain with 12-month recurrence after first radiofrequency ablation for persistent atrial fibrillation: evaluation of a combined model

Left atrial low-voltage area (LVA) burden and reservoir strain characterize the electrophysiological substrate and mechanical function. We assessed their associations with 12-month recurrence after first radiofrequency ablation for persistent atrial fibrillation (AF) and their prognostic value beyond clinical variables. This retrospective study included 312 adults with complete 365-day outcomes; seven patients with shorter follow-up were included in complementary analyses. Reservoir strain was measured during AF before ablation, and LVA burden was measured during sinus rhythm before radiofrequency delivery. Recurrence was defined as documented AF, atrial flutter, or atrial tachycardia lasting ≥ 30 s after an 8-week blanking period through day 365. We fitted fixed-specification logistic models and an expanded Cox model, performed paired bootstrap validation and monitoring-restricted analyses, and estimated performance differences using nested bootstrapping. The model is intended for use after ablation once all predictors are available, preferably before discharge. Recurrence occurred in 118 patients (37.8%). Each doubling of LVA burden was associated with higher odds of recurrence (adjusted odds ratio [aOR], 3.841; 95% confidence interval [CI], 2.332–6.326; P < 0.001), whereas each 1-percentage-point increase in reservoir strain was associated with lower odds (aOR, 0.743; 95% CI, 0.593–0.932; P = 0.010). The combined model had an apparent area under the receiver operating characteristic curve (AUC) of 0.871 and an optimism-corrected AUC of 0.851. The corrected difference in AUC (ΔAUC) was 0.093 versus the clinical model, 0.003 (95% CI, − 0.003 to 0.018) versus the clinical model plus LVA burden, and 0.046 (95% CI, 0.018–0.078) versus the clinical model plus reservoir strain. Higher LVA burden and lower reservoir strain were independently associated with late recurrence. The combined model improved performance over the clinical model and the clinical model plus reservoir strain, but its additional performance over the clinical model plus LVA burden remained uncertain. Independent external validation and recalibration are required. Not applicable.