Adjunctive Intracardiac Echocardiography to Guide Slow Pathway Ablation in Human Atrioventricular Nodal Reentrant Tachycardia
Background Because of the inability of fluoroscopy to image intracardiac structures, the precise anatomic location of successful slow pathway (SP) ablation is controversial. We hypothesized that adjunctive intracardiac echocardiography (ICE) in concert with conventional fluoroscopy and electrogram guidance could identify the anatomic site of successful SP ablation.
Methods and Results In 25 patients, radiofrequency ablation was performed in the triangle of Koch directed by biplane fluoroscopy and a 6.2F, 12.5-MHz ICE catheter positioned adjacent to the triangle of Koch. Persistent SP conduction, number of radiofrequency applications, presence of junctional tachycardia, and fluoroscopy times were evaluated. As demonstrated by ICE, anterograde SP ablation was achieved between 2 and 7 mm from the tricuspid valve in imaging planes containing the AV muscular septum in all cases. Radiofrequency energy applications applied at other sites within the triangle of Koch failed to interrupt SP conduction. A mean of three radiofrequency energy applications (3±2; range, 1 to 12) successfully ablated all evidence of anterograde SP conduction in all patients studied. Junctional tachycardia was seen in 96% (71/74) of the radiofrequency energy applications.
Conclusions Radiofrequency ablation at the tricuspid valve’s insertion into the AV muscular septum as identified by ICE reliably terminates anterograde SP conduction, supporting the hypothesis that the SP consistently traverses this anatomic location.
During AV nodal modification, the positioning of the ablation catheter is guided by fluoroscopic imaging combined with interpretation of electrogram morphology.1–9 However, fluoroscopy cannot directly visualize endocardial structures, and a solely electrogram-guided approach to slow pathway AV nodal modification is neither sensitive nor specific for identifying an effective ablation site.10 ICE has demonstrated utility in imaging ablation catheter position, stability, and lesion formation during RF ablation.11–13 Chu et al11 described the use of ICE for slow pathway modification in two patients. Because of their limited number of patients and difficulty imaging more anterior cardiac structures near the triangle of Koch due to their use of a nondeflectable ICE catheter, sufficient data regarding the anatomic location of successful slow pathway ablation are not available.
In the present study, the anatomic location of successful and unsuccessful ablation sites as imaged by ICE during conventional slow pathway modification of typical AVNRT was studied with small, nondeflectable 6.2F, 12.5-MHz ICE catheters angled through long vascular sheaths to image structures near the triangle of Koch.14 We hypothesized that ICE catheters directed through preformed long vascular sheaths would identify the effective anatomic ablation location during slow pathway modification for typical AVNRT and that the effective site would be adjacent to the tricuspid annulus. To test these hypotheses, a series of prospectively enrolled patients with typical AVNRT undergoing conventional fluoroscopic and electrogram-guided slow pathway AV nodal modification were studied with adjunctive ICE to anatomically define the effective ablation sites. Ablation characteristics, including safety, electrograms and fluoroscopic locations at the successful site, presence of residual slow pathway conduction, and follow-up, were evaluated.
Twenty-five consecutive patients (11 women, 14 men; mean age, 43±18 years) undergoing electrophysiological study for typical AVNRT were prospectively enrolled, and informed consent was obtained under a protocol approved by the Investigational Review Board at the National Naval Medical Center.
Quadrapolar catheters were placed in the high right atrium, right ventricular apex, and His bundle positions from the femoral veins under fluoroscopic guidance. A decapolar coronary sinus catheter was placed via the left subclavian vein. ECG leads I, II, aVF, and V1 and intracardiac electrograms were recorded on a MIDAS system (Electronics for Medicine). The stimulation protocol consisted of atrial and ventricular incremental pacing and extrastimulation. Dual AV nodal physiology was confirmed by conventional criteria (≥50-ms increase in H1-H2 interval for a 10-ms decrease in A1-A2). Isoproterenol was administered to patients without inducible tachycardia at baseline. AVNRT was confirmed by ventricular extrastimulus testing during tachycardia when the His bundle was refractory. Reproducible tachycardia induction was confirmed before ablation.
