Aims Septal reduction is needed for hypertrophic obstructive cardiomyopathy (HOCM) patients with severe left ventricular outflow tract (LVOT) gradients and symptoms despite medication. Myectomy cannot be performed in all. Alcohol septal ablation cannot be performed in 5–15% due to technical difficulties. A method of delivering percutaneous tissue damage to the septum that is not reliant on coronary anatomy is desirable. To directly ablate the interventricular septum at the mitral valve (MV) systolic anterior motion (SAM)-septal contact point using radiofrequency (RF) energy guided by CARTOSound.
Methods and results Five patients underwent RF ablation (RFA); we describe follow-up at 6 months in four patients. Intracardiac echocardiography (ICE) images are merged with CARTO to create a shell of the cardiac chambers. The SAM-septal contact area is marked from ICE images and mapped on to the CARTO shell; this becomes the target for RF delivery. Conduction tissue is mapped and avoided where possible. Twenty-eight to 42 min of RF energy was delivered to the target area using retrograde aortic access and SmartTouch catheters. Resting LVOT gradient improved from 64.2 (±50.6) to 12.3 (±2.5) mmHg. Valsalva/exercise-induced gradient reduced from 93.5 (±30.9) to 23.3 (±8.3) mmHg. Three patients improved New York Heart Association status from III to II, one patient improved from class III to I. Exercise time on bicycle ergometer increased from 612 to 730 s. Cardiac magnetic resonance shows late gadolinium enhancement up to 8 mm depth at LV target myocardium. One patient died following a significant retroperitoneal haemorrhage.
Conclusion Radiofrequency ablation using CARTOSound® guidance is accurate and effective in treating LVOT gradients in HOCM in this preliminary group of patients.
Hypertrophic obstructive cardiomyopathy
An advancement in understanding of the mechanism of LVOT gradients in HOCM; anatomical mapping of the contact area of the anterior MV leaflet on the interventricular septum is a novel description.
The use of the integrated technologies of ICE and CARTO mapping allows accurate delivery of RF energy to the basal septum to treat LVOT gradients in HOCM.
A small area of myocardial injury delivered by RFA at the correct location may lead to successful treatment of LVOT gradients in HOCM.
Hypertrophic cardiomyopathy (HCM) is an inherited disease characterized by otherwise unexplained hypertrophy of the myocardium. Significant left ventricular outflow tract (LVOT) gradients (usually defined as peak gradient ≥30 mmHg) are present in 20–30% of subjects at rest and in up to 70% with exercise provocation.1–3 Left ventricular outflow tract obstruction is associated with greater levels of dyspnoea, a greater incidence of stroke, and higher mortality.1 Removal of LVOT gradients has been shown to improve symptoms and potentially prognosis.4
Asymmetric septal hypertrophy narrows the LVOT, partially obstructing blood flow. In most cases, this is exacerbated by ‘systolic anterior motion’ (SAM) of the mitral valve (MV) apparatus. Severe SAM resulting in contact of the anterior MV leaflet with the interventricular septum is integral to the development of significant LVOT obstruction.5 Once the anterior MV leaflet comes into contact with the septum, a positive amplifying feedback loop further increasing LV pressures is initiated, interrupting this loop by reducing hypertrophy and contractility of this SAM-septal contact area is the goal of septal reduction techniques. Accuracy of septal reduction therapy is paramount to success, and precise targeting of the SAM-septal contact area is imperative.
The ability to perform alcohol septal ablation (ASA) to treat LVOT gradients in hypertrophic obstructive cardiomyopathy (HOCM) depends on suitable septal arterial anatomy. International literature has estimated that ASA cannot be performed in 5–15% HOCM patients due to restrictions of septal anatomy.6,7 These ‘technical failures’ generally fall into two categories: an inability to locate the artery supplying the target for ablation and the inability to instrument this artery due to unfavourable size, angles, and pressures. An alternative method to damage the basal septum that does not rely on arterial anatomy would bypass these restrictions.
