Impact of saline irrigation on haemolysis, silent cerebral lesion incidence, thermal dynamics, and bubble formation in pulsed field ablation with a variable-loop circular catheter

Abstract

Aims

Though pulsed-field ablation (PFA) has demonstrated an excellent safety profile in reducing collateral injury to the oesophagus and phrenic nerve, it is still associated with specific effects, including electrode heating, haemolysis, and electrolysis due to excessive energy dispersion. This study aims to assess whether saline irrigation during PFA application could mitigate these risks.

Methods and results

To comprehensively evaluate the effect of irrigation with the variable-loop circular catheter (VLCC), the following experiments were performed: (i) ex-vivo potato model: to evaluate the lesion depth, bubble formation, and thermal effects in different irrigation regimens; (ii) in vitro blood pool and cardiac ablation: to determine the haemolysis status and tissue temperature change after PFA; (iii) in vivo swine ablation (n = 8), and (iv) clinical randomized trial (n = 25): to compare the efficacy and safety profile between low (4 mL/min) and high (30 mL/min) flow irrigation using the VLCC. Though peak core temperatures at 5 mm depth were all < 50°C under low- and high-irrigation, high irrigation significantly mitigated the instant electrode and deep tissue heating both in the potato and isolated cardiac models. Ex vivo potato slices showed that high-flow irrigation produced the deepest lesion sets when compared to low-flow irrigation (5.94 ± 0.29 mm vs. 5.36 ± 0.33 mm, P = 0.043). Assessment from a high-speed camera and bubble detector demonstrated that high-flow irrigation significantly reduced the total number of gaseous bubbles (54.50 IQR 53.00–56.75 vs. 82.00 IQR 72.00–83.00, P < 0.001) and eliminated the occurrence of larger bubbles. The high-flow irrigation group showed a smaller increase in the level of free haemoglobin immediately after the procedure across the blood pool, swine, and clinical models. Haptoglobin and lactate dehydrogenase levels were also attenuated by high irrigation in the in vivo swine model and clinical trial. One swine in the low-irrigation group developed an acute cerebral lesion (3 mm). The clinical trial confirmed that the incidence of silent cerebral lesions was significantly lower in the high-flow irrigation group (16.7% vs. 66.7%, P = 0.036).

Conclusion

Proper saline irrigation during PFA with VLCC may mitigate electrode-associated haemolysis, reduce electrode and tissue temperature, limit bubble aggregation, and be associated with a lower incidence of silent cerebral lesions, the clinical significance of which remains unclear.

Graphical Abstract

Graphical Abstract.

What's new?

• Saline irrigation during PFA could reduce haemolysis by ‘surrounding’ the electrodes during energy delivery

• In addition, high-flow saline irrigation seems to slightly increase lesion depth, which may be beneficial for ablation in thick-wall myocardial tissue

• High flow irrigation may flush microbubbles formed on the electrodes, thereby preventing their coalescence into clinically significant large bubbles or emboli, and is associated with a lower incidence of silent cerebral lesions.

• Irrigation could reduce electrode and surface heating, which may be associated with a lower risk of thermal injury and char formation.

Introduction

Pulsed-field ablation (PFA) is an innovative, minimally thermal energy modality that delivers transient, high-voltage electrical pulses to induce irreversible electroporation in cardiac myocytes.^1,2 Despite clinical evidence supporting PFA’s efficacy and safety due to its cardiac tissue selectivity,^3–9 inherent risks—such as electrode heating, haemolysis, and electrolysis^6,10–15—remain because of the complex electrochemical reactions involved in energy delivery. Although the clinical implications of these biophysical phenomena are not fully determined, their potential correlation with thromboembolic events and renal impairment remains a significant safety concern.

Irrigation is the standard protocol for radiofrequency ablation in atrial fibrillation (AF) treatment, utilized to cool the catheter–tissue interface and prevent thrombus and char formation.¹⁶ In the context of PFA, particularly when using the variable loop circular catheter (VLCC), the rationale for irrigation is increasingly relevant. Recent studies indicate that electrode heating during VLCC PFA is non-negligible, and active irrigation can effectively mitigate this temperature rise.^17,18 Furthermore, a similar protective mechanism may apply to haemolysis. Active irrigation could theoretically dilute the local blood concentration and flush red blood cells away from the electrodes,¹⁹ thereby potentially attenuating haemoglobin release. However, critical knowledge gaps remain. It is unknown whether a higher irrigation rate during PFA may influence microbubble aggregation, the incidence of subclinical neurovascular events, or inadvertently compromise lesion formation by altering the electrical field distribution. Therefore, this study was conducted to comprehensively evaluate the biophysical and clinical effects of irrigation using the VLCC catheter, focusing on the thermal profile, microbubble formation, haemolysis, and lesion characterization from ex vivo models to clinical application.

Methods

Study design

The current study consisted of four parts: (i) ex vivo potato model, (ii) in vitro blood pool and cardiac experiment, (iii) in vivo swine ablation, and (iv) randomized clinical trial (Figure 1). A variable-loop circular catheter (VLCC) with an adjustable irrigation system (VARIPULSE, Biosense Webster, Irvine, CA, USA) and a three-dimensional electro-anatomical system (Carto 3, Biosense Webster, Irvine, CA, USA) was used throughout the experiments. The catheter contact was carefully monitored by impedance-based tissue proximity indication (TPI) during PFA applications.

Figure 1.

Study design. (A) The current study comprised a series of experiments to evaluate the safety and efficacy of irrigation in VLCC. The corresponding examination was listed in the table. (B) The irrigation effect through 0 to 60 mL/min was presented by infusing a dye-colored solvent under a high-speed camera (1000 fps) in the same moment. ICE, Intracardiac echocardiography; MRI, Magnetic resonance imaging.

