138 Patients with symptomatic paroxysmal (38%) or persistent AF (62%) were analyzed, 101 consecutive patients in Hamburg and 37 consecutive patients in Bad Oeynhausen (Fig. 1). Mean age was 67 ± 12 years with male predominance (66%), CHA2DS2–VASc Score of 2.7 ± 1.7, and with moderate LA enlargement (diameter 43 ± 5 mm, volume index 37 ± 13 ml/m2). Detailed patient characteristics are given in Table 1.

Flow chart of patients included in this prospective observational study (STROBE format). LC = left common, PFA = pulsed-field ablation, PV = pulmonary vein, AF = atrial fibrillation, AFl = atrial flutter, AT = atrial tachycardia

Mean procedure time (defined as time from femoral access until sheath removal or begin of a secondary ablation) was 78 ± 22 min including pre- (15 ± 8 min) and post-PVI (11 ± 5 min) voltage mapping time (Fig. 2C). Mean LA dwell time was 60 ± 20 min, PFA catheter dwell time was 24 ± 10 min (Table 2). The mean total number of PFA-applications was 33 ± 3/patient corresponding to a PVI mean ablation time of 83 ± 8 s/patient. Most additional applications were applied at right-sided PVs (23/277) vs. left-sided PVs (6/269; p < 0.001, Fig. 3). During the first 101 cases in Hamburg the PFA catheter dwell time was reduced from 32 ± 10 min (first 30 pts) to 23 ± 8 min (p < 0.001; 61–101st patient, Fig. 4), which was not significantly different from patient 31–61st (24 ± 10 min, p = 0.951). This reflects a learning curve reaching a plateau after approximately 30 procedures.

A Example of pulsed-field ablation catheter visualization in CARTO3 (Biosense Webster) at the LIPV in posterior–anterior (left) and left lateral projection (right). B Example of disappearance of pulmonary vein signals at the left superior pulmonary vein after the first pulsed-field application. C Three-dimensional electroanatomic voltage map of the left atrium in postero-anterior before (left) and after PFA-based pulmonary vein isolation (right)

Quantification of pulsed-field applications additional to the recommended 8 applications per pulmonary vein. Green represents additional application due to positioning on operator’s discretion. Red represents additional applications due to reconnection of pulmonary vein. LSPV = left superior PV, LIPV = left inferior PV, LC = left common, PFA = pulsed-field ablation, PV = pulmonary vein, RIPV = right inferior PV; RSPV = right superior PV

Single-center learning curve illustrated by PFA catheter LA dwell time during the first 101 cases in Hamburg. Each dot shows LA dwell time of the PFA catheter, the line is a best fitted exponential graph

Successful electrical PVI was achieved in 546/546 PVs (100%), including 8 left common PVs and 1 left and 1 right middle PV. Of note, PV signals recorded on the pentaspline PFA catheter disappeared after the first application (2.5 s) in 544/546 PVs (99.5%, for example, see Fig. 2B). A total of 515/546 PVs (94.3%) was treated by 8 applications (4 applications each in the basket and flower configuration, Fig. 5) according to the current recommendations. In 17/138 patients (12.3%) and 20/546 PVs (3.7%) additional applications were deployed due to adapted positioning of the catheter at the respective PVs (Table 3). Acute PV reconnection was observed during remapping in 9/138 patients (7%) and in 10/546 PVs (1.8%, Fig. 3, Table 3). All these PVs were re-isolated by additional PFA applications in adapted catheter positions. No RF touch-up was necessary. PFA induced wide antral circumferential lesions around the ipsilateral PVs (Fig. 2), even in left common PVs (S-Fig. 1). In a subset of post-PVI voltage maps (n = 33), posterior distance between left and right low-voltage areas was reduced from 69 ± 12 mm prior ablation to 25 ± 10 mm (p < 0.001) after PVI at superior PVs and from 61 ± 10 mm to 25 ± 10 mm (p < 0.001) at inferior PVs with a remaining non-ablated LA posterior wall area of 11.9 ± 4.7 cm2.