Endocardial radiofrequency (RF) ablation (RFA), widely used in the management of arrhythmias, has previously been used to ablate the hypertrophied septum of HOCM patients.8,9 Techniques have involved the visualization of the septal hypertrophy by transoesophageal echocardiography or the use of an electroanatomic electrophysiology (EP) mapping system (CARTO®) to guide and mark ablation lesions. Intracardiac echocardiography (ICE) provides high quality images and has been used successfully in other cardiac procedures. Intracardiac echocardiography and CARTO technologies have inherent advantages and a technology that seamlessly integrates both exists in the form of CARTOSound® (Biosense Webster, Diamond Bar, CA, USA). The CARTOSound® module and Soundstar™ catheter integrate real-time ICE images into the CARTO® mapping system. The Soundstar™ catheter tip also contains a navigation sensor and can be visualized on CARTO maps®, and RFA catheter tips can be seen in the live ICE image. Intracardiac echocardiography has been trialled in ASA and has been shown to detail the SAM-septal contact area well.10
We report the first series of CARTOSound®-guided RFA as septal reduction therapy for HOCM.
Percutaneous septal reduction in the form of ASA has been available at our institution since 2000. This has been the primary method of septal reduction, with myectomy performed in cases where additional cardiac pathology needed addressing (MV surgery, coronary artery bypass grafting). Since then, 111 patients have entered our laboratory with the intention of delivering alcohol. Six patients (5%) did not receive alcohol at any procedure as a suitable target, for alcohol delivery could not be identified or instrumented. A further 18% of those that received alcohol had a persisting significant LVOT gradient at the end of ASA treatment options. The inability to deliver a further dose of alcohol was due to anatomical restrictions; these included the inability to identify a suitable septal artery, inaccurate location of myocardial contrast on injection, or the inability to access an identified septal artery with an over-the-wire balloon.7 Three of these patients underwent septal myectomy. Six patients who could not undergo myectomy remained under our care. One patient was deemed unfit for RFA due to advancing age and cerebrovascular disease. This patient subsequently died within 1 month of assessment. The remaining five patients could not undergo myectomy because of surgical risk2 and patient choice3 (see Table 1).
Demographics, echocardiographic data, and alternative septal reduction method details
Resting (provoked) gradient (mmHg)
Interventricular width in diastole
ASA attempts (doses delivered)
Reason myectomy not performed
Five successive patients underwent CARTOSound®-guided RFA are described; 80% female, mean age 59.20 (44–79) years (Table 1). All had a resting or exercise-provoked LVOT gradient >50 mmHg associated with significant SAM. All patients were suffering with New York Heart Association (NYHA) class III dyspnoea.
Appropriate permissions to perform RFA in these patients were granted by the research and development board of Liverpool Heart and Chest Hospital. All procedures were performed by a single operator (S.M., assisted by R.M.C.) under full informed consent.
Patients underwent resting ± exercise stress echocardiography for LVOT gradient assessment pre- and 6 months post-procedure (Phillips IE33 scanner, Phillips S5-1 probe). Exercise stress echocardiography was performed as part of cardiopulmonary exercise testing at peak exercise in those that had a resting gradient <50 mmHg. Cardiac magnetic resonance (CMR) was performed prior to and 6 months after RFA. Examinations were performed using a 1.5-T scanner (Magneton AERA; Siemens, Medical Imaging, Erlangen, Germany). Left ventricular volumes, ejection fraction, and LV mass were determined using CMRtools (CVIS, London, UK). For late gadolinium enhancement (LGE) imaging, 0.1 mmol/kg gadolinium–diethylenetriamine pentaacetic acid (Gadovist, Bayer Schering, Berlin, Germany) was administered intravenously and standard breath-hold inversion recovery imaging was performed. Late gadolinium enhancement short axis images were analysed using previously validated thresholding and planimetry methods.11
Cardiopulmonary exercise testing was performed using a bicycle ergometer protocol with 10 W minutely increments. Euroqol EQ5D-5L quality-of-life questionnaires were completed in outpatient clinics at first consultation and prior to knowledge of haemodynamic and imaging outcome at follow-up at 6 months.
Radiofrequency ablation using CARTOSound®
All procedures were performed under general anaesthesia using propofol and atracurium followed by isoflurane titrated to bispectral index monitoring. Vasoconstrictors were used as required and haemodynamic monitoring was with invasive arterial blood pressure only. The SoundStar™ catheter was inserted via the right femoral vein and manipulated into the right ventricle (RV) inlet. The phased array probe was used to produce ICE images of the RV, LV, and aorta. Endocardial borders, papillary muscles, aortic cusps, and coronary ostia were delineated and transferred into the CARTO® system (Figure 1, Supplementary material online, File S1). The structural borders were manually contoured and recorded in the traditional method at the end of diastole. A new high density, multiplanar CARTO map was then created of the multiple regions of contact of the anterior MV leaflet and the hypertrophied septum (SAM-septal contact map, Figure 2). Owing to physiological restrictions (i.e. contact only during systole), this map was created in systole. The SAM contact map was superimposed on the LV shell and was the target for RF energy delivery. Quadripolar diagnostic EP catheters were also used to allow RV apical pacing and demonstrate the location of the His bundle (Figure 3).