We categorized the irrigation flow as follows: 0 mL/min irrigation group, 4 mL/min irrigation group (low-flow irrigation, as recommended by the manufacturer during the study period), 15 mL/min irrigation group, 30 mL/min irrigation group (high-flow irrigation), and 60 mL/min irrigation group (see Supplementary material online, Video S1-S5). To minimize irrigation volume, irrigation in the 30 mL/min and 60 mL/min group was manually controlled to last 1–2 s before and after each PFA application. Based on the observation from ex-vivo and in vitro experiments, further comparison was performed under 4 and 30 mL/min in the swine and clinical models.

Ethics approval was obtained from the local ethics committees, and all study procedures were approved by the institutional review boards (AZ2025LLA001, 2025221x). Written informed consent was obtained from all study participants.

The integrated PFA system

A fully integrated PFA system, including the VLCC (VARIPULSE, Biosense Webster, US) with a PFA generator (TRUPULSE), an adjustable irrigation system, and a three-dimensional electro-anatomical system (CARTO 3, Biosense Webster, Irvine, CA, USA) was used throughout the duration of the experiment. This circular catheter contains 10 electrodes, each surrounded by 10 evenly distributed irrigation ports (100 ports in total) to ensure uniform saline delivery. The irrigation system additionally includes an adjustable-flow pump (nGEN, capable of controlling and recording), standard irrigation tubing, and a 0.9% normal saline reservoir, which together provide controlled and consistent saline flow throughout all applications. Catheter contact was carefully monitored by TPI during PFA applications.

Potato experiment

A total of 30 potato models were immersed in a thermostatic saline bath. A robotic arm was used to position the catheter perpendicular to the model surface. Adjustments were made based on the TPI value to ensure sufficient contact of all electrodes (impedance: 180–200 Ω, conductivity: 6.5 ± 0.5 mS/cm). PFA applications were performed under different irrigation flow rates, with each flow rate being tested three times. After ablation, the models were incubated in 1% triphenyl tetrazolium chloride solution, and maximum lesion depth was measured by two independent investigators.

A water bath circulatory system was used for qualitative detection of bubble generation and temperature measurement under different irrigation conditions (see Supplementary material online, Figure S1A).²⁰ A custom air bubble indicator (SMARTABLATE™ Irrigation Pump) was placed at a distance corresponding to approximately 10 s of circulation time from the catheter, to simulate the transit time from the atrium to the cerebral circulation (circulation speed: 100 mL/min). The system operated continuously for 5 min as one complete test and was repeated three times to record the number of bubble-positive alerts. The bubble which triggered the alert within the tube was examined and its diameter was estimated. A high-speed camera (1000 fps, Phantom Flex 4 K, Wayne, NJ, USA) was used to detect bubble formation at each electrode of the circular catheter under different irrigation conditions.²⁰ Bubble counting was performed with each PFA application considered as one test; each test was repeated three times. The camera shot was oriented towards the electrode-facing side of the catheter to visualize bubble formation associated with the 10 electrodes. In addition to the micro bubbles produced by energy delivery, large air bubbles were defined as those with diameters >0.5 mm. The total number of bubbles and the occurrence of large bubbles were independently interpreted by two physicians. Large bubbles attached to the electrodes and formed by the aggregation of microbubbles were specifically counted.

In the same circulatory system, electrode temperatures were measured using a thermal camera (Hikmicro, −22°C to 550°C, 25 fps) and tissue temperatures were measured using a fibre-optic temperature sensor (OFSCN, −40°C to 120°C, 10 Hz)(see Supplementary material online, Figure S1B). The temperature probe was secured at 0, 3 and 5 mm from the electrode. The peak temperature and its variation were recorded after three successive PFA applications, each separated by a 10-s interval. The cumulative tissue heating effect was tested after nine applications, and the final temperature was recorded.

Blood pool and cardiac ablation

Thirty separate experiments were performed using freshly collected, fully heparinized blood from three swine (200 mL per experiment). Each irrigation group was separately tested under two catheter positions: (i) floating in blood (no contact) and (ii) in contact with excised fresh swine myocardial tissue, submerged within the blood (in contact). Each test was repeated three times. After ablation, the blood was diluted to an equal volume. Blood samples were collected to measure free haemoglobin (fHb) levels before application and after the 48 (=16 ablation) and 72 (=24 ablation) applications (see Supplementary material online, Figure S1C). The temperature experiment was repeated as previously described using fresh cardiac ventricular tissue in the circulatory bath.

PFA procedure in swine

Eight Yorkshire swine (45.64 ± 2.95 kg) were 1:1 randomly assigned to 4 mL/min irrigation group and 30 mL/min irrigation group. The procedures were performed under general anaesthesia with endotracheal intubation and continuous electrocardiographic monitoring. Heparin was administered immediately after femoral access and transseptal puncture, targeting an activated clotting time (ACT) ≥ 350 s, and was maintained throughout the procedure.

A total of 48 PFA applications (=16 ablations) were delivered to the right/inferior pulmonary veins (PV) and posterior wall in all swine. Subsequently, an additional 24 applications (=8 ablations) were performed at the right atrium (including superior vena cava, right atrial appendage, and lateral wall). Blood samples were collected at baseline, after 48 and 72 applications, and after 24 h post-ablation. Swine were euthanized humanely 14 days post-ablation. Lesion tissues were submitted for histopathology assessments to evaluate lesion depth. Cerebral magnetic resonance imaging (MRI) scanning within 24 h post-ablation procedure was performed to document any new-onset cerebral lesions.¹⁵

Clinical PFA procedure

The current study intended to include 40 patients undergoing PFA for drug-refractory symptomatic paroxysmal AF using VLCC with 1:1 randomization into 4 mL/min irrigation group (control) or 30 mL/min group (experimental). Because a higher incidence of silent cerebral lesion (SCL) events was observed in the control group after the enrolment of 25 patients (12 for the control and 13 for the experimental group), early termination of the study was approved by the independent data and safety monitoring committee.

All procedures were performed under general anaesthesia guided by three-dimensional electro-anatomical system. For each PV, two ostial and two antral ablations were delivered using all 10 electrodes with three applications in each ablation. Supplemental PFA applications were limited to posterior wall/roofline/superior vena cava with a single ablation per site. The total number of applications was less than 72 (=24 ablations) per patient. Linear ablation at the mitral/tricuspid isthmus was performed using radiofrequency energy (STSF, Biosense Webster, Irvine, CA, USA), if necessary.

