A Review of Potential Complications of Pulsed Field Ablation for AF: Beneath the Surface

Pulsed field ablation (PFA) has been rapidly adopted for catheter ablation of AF globally. With supportive clinical trial data demonstrating non-inferiority compared with traditional thermal ablation modalities as well as significantly improved procedural workflow, more than 500,000 cases have been performed and it is likely to capture a significant market share moving forward.^1–3 Although PFA has often been described as ‘non-thermal’ based on increased cell membrane permeability from electroporation, it still has thermal consequences, but is more selective towards proximate cardiac tissue depending on how the pulsed electric field is delivered (i.e. amount of energy delivered, the pulse duration and number of applications).⁴ Such factors also influence the degree of reversible electroporation, which in turn affects lesion durability.⁵ Reassuringly, there have not been reports of injury to certain surrounding non-cardiac structures such as the oesophagus.⁶ Nevertheless, PFA is not without risk – the safety profile is simply different (Table 1 and Figure 1 ). Our understanding of its effects in and around the heart is still being better understood as large post-approval registry data accumulate and less frequent complications come to light beyond initial small investigational device exemption studies.

In this review, we aim to highlight the potential complications associated with PFA for AF and steps towards identification, management and prevention for clinical cardiac electrophysiologists.

Pericardial Tamponade

Pericardial tamponade is an uncommon, yet potentially fatal, complication of catheter ablation for AF that persists with PFA. Early clinical studies using FARAPULSE (Boston Scientific) reported 1.6% (2/121) risk of pericardial effusion/tamponade, while the inspIRE study using a variable loop-catheter (VARIPULSE, Johnson & Johnson MedTech) had no pericardial tamponade (126 patients).^7,8 These estimates are limited by small sample size and being in a clinical trial context. Large registry data from MANIFEST-17K and FRANCE-PFA report the rate of PFA-associated tamponade to be 0.36% and 0.40%, respectively.^6,9 Conversely, data from the EU-PORIA registry, consisting of seven European centres with 1,233 patients, showed pericardial tamponade in 1.14% of cases.¹⁰ This places pericardial tamponade as the most common serious adverse event, occurring more frequently than major vascular access complications.^6,9 In comparison to thermal ablation, a meta-analysis of randomised and non-randomised studies with 18 studies and 4,998 patients undergoing pulmonary vein isolation (PVI) ablation found significantly higher rates of cardiac tamponade with PFA compared with thermal ablation (OR 2.98; 95% CI [1.27–7.00]).¹¹ There is also a possible signal of increased risk of pericardial tamponade in women compared with men undergoing PFA.¹²

The mechanism of cardiac tamponade from AF ablation typically results from myocardial perforation or vascular injury during catheter manipulation.¹³ This is likely multifactorial, including patient factors, catheter design, operator experience and therapeutic anticoagulation. Patient factors include small left atrial size and increased wall fragility. Catheter design factors include sheath size and steerability, especially for transseptal crossing and use of over-the-wire design. Over-the-wire catheters, such as FARAPULSE, PulseSelect (Medtronic) and Sphere-360 (Medtronic), require the catheter and sheath to be placed over a guidewire for navigation into pulmonary veins (PVs). When wiring in the left atrium, there remains a risk of inadvertent cardiac injury, such as to the left atrial appendage, or snaring within the PV sleeve even with J-tip guidewires.¹¹ Additionally, the risk of myocardial injury can be influenced by catheter tip design. The various catheter tip designs each create distinct lesion profiles and lesion depth.¹⁴ Design parameters include catheter shape (e.g. pentaspline, circular, loop configurations) and footprint size. Larger and single-shot footprints may have greater propensity for mechanical myocardial damage.¹⁵ Inadvertent myocardial injury can also occur from catheter manipulation and some systems have integrated contact force sensing function in combination with 3D electroanatomic mapping (e.g. VARIPULSE and Volt [Abbott Cardiovascular]) to mitigate excessive force application.^16,17 Some systems use a tight locking function between the dilator and the sheath, where dilator unlocking can lead to unintentional sheath progression.¹¹ Ongoing improvement in catheter system design and operator experience will likely reduce risk of inadvertent cardiac injury. Finally, ablation protocols often aim for higher activated clotting time (ACT) targets between 300–400 seconds for stroke prophylaxis, which must be balanced with increased bleeding risk and potentially worse tamponade.^6,11,18

