Echocardiographic predictors of low-voltage zones in the left atrium in patients undergoing atrial fibrillation ablation

Abstract

Reaching the left atrial arrhythmogenic substrate isolation is an important procedural target in patients with atrial fibrillation (AF) and advanced atrial remodelling. While the presence of low-voltage zones (LVZs) can be identified during the procedure using electroanatomic mapping (EAM), this technique is not applied routinely. To properly plan the procedural approach and select patients who should undergo EAM during AF ablation, we need non-invasive and easily obtainable predictors of LVZs. Echocardiography, with its ability to assess morphology and presence of left atrial thrombus can potentially be supplemented by the velocity in the left atrial appendage (LAAV) obtained in preprocedural transoesophageal echocardiography (TEE) - as reduced LAAV has previously been reported to be strongly associated with atrial remodelling. The aim of this study was to assess the predictive value of LAAV indicating LVZs presence in the left atrium compared with parameters routinely tested in transthoracic echocardiography (TTE).

Consecutive patients who underwent catheter ablation for AF were retrospectively and prospectively recruited in a single site participating in the multicentre IMAGE-AF registry. The study population was divided into two subgroups based on LVZs presence (areas with voltage ranges of 0.1-0.5 mV) on EAM performed during the procedure. Demographic, clinical, echocardiographic and procedural data were included into a multivariable logistic regression model. The analysed echocardiographic parameters included: left atrial volume index (LAVI), left ventricular ejection fraction (LVEF), E/e’, left ventricular septal and lateral e’ in TTE, and left atrial appendage velocity (LAAV) in TEE. Additionally, we trained and tested in 5-fold cross validation to Xgboost models - one of which included LAAV and another excluding it.

The study group consisted of 132 patients (80 males) with a median age of 67 years (IQR 60-73). LVZs were observed in 36 patients (27.3%). Persistent AF was present in 62 patients (47.0%). LVZs were observed in 11 (15.7%) patients with paroxysmal AF and in 25 (40.3%) patients with persistent AF. Significant predictors of LVZs (p<0.05) identified in the multivariable logistic regression are presented in Table 2. Apart from higher age, only a decrease in LAAV was a significant predictor of an increased risk of LVZs – almost a 2-fold increase per 10 cm/s (Figure 1). ROC curves (Figure 2) revealed no differences between models with (AUC=0.824; 95% CI=0.753-0.895) and without LAAV inclusion (0.784; 0.707-0.861) in LVZs prediction (DeLong p=0.064).

Preprocedural echocardiographic imaging allows the risk assessment of LVZs presence and may assist in the decision to perform a non-pulmonary vein substrate ablation. Importantly, LAAV emerged as the single strongest predictor of LVZ and its addition to a machine learning model markedly improved its discriminatory ability.