ICE was performed with a 6.2F, 12.5-MHz blunt-tipped Sonicath catheter (Boston Scientific Corp) paired with a Hewlett-Packard SONOS 2000, model 2400A ultrasound system. The ICE catheter was inserted into a 60-cm SL2 sheath (Daig Corp) placed in the left femoral vein, permitting direction of the ICE catheter to the triangle of Koch. As the ICE catheter was withdrawn from the right ventricular outflow tract to the coronary sinus ostium, imaging was performed in planes perpendicular to the interatrial and interventricular septa (Fig 1⇓). ICE images were recorded on videotape with an audio soundtrack. Care was taken to image the ablation catheter tip by orienting the ICE catheter parallel with and in close proximity to the ablation catheter before each energy application with both ICE and biplane fluoroscopy. ICE measurements were made with an on-line measurement package. The diameter of the coronary sinus ostium was measured at the level of the thebesian valve. Imaging planes were categorized as A, B, C, or D (Fig 1⇓). Plane A was defined by the presence of the coronary sinus ostium, fossa ovalis, or both. Plane B was defined by the presence of the AV muscular septum and the posterior leaflet of the mitral valve without visualization of the membranous septum or the coronary sinus. Plane C was defined by the presence of the membranous and muscular interventricular septa without the AV muscular septum. The plane incorporating both the AV muscular septum and a portion of the membranous septum was designated BC (Fig 2C⇓) because it was intermediate between planes B and C. Plane D was defined by the presence of the right ventricular outflow tract. Catheter stability was assessed by ICE and recorded for each ablation location. The distance of the ablation catheter tip from the insertion of the tricuspid leaflet into the tricuspid annulus was measured in each imaging plane.
RF ablation was performed with a 4-mm–tip standard curve steerable catheter (EP Technologies) that was passed through an SR0 long sheath (Daig Corp) placed in the right femoral vein. The ablation catheter tip was positioned in the midseptal region of the tricuspid annulus by use of biplane fluoroscopy. Fluoroscopic angulations were standardized to the 45° RAO and 45° LAO projections. Fluoroscopic images of the successful ablation sites were recorded cineradiographically and classified on the basis of the anterior, middle, and posterior thirds of the septum as measured between the His bundle catheter and coronary sinus catheter in a scheme proposed by Jazayeri et al.3 Ablation catheter tip location was similarly recorded for LAO projections. Therefore, the final successful fluoroscopic location was recorded as a two-letter code, with the first letter (A, M, or P) denoting the anterior, middle, or posterior third of the distance between the His bundle and coronary sinus catheters in the RAO view and the second letter (A, M, or P) denoting the LAO positioning. The catheter tip was localized by advancing or withdrawing the ultrasound catheter until the characteristic bright hyperechoic reflectance artifact of the ablation catheter tip electrode just entered the ultrasound imaging plane.11,13 Catheter stability was assessed with both fluoroscopy and ICE imaging. ICE documented catheter stability by noting when the distal ablation catheter tip moved in concert with the endocardial surface before each RF energy application. If catheter stability was not obtainable, then a different ablation catheter or sheath was used before application of RF energy. RF energy was administered in a unipolar fashion with 20 to 30 W for 10 to 60 seconds via either a Radionics RFG-33 RF Lesion Generator System or an EPT 1000 Cardiac Ablation Controller. After each RF energy application, repeat attempts were made to induce AVNRT. After 30 minutes, repeat atrial and ventricular extrastimuli were administered to assess for evidence of dual AV nodal physiology or inducible AVNRT. If more than one AV nodal echo beat was identified, further RF applications were performed. Otherwise, the study was terminated.
In addition to the ultrasound-measured catheter positions from the tricuspid annulus, tachycardia cycle length, fast and slow pathway effective refractory periods preablation and postablation, presence or absence of junctional tachycardia with each RF application, the presence of AV or VA block during the RF applications, the time from the His bundle atrial electrogram to the distal ablation catheter bipolar electrogram (AHis-AAbl), total number of RF applications, the ratio of atrial to ventricular electrogram amplitudes, catheter stability at each ablation site, presence or absence of visible lesions (and their sizes), coronary sinus ostium diameter at the level of the thebesian valve, and the fluoroscopic catheter position at the successful ablation site were recorded.
For comparison of differences between groups, a χ2 test was used for discontinuous variables. Results are expressed as mean±SD. A value of P<.05 was considered significant. Correlations were performed with a Pearson product-moment correlation coefficient with Sigmastat software version 1.01 (Jandel Scientific).
