Local conduction velocities determined by non-invasive electrocardiographic imaging predict arrhythmia-free survival after pulmonary vein isolation

Abstract

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – EU funding. Main funding source(s): This project has received funding from the European Union's Horizon Research and Innovation Programme under the Marie-Sklodowska Curie Grant Agreement No. 860974 Background Recurrence rates after atrial fibrillation (AF) ablation are still unsatisfactory. As catheter ablation is primarily targeting pulmonary vein (PV) ectopic activity, it is not surprising that extra-PV arrythmogenic substrate is a key determinant of arrhythmia recurrence. Against this background, several studies have proposed assessment of extra-PV substrate in terms of atrial fibrosis or locally reduced conduction velocities (CV) to guide treatment. However, to date no non-invasive method directly assessing electrical arrhythmogenic substrate has been established in clinical practice, and treatment decisions are commonly based on crude surrogates like AF type or left atrial size. Here we establish and validate a novel non-invasive method based on electrocardiographic imaging (ECGi) to determine atrial arrhythmogenic substrate in terms of reduced local CVs and its predictive value regarding arrhythmia recurrence after PVI. Methods and results 52 consecutive patients scheduled for AF ablation (PVI-only) and 19 healthy controls were prospectively included and received ECGi to assess left and right atrial arrhythmogenic substrate. This ECGi-based method uses 64 electrodes placed on the torso. Subsequently, a 3D model of the torso is acquired as an anatomical reference using a 3D reconstruction camera. A personalised 3D atrial geometry is then derived from a database of human atria using an artificial intelligence-based algorithm. Finally, unipolar surface electrograms are projected onto the cardiac geometry and local CVs are estimated. Mean ECGi-determined atrial CVs were significantly lower in AF patients than in healthy controls, both in a global analysis (1.45±0.15 m/s vs. 1.64±0.15 m/s; p<0.0001) and a regional analysis of 19 predefined left and right atrial segments (Fig. 1). Considering only the segments with the lowest average CV in each patient, differences in CVs were more pronounced (0.80±0.22 vs. 1.08±0.26 m/s; p<0.0001). Multivariate logistic regression analyses combined with c-statistics including other previously proposed predictors found this mean CV of the "slowest" segment to be the strongest predictor of arrhythmia recurrence. A ROC analysis revealed that a cut-off for this variable of 0.72 m/s best discriminates PVI responders from non-responders: patients with a mean CV >0.72 m/s in all atrial regions showed a 6-months arrhythmia-free survival of 90.9%, whereas patients with one or more atrial regions with a mean CV <0.72 m/s had a poor outcome with an arrhythmia-free survival rate of only 57.9% (Fig. 2). Conclusion This was the first study to investigate local atrial CV non-invasively and to validate their predictive value regarding outcome after PVI. The absence of ECGi-determined areas of slow conduction well discriminated PVI responders from non-responders. Such non-invasive assessment of electrical arrhythmogenic substrate may guide treatment strategies and be an important step towards personalised AF therapy.