The catheter was manipulated meticulously to avoid stacking of ablations at the same site. A 10-s waiting-time interval and pull-out-and-push-in reposition of the catheter between each ablation was employed to avoid a cumulative heating effect. Targeted ACT was ≥350 s and was maintained throughout the procedure. No peri-procedural hydration was performed. An intracardiac echocardiography was used to monitor gaseous bubble formation during ablation. Haemolysis biomarkers were collected at baseline, immediately post-ablation, and 12–24 h post-ablation. Direct (fHb) and indirect biomarkers [haptoglobin, lactate dehydrogenase (LDH), and indirect bilirubin (IBIL)] were used to examine the severity of haemolysis.^10–12,21 Cerebral MRI was performed 12–24 h after the procedure.

Statistical analysis

Continuous variables were reported as mean ± standard deviation, or median (interquartile range), and categorical variables as numbers (percentages). Continuous variables were compared using ANOVA for parametric data and Kruskal–Wallis or Mann–Whitney testing for non-parametric data, while categorical variables were compared using Chi-square or Fisher’s exact testing. Spearman’s rank correlation was used to assess the association of fHb levels with contact conditions and the number of PFA applications. Differences in time-course patterns between low- and high-speed irrigation groups were examined using linear mixed-effects models, with time, group, and their interaction (time × group) treated as fixed effects. Because repeated biomarker measurements were obtained from the same subject across multiple timepoints, time was modelled as a within-subject factor. To account for individual-level baseline variability, a random intercept was included for each subject, resulting in a random-effects structure of (1 | number). Post-hoc comparisons of estimated marginal means at each timepoint (T1–T4) between the two irrigation groups were performed using the Šídák correction for multiple testing. A P-value <0.05 was considered statistically significant, and all analyses were completed using R software (version 4.4.1).

More detailed descriptions of the methodology are provided in the Supplementary Methods.

Results

Potato experiment

Thermal effect

Using the thermal camera (Supplementary material online, Figure S2), the 30 and 60 mL/min irrigation groups demonstrated lower peak electrode temperatures (55.30 ± 0.30 °C vs. 51.20 ± 2.62 °C, P = 0.156) compared to the 0, 4, and 15 mL/min groups (61.97 ± 0.45 °C, 62.97 ± 3.17 °C, and 61.87 ± 1.50 °C, respectively P = 0.78; overall P < 0.001). For tissue temperature, the 30 mL/min group exhibited the lowest peak surface temperature (45.45 ± 0.85 °C), which was significantly lower than that in the 0, 4, and 15 mL/min groups (all P < 0.005), but not significantly different from the 60 mL/min group (43.67 ± 0.89 °C, P = 0.192). A similar trend between the 4 mL/min and 30 mL/min irrigation group was also observed at 3 and 5 mm depth (Table 1). Peak tissue temperatures were achieved at the third PFA application, all of which remained below 60°C. All deep tissue temperatures returned to below 50°C within 2 s, as provided in the Supplementary material online, Table S1.

Table 1.

Peak temperature of electrode and tissue

Peak temperature
	0 mL/min	4 mL/min	15 mL/min	30 mL/min	60 mL/min	P
Potato
electrode temperatures,°C	61.97 ± 0.45	62.97 ± 3.17	61.87 ± 1.50	55.30 ± 0.30	51.20 ± 2.62	<0.001
tissue temperatures—0 mm,°C	58.13 ± 0.89	54.75 ± 0.44	49.02 ± 1.28	45.45 ± 0.85	43.67 ± 0.89	<0.001
tissue temperatures—3 mm,°C	54.56 ± 0.14	53.43 ± 0.78	49.61 ± 0.57	43.42 ± 0.11	40.91 ± 0.59	<0.001
tissue temperatures—5 mm,°C	40.85 ± 0.30	40.57 ± 0.20	39.13 ± 0.47	39.13 ± 0.77	39.38 ± 0.55	0.003
Isolated Cardiac
electrode temperatures,°C	66.60 ± 2.52	65.30 ± 4.97	59.33 ± 3.97	55.07 ± 1.16	52.70 ± 2.62	0.001
tissue temperatures—0 mm,°C	57.38 ± 0.71	57.41 ± 0.63	52.84 ± 1.56	52.65 ± 0.59	48.43 ± 0.68	<0.001
tissue temperatures—3 mm,°C	54.18 ± 0.71	54.53 ± 0.99	49.96 ± 0.82	47.05 ± 0.66	46.58 ± 1.03	<0.001
tissue temperatures—5 mm,°C	39.69 ± 1.18	38.94 ± 0.19	38.32 ± 0.30	37.81 ± 0.16	38.00 ± 0.09	0.012

Lesion depth

After three applications at the same site (one standard ablation), an acute lesion was produced with an average depth of 5.49 ± 0.41 mm. The 30 mL/min irrigation group produced the deepest lesions (5.94 ± 0.29 mm), which were significantly deeper than that observed in the no irrigation group (5.17 ± 0.37 mm, P = 0.004) and 4 mL/min irrigation group (5.36 ± 0.33 mm, P = 0.043). There was no significant difference from the 15 mL/min irrigation group (5.52 ± 0.40 mm, P = 0.218) and 60 mL/min irrigation group (5.48 ± 0.29 mm, P = 0.149) (Figure 2).

Figure 2.

Lesion depth under different irrigation. (A) The longitudinal sliced (≈2 mm) vegetal sample after TTC staining was presented. For each lesion set, one single complete ablation was performed, comprising 3 separate applications delivered ≈30 s, using all 10 electrodes with maximum loop diameter. (B) Box plot showing the average lesion depth after three repeated experiments. The 30 mL-irrigation group yielded the deepest lesion sets (5.94 mm ± 0.29 mm), as compared to no- (5.17 mm ± 0.37 mm, P = 0.004) and 4 mL/min- (5.36 mm ± 0.33 mm, P = 0.043) irrigation groups. Asterisk (*) indicates that the difference reaches statistical significance when comparing with 30 mL/min. Significance level: * P < 0.05, ** P < 0.01.