The treatment of pericardial tamponade involves pericardiocentesis, blood aspiration, administration of protamine and cardiac surgery where appropriate. However, there are currently no consensus guidelines on the timing and need for escalation on these treatments. Centres without additional cardiac surgery support may have lower thresholds to initiate protamine infusions for additional haemostasis, with dosing often guided by pre-procedural ACT levels or the dose of heparin used.¹⁹ Clotting factors are not commonly administered due to risk of pericardial clotting.¹⁹ Fresh aspirated blood is typically reused for autotransfusion as a quick, safe and cost-effective method of providing haemodynamic support. For centres with cardiac surgery support, timing of referrals for repair are dependent on situational factors (e.g. degree of blood loss, stability of patient).¹⁹ Regardless, it is important that PFA centres have cardiac surgery backup, whether that be permanently on site or at an accessible affiliated centre.

In terms of tamponade prevention, pivotal time points of potential myocardial injury are during transseptal puncture and engagement of pulmonary vein with guidewire. Operator experience with transseptal technique, whether that be by fluoroscopic guidance, transoesophageal guidance or intracardiac echocardiography, is crucial to reduce risk of complications. Successful left atrial access is verified through introduction of the guidewire into the left atrium and usually the left-sided PVs.¹⁹ During guidewire movement for PV engagement, caution must be emphasised against pushing or pulling against resistance. Additionally, when moving the catheter within the left atrium (e.g. from left-sided to right-sided PVs), retraction of the guidewire into the catheter lumen can reduce risk of inadvertent injury from the guidewire tip.

Coronary Artery Spasm

Coronary artery spasm involves transient constriction of a coronary artery and is a clinical concern due to potential myocardial ischaemia. The MANIFEST-17K, FRANCE-PFA and EU-PORIA registries report incidences of between 0.08% and 0.14%.^6,9,10 Individuals with underlying coronary artery disease or underlying endothelial dysfunction may have increased risk of developing ischaemia with coronary spasm. The precise mechanisms behind PFA-related coronary artery spasm remain uncertain but multiple reasons have been proposed. First, the high-voltage energy delivery may stimulate intracellular calcium influx and trigger vascular smooth muscle constriction, which is further exacerbated by direct cell stunning from electroporation.²⁰ Additionally, PFA-induced cell injury can trigger the release of local vasoactive mediators and cytokines that promote vascular tone.²⁰

The risk of coronary artery spasm is related to the number of applications and more importantly, where the application is delivered anatomically within the heart.⁶ Off-label use of PFA, such as ablation at the cavotricuspid isthmus (CTI) and mitral isthmus, is associated with significantly greater risk. A single-centre retrospective analysis of coronary angiography before, during and after PFA applications reported that all cases of coronary artery spasm occurred after CTI ablation compared with zero cases with PV and left atrial posterior wall (PW) applications.²¹ Similar reports were found in a separate prospective study where several patients who underwent PFA ablation within the mitral isthmus line experienced spasm in the adjacent left circumflex artery.²² Nevertheless, cases of remote diffuse coronary spasm even in the context of PVI and PW application (without CTI or mitral isthmus line) have been described and operators must remain vigilant with regards to this potential risk even with standard lesion sets.^23,24

In terms of clinical presentation, coronary artery spasm tends to be asymptomatic but can present with ST-segment elevation on ECG, ventricular arrhythmias and/or cardiac arrest.²⁵ Of note, only two of 25 patients (8%) with coronary spasm had chest pain in MANIFEST-17K, although ECG changes were noted in 92%.⁶ In most cases, this phenomenon does not typically cause clinically significant myocardial ischaemia as there was no difference in levels of cardiac biomarkers between coronary artery spasm subjects with and without ST-segment changes.¹⁵ This is likely due to coronary artery spasm having rapid and effective responses to nitroglycerin (GTN) therapy. Instead, the post-procedural troponin elevation may be a separate mechanism related to direct myocardial injury from pulsed energy.¹⁵ Therefore, the role of routine post-procedural cardiac biomarkers for monitoring coronary complications post-PFA is limited. Rather, typical signs of myocardial ischaemia that warrant immediate coronary assessment include persistent ischaemic ECG changes without vasodilator response, haemodynamic instability with other causes excluded (e.g. vagal response, cardiac tamponade or other bleeding) and ventricular arrhythmias.