The ICE probe is situated in the RV inlet and rotated to face the interventricular septum. Several planes of ultrasound are used to create a complete image of the LV (RV and aorta shells can also be created). (A) Left ventricular CARTO map being constructed. (B) Corresponding endocardial ICE contours (green lines, i–iv) used to create the CARTO map.
(A) Left ventricular, aorta, and SAM-septal contact maps in the CARTO shell. The ICE probe is located in the RV inlet and directed towards the interventricular septum. The plane of ultrasound can be seen on the CARTO image to allow the operator to know exactly what level of LV and SAM contact is imaged. (B) Corresponding ICE image. The green line is annotated by the operator to mark the line of SAM-septal contact in various ICE planes, creating a full SAM-septal contact map (pink). (A and B)Very basal area of SAM-septal contact. The ICE probe is realigned and the process is repeated (C and D, and E and F) with each contour adding to the SAM-septal CARTO map. Once the SAM-septal contact map is completed, we can accurately estimate the area of SAM-septal contact in systole, in this case 3.2 cm2 (G).
Markers denoting His bundle (yellow), left anterior and posterior fascicles (white), and ablation lesions (red) are seen in the CARTO image (A and B). Mapping catheter pressure over the left bundle within the pink SAM-septal contact area leads to the development of LBBB (C).
Retrograde aortic access was used to enter the LV. After introduction of the ablation catheter to the LV, intravenous heparin was administered to keep activated clotting time >200 s. The His conduction tissue, left bundle, left anterior and posterior fascicles were directly mapped and annotated on the CARTO shell, and their positions noted in relation to the SAM-septal contact area (Figure 3). Radiofrequency energy was delivered via Navistar and THERMOCOOL® catheter (Biosense Webster, Diamond Bar, CA, USA). Catheter stability was initially difficult due to the dynamic septum and turbulent LVOT blood flow, as ablation progressed and wall motion abnormalities were induced, stability became easier. Trans-atrial septal access was attempted in one patient in an attempt to improve catheter stability; this was unsuccessful and difficult to manipulate. Retrograde aortic access was more stable and therefore adopted for this and other procedures. THERMOCOOL® SMARTTOUCH catheters were used for Patients 3–5 to allow estimation of endocardial contact. Using a combination of CARTO and intracardiac echo navigation, RF energy lesions were placed over the SAM-septal contact area (Figure 4, Supplementary material online, File S1). If possible, specialized conduction tissue (e.g. left bundle/fascicles) was avoided unless it coursed through the central SAM-septal contact area. Stable contact forces of >10 g/>2 bars were sought before RF application. Catheter stability was improved with the onset of rapid right ventricular apical pacing in Patient 5. Radiofrequency ablation powers of 50–60 W limited to temperatures of 60°C were used with saline irrigation at 30 mL/min. Pending catheter stability 2-min application times were used per lesion. A mean of 33.60 (28–42) min of RF energy was applied (Table 1). Procedural endpoints included complete coverage of SAM-septal contact area and basal septal akinesia. Myocardial oedema could be seen up to 10 mm from the endocardial LV surface at cessation of energy delivery. Resolution of LVOT gradient could not be used as this paradoxically increases immediately after RFA. The SAM-septal contact area was 2.1 cm2 (0–3.2 cm2). Ablated area was 14.6 cm2 (7.5–23.1 cm2), representing 9.9% of total LV endocardial surface (5.6–14.2%) (Table 2). Anticoagulation was reversed at the end of the case using intravenous protamine.
Paradoxical increase in LVOT gradient. LBBB and late complete heart block
Retrograde aortic access was the preferred access route in all patients, the attempt at transseptal access in Patient 3 was unsuccessful due to unstable catheter position. Systolic anterior motion-septal contact areas are on table and under General anaesthetic, Patient 2 had SAM-septal contact on exercise testing pre-procedure.