Bubble formation

The high-speed camera showed that there were two types of gaseous bubbles produced by PFA: large and micro bubbles. The former’s diameter was estimated to be over 0.5 mm. At 0–4 mL/min irrigation rates, large bubbles mostly originated from microbubble aggregation and were able to attach to the electrodes in every independent test (as shown in Supplementary material online, Video S6). Large bubbles from microbubble aggregation were absent under 30–60 mL/min irrigation.

In quantitative analysis, the total number of bubbles in the 30 mL/min irrigation group [54.50(53.00,56.75)] was significantly less than that in lower irrigation groups [0 mL/min, 82.00(72.00,83.00), P < 0.001; 4 mL/min, 71.00(66.25,75.75), P = 0.004; 15 mL/min, 68.00(61.50,72.25), P = 0.031]. No significant difference was observed between the 30 mL/min irrigation group compared to the 60 mL/min irrigation group [42.50(38.00,49.25), P = 0.365] (Figure 3C). The number of large bubbles in the 30 mL/min irrigation group (3.17 ± 1.47) was significantly less than that in 0 mL/min (7.50 ± 2.59, P = 0.003) and 4 mL/min irrigation groups(6.33 ± 2.16, P = 0.040); but, no significant difference was observed compared to the 15 mL/min (4.00 ± 1.26, P = 0.929) and 60 mL/min irrigation group (3.00 ± 1.10, P = 1.000) (Figure 3D).

Figure 3.

Bubble formation under high-speed camera recording. (A) Two types of air bubbles were produced when PFA was delivered. As indicated, most of them were microbubbles, which were rapidly flushed away. Some bubbles are significantly larger (>2 times the diameter of micro bubbles), and they were easy to adhere to the electrodes. (B) The bubble formation under different irrigation regimes. Of note, a large bubble did not appear at the 30 mL/min and 60 mL/min irrigation groups. (C) Total number of bubbles, including micro and large bubbles, produced by a single ablation in all 10 electrodes. (D) Number of large bubbles produced by a single ablation in all 10 electrodes. An asterisk (*) indicates that the difference reaches statistical significance when comparing with 30 mL/min. Significance level: * P < 0.05, ** P < 0.01, *** P < 0.001.

In qualitative analysis, bubble detectors identified visibly large air-bubbles (diameter ≥ 1 mm) within the tubing in the no irrigation and 4 mL/min irrigation group after a continuous 5-minute circulatory cycle (Table 2). No large air-bubbles were detected in the 30–60 mL/min irrigation group.

Table 2.

Qualitative detection of bubble formation

	0 mL/min	4 mL/min	15 mL/min	30 mL/min	60 mL/min
Test-1	a	a	c	c	c
Test-2	b	a	c	c	c
Test-3	a	c	b	c	c

^aVisibly large bubble was detected by bubble detector during the continuous flush.

^bThe bubble detector did not detect any bubbles, but bubbles were visibly attached to the catheter.

^cThe bubble detector did not detect any bubbles and no bubbles were attached to the catheter.

Blood pool and isolated cardiac model

Thermal effect

In the isolated cardiac model, peak temperature, cumulative tissue heating, and recovery time below 50°C were similar to those in the potato model (Table 1). The maximum tissue temperatures at 3 mm depth with 4 mL/min and 30 mL/min irrigation were recorded as 67.39°C and 54.72°C, respectively, following nine complete applications. Of note, the surface tissue exhibited visible lesions after a single ablation under irrigation rates of 0–15 mL/min (see Supplementary material online, Figure S3).

Haemolysis

Before ablation, the plasma appeared clear, and visible haemolysis was observed in all 60 (100%) samples post-ablation. After dilution to the same volume (400 mL), the concentration of fHb under different irrigation flow rates was as follows: 7.93 ± 0.91 (0 mL/min), 6.47 ± 0.98 (4 mL/min), 4.79 ± 0.76 (15 mL/min), 3.27 ± 0.63 (30 mL/min), and 2.53 ± 0.84 g/L (60 mL/min) (P < 0.001). The fHb concentration of the 30 mL/min-irrigation group was significantly lower compared to the lower irrigation groups but showed no significant difference from the 60 mL/min irrigation group (Figure 4A).

Figure 4.

Haemolysis in blood pool experiment. The blood sample was collected at baseline (T1), after 48 applications (T2), and after 72 applications (T3). (A) Box plot showing the average free haemoglobin level under different irrigation flows. The free haemoglobin concentration of the 30 mL/min irrigation group was significantly lower than in other groups, but showed no significant difference from the 60 mL/min irrigation group. (B) The bar graph showed that the level of free haemoglobin was significantly reduced with increasing irrigation flow rate regardless of catheter contact. This effect was equivalent after 48 and 72 applications. Though the level of free haemoglobin was higher under the no-contact condition than in-contact across all irrigation groups, the extent of haemolysis could still be mitigated by increasing irrigation flow. An asterisk (*) indicates that the difference reaches statistical significance when comparing with 30 mL/min. Significance level: ** P < 0.01, *** P < 0.001.

Additional analysis was performed to assess the impact of irrigation on haemolysis under different contact conditions and number of applications. A strong negative correlation was observed between irrigation flow rate and fHb levels for both in-contact (ρ=−0.939, P < 0.001) and no-contact conditions (ρ=−0.931, P < 0.001). A similar effect was also observed after 48 applications (ρ=−0.896, P < 0.001) and 72 applications (ρ=−0.923, P < 0.001). Notably, this protective effect observed with higher irrigation was not affected by the number of applications: fHb levels decreased by 0.081 g/L per 1 mL/min increase in flow at 48 applications and by 0.078 g/L at 72 applications (P = 0.814) (Figure 4B).