The treatment and prevention of coronary artery spasm involves peri-procedural GTN and typically avoidance of PFA in certain anatomical locations such as the CTI and mitral isthmus. A small prospective study of patients undergoing focal PFA of the PVs, PW and CTI reported that 80% of patients (four out of five patients) without pre-procedure GTN developed moderate to severe coronary artery spasm of the right coronary artery on invasive angiography.²⁶ The ADVANTAGE AF Phase II trial implemented a prophylactic GTN protocol to reduce risk of coronary artery spasm during CTI ablation with PFA.²⁷ This regimen consisted of a 3 mg bolus 1 minute prior to the first PFA application with an additional 2 mg every 2 minutes until the set was completed.²⁷ However, clinically significant coronary artery spasms have still been reported despite prophylactic GTN at doses substantially higher than standard acute coronary syndrome care.²⁸ Our centre does not perform PFA in areas near coronary arteries. However, if choosing to do so, it is prudent to routinely administer high-dose prophylactic GTN to mitigate risk with pre-emptive awareness of associated systemic hypotension and appropriate vasopressor support, and observe a period of longer post-operative cardiac monitoring as delayed coronary spasm has been reported.²³

The long-term implications of PFA near coronary arteries remain uncertain. A pre-clinical study demonstrated histological evidence of intimal hyperplasia and tunica media fibrosis with surrounding leukocytic infiltration as early as a few days post-endocardial ablation.²⁹ Epicardial PFA lesions in swine also showed chronic coronary artery stenosis from similar microscopic changes.³⁰ These chronic effects are also seen in human coronary arteries associated with off-label uses in CTI and mitral isthmus lines. Mild arterial stenosis persisted at 3 months post-PFA in a prospective study of 21 patients using coronary angiography and optical coherence tomography assessment.³¹ Such longer-term findings may not be related to peri-procedural acute coronary spasm and requires further investigation. Nevertheless, there remains a risk of coronary artery injury despite prophylactic high-dose GTN protocols and due caution, if not avoidance, of certain higher-risk anatomical regions is advised until further long-term data and improved focal PFA technology and design are available.

Haemolysis and Acute Kidney Injury

Haemolysis describes erythrocyte destruction within a vascular structure, resulting in the release of free haemoglobin and activation of scavenging proteins in the plasma to neutralise its toxicity. This phenomenon is unique to PFA with an incidence of up to 94% biochemically and has not been previously recognised with conventional thermal ablation.³² This is due to rapid and intense electroporation with unintentional injury to red blood cells in the surrounding blood pool, especially when applied to areas of high blood flow. This effect is dose dependent, where detection of free haemoglobin can occur with as few as two PFA applications and increased haemolysis observed with more applications.^32,33 Other factors affecting the degree of haemolysis include pulse amplitude, catheter design, anatomical site of energy delivery and patient comorbidities.^32,33 Patients with baseline chronic renal impairment experience a greater burden of haemolysis due to impaired clearance of free haemoglobin filtered through the kidneys with resultant haemoglobinuria, cast formation and haem uptake in proximal tubular cells that can result in acute tubular necrosis.³⁴ This phenomenon may be subclinical, where the expected laboratory findings include elevated plasma free haemoglobin, lactate dehydrogenase, bilirubin, and reduced haptoglobin.³² Clinically, the only marker may be evidence of haemoglobinuria.

While PFA-associated haemolysis has not been shown to induce significant anaemia, haemoglobinuria and subsequent azotaemia with acute kidney injury (AKI) have been described. The FRANCE-PFA registry reported no cases of haemolysis with AKI, whereas MANIFEST-17K reported a rate of 0.04% with cases likely occurring in the context of pre-existing renal impairment, hypotension and dehydration.^6,9 The renal system is particularly susceptible to the toxic effects of haem compounds such as tubular obstruction, oxidative stress and direct nephrotoxicity because the kidneys are the predominant organs of haemoglobin clearance when the plasma scavenging compounds become saturated.^32,35

In rare instances, clinically significant AKI from PFA-induced haemolysis can be severe and require dialysis.^6,15,36 The risk can be mitigated with consideration of number of PFA applications and hydration protocols. Moderating the number/intensity of PFA applications and optimising electrode parameters appropriately for its desired target lesion reduces haemolysis risk.^32,33,37 Operators can also limit unnecessary energy delivery towards surrounding high-flow blood pools by applying greater contact with the catheter during application delivery.³⁸ This is likely to improve with integration of 3D electroanatomic mapping systems and contact-force sensing catheters moving forward. Furthermore, centres performing PFA should establish hydration protocols and limit contrast use where possible.