Intracardiac Ultrasound Imaging
Using the 12.5-MHz ICE catheter, the image region extended up to 40 mm radially from the ultrasound catheter in all patients. The SL2 long sheath permitted the ICE catheter to be directed anteriorly and medially toward the central fibrous body of the heart (Fig 1⇑). From this position, structures within and near the triangle of Koch were well visualized. The parallel orientation of the ICE catheter to the ablation catheter allowed direct imaging of the ablation catheter tip during all energy applications (Figs 2A⇑, 2B⇑, 3A⇓, and 3B⇓). Catheter shafts without metallic electrodes caused an acoustic shadowing artifact that obscured imaging beyond the shaft. The platinum electrodes on the catheters caused a hyperacoustic “spray” artifact, with the largest artifact produced by the large 4-mm ablation catheter tip electrode (Figs 2C⇑ and 3C⇓). Imaging during RF energy delivery with the Radionics RF generator did not interfere with image quality, but the EPT RF generator system caused significant distortion of the ICE image during RF energy delivery in patient 4 and was abandoned for all subsequent patients.
Successful Ablation Site Locations
The ablation catheter tip position at the onset of RF energy delivery is demonstrated in Table 1⇓⇓. Fifteen energy applications were performed in the imaging plane at the AV septum just posterior to the membranous septum (imaging plane BC, Fig 2⇑), 43 in the imaging plane of the AV muscular septum anterior to the coronary sinus and posterior to the membranous septum (imaging plane B, Fig 3⇑), and 16 applications between the tricuspid annulus and coronary sinus (imaging plane A). Successful slow pathway ablation was achieved between 2 and 7 mm from the tricuspid annulus with the catheter overlying the AV muscular septum in imaging plane B or BC in all patients (Fig 4⇓). All energy applications placed between the coronary sinus and the tricuspid annulus (imaging plane A) failed to terminate slow pathway conduction, regardless of proximity to the tricuspid annulus. All sites closer than 2 mm from the insertion of the tricuspid valve leaflet to the annulus dislodged into the right ventricle during energy application and were ineffective. ICE also disclosed significant anatomic variability in the maximal diameter of the coronary sinus (16±5 mm; range, 7 to 25 mm).
Visible lesions were identified in 9 of 25 patients (36%) and averaged 6±2 mm. All of the lesions had relatively hyperechoic borders but varied slightly in appearance among patients: 5 lesions had a hyperechoic border with a central pit from the ablation catheter (Figs 5⇓ and 6⇓), 2 without a visible pit, and 1 lesion consisted of multiple micropits from which the ablation catheter appeared to have “bounced” on the myocardial surface. Six visible lesions occurred at the AV muscular septum within 4±2 mm (range, 2 to 7 mm) of the insertion of the tricuspid valve and were uniformly effective at terminating slow pathway conduction. Three lesions measuring 6 to 8 mm placed between the coronary sinus and tricuspid annulus were each ineffective at terminating the patients’ tachycardia (Fig 6⇓). In those lesions that were visible, the sizes of the effective lesions (6±3 mm) were not significantly different from the ineffective lesions (6±2 mm).
In four patients, visualization of a significant hyperechoic lesion prompted discontinuation of RF energy delivery. The largest visualized lesion (11 mm) was identified in patient 1 after 30 W was applied for 30 seconds (Fig 5⇑). Because of the size of this lesion and its proximity to the compact AV node (15 mm from the His bundle catheter), all other lesions in subsequent patients were applied at 20 to 25 W power. In no patient in whom a visible lesion was identified involving the tissue overlying the insertion of the tricuspid valve into the AV muscular septum was persistent slow pathway conduction found after ablation.
ICE permitted demonstration of movement of the ablation catheter tip from its initial position during RF energy delivery. Of the 74 RF energy applications, only 37 (50%) achieved constant tissue contact throughout energy application (Table 1⇑). Of the unstable catheter appearances, catheter “bouncing” or “slipping” near the initial placement site was the most common movement seen (25 of 74 applications, 34%), followed by dislodgment of the catheter tip into the coronary sinus (16 of 74 applications, 22%) or tricuspid valve (7 of 74 applications, 9%). Coronary sinus migration was always seen on LAO fluoroscopic images and on ICE by loss of the ablation catheter tip artifact from the ultrasound imaging plane. Catheter “bouncing” or migration into the right ventricle (with the ablation catheter tip over the tricuspid valve) was not appreciated by fluoroscopy.
