A morphology-based ECG algorithm for identifying left atrial fibrosis in atrial fibrillation

Background

Left atrial (LA) fibrosis represents a critical factor contributing to the recurrence of atrial fibrillation (AF) beyond pulmonary vein triggers, yet its non-invasive identification remains a major clinical challenge. While invasive voltage mapping provides direct evidence of LA fibrosis through the presence of low-voltage areas (LVAs), it is impractical for widespread screening. As the ECG is a functional evaluation of intracardiac voltages, we aimed to develop surface ECG-based criteria for detecting LA fibrosis in patients undergoing AF ablation.

Methods

We retrospectively analyzed 12-lead ECGs of 78 patients who underwent AF ablation with invasive endocardial LA voltage mapping in sinus rhythm or during proximal CS pacing. LA fibrosis was defined as the presence of LVAs with bipolar voltage of <0.50 mV on invasive mapping. An algorithm incorporating one clinical criterion and three morphological criteria was evaluated: (1) history of atypical atrial flutter, (2) a positive-to-negative deflection ratio >2 in the P-wave of lead V1, (3) presence of typical or atypical interatrial block (types I–IV), and (4) absent terminal P-wave progression in V4-V6, e.g. a negative or iso-electric segment in the second half of the p-wave in 2 or more leads, corresponding to delayed left atrial activation. The presence of at least one of these criteria was considered indicative of LA fibrosis. Diagnostic performance was assessed against invasive voltage mapping as the reference standard. The ECG was manually evaluated by two electrophysiologists using amplified P-wave analysis (50-100 mm/sec and 40mm/mV for scale) blinded for voltage mapping.

Results

Of the 78 patients analyzed, 35 (13 paroxysmal, 22 persistent) demonstrated LA fibrosis during electrophysiologic investigation, while in 43 (29 paroxysmal, 14 persistent), there were no LVAs. The ECG-based algorithm showed excellent diagnostic performance, with a sensitivity of 94%, a specificity of 95%, a positive predictive value of 94%, and a negative predictive value of 95% resulting in a general accuracy of 95%. Morphological abnormalities consistent with interatrial block and absent p-wave progression were the most frequent predictors among patients with confirmed LVAs.

Conclusions

This study demonstrates that specific P-wave morphological features derived from a standard amplified ECG, when integrated into a structured algorithm, may provide a highly accurate, non-invasive means of detecting left atrial fibrosis in patients with AF. The proposed four-parameter model correlated strongly with the invasive electrophysiologic study and may offer a practical tool for pre-procedural substrate screening and individualized ablation strategy planning (e.g. PVI vs PVI+).

A multicentric study is planned to validate these findings in a larger patient cohort, as well as to explore the predictive value of this algorithm in identifying PVI-only responders.