In a single-centre analysis, 75% of cases without post-ablation hydration demonstrated biochemical haemoglobinuria and significantly increased serum creatinine levels compared to the group that received a post-ablation hydration regime.³⁷ Another comparative analysis trialled a regimen of 1 l IV fluids pre-procedurally followed by another 1 l post-procedure if exceeding 100 PFA applications, compared with the alternative of pre-emptive 2 l IV fluids pre-procedurally plus another 500 ml if exceeding 100 PFA applications; post-procedural AKI did not occur in the latter group, but there was a rate of 9% of AKI in the former group.³⁹ These results suggest that significant volume replacement is likely required pre-procedurally, and sometimes post-procedurally, to clear potentially harmful free haemolytic metabolites generated during PFA. However, significant fluid administration must be balanced with a patient’s competing risk of fluid overload especially in the context of pre-existing chronic heart failure or renal failure. Short-term oral diuresis can be prescribed to mitigate iatrogenic overload if needed post-procedure.³⁹ Our practice in patients with normal left ventricular systolic function and normal renal function has been routine commencement of IV crystalloid solution pre-procedure, amounting to an approximate total of 2 l without routine placement of a Foley urinary catheter to help facilitate same-day discharge. However, this may be over-cautious in patients who receive limited PFA applications.

Stroke and Silent Cerebral Emboli

Stroke is one of the most feared complications of any intracardiac procedure. It may result from thrombus formation on catheters, air embolism, or dislodgement of pre-existing atrial thrombi in AF. Thus far, overt and symptomatic strokes following PFA are rare, where registries have reported rates of clinically apparent stroke between 0.10% and 0.41%.^6,9,10 These rates appear to be favourable compared with thermal ablation data.⁴⁰ Despite low incidence of clinical stroke, a more subtle and potentially underappreciated concern is the occurrence of silent cerebral emboli (SCE), which are small and typically asymptomatic embolic events only visible on diffusion-weighted MRI. Such sub-studies using PFA have demonstrated SCE rates from 9% to 12%, while an ADVENT sub-study found similar incidences of silent cerebral lesions between PFA and thermal ablation post PVI.^8,41,42 Meanwhile, a small, single-centre study reported half (8/16; 50%) of their PFA cohort developed SCE lesions and suggested that various catheter systems may alter the risk based on their inherent configurations.⁴³

The exact mechanism of stroke and SCE during PFA likely has multiple aetiologies. Gas embolism formed from microbubbles during pulse delivery, air entrainment during catheter exchanges and/or thrombus dislodgement during transseptal access are all potential contributing factors.⁴⁴ Another proposal is PFA-induced haemolysis, where electroporation of blood pools (particularly when electrodes have poor myocardial tissue contact) causes red blood cell destruction and release of free toxic haem molecules, which promote thrombo-inflammation and embolic potential.⁴⁵ This may explain the direct platelet activation induced by PFA, although this is likely a different biological mechanism compared with thrombosis induced by thermal ablation.⁴⁶ Additionally, given the more rapid workflow in PFA, the peri-procedural clotting activation may not reach target. A retrospective analysis suggested that PFA was associated with longer times to reach target ACT and fewer ACTs in therapeutic range compared with radiofrequency ablation. This was attributed to the shorter procedural time, fewer additional heparin units administered and variability in saline irrigation between different PFA catheters.⁴⁷

SCE can also be dependent on imaging timing and procedural factors. For instance, the inspIRE study had 39 of their subjects who completed PV isolation via PFA undergo pre- and post-procedural MRI screening, where four SCE lesions were detected in the first six subjects. With procedural modifications including longer pauses between applications, reducing catheter exchanges and enforcing a stricter ACT target, the subsequent 33 subjects had four cases (12%) with detected SCE.⁸ Thus far, there has been no reported evidence of significant neuropsychiatric morbidity from PFA-related SCE, with most lesions being asymptomatic and self-resolving in the intermediate term. However, the long-term sequelae of SCE remain unknown.^8,44 A prospective observational study of AF patients undergoing left atrial appendage occlusion showed a 39% incidence of procedure-related SCE which resulted in significant and irreversible neurocognitive decline up to 1-year follow-up.⁴⁸ Similar long-term imaging studies are necessary to ascertain this concern with PFA.

To mitigate stroke and SCE risk, peri-procedural anticoagulation is critical. Current PFA studies have been maintaining an ACT ≥300 seconds.^7,15,18,49 For patients on oral anticoagulation for AF, it is highly recommended to be on therapy for at least 3 weeks prior to the procedure, where it is generally safe to have uninterrupted therapy during the peri-procedural period, or at most withholding an anticoagulation dose during the morning of the procedure.⁵⁰ Additionally, judicious use of the number of PFA applications to reduce haemolysis risk and potential downstream thromboembolism is recommended. Other measures to reduce embolic burden are being actively explored in newer catheters, such as management of air entrainment and optimising contact-sensing systems.