(A and B) An Right anterior oblique (RAO) projection of the process of RF energy delivery over the SAM-septal contact area. Ablation is also performed around the perimeter of the SAM map to compensate for the ‘systole-acquired’ SAM map. (C) The automatic ‘tracking’ of the RF catheter tip on the live ICE screen (green halo). The papillary muscles can also be seen in the ICE images and marked on CARTO. (D and E) The final RAO and left anterior oblique projections of RF delivery. The medial displacement of the ablation lesions compared with the SAM map (E) is a function of systole vs. diastole-acquired points.
All four surviving patients reported an improvement in symptoms at 6 months. Three patients had improved from NYHA class III to II and one patient had improved from NYHA class III to I. Chest pain was present in two patients prior to RFA and this resolved in both subjects.
Echocardiogram was performed at 6 months in all surviving patients. Average peak resting gradients improved from 64.25 (±50.60) to 12.25 (±2.50) mmHg (Figure 5). Valsalva- or exercise-induced gradient improved from 93.50 (±30.88) to 23.25 (±8.30) mmHg. Visual estimation of SAM improved in all patients. All patients demonstrating rest or exercise SAM-septal contact did not demonstrate this post-RFA. Basal septal diameter in diastole reduced from 18.25 (±1.89) to 16.75 (±2.50) mm. Basal septal hypokinesia was seen in all. Left atrium size reduced from 48.75 (±6.50) to 44.75 (±8.30) mm. Left ventricular diameter did not change in diastole (47.25 ± 8.30–47.25 ± 8.50 mm) or systole (30.75 ± 4.86–29.5 ± 4.51 mm). The grade of SAM-associated MR reduced in two patients, one from moderate to mild, and one from mild to none. Two patients with mild MR did not change post-procedure.
Pre- and post-RFA resting LVOT gradient, provoked LVOT gradient, and % predicted peak VO2. Solid lines represent a mean value of the four surviving patients. The dotted lines represent individual patient values.
Cardiac magnetic resonance imaging
Two patients had non-magnetic resonance imaging safe implantable cardiac defibrillators in situ that precluded CMR imaging. Late gadolinium enhancement could be seen in the basal septum in both patients who underwent CMR scanning 6 months after the procedure (see Figure 5). Late gadolinium enhancement was seen up to 8 mm from the LV endocardial surface. This was similar to the depth of tissue oedema visible on ICE during the procedure. Scar measured 6.3 and 2.2 g, respectively, representing 2.4 and 1.1% of total LV mass. Left ventricular mass prior to ablation measured 198 and 259 g, respectively. This reduced to 160 and 236 g, representing a 9 and 19% reduction in total LV mass, respectively.
Cardiopulmonary exercise test was performed prior to and 6 months after ablation in three patients. Total exercise time increased from 558 (±129.87) to 730 (±63.64) s. Pre-procedural peak VO2 measured 15.48 (±2.27) mL/min/kg, this improved to 16.53 (±5.16) mL/min/kg.
EQ5D-5L quality-of-life index value increased from 0.57 (±0.17) to 0.65 (±0.18). Health score improved from 44 (±18.93) to 70 (±3.54).
One patient underwent successful RFA but developed a significant retroperitoneal haemorrhage and cardiovascular collapse following sheath removal. Urgent surgical repair of the right femoral artery was initially successful but a secondary bleed within 24 h led to mesenteric ischaemia and ultimately death.
One patient developed a paradoxical increase in LVOT gradient immediately following ablation resulting in pulmonary oedema. Reintubation, intravenous dexamethasone, and right ventricular apical pacing were employed to reduce the gradient. She recovered well and was discharged. As the left bundle branch block (LBBB) was seen to pass through the SAM-septal contact area in this patient, peri-procedural LBBB with subsequent late complete heart block at 6 months post-procedure was also seen.
She showed significant improvement symptomatically and haemodynamically at 6-month assessment, with clear hypokinesia of the basal septal myocardium.
The course of the LBBB also fell within the SAM-septal contact area in one patient. Left bundle branch block was induced on LV conduction tissue mapping (and marked on CARTO shell) but did not recover before ablation was eventually delivered in this area. Left bundle branch block persisted to latest follow-up with no evidence of atrioventricular (AV) block on ambulatory electrocardiogram monitoring.