PFA procedure in swine

Haemolysis

Consistent with PFA in the blood pool, all plasma samples appearance showed visible haemolysis immediately after ablation and became clear 24 h after ablation (Figure 5A). The immediate level of fHb concentration was significantly higher in the 4 mL/min group than in the 30 mL/min group, both after 48 applications (0.46 ± 0.16 g/L vs. 0.24 ± 0.06 g/L, P = 0.047) and 72 applications (0.60 ± 0.13 g/L vs. 0.36 ± 0.08 g/L, P = 0.021).

Figure 5.

Observations from swine ablation (n  =  8). The blood sample was collected at baseline (T1), after 48 applications (T2), after 72 applications (T3), and 24 h after procedure (T4). (A) The plasma sample was presented after 48 applications under 4 mL/min (left) and 30 mL/min (right) irrigation flow. The colour of plasma under 30 mL/min irrigation was significantly brighter than that under 4 mL/min irrigation. (B) The degree of haemolysis was attenuated by 30 mL/min vs. 4 mL/min irrigation (all P for interaction < 0.05), as indicated by the temporal changes in biomarkers of haemolysis, including free haemoglobin, haptoglobin, LDH, and IBIL. (C) Voltage remapping suggested an extensive low-voltage area within the ablated region using 30 mL/min irrigation. (D) A cerebral lesion was observed in one swine that belonged to 4 mL/min group. LDH, lactate dehydrogenase; IBIL, indirect bilirubin; significance level: * P < 0.05, ** P < 0.01.

The mixed-effect model was used to evaluate the temporal trends and intergroup differences under two irrigation conditions. The results showed that fHB (P < 0.01), haptoglobin (P < 0.001), LDH (P < 0.05), and IBIL (P < 0.05) levels were statistically significant between irrigation groups (Figure 5B), while blood cell counts and renal function biomarkers did not differ between the 4 mL/min and 30 mL/min irrigation conditions (see Supplementary material online, Figure S4).

Lesion depth

Remapping at 14 days post-ablation demonstrated extensive, low-voltage regions within the ablated areas (PV ostium and posterior wall) across two groups (Figure 5C). Histopathological examination showed transmural lesion formation (see Supplementary material online, Figure S5).

Acute cerebral microinfarct

An acute cerebral infarction lesion was found in one swine (4 mL/min). The lesion diameter was 3 mm, located near the occipital lobe of the brain. Other swine did not show any abnormalities (Figure 5D and S6). Bubble formation within the LA chamber under 4 mL/min and 30 mL/min irrigation was visualized by an ICE catheter, as shown in Supplementary material online, Video S7–S8.

PFA procedure in patients

Clinical PFA procedure

Enrolment was prematurely terminated after 25 patients due to a higher incidence of SCLs in the 4 mL/min group. Twelve patients received VLCC PFA with 4 mL/min irrigation and 13 patients received PFA with 30 mL/min irrigation. After randomization, baseline and procedural characteristics were well balanced (Table 3). The total procedural duration (70.25 ± 20.58 vs. 83.54 ± 14.26 min, P = 0.078) and total PFA applications (59.00 ± 4.67 vs. 58.38 ± 5.56, P = 0.767) were comparable between the two groups. Ablation beyond PV was performed in 6 patients in the 4 mL/min irrigation group and 10 patients in the 30 mL/min group (50% vs. 76.9%, P = 0.226). Adjunctive linear ablation using radiofrequency energy was performed in 4 and 7 patients, respectively (33.3% vs. 53.8%, P = 0.428).

Table 3.

Baseline and procedural characteristics

	4 mL/min (n = 12)	30 mL/min (n = 13)	P
Baseline characteristics
Age, year	70.50 [63.00, 73.00]	63.00 [55.00, 69.00]	0.164
Male, n (%)	9(75.0%)	11(84.6%)	0.645
Body mass index, kg/m2	25.14 ± 2.86	26.03 ± 2.25	0.399
Hypertension, n (%)	8 (66.7%)	7 (53.8%)	0.688
Diabetes, n (%)	5(41.7%)	3(23.1%)	0.411
Vascular disease, n (%)	4(33.3%)	3(23.1%)	0.673
Previous stroke/transient ischemic attack, n (%)	1(8.3%)	2(15.4%)	1.000
Heart failure, n (%)	0(0.0%)	1(7.7%)	1.000
CHA2DS2VASC score	3.00 [1.75, 3.25]	2.00 [1.00, 2.00]	0.132
Anaemia, n (%)	1(8.3%)	3(23.1%)	0.593
Impaired kidney function (baseline GFR <50 mL/min), n (%)	0(0.0%)	2 (15.4%)	0.480
Procedural data
Procedure duration, min	70.25 ± 20.58	83.54 ± 14.26	0.078
Left atrial dwell time, min	31.25 ± 14.69	36.69 ± 10.04	0.296
Sheath exchange ≥ 1, n (%)	4 (33.3%)	4 (30.8%)	1.000
Catheter exchange ≥ 1, n (%)	8 (66.7%)	12 (92.3%)	0.160
Catheter exchange ≥ 2, n (%)	4 (33.3%)	8 (61.5%)	0.238
Atrial ablation beyond PVI (PVI+), n (%)	6 (50.0%)	10 (76.9%)	0.226
PFA applications (PVI)	56.25 ± 5.14	53.77 ± 5.40	0.251
PFA applications (PVI+)	3.00 [3.00, 7.50]	6.00 [3.00, 8.25]	0.641
PFA applications (Total)	59.00 ± 4.67	58.38 ± 5.56	0.767
Radiofrequency ablation, n (%)	4 (33.3%)	7 (53.8%)	0.428
Sinus rhythm before ablation, n (%)	11 (91.7%)	9 (69.2%)	0.322
Electrical cardioversion, n (%)	0 (0.0%)	2 (15.4%)	0.480
Average ACT, S	480.67 ± 136.51	453.77 ± 74.87	0.554

PVI, pulmonary vein isolation; PVI+, atrial ablation beyond PVI; ACT, activated clotting time.