Femoral Vascular Access Complications

Vascular access site complications are a potential adverse sequela of any interventional electrophysiology procedure. This is relevant for thermal catheter ablation as well as PFA for AF. Recent data suggest that despite an overall reduction in procedural complications that has been observed in more recent years for AF ablation, the rate of vascular complications remains largely unchanged.⁵¹ Major vascular access site complications from PFA, often defined as those requiring intervention (e.g. surgical or endovascular intervention) or blood transfusion support, remain relatively uncommon, with registry data reporting rates of 0.2% and 0.3%, respectively.^6,9 The rate of minor vascular access complications varies, and depends on definition, such as need for further manual compression or non-invasive intervention (e.g. FemoStop, Abbott Cardiovascular) or unplanned interruption of anticoagulation post-procedure. Regardless, with growing suggestions to aim for a higher ACT target (e.g. 350 seconds) and a strategy of continued anticoagulation during the perioperative period, vascular complications are potentially more likely to occur. Certain patient factors, such as female sex and concomitant antiplatelet therapy, could also contribute, alongside operator experience, routine ultrasonography guidance, catheter sheath size and method of femoral venous closure at completion of procedure.^52–54

Mitigation of vascular access site complications is critical for assessment of procedural success from both from a complication viewpoint but also patient satisfaction. Some centres practice same-day discharge and concerns for vascular complications can delay discharge and lead to unplanned hospitalisation. Practical consideration to reduce vascular access complications include appropriate patient selection, use of ultrasonography guidance for femoral venous puncture and vascular access closure device over figure-of-eight suture.^55,56 Prompt recognition of an impending haematoma is crucial in the recovery area, which requires nursing education and awareness. Additionally, patient education regarding restricted physical movements post-procedure can also help reduce complications and avoid interruption of therapeutic anticoagulation.

Vagal Responses

PFA is well-recognised to induce a vagal response during the procedure. This tends to occur when applied in proximity to the cardiac ganglionated plexuses within the antrum of the PVs. A separate intracardiac catheter can be inserted, often into the coronary sinus, to both guide fluoroscopic transseptal puncture as well as provide atrial pacing support, particularly with concurrent electroanatomic mapping. However, the catheter would otherwise need to be repositioned in the case of vagally mediated atrioventricular block. Prophylactic parasympatholytic drug administration (usually atropine) can pre-empt these potential vagal responses; however, this is associated with adverse drug events and higher risk of AF.⁵⁷

Given the sheath direction across the transseptal puncture, it is natural to commence ablation on the left-sided PVs first. However, randomised data have shown that vagal responses are observed less frequently when performing circumferential PVI with radiofrequency from the right-sided PVs first, and specifically the right anterior aspect.⁵⁸ It is thought that the right anterior ganglionic plexus has a critical role in regulating autonomic nervous information of other ganglionic plexi in the left atrium. This has subsequently been confirmed in a study concerning PFA, where vagal responses were observed less frequently when commencing ablation first at the right superior pulmonary vein compared with left superior vein.⁵⁹

Phrenic Nerve Injury

Transient phrenic nerve injury has been observed with PFA, although rarely persistent. Registry data have reported rates of transient phrenic nerve injury between 0.04% and 0.3%, with only one patient among these registries having persistent phrenic nerve injury.^6,9,10 Conversely, a prospective single-centre observational study used compound motor action potential monitoring to systematically evaluate phrenic nerve injury during right pulmonary vein ablation with PFA and reported an incidence of 40% phrenic nerve injury, of which the overall incidence of phrenic nerve injury at discharge was 24% (6/25).⁶⁰ Cautious interpretation of these data is warranted, given that there was no systematic evaluation of symptoms or quality of life, fluoroscopic assessment of phrenic nerve injury was based on an arbitrary change of more than 15% lower hemidiaphragm compared with the contralateral side, and most centres do not use compound motor action potential monitoring.

Conclusion

The advent of PFA has changed the treatment landscape for AF. More patients will be able to access better rhythm control through catheter ablation given the vast improvement in procedural efficiency and throughput. The safety data discussed in this review show our actively evolving understanding of this technology and highlight its different safety profile and unique complications compared with traditional thermal ablation techniques. We can anticipate that competitive industry iterations of catheter design and procedural workflow will lead to constant improvements. We must nevertheless remain vigilant regarding potential adverse sequelae of any new technology and ensure we first do not cause harm to our patients.