PFA for Atrial Arrhythmias

Definition

Pulsed field ablation for atrial arrhythmias is the delivery of high-voltage, short-duration electrical pulses to atrial myocardium to achieve irreversible electroporation — in practice, durable pulmonary vein isolation for atrial fibrillation, with extensions to the posterior wall, superior vena cava, mitral isthmus, left atrial appendage and cavotricuspid isthmus. It is the first ablation energy that does not depend on heating or freezing tissue, and the first in two decades to change the safety ceiling of the procedure: more than 500,000 pentaspline procedures have been performed worldwide without a single reported atrio-oesophageal fistula, against a thermal incidence of 0.025% and 66% mortality. (sources/pfa-af-ehra-2026, rating: very high) The ventricular application is a separate problem with a different evidence base and its own constraints — see concepts/Ventricular-PFA.

Key Concepts

Mechanism and what selectivity means

• The field is highest at the electrode and decays rapidly with distance; the poorly conductive lipid bilayer amplifies the transmembrane field (~0.5 V over 5 nm ≈ 10⁸ MV/m) until nanoscale pores form, admitting water and reactive oxygen species, oxidising lipids and damaging transmembrane proteins. (sources/pfa-af-ehra-2026, rating: very high)
• Reversible electroporation precedes cell death. Permeability rises across an area larger than the eventual lesion, rendering it unexcitable and collapsing bipolar electrogram amplitude, but resealing restores excitability within minutes and unpredictably. This is why "stunned" but conducting tissue, dormant conduction and recovering lesions occur.
• Selectivity is relative. Any cell can be electroporated at sufficient field strength; myocardium is simply more susceptible than oesophagus, phrenic nerve and vessel wall. The correct characterisation is that PFA is tissue-selective, contact-dependent, minimally thermal and voltage-dependent — not "cardioselective".
• It is not purely non-thermal. At the catheter–tissue interface current density is highest, tissue heats according to duty cycle, and the endothelium may be injured thermally. Blood conducts better than myocardium, so the field extends into the blood pool, electroporating blood cells — the mechanism of dose-dependent haemolysis.
• Healing differs from thermal ablation. Electroporation is a membrane phenomenon and leaves the extracellular matrix intact; transient microvascular constriction and oedema occur, but preserved microcirculation allows faster lesion maturation.

(Table: sources/pfa-af-ehra-2026)

• The depth ceiling defines the boundary with ventricular work: contemporary platforms create atrial lesions easily but reach only 6–8 mm in thicker tissue irrespective of waveform or catheter design. That is sufficient for atrium and not for the left ventricular wall.

Device classification and platforms

The consensus proposal classifies catheters by the largest dimension in contact with tissue, replacing "single-shot", "wide-area focal" and "lattice-tip":

Functionally the systems split three ways: circumferential or large-area devices for rapid PVI (simplest, best safety data); large-footprint catheters integrated with 3D mapping (adaptable to linear and ventricular lesions, but mapping-dependent); and contact-force RF catheters repurposed to deliver PFA (versatile and familiar, but point-by-point). (sources/pfa-af-ehra-2026)

Adagio, Lotos, nsPFA, Optishot and Arga systems are in development; no complete list is possible because devices change continuously, and waveform parameters are rarely disclosed, which the writing committee names as a barrier to comparison. (sources/pfa-af-ehra-2026)

Energy delivery and dosing

• The dose is set by amplitude (typically 500–3000 V), pulse duration (ns–µs), pulse number, repetition frequency and waveform; these determine the cellular effect and the lesion's size and shape.
• Bipolar delivery and biphasic waveforms reduce charge build-up and skeletal muscle stimulation; monopolar delivery and monophasic waveforms create deeper lesions.
• Irrigation alters thermal management, not the primary mechanism.
• ECG gating has been abandoned in most atrial platforms, because biphasic pulses proved minimally arrhythmogenic in large studies.
• Dosing is manufacturer-defined as a minimum number of applications per vein, and the evidence behind those minima is weak: clinical studies have not shown a clear association between application number and long-term durability. DOPPIO (NCT07021313) is testing escalation.