Haemolysis

Baseline haemolysis biomarkers, blood cell count, and renal function were comparable between the two groups (all P > 0.05, Table 4). At the end of the procedure, haemolysis (fHb > 0.2 g/L) was observed in seven patients in the 4 mL/min irrigation group and in one patient in the 30 mL/min group (58.3% vs. 7.7%, P = 0.011). In addition, mean fHb levels were significantly higher in the 4 mL/min irrigation group (0.22 ± 0.12 vs. 0.09 ± 0.04 g/L; P = 0.006), albeit the difference was no longer detectable 24-h after ablation [0.01(0.01,0.01) vs. 0.01(0.01,0.01) g/L; P = 0.341] (Table 4). Although the absolute levels of haptoglobin[0.38 (0.26, 0.56) vs. 0.42 (0.34, 0.80) g/L, P = 0.135], LDH (256.08 ± 33.64 vs. 244.97 ± 20.30 U/L, P = 0.335) and IBIL (19.34 ± 9.04 vs. 20.82 ± 7.31 µmol/L, P = 0.660) did not differ significantly between the two groups (all P > 0.05), further analysis revealed that the temporal changes in fHb (P < 0.001), haptoglobin (P < 0.05), and LDH (P < 0.001) levels differed significantly between the two irrigation groups (Figure 6A), with fHb (0.20 ± 0.12 vs. 0.08 ± 0.04 g/L, P = 0.005) and LDH (94.67 ± 14.79 vs. 58.82 ± 25.47 U/L, P < 0.001) showing significantly different magnitudes of increase (Table 4). No patient experienced acute kidney injury or visible haematuria after the index procedure. An increase in creatinine over 5 μmol/L was observed in five patients in the 4 mL/min irrigation group and two patients in the 30 mL/min group (41.7% vs. 15.4%, P = 0.202).

Table 4.

Blood test at baseline and post-ablation

	4 mL/min(n = 12)	30 mL/min(n = 13)	P
Baseline
fHb (g/L)	0.01 [0.01, 0.02]	0.02 [0.01, 0.02]	0.183
Haptoglobin (g/L)	1.26 ± 0.35	1.21 ± 0.45	0.721
LDH (U/L)	161.42 ± 23.77	186.15 ± 37.05	0.059
IBIL (µmol/L)	8.44 ± 3.15	11.78 ± 4.80	0.051
RBC (*1012/L)	4.67 ± 0.43	4.75 ± 0.55	0.676
HB (g/L)	142.50 [133.00, 152.00]	150.00 [138.00, 153.00]	0.785
MCV (fL)	90.25 [88.38, 91.65]	88.10 [85.90, 89.10]	0.103
MCH (pg)	30.65 ± 1.66	29.63 ± 2.11	0.191
MCHC (g/L)	343.17 ± 6.56	339.69 ± 10.44	0.327
Creatinine (µmol/L)	74.33 ± 11.90	78.52 ± 15.01	0.446
Urea (mmol/L)	6.10 ± 1.34	5.92 ± 1.98	0.795
eGFR (mL/min/1.73 m2)	88.14 ± 8.40	88.77 ± 11.15	0.873
Post-ablation (immediately)
haemolysis, n (%)	7 (58.3%)	1 (7.7%)	0.011
fHb (g/L)	0.22 ± 0.12	0.09 ± 0.04	0.006
Haptoglobin (g/L)	0.73 ± 0.29	0.91 ± 0.40	0.204
LDH (U/L)	220.33 ± 45.02	207.98 ± 40.52	0.480
IBIL (µmol/L)	12.51 ± 8.16	13.76 ± 6.69	0.680
Creatinine (µmol/L)	78.20 [62.42, 82.17]	68.90 [68.30, 92.20]	0.644
Urea (mmol/L)	5.47 [5.08, 6.38]	5.75 [5.16, 6.73]	0.605
eGFR (mL/min/1.73 m2)	88.60 ± 11.29	89.45 ± 10.90	0.850
Post-ablation (24 h)
fHb (g/L)	0.01 [0.01, 0.01]	0.01 [0.01, 0.01]	0.341
Haptoglobin (g/L)	0.38 [0.26, 0.56]	0.42 [0.34, 0.80]	0.135
LDH (U/L)	256.08 ± 33.64	244.97 ± 20.30	0.335
IBIL (µmol/L)	19.34 ± 9.04	20.82 ± 7.31	0.660
RBC (*1012/L)	4.38 ± 0.50	4.41 ± 0.58	0.896
HB (g/L)	132.67 ± 18.64	130.23 ± 21.92	0.767
MCV (fL)	89.55 [87.83, 91.03]	87.40 [85.90, 89.40]	0.277
MCH (pg)	30.21 ± 1.71	29.34 ± 1.75	0.221
MCHC (g/L)	339.92 ± 6.91	336.54 ± 8.66	0.291
Creatinine (µmol/L)	78.91 ± 12.85	81.28 ± 16.00	0.686
Urea (mmol/L)	6.31 ± 1.59	6.98 ± 1.76	0.329
eGFR (mL/min/1.73 m2)	86.20 [79.47, 89.02]	83.79 [76.26, 97.48]	0.683
Change value after procedure
fHb increase (g/L)	0.20 ± 0.12	0.08 ± 0.04	0.005
Haptoglobin decrease (g/L)	0.88 ± 0.33	0.59 ± 0.43	0.067
LDH increase (U/L)	94.67 ± 14.79	58.82 ± 25.47	<0.001
IBIL increase (µmol/L)	10.90 ± 7.48	9.04 ± 3.79	0.451

fHb, free haemoglobin; LDH, lactate dehydrogenase; IBIL, indirect bilirubin; RBC, red blood cell; HB, haemoglobin; MCV, mean corpuscular volume; MCH, mean corpuscular haemoglobin; MCHC, mean corpuscular; eGFR, estimated glomerular filtration rate.

Figure 6.

Observations from clinical RCT (n  =  25). The blood sample was collected at baseline (T1), immediately after ablation (T2), and 24 h after ablation (T3). (A) Dynamic changes in free haemoglobin, haptoglobin and LDH also indicated that the level of haemolysis was attenuated by high-flow irrigation (P < 0.001). (B) Voltage remapping revealed an extensive low-voltage area within the ablated region (bilateral PV and roofline) using 30 mL/min irrigation. (C) Two representative images of SCL after the procedure. Multiple SCLs with a diameter over 3 mm were observed in one patient with low-flow irrigation (left). One single minimal SCL was detected in one patient with high-flow irrigation (right). LDH, Lactate dehydrogenase; IBIL, Indirect bilirubin; PV, pulmonary veins; significance level: ** P < 0.01.