Junctional tachycardia was seen in 71 of 74 RF applications (96%). Stable catheter positioning on the atrial aspect of the tricuspid annulus over the AV muscular septum resulted in continuous junctional tachycardia, whereas intermittent junctional tachycardia occurred when the catheter tip “bounced” on the myocardial surface or dislodged anteriorly into the right ventricle or posteriorly into the coronary sinus (Table 1⇑). Seven RF energy applications were terminated prematurely by the operator because of concern about the rapid nature of the junctional tachycardia without sequelae. Three RF applications (4%) failed to precipitate junctional tachycardia: ultrasound imaging disclosed that two were applied with the ablation catheter tip on the tricuspid valve leaflet and one was applied before the catheter was stabilized against the myocardium and resulted in microbubble formation. Each of these three applications occurred despite apparently adequate electrogram and fluoroscopic appearances of the catheter placement.
Electrophysiological and Procedural Data
The patients’ electrophysiological variables are outlined in Table 2⇓. Although no anterograde slow pathway conduction was present after ablation in any patient, one patient (patient 9) had evidence of persistent single atypical AV nodal echoes after ablation. This same patient had evidence of both typical and atypical forms of AVNRT before ablation. There was no difference between preablation and postablation AH intervals in patients after slow pathway modification. At the successful ablation site, the average A:V electrogram amplitude ratio was 0.29±0.15 and the AHis-Abl was 35±10 ms. However, in the evaluation of all the RF energy application electrograms, no significant correlation existed between the measured A:V electrogram ratio and the measured distance of the ablation catheter tip to the tricuspid valve as visualized by ICE (Table 1⇑, r=−.083, P=.48).
Six of 74 RF lesions (8%) resulted in retrograde VA block during application of RF energy. Two patients developed transient first-degree AV block after cessation of the RF energy, but this resolved in <10 minutes after the RF application and in both patients was preceded by VA block. The AHis-AAbl intervals in these two RF energy applications were 10 and 20 ms, respectively, and occurred in ultrasound imaging plane BC. Both patients had inducible AVNRT after these episodes.
Complications and Follow-up
No procedural complications or episodes of complete heart block occurred. No clinical recurrence of typical AVNRT has occurred after a mean of 11.0±3.4 months.
This is the first study using ICE in vivo to systematically image the anatomic sites of ablation during AV nodal modification in humans. ICE provided clear visualization of the septal structures within the triangle of Koch. Successful slow pathway ablation was uniformly achieved when RF energy was applied in close proximity to the insertion of the tricuspid leaflet into the AV muscular septum. Ablation sites >7 mm posterior to the tricuspid annulus within the triangle of Koch did not result in interruption of the AVNRT circuit. Although the area of the tricuspid annulus has previously been suggested as the effective site for RF ablation of the slow pathway,1–4,6,15 ICE confirmed the importance of targeting periannular cells over the AV muscular septum to achieve success.
Previous investigators have suggested that slow pathway ablation can occur from within the coronary sinus.1,2,6 Our data do not exclude the possibility that such sites may also result in successful ablation. However, lesions applied between the coronary sinus and tricuspid annulus in the present study were uniformly ineffective at achieving slow pathway ablation, even when visible lesions were identified with ICE. It is possible that lesions placed within the coronary sinus in other studies interrupted a separate portion of the AV nodal reentrant circuit, encompassed all posterior inputs to the AV junctional cells, directly incorporated tricuspid annular cells because of larger lesions from the higher ablation energies administered, or altered slow pathway conduction sufficiently to preclude sustained reentry.
ICE disclosed identifiable lesions in 9 of 74 RF applications (12%). Changes in tissue characteristics after ablation were difficult to appreciate in the majority of lesions imaged with the 12.5-MHz ICE transducer. Cardiac motion, episodes of ablation catheter instability, and low ablation energies may have contributed to the inability to identify discrete lesions. However, identification of a lesion over the AV muscular septum adjacent to the tricuspid annulus reliably correlated with complete elimination of slow pathway conduction. The small size of visible lesions after effective ablation suggests that periannular atrial cells over the AV muscular septum play an important role in slow pathway conduction.