(A) A CMR three-chamber image taken at Day 2 following ablation. Microvascular obstruction (MVO) is highlighted in the basal septum on this early gadolinium-enhanced image. (B) A 6-month three-chamber LGE image with LGE in the basal septum. (C and D) Short axis LGE images one cut (8 mm) below the LVOT, at the level of the body of the MV. Late gadolinium enhancement can be seen in the LV endocardium in the target area.
There remains a minority of patients in whom conventional methods of septal reduction either fail or cannot be attempted. Individual centres will vary in their availability of septal reduction techniques with surgical myectomy remaining the gold standard when performed in high-volume centres,12 although myectomy and ASA are offered on an equal recommendation in recent ESC guidance.3 Outcomes from myectomy in high-volume centres are excellent with good procedural success and low mortality. This is not reflected in centres with a low volume of patients.13 Myectomy is major surgery and carries risks, especially in those with co-morbidities. Patients will often choose percutaneous solutions if they are on offer.14
Many centres will offer ASA as the first line septal reduction therapy. The main procedural risk is AV block requiring pacemaker implantation.15 Most series present outcome data as mean values; however, this can hide individual failure. In series with individualized results up to a third of patients have an unsatisfactory outcome as measured by symptom and LVOT gradient parameters.7,16 These results are in part due to the constraints of septal arterial anatomy. If alcohol-induced infarction is not at the SAM-septal contact area, LV haemodynamics will not change significantly.17 In addition to those with inaccurate location of infarct and poor outcome, we must consider the 5–15% of patients who cannot receive alcohol due to an inability to locate an appropriate septal vessel.6,7 There would be a clear advantage of a percutaneous septal reduction technique that afforded accurate targeting of myocardial damage, independent of arterial anatomy.
Radiofrequency septal ablation in HOCM has been shown to be feasible8,9; however, the novel use of CARTOSound® technology in our patients defines the ablation target with previously unparalleled accuracy. The live ICE image and the ability to incorporate this information into the electroanatomic mapping system display the SAM-septal contact target clearly. Although the volume of tissue damage appears small compared with other forms of septal reduction, the accuracy seems to be sufficient to affect SAM and thus reduce LVOT gradients.
The required size of infarct to produce symptom resolution in non-surgical septal reduction is not known. Cardiac magnetic resonance studies have highlighted an alcohol-induced scar of up to 25 g post-ASA, with a total LV mass reduction of 14–16 g.17 This was often described as a transmural infarct that was associated with right bundle branch block. As CMR was only possible in two patients following treatment, we could not make secure conclusions about the appearance of scar, but there are signals that the damage delivered is smaller than that in ASA. The scar size seen in these two patients measured just 4.2 g (Figure 6). This was re-iterated by the modest reduction in septal width measured by echocardiography, a 1.5 mm reduction is less marked than the change reported in ASA series, which can be up to 5–6 mm.7,15 This modest reduction in septal width could perhaps lend itself well to those with less marked hypertrophy who are thought to be at risk of ventricular septal defect with traditional methods. A small amount of damage accurately delivered can interrupt the SAM-septal feedback mechanism and reduce LVOT gradients effectively. The location of tissue damage seems to be at least as important as the size.
This small series highlights our early experience of a novel technique for septal reduction in a highly symptomatic population with advanced, obstructive cardiac failure. Failed trial of medications, ASA, and unsuitability for myectomy leaves this cohort with no other traditional therapeutic option for reduction of LVOT gradients. The fragility of patients with advanced obstructive cardiac failure is evidenced both by the early death of our patient who was deemed too frail for an RF septal reduction and by our patient who died following a retroperitoneal haemorrhage. While retroperitoneal haemorrhage is a recognized risk in any percutaneous procedure,18 it is likely to be poorly tolerated in these patients due to an inability to cope with the acute preload reduction.
The intolerance of reduced preload is inherent in any obstructive HCM patient but may be worsened transiently following direct RFA of the SAM-septal contact area due to tissue oedema. Myocardial oedema is well recognized in CMR studies following RFA in other procedures.19 The mechanism of cell death in RFA is not the same as alcohol-induced cell membrane desiccation. The cell injury created by the hyperthermic effects of RFA is thought to be mediated by injury to the sarcolemmal membrane and subsequent calcium overload. The different processes involved in RFA and ASA may explain the disproportionate tissue oedema, as the total cell necrosis by LGE CMR scanning at 6 months is substantially less in RFA.