Efficacy of ablation

Acute procedural success was achieved in all patients, including 100% PV isolation (n = 25) and roofline block or posterior wall isolation (n = 16), which was confirmed by meticulous remapping (Figure 6B). After a median of 113.12 ± 20.56 days of follow-up (four patients were within blanking period), early recurrence of atrial tachyarrhythmia was recorded in three patients (two from the low-irrigation group and one from the high-irrigation group). The patient with newly onset peri-mitral atrial tachycardia underwent a repeat ablation and the bilateral PV and posterior wall remained isolated in the second procedure. No other repeat ablations were performed within the study period.

Silent cerebral lesions

Cerebral MRI was performed in 24 patients (96%) within 24 h post-ablation, revealing a total of 30 SCLs in 10 patients (Figure 6C). The incidence of SCLs was significantly lower in the high-speed-irrigation group compared to the low-speed-irrigation group [2/12(16.7%) vs. 8/12(66.7%), P = 0.036] (Table 5). Although the incidence of lesions >3 mm was lower in the high-speed irrigation group, the difference was not statistically significant [2/12(16.7%) vs. 7/12(58.3%), P = 0.089]. Notably, one lesion >10 mm was observed in the 4 mL/min group, but no such lesion was found in the high-speed group. Among patients with SCLs, six exhibited multiple lesions, of whom only one belonged to the 30 mL/min irrigation group (8.3% vs. 41.7%, P = 0.155). The maximum lesion diameters were comparable between the two groups [4.45 (3.78, 5.12) vs. 5.05 (4.27, 6.78) mm, P = 0.711]. No patient experienced a symptomatic embolism event after ablation. Bubble formation within the LA chamber and ascending aorta was visualized by ICE catheter using 4 mL/min and 30 mL/min irrigation, as shown in Supplementary material online, Video S9–S12.

Table 5.

Cerebral MRI finding

	4 mL/min(n = 12)	30 mL/min(n = 12)	P
SCL, n (%)	8 (66.7%)	2 (16.7%)	0.036
Multiple SCL, n (%)	5 (41.7%)	1 (8.3%)	0.155
Lesion size >3 mm, n (%)	7 (58.3%)	2 (16.7%)	0.089
Lesion size >10 mm, n (%)	1 (8.3%)	0 (0.0%)	1.000
Number of SCLs, n (%)	2.00 [1.00, 3.25]	1.50 [1.25, 1.75]	0.585
Maximum diameter, mm	5.05 [4.27, 6.78]	4.45 [3.78, 5.12]	0.711

SCL, silent cerebral lesion.

Discussion

This study investigated the effect of irrigation in PFA using a VLCC catheter, with the following major findings: (i) high-flow irrigation may significantly mitigate electrode heating during PFA, although these instances of electrode heating have only a minimal accumulative effect on deep tissue; (ii) small gaseous bubbles are common around the electrodes during pulse delivery especially under low-flow irrigation, while high-flow irrigation prevents bubble retention and aggregation; (iii) proper high-flow irrigation slightly increases the lesion depth as compared to low-flow irrigation; (iv) high-flow irrigation reduces the extent of haemolysis, and this protective effect persists even with increased PFA applications and poor catheter contact; (v) clinically, proper saline irrigation was associated with a lower incidence of immediate haemolysis and SCLs.

Thermal effect and lesion depth

In contrast to thermal ablation, lesion formation in PFA is not mainly driven by tissue heating^1,3; thus, irrigation does not seem to significantly affect ablation efficacy. However, electrode heating is non-negligible when using VLCC with minimal irrigation. In this study, we repeated the experiment as conducted by Sauer et al¹⁸ and Zito et al¹⁷ in potato and swine myocardium. Although peak electrode surface temperature reached nearly 70°C with 4 mL/min irrigation, it returned to <50°C in 1.68 ± 1.39 s, and cumulative core temperatures remained below 55°C following both irrigation regimens, making protein denaturation or oesophageal injury unlikely. Based on these observations, the increased lesion depth with 30 mL/min irrigation, compared to 4 mL/min irrigation (5.94 ± 0.29 mm vs. 5.36 ± 0.33 mm), is likely not due to tissue heating but rather from the altered dielectric properties at the electrode-tissue interface (blood vs. saline). Notably, the observed differences in thermal behaviours between 0–15 mL/min and 30 mL/min irrigation rates should be considered exploratory and not interpreted as evidence of an irrigation flow threshold. Proper high-flow irrigation might be an alternative approach to targeting thick-walled myocardium (e.g. ventricular substrates and mitral isthmus). Nonetheless, excessive irrigation did not further increase lesion depth (5.94 ± 0.29 mm vs. 5.48 ± 0.29 mm under 60 mL/min), as high velocity flow may induce catheter instability by perturbing catheter-tissue contact. It should also be noted that the thermal and lesion depth data—particularly from the potato model—should be considered qualitative rather than quantitative, given the substantial differences in thermal conductivity across tissues.

Effect of haemolysis

Haemolysis is a unique collateral injury in intracardiac PFA.^11,13,22,23 Although the erythrocyte destruction threshold via electroporation far exceeds that of cardiomyocytes,²⁴ the non-uniform electric field distribution around PFA electrodes frequently generates localized field intensity exceeding 1500 V/cm.¹⁹ Both in silico and clinical evidence demonstrate that each PFA energy delivery induces a quantifiable degree of erythrocyte destruction, with its severity correlated with electrode design,^21,25 catheter-tissue contact,²⁶ and PFA applications.¹⁰ We hypothesized that during PFA delivery, proper saline irrigation may transiently ‘surround’ the electrode, thereby attenuating erythrocyte lysis. As expected, our serial, non-clinical experiments revealed that high-flow irrigation consistently reduced all immediate biomarkers, including fHb, LDH, and haptoglobin. Although the immediate levels of haptoglobin and LDH did not reach statistical difference, the clinical study showed that high-flow irrigation effectively mitigated the immediate level of fHb and temporal changes in haptoglobin, LDH, and fHB. The observed difference in the effect of irrigation between non-clinical and clinical ablation might be attributed to the highly controlled condition (optimal catheter contact and limited PFA applications) and the small sample size. Notably, the protective effect of irrigation still exists with increasing PFA applications and in poor contact conditions, as indicated by the blood pool experiment. It could be translated into a common clinical scenario: extensive ablation beyond PVI with excessive energy deliveries and a higher risk of haemolysis.