A:V Electrogram Amplitude Ratio
As seen in other studies, a small A:V electrogram amplitude ratio was noted at the effective site.3,5 Contrary to expectation, there was a poor correlation between this ratio and the distance from the ablation catheter tip to the tricuspid annulus. Factors explaining this finding include error introduced by the proximal position of the recording bipole compared with the distal ablating electrode, the interaction of the endocardial activation vector to the recording bipole vector, and the fact that left ventricular muscle lies adjacent to right atrial tissue over a broad region within the triangle of Koch (Fig 7⇓).16 In addition, the proximity of the coronary sinus to the left ventricle permitted small A:V electrogram ratios to be recorded within the coronary sinus outside the triangle of Koch.
Catheter/Tissue Interface Assessment and Safety
The ability to directly assess catheter tip contact with endocardium most likely contributed to successful slow pathway ablation and a higher incidence of junctional tachycardia in this series compared with the fluoroscopically guided series of Epstein et al5 (119 of 181 applications, 67%) or Hintringer et al17 (172 of 540 applications, 32%). Improved assessment of the ablation catheter location may also have contributed to the fewer episodes of VA block seen in this study (6 of 74, 8%) compared with Hintringer et al (46 of 172, 26.9%, P<.002) despite similar ablation energies being used (20 to 30 W).
No episodes of inadvertent sustained first-degree or complete heart block occurred in this series of patients. However, temporary first-degree AV block lasting <10 minutes was noted after two RF applications at sites near the membranous AV septum (imaging plane BC) with evidence of unstable catheter contact. Both episodes were heralded by VA dissociation during junctional tachycardia and short AHis-AAbl intervals. These data are similar to the findings of other series in which monitoring of electrophysiological variables was important for safe slow pathway modification.15,17 Ablating in a more posterior imaging plane (ICE imaging plane B) never caused AH prolongation.
Slow Pathway Ablation
No evidence of postablation anterograde slow pathway conduction was present in any patient studied. Although complete slow pathway ablation may not be required for initial resolution of the patient’s tachycardia, the presence of residual slow pathway conduction may be associated with an increased recurrence rate.18–20 Previous studies using either electrogram markers or an anatomic approach to AV nodal modification have demonstrated persistent slow pathway conduction in 7% to 64% of patients.1–4,7,8 In the present study, one patient with typical and atypical AVNRT before ablation had evidence of persistent retrograde slow pathway conduction after ablation, as evidenced by the presence of a single atypical AV nodal echo during postablation ventricular extrastimulus testing. This may have occurred because of incomplete ablation of the slow pathway with residual anterograde unidirectional block or because of the existence of more than one slow pathway input to the AV node in this patient.21
With the exception of four cases, no attempt was made to target other posterior sites in the triangle of Koch. It is possible that ablation at other more posterior sites may be equally effective in terminating slow pathway conduction, particularly in patients with the uncommon form of AVNRT. However, the consistent ability to ablate anterograde slow pathway conduction at the AV muscular septum near the tricuspid annulus lends credibility to this technique and provides insight into the anatomic course of slow pathway inputs to the compact AV node.
Larger randomized series with long-term follow-up are required to assess the cost-effectiveness of this technique. Fluoroscopy remains essential for safe introduction of the ultrasound catheter into the heart and the initial placement of electrophysiological catheters.
ICE with a 6.2F, 12.5-MHz ICE catheter directed through a long sheath to image the region of the triangle of Koch provides a safe, effective, and novel means of anatomically defining a successful site of slow pathway ablation in patients with typical AVNRT. RF energy application directed by ICE to atrial tissue from 2 to 7 mm posterior to the insertion of the tricuspid valve into the AV muscular septum terminated anterograde slow pathway conduction in all patients. In contrast, RF energy applications delivered more posteriorly between the coronary sinus ostium and tricuspid annulus resulted in junctional tachycardia in all RF energy applications but failed to result in slow pathway ablation. Therefore, a critical portion of the slow pathway exists adjacent to the tricuspid valve overlying the AV muscular septum. Although conventional methods of targeting the slow pathway in patients with typical AVNRT are effective, adjunctive ICE may facilitate slow pathway ablation in patients with complicated cardiac or intrathoracic anatomy.