In our patients, despite successful ablation of the SAM-septal contact area, we experienced no acute change in LVOT gradients when assessed invasively at the end of the procedure. The patient with a paradoxical increase in LVOT gradient had most likely generated greater tissue oedema over the narrowest region of the LVOT. This was probably more apparent as a result of improving accuracy, catheter stability, and contact as we learned the procedure.
Paradoxical increase in LVOT gradients was also observed in the paediatric HOCM population treated with RFA (with fatal consequence) and potentially stands to be a challenge in the adoption of this technique into the mainstream.8 Peri-procedural intravenous dexamethasone has now been employed in our patients as empirical therapy to reduce this tissue oedema effect but subsequent numbers are insufficient to assess its efficacy.
Although such a small preliminary study cannot accurately assess the morbidity and mortality of a novel technique, recent analyses of complications during RFA for ischaemic Ventricular tachycardia found a procedure-related death rate of between 0.4 and 3.0%, some relating to retroperitoneal haemorrhage.20 This is probably the closest representation of procedural risk in our patients due to the similarities in LV-based ablation techniques.
The ability of CARTO to identify specialized conduction tissue enables the operator to deliver tissue damage to the target area with the knowledge of the location of specialized conduction tissue. While it was hoped that the technical accuracy of the CARTO map would prevent this collateral damage to this tissue, there are clearly cases where the LBBB courses directly through the target area of SAM-septal contact. While it is accepted that the development of LBBB is generally undesirable, it is a common finding in post-surgical myectomy. Surgical myectomy has been suggested to actually improve patient-related outcome, suggesting that LBBB in this context has no detrimental effect.12 In order to deliver effective treatment to the SAM-septal contact area, we may be forced to accept LBBB as an outcome from the procedure. The change in LV conduction and hence contraction may be relevant when considering the mechanism of resolution of gradient in two patients, one had LBBB and one was RV paced at 6-month follow-up. There was, however, clear akinesia of the basal septum on echocardiogram, suggesting myocardial damage and structural change rather than the dyskinetic contraction seen in LBBB. In addition, if rapid RV apical pacing is used to stabilize catheter movement, then real-time electrogram visualization of Purkinje potentials is forfeited, and the operator must rely solely on the CARTO-mapped specialized conduction tree.
This is a preliminary study to explore the use of combined ICE and CARTO technology to guide RFA to treat LVOT gradients in HOCM. The study population was therefore small. The ability to make secure conclusions about outcome is obviously restricted by the small numbers. The follow-up period was also short. Although these are restrictions to the long-term application of CARTOSound-guided RFA, the aim of this study was to implement and describe a new technology. Further enrolment and follow-up of appropriate patients will continue.
Radiofrequency ablation to treat LVOT gradients in HOCM using CARTOSound® shows significant promise. The unprecedented accuracy of tissue damage interrupts the SAM-septal contact cycle and results in improvements in LVOT gradients, symptoms, and quality of life in this preliminary group of patients. Further experience of this technique is required before it can be considered equivalent to standard septal reduction techniques.
. 2014 ESC Guidelines on diagnosis and management of hypertrophic cardiomyopathy: the Task Force for the Diagnosis and Management of Hypertrophic Cardiomyopathy of the European Society of Cardiology (ESC). Eur Heart J 2014;35:2733–79.
. Alcohol septal ablation for hypertrophic obstructive cardiomyopathy: safe and apparently efficacious but does reporting of aggregate outcomes hide less-favorable results, experienced by a substantial proportion of patients? J Invasive Cardiol 2015;27:301–8.
. Comparative in hospital safety and cost of alcohol septal ablation versus isolated surgical myectomy for symptomatic hypertrophic obstructive cardiomyopathy: insights from the multi center PREMIER database. TCT Oral abtracts session 2011. 2012.
. Myocardial infarction after percutaneous transluminal septal myocardial ablation in hypertrophic obstructive cardiomyopathy: evaluation by contrast-enhanced magnetic resonance imaging. J Am Coll Cardiol 2004;43:27–34.
. Acute pulmonary vein isolation is achieved by a combination of reversible and irreversible atrial injury after catheter ablation: evidence from magnetic resonance imaging. Circ Arrhythm Electrophysiol 2012;5:691–700.