Dual-energy systems

A single catheter and generator switching between RF and high-voltage PF trains under unified power management, with impedance, temperature and voltage feedback. The rationale is complementarity — PFA for thin-walled structures, RF for thicker myocardium and where coronary spasm is a concern, and combined PF+RF on one trajectory ("lesion stacking") for greater depth. Named systems are the Affera Sphere-9 and the ThermoCool SmartTouch SF Dual Energy with TRUPULSE. Comparative outcome data against single-energy PFA are awaited. (sources/pfa-af-ehra-2026)

Anaesthesia

• Early monophasic systems required general anaesthesia for skeletal muscle capture; biphasic waveforms allow deep sedation.
• >80% of MANIFEST-PF procedures (>17,000 patients) were performed under deep sedation, and in EU-PORIA (1,233 patients, 80% sedation vs 20% GA) sedation was associated with shorter skin-to-skin and fluoroscopy times, identical serious adverse events and essentially identical 1-year freedom from AF/AT (~74%).
• No randomised evidence favours GA. GA suits poor sedation tolerance, sleep apnoea or obesity, and provides immobility early in the learning curve. Continuous capnography is advised during sedation.
• Terminology is inconsistently reported; recording agents and whether neuromuscular blockade was used would improve comparability.

Imaging, mapping and ICE

• Pre-procedural CT/MRI characterises PV anatomy, can be fused with mapping or fluoroscopy and may guide device sizing, though the latter lacks evidence; it is most useful early in the learning curve and in complex anatomy.
• Fluoroscopy-only workflows shorten procedure time without affecting safety or outcomes, at the cost of radiation dependence; contact is inferred from spline indentation or shape change and can be optimised by pacing from the catheter and adjusting to atrial capture.
• 3D mapping ranges from simplified impedance display to full integration with lesion tagging, and is most valuable in redo procedures and atypical anatomy — it is what makes lesion overlap assessable, the practical determinant of durable lesion sets.
• ICE is not inferior to TEE for excluding appendage thrombus, enables safe puncture in lipomatous septal hypertrophy, septal aneurysm or prior repair, reduces radiation and procedure time, confirms ostial position and contact, and in several reports improves PVI durability with fewer reconnections. It is invasive and carries additional access risk.

Device interactions

• CIEDs: large series show no significant interference, lead dislodgement or malfunction; avoid ablating immediately adjacent to leads and interrogate before and after. Reported exceptions: permanent CRT-D failure and transient software reset with apparent battery depletion of a leadless atrial pacemaker during monopolar PFA in presumed contact with an SVC coil, and PFA-induced VF in a dual-coil ICD patient from likely unintended capture via electromagnetic induction near the RV lead.
• LAA occlusion: plug-type devices and surgical clips appear compatible; lobe-and-disc occluders interfere — one-third of such patients had aborted left-sided applications because the protruding disc distorted the field and triggered the system's safety interruption. Perform PFA before implantation where possible.
• Other implants: ASD/PFO occluders, AtriClip and prosthetic valves have been treated safely, with arcing, electromagnetic interference and difficulty delivering energy near metallic valve components as the residual risks.

Step-by-step

1. Access: ultrasound-guided femoral venous puncture; uninterrupted oral anticoagulation where possible; ACT ≥300–350 s (≥350 s before left atrial manipulation for some systems).
2. Guidewire: a standard J-tip suffices for over-the-wire devices; extra-stiff wires add support and perforation risk; straight-tip wires have been associated with asymptomatic bronchial bleeding.
3. Transseptal puncture: posterior is standard, anterior can ease right inferior PV access when the LA is not enlarged; a single over-the-needle puncture with the PFA sheath ("zero-exchange") has been described in 100 patients.
4. Air embolism prevention: meticulous de-airing, continuous saline flush, and zero-exchange designs that also shorten fluoroscopy and LA dwell time. Negative LA pressure under sedation may favour air ingress.
5. Vagal response: 30–70% incidence, mostly at the left superior PV, usually self-limited; atropine 1.0 mg, rarely temporary pacing. Prophylactic parasympatholytics are contraindicated in glaucoma and benign prostatic hyperplasia. RSPV-first ablation reduced vagal reactions from 78% to 13%.
6. Contact: coaxial antral alignment, never deep in the vein; assess by tactile feedback, fluoroscopic indentation, pacing capture, ICE and impedance-based mapping markers (good contact gave 100% transmurality in vivo). PFA is less forgiving of intermittent contact than RF, because ultra-short delivery has no thermal latency and proximity must exist at the instant of delivery.
7. Projections: LAO for left veins, RAO for right veins; biplane may help.
8. Interruption troubleshooting: automatic cut-off occurs with unsafe geometry, abnormal impedance or metallic interaction — touching splines, loop diameter below the validated range, a wedged lattice tip (temperature overshoot) or balloon malposition. Arcing is rare and requires electrode–electrode or electrode–metal contact.
9. Endpoint: near-field/far-field discrimination is harder after PFA and varies with electrode size and spacing; durable entrance and exit block remain the universal endpoint, adenosine has low yield, and a standard waiting time adds little.
10. Post-procedural care: uninterrupted anticoagulation ≥2 months, then by CHA₂DS₂-VA; most centres have abandoned routine proton-pump inhibition; same-day discharge is supported in selected patients by registries, an admIRE sub-analysis and an EHRA meta-analysis, and is inappropriate after complications.