Pulsed-field ablation catheter (FARAPULSE) in its basket (A) and flower (B) configuration. Corresponding fluoroscopic views of the pulsed-field ablation catheter at a right superior pulmonary vein in basket (C, right anterior oblique 30°) and flower (D, left anterior oblique 40°) configuration. Red arrow indicates subequatorial electrode exemplary for one spline, which can be visualized in 3D mapping systems (Fig. 3)

Phrenic capture during PFA pulses was observed in 90/93 (96.8%) patients and was more frequently seen along the right than left PVs (88/93 vs. 30/93, p < 0.001) and more often at the superior vs. inferior PVs (85/93 vs. 75/93, p = 0.035). In detail, phrenic nerve capture occurred at 83/93 RSPVs (89.2%), 71/93 RIPVs (76.3%), 25/93 LSPVs (26.9%) and 15/93 LIPVs (16.1%). Atropine was administered in a total of 88/138 patients to reduce vagally induced bradycardia due to sinus pause (S-Fig. 2) or atrioventricular block (S-Fig. 3). 4/138 patients received atropine after previous distinct vagal reactions following first PFA pulses. Asystole was much shorter in patients with previous atropine administration (2.0 ± 0.7 s, n = 10) than without (5.9 ± 3.3 s, n = 10, p = 0.018). 49/138 patients did not receive atropine due to the operator’s discretion or due to implanted pacemakers with a ventricular lead or due to the ability to pace the ventricle via the coronary sinus catheter.

The PFA pentaspline catheter could be visualized within the 3D map with both mapping systems (CARTO3, Bisosense Webster, Fig. 2 and EnSite™NavX™, Abbott, S-Fig. 1), helping to navigate the catheter. The subequatorial electrode of each spline (third electrode) could be connected to the mapping system (Fig. 5), resulting in a ring in the 3D map (Fig. 2A), representing the external dimension of the catheter in basket or flower configuration. During PFA pulses it is recommended to pause the catheter visualization. When visualization was not paused, the catheter disappeared during PFA pulses and re-appeared within a few seconds afterward, but in few cases 3D mapping matrix was lost. With the Ensite mapping system, the pentaspline PFA catheter could also be used to create a 3D voltage map of the LA recorded by 5-electrodes in total, although mapping resolution was reduced compared to dedicated mapping catheters.

Two patients developed an atrial tachycardia during the procedure and according to entrainment maneuvers the lateral anterior wall was involved in the circuit. PFA applications (4 applications in basket configuration in the first patient and 8 applications, 4 in basket- and 4 in flower configuration, in the second patient) at the anterior wall terminated the tachycardias. However, bidirectional conduction block was not assessed. In one patient with persistent AF, the posterior wall was isolated by 4 applications in flower configuration.

One patient with persistent AF suffered from repetitive onset of AF after multiple electrocardioversions after PFA-based PVI. The trigger origin was identified within the superior vena cava (SVC). The PFA catheter was then placed into the SVC and 4 PFA applications were conducted. During these applications sinus arrest occurred and after 5 min of atrial pacing sinus rhythm returned. After SVC ablation AF did not re-occur. However, isolation of SVC was not assessed.

In 4/138 patients (2.9%) bidirectional block of the CTI was performed after PVI with PFA by 9 ± 2 applications and in 9/138 patients (6.5%) by RF ablation. One patient received slow-pathway modulation by RF after induction of atrioventricular nodal reentry tachycardia.

Periprocedural complications were one cardiac tamponade requiring pericardiocentesis (blood gas analysis suggested venous bleeding, potentially indicating an association with a difficult insertion of the CS-catheter in this procedure), one patient experienced transient ST-elevation and concomitant atrioventricular-block for 3 min which resolved spontaneously and three minor groin site complications which were treated conservatively. One patient described minor neurological missperceptions and underwent cerebral MRI without any pathological signs, while symptoms disappeared within hours. No other adverse events occurred.

129 of 137 (94%) patients could be included into follow-up analysis with a mean follow-up time of 249 ± 90 days after PVI. The single procedure Kaplan–Meier estimates for freedom from AF, AFl or AT was 60 ± 10% for patients with persistent AF (n = 82) vs. 90 ± 6% in patients with paroxysmal AF (n = 47, p = 0.015, Mantel–Cox test, Fig. 6). At last follow-up, class I or III antiarrhythmic drug therapy was continued in 2.4% patients with persistent AF and in 2.1% with paroxysmal AF.

Kaplan–Meier curve showing freedom from AF/AFL or AT in paroxysmal (blue) and persistent (red) atrial fibrillation after initial pulmonary vein isolation by pulsed-field ablation during 1-year follow-up. AFl indicates atrial flutter; AF, atrial fibrillation and AT, atrial tachycardia