It should be noted that, irrigation reduces—but does not eliminate—haemolysis, as informed by all groups exhibiting elevated biomarkers of haemolysis compared with baseline. Although most elevations were transient (resolving within 24 h), the clinical implications of such haemolysis on end-organ function (e.g. renal injury) remain unclear due to its low incidence. As PFA continues to advance as an innovative ablation modality, a comprehensive safety evaluation remains essential. Additionally, because the study was conducted exclusively using the VLCC catheter, further investigation is warranted to determine whether these findings can be extrapolated to other catheters.

Bubble formation and silent cerebral events

Gaseous microbubble formation during PFA primarily arises from direct electrochemical reactions with blood under high local current densities, with additional contributions from localized medium boiling.²⁷ Although Joule heating is relieved under high-flow irrigation, the release of dissolved gases from blood could also contribute to microbubble formation. To prove this assumption, both a high-speed camera and a bubble detector were used in the current study. The high-speed camera revealed that the number of micro and large gaseous bubbles generated is significantly reduced by increasing irrigation flow under identical PFA settings. Furthermore, low-flow irrigation resulted in microbubble retention and aggregation around electrodes, whereas high-flow irrigation rapidly dispersed these microbubbles into the circulatory system. Qualitative analyses demonstrated that no/low-flow irrigation occasionally led to potentially clinically significant bubble detection after multiple PFA applications, whereas moderate-to-high flow irrigation prevented such instances. As reported in the previous PFA cohort, the incidence of SCLs was reported from 3.3% to 85.7%.^7,9,28–37 However, it remains unclear how these microbubbles might translate into air embolism and SCL. Of note, microbubbles are also common in irrigated radiofrequency ablation, where abundant bubble showers are visualized by ICE during energy delivery, although they rarely result in embolic events. Based on Doppler ultrasound data, the lifetime of these bubbles in the circulatory system depends on their size, with those larger than 0.1 mm persisting for 3 min in the bloodstream.³⁸ The elimination of visible large bubbles by high-flow irrigation may be helpful to reduce the relevant embolic risk.

In addition to air embolism, thermal effects may also play a role in causing these SCL events. As demonstrated by our thermal experiment, single low-irrigation ablations produce visible lesions on the cardiac surface with an instant peak temperature reaching 65.30 ± 4.97°C. This significant Joule heating phenomenon may result in endothelial damage and tissue charring. Not surprisingly, the incidence of SCLs in the low-speed-irrigation group was as high as 66.7%, despite the strict anticoagulant and manipulation protocol. Similar results were reported by a Japanese cohort comprising seven patients undergoing VLCC ablation.²⁸ Although the irrigation speed was not explicitly stated in that study, it is presumed to be 4 mL/min, as was the initial recommendation by the manufacturer. In contrast, in the current study, SCL incidence reduced to only 16.7% with high-speed irrigation. High-speed irrigation appeared to reduce the incidence of SCLs in our clinical ablation, which led us to prematurely terminate clinical enrolment. In this context, irrigation seems essential to provide sufficient passive cooling during VLCC PFA.

Limitation

This study has several limitations. First, our findings are specific to the VLCC PFA system. Given the significant heterogeneity in catheter designs and waveform parameters across different platforms, these irrigation results may not be directly generalizable to other PFA systems. Second, regarding the clinical trial, the sample size was calculated based on the anticipated incidence of haemolysis rather than SCLs. Furthermore, the early termination of the study resulted in a smaller sample size. Consequently, the study was underpowered, and the observed differences in SCL incidence should be interpreted with caution and considered hypothesis-generating. Third, a true non-irrigated control arm was absent in both in vivo and clinical procedures due to manufacturer recommendations and ethical concerns. While the low-flow (4 mL/min) group served as a proxy, a direct comparison between irrigation and non-irrigation remains unassessed. Fourth, although catheter-tissue contact is a critical determinant of energy delivery and bubble dynamics, this variable was not systematically quantified in our in vitro model, limiting mechanistic insights into the interplay between contact force and microbubble formation. Fifth, despite restricting high-flow irrigation to the duration of energy delivery (yielding an average total volume <300 mL), potential haemodilution may still influence laboratory measurements. From a clinical perspective, although the volume load was manageable, caution is still advisable for patients with limited fluid tolerance, such as those with decompensated heart failure. Sixth, clinical efficacy was defined by acute procedural success and remapping; long-term electrophysiological follow-up is required to determine whether the increased lesion depth associated with high-flow irrigation translates to durable isolation. Finally, ablation was confined to pulmonary vein isolation and the posterior wall. Because the risks of haemolysis and SCLs are correlated with the total pulse burden, the protective impact of irrigation might be underestimated compared to more extensive procedures requiring higher numbers of PFA applications.

Conclusion

Proper irrigation during PFA with the VLCC catheter is associated with lower electrode and tissue temperatures, reduced bubble formation, and a lower incidence of haemolysis and silent cerebral lesions. These findings align with the recent adjustment of the recommended irrigation flow from 4 mL/min to 30 mL/min and provide valuable mechanistic insight.

Supplementary material

Supplementary material is available at Europace online.

Funding

This study was supported by the National Science Foundation of China (82400378, 82151306), the Beijing Physician Scientist Training Project and the Cultivation Program for the Clinical-Basic Collaborative Platform of Capital Medical University (JLPYPT2025006).

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.