(All: sources/pfa-af-ehra-2026)

Training and the learning curve

• Procedures should be led by physicians with recognised EP certification (e.g. EHRA EP certification), in centres providing structured team education and registry participation.
• The pathway is didactic biophysics and device specifics → simulators or animal models → supervised practice with a proctor, which prevents unsafe extrapolation from RF or cryoballoon experience.
• Acute PVI is reproducible after few procedures and junior and senior operators obtain similar registry outcomes, but the real curve continues through ablation beyond the veins, redo procedures and non-PV arrhythmias. The cryoballoon precedent is the warning: a simple workflow does not compensate for poor case selection, incomplete anatomy or abandoned endpoints.

Clinical Evidence

The five randomised trials

• Non-inferiority was met in every trial; superiority only in SINGLE SHOT CHAMPION (P=0.046, borderline). AF burden, quality of life and healthcare utilisation did not differ, and procedure times were shorter with PFA in all five. (sources/pfa-af-ehra-2026, rating: very high)

Trial detail worth retaining

• ADVENT (first RCT): non-inferiority met for freedom from treatment failure (73.3% vs 71.3%; posterior probability >0.999) and for safety (2.1% vs 1.5%); PFA superior for PV preservation (−0.9% vs −12.0% cross-sectional area change; RF subgroup −19.5%, cryo −3.3%); shorter procedure (105.8 vs 123.1 min) but longer fluoroscopy (21.1 vs 13.9 min); PVI success 99.6% vs 99.8%; one procedure-related death from catheter manipulation; 3/33 (9.1%) asymptomatic MRI cerebral lesions vs 0/37 in a small substudy. (sources/pfa-advent-nejm-2023, rating: very high)
• SINGLE SHOT CHAMPION: with continuous implantable cardiac monitor monitoring, PFA met non-inferiority and formal superiority (37.1% vs 50.7% recurrence days 91–365; −13.6 pp; 95% CI −26.9 to −0.3); recurrence was also lower during the blanking period (−20.0 pp), and day-1 troponin was higher with PFA (1920 vs 1114 ng/L), supporting a broader ablation zone; repeat ablation was numerically more frequent (16 vs 10; rate ratio 1.60; NS). (sources/pfa-cryo-singleshotchampion-nejm-2025, rating: high)
• AVANT GUARD (first-line persistent AF): PFA (PVI + posterior wall) vs antiarrhythmic drugs in 310 previously untreated patients: treatment success 56% vs 30% at 12 months (HR 0.46; 95% CI 0.33–0.65); safety endpoint met (5.1% serious adverse events vs a 12% performance goal) but 6 procedure-related strokes (2%) prompted a pause and protocol changes, after which there were none. (sources/PFA-LPeAF-AVANT-GUARD-NEJM-2026, rating: very high)

Registries and non-randomised comparisons

• Aggregated 1-year freedom from atrial tachyarrhythmia across pivotal and comparative studies is 55–80% (paroxysmal 66–82%, persistent 55–71%), with no significant difference versus thermal energy (RR 0.85; 95% CI 0.64–1.14; P=0.29). (sources/pfa-af-ehra-2026, rating: very high)

• PULSED AF and PersAFOne confirmed safety and efficacy in paroxysmal and persistent AF; a European all-comer registry showed safe and effective PVI plus extra-PV lesions. (sources/ca-af-ehj-2024, rating: high)

Persistent AF — the weakest evidence

• Early persistent AF with minimal remodelling: PVI alone is appropriate. More advanced disease: PVI-only remains acceptable but yields lower success and may need additional lesions.
• ADVANTAGE AF (pentaspline PVI + posterior wall, continuous monitoring in phase 2) reported 1-year effectiveness 65–75% with low adverse events; SPHERE Per-AF showed non-inferior efficacy and safety with greater efficiency than RF. Whether additional lesion sets beyond PVI add value is being tested in PIFPAF-PFA (NCT05986526).

Redo procedures

• PVI durability after PFA ranges from very high to equivalent to thermal ablation; reconnection clusters at the LSPV and RIPV, with differences between systems. Posterior wall isolation with PFA shows high durability in early reports.

Blanking period

• Thermal blanking has been shortened from 3 months to 8 weeks; PFA's inflammatory profile appears less pronounced. Recurrence during the blanking period was far lower with PFA than cryoballoon (38.1% vs 58.1%), and PULSED AF showed early recurrence predicts late recurrence — so a 1-month blanking period may be reasonable after PFA, and the use of a blanking period in PFA trials should be reconsidered.