Selected Abbreviations and Acronyms
|A:V||=||atrial to ventricular|
|AVNRT||=||AV nodal reentrant tachycardia|
|ICE||=||intracardiac echocardiography, echocardiographic|
|LAO||=||left anterior oblique|
|RAO||=||right anterior oblique|
The Chief, Navy Bureau of Medicine and Surgery Washington, DC, Clinical Investigation Program sponsored this study, grant B95–078. The authors would like to acknowledge Mark C.P. Haigney, MD, for his review of this manuscript and the outstanding clinical and technical support provided by HM1 David W. Neff, HM2 C. Michael Fairman, HM2 Ral S. Wheeler, HM3 John A. Sral, HM3 Richard E. Stone, HM2 Gina L. Morosky, and other members of the cardiac electrophysiology laboratory at the National Naval Medical Center, Bethesda, Md, in completion of this research project.
Reprint requests to Westby G. Fisher, MD, Cardiac Electrophysiology Associates, 2123 Auburn Avenue, Suite 440, Cincinnati, OH 45219.
The views expressed in this article are those of the authors and do not reflect the official policy or position of the US Navy, Department of Defense, or the US Government.
- Received March 5, 1997.
- Revision received May 29, 1997.
- Accepted June 6, 1997.
- Copyright © 1997 by American Heart Association
Jackman WM, Beckman KJ, McClelland JH, Wang X, Friday KJ, Roman CA, Moulton KP, Twidale N, Hazlitt HA, Prior MI, Oren J, Overholt ED, Lazzara R. Treatment of supraventricular tachycardia due to atrioventricular nodal reentrant tachycardia by radiofrequency catheter ablation of slow pathway conduction. N Engl J Med. 1992;327:313–318.
Kay GN, Epstein AE, Dailey SM, Plumb VJ. Selective radiofrequency ablation of the slow pathway for the treatment of atrioventricular nodal reentrant tachycardia: evidence for involvement of perinodal myocardium within the reentrant circuit. Circulation. 1992;85:1675–1688.
Jazayeri MR, Hempe SL, Sra JS, Dhala AA, Blanck Z, Deshpande SS, Avitall B, Krum DP, Gilbert CJ, Akhtar M. Selective transcatheter ablation of the fast and slow pathways using radiofrequency energy in patients with atrioventricular nodal reentrant tachycardia. Circulation. 1992;85:1318–1328.
Haissaguerre M, Gaita F, Fischer B, Commenges D, Montserrat P, d’Iversois C, Lemetayer P, Warin JF. Elimination of atrioventricular nodal reentrant tachycardia using discrete slow potentials to guide application of radiofrequency energy. Circulation. 1992;85:2162–2175.
Langberg JJ, Leon A, Borganelli M, Kalbfleisch SJ, El-Atassi R, Calkins H, Morady F. A randomized, prospective comparison of anterior and posterior approaches to radiofrequency catheter ablation of atrioventricular reentrant tachycardia. Circulation. 1993;87:1551–1556.
Niebauer MJ, Daoud E, Williamson B, Man KC, Strickberger A, Hummel J, Morady F. Atrial electrogram characteristics in patients with and without atrioventricular nodal reentrant tachycardia. Circulation. 1995;92:77–81.
Chu E, Fitzpatrick AP, Chin MC, Sudhir K, Yock PG, Lesh MD. Radiofrequency catheter ablation guided by intracardiac echocardiography. Circulation. 1994;89:1301–1305.
Olgin JE, Kalman JM, Fitzpatrick AP, Lesh MD. Role of right atrial endocardial structure as barriers to conduction during human type I atrial flutter: activation and entrainment mapping guided by intracardiac echocardiography. Circulation. 1995;92:1839–1848.
Schwartz SL, Gillam LD, Weintraub AR, Sanzobrino BW, Hirst JA, Hsu T-L, Fisher JP, Marx G, Fulton D, McKay RG, Pandian NG. Intracardiac echocardiography in humans using a small-sized (6F), low frequency (12.5 MHZ) ultrasound catheter: methods, imaging planes and clinical experience. J Am Coll Cardiol. 1993;21:189–198.
Wilcox BR, Anderson RH. Surgical Anatomy of the Heart. 2nd ed. London, UK: Gower Medical Publishing; 1992:7.9.
Hintringer F, Hartikainen J, Davies DW, Heald SC, Gill JS, Ward DE, Rowland E. Prediction of atrioventricular block during radiofrequency ablation of the slow pathway of the atrioventricular node. Circulation. 1996;92:3490–3496.