Predictive Value of an Atrial Fibrillation Burden-Based Model for Early Recurrence After Circumferential Pulmonary Vein Ablation for Paroxysmal Atrial Fibrillation

Background: To explore the importance of a predictive model, we combined the proportion of time in atrial fibrillation (AF) recorded by 24-hour dynamic electrocardiogram (DCG) with left ventricular function data. This study evaluated the probability of premature recurrence following circumferential pulmonary vein ablation (CPVA) in individuals with paroxysmal atrial fibrillation (PAF). Methods: A total of 237 PAF patients who were scheduled to undergo the first CPVA were selected, and the following patient data were collected: (1) Transthoracic echocardiographic assessment of left ventricular ejection fraction (LVEF) using the biplane Simpson technique; (2) Atrial fibrillation burden parameters obtained by 24-hour DCG, including the total time of AF, number of AF arrays, and proportion of time in AF; (3) Based on DCG findings obtained 3–6 months post-procedure, participants were categorized into either the sinus rhythm cohort or the recurrence cohort. Multivariate logistic regression analysis was subsequently performed to identify significant predictors influencing PAF recurrence following CPVA intervention; (4) An integrated forecasting framework was developed, with predictive capabilities evaluated through receiver operating characteristic (ROC) curve assessments to determine model efficacy. Results: (1) Among 224 participants undergoing CPVA for PAF, follow-up assessments at 3–6 months identified 20 cases (8.9%) with arrhythmia recurrence. Comparative analysis revealed elevated values for brain natriuretic peptide (BNP), left ventricular diastolic dimension (LVDD), SDNN Index (representing two R waves, RR interval variability), and atrial fibrillation burden in the recurrence cohort versus those maintaining sinus rhythm (all p < 0.05). Conversely, the recurrence group demonstrated reduced LVEF, minimum heart rate (MinHR), maximun heart rate (MaxHR), and minute meter slowest heart rate compared to the rhythm-controlled population (all p < 0.05). (2) Multivariate logistic regression analysis showed that BNP, MaxHR, and proportion of time in AF were independent influencing factors for PAF recurrence (all p < 0.05). (3) An integrated forecasting framework incorporating biomarkers including BNP, LVEF, and atrial fibrillation burden metrics was developed. Receiver operating characteristic analysis demonstrated superior predictive capability for the composite model (area under the curve, AUC = 0.874) when compared to atrial fibrillation burden assessment alone (AUC = 0.744) (p = 0.014). Conclusion: The proportion of time in AF combined with LVEF has good predictive value for early recurrence in PAF patients after CPVA.