Applications Beyond Pulmonary Vein Isolation

(All: sources/pfa-af-ehra-2026, rating: very high)

• The pattern is consistent: acute success approaches 100% at every non-PV target, and durability is the limiting problem. Linear and appendage lesions behave less well than the vein encircling lesions the technology was designed for.

Safety

(Table: sources/pfa-af-ehra-2026, rating: very high)

Platform-specific and delayed concerns

• Haemolysis is dose- and system-related; mindful application numbers and pre-ablation hydration are advised, especially in chronic kidney disease.
• Coronary vasospasm begins within seconds to minutes, but delayed onset is described — relevant to same-day discharge. Nitroglycerine prevents or reverses it (1–2 mg intracoronary for treatment; 1 mg intracoronary or 1–2 mg intravenous for prophylaxis); the long-term coronary effect is unknown, and mechanisms (endothelial or smooth-muscle electroporation, autonomic effects, circulating vasoactive factors) are unresolved.
• Conduction system: transient AV block or sinus node dysfunction is rare after atrial flutter or SVC ablation; the full effect of PFA on the conduction system is unexplored.
• Left atrial function: acute stunning with reduced compliance after PVI and after more extensive ablation, recovering in the long term — though persistent contractile impairment at 3 months has been described, whereas RF with extensive LA ablation causes a persistent compliance decline.
• Autonomic effects: vagal responses in 30–70% acutely; chronically, unlike RF or cryo-PVI, PFA-PVI produces no post-procedural heart-rate rise and no increase in neuronal injury markers (e.g. S100B) — durable neural injury appears minimal.
• Silent cerebral events should not be attributed to electroporation alone: air embolism during exchanges, sheath or electrode thrombus, transient hypotension and anticoagulation management all contribute, and risk varies between platforms. Post-market evaluation of a variable-loop catheter led to updated instructions after a higher early stroke/TIA rate.
• Rare delayed malignant arrhythmias and sudden death after ablation beyond the PVI have been reported; causality is unproven but vigilance is advised.

Contradictions / Open Questions

• Equivalence is established; superiority is not. Four of five trials met only non-inferiority, and the single superiority result was borderline in a 210-patient trial using continuous monitoring. (sources/pfa-af-ehra-2026, rating: very high)
• "Cardioselective" was wrong. Early claims have given way to tissue-selective, contact-dependent and minimally thermal; phrenic capture, coronary spasm, haemolysis and conduction-system injury are all recognised field effects.
• The zero-fistula claim has no denominator. Absence in >500,000 voluntary registry and survey procedures cannot establish a true rate of zero, and the committee says so.
• No validated lesion marker exists. Contact force, impedance drop and temperature predict RF lesions; no PFA variable predicts irreversible transmurality, so durability cannot be confirmed at delivery.
• Dosing minima are arbitrary. Application-number thresholds are manufacturer-defined without dose-response evidence; DOPPIO is the first direct test.
• The 1-month blanking period is an inference. It rests on one trial's early-recurrence data and on the expectation that early PFA recurrence is benign — the opposite of what PULSED AF found.
• "Non-inferior" means different things across trials. Endpoint definitions, monitoring and comparators differ between the five trials, so the 62.9–77.2% range of freedom from AF is not one measurement.
• Non-PV durability may not be fixable by energy choice. Acute success is near universal, yet LAA isolation reconnects in 95% and mitral isthmus block persists in 5.5%; adjunctive posterior wall ablation did not improve 12-month outcomes in MANIFEST-PF.
• Platform heterogeneity undermines the class claim. More than a dozen systems with undisclosed waveforms and no cross-vendor standardisation mean that "PFA" trials are largely platform trials, and results may not transfer. (sources/pfa-af-ehra-2026)

Connections

• Related to concepts/Ventricular-PFA — the ventricular application, with its separate depth constraints
• Related to concepts/Catheter-Ablation-AF — where PFA sits among ablation energies and strategies
• Related to concepts/AF-Burden — the rhythm endpoint these trials measure
• Related to entities/Atrial-Fibrillation — the target arrhythmia
• Related to concepts/Epicardial-VT-Ablation — epicardial delivery and its distinct coronary hazard
• Related to concepts/Late-Gadolinium-Enhancement — imaging lesion extent after ablation
• Related to entities/ICD — device interaction during delivery
• Related to entities/LAAO — lobe-and-disc occluders distort the field and abort left-sided applications, so PFA should precede implantation

Sources

• sources/pfa-af-ehra-2026
• sources/pfa-advent-nejm-2023
• sources/pfa-cryo-singleshotchampion-nejm-2025
• sources/PFA-LPeAF-AVANT-GUARD-NEJM-2026
• sources/ca-af-ehj-2024