Author + information
- Received November 6, 2005
- Revision received March 20, 2006
- Accepted April 4, 2006
- Published online August 1, 2006.
- Mark J. Earley, MRCP,
- Dominic J.R. Abrams, MRCP,
- Simon C. Sporton, MD and
- Richard J. Schilling, MD, FRCP⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Richard J. Schilling, Cardiology Research Department, Dominion House, St. Bartholomew’s Hospital, London, EC1A 7BE United Kingdom.
Objectives The aim of this study was to validate noncontact mapping (NCM) in the left atrium (LA) during sinus rhythm and atrial fibrillation (AF).
Background Understanding the mechanisms of AF is crucial to the development of novel and effective treatments. Noncontact mapping records global electrical activation simultaneously and therefore has the potential to elucidate these mechanisms.
Methods Patients underwent catheter ablation of permanent AF guided by NCM. Virtual and contact unipolar electrograms were recorded simultaneously during sinus rhythm and AF from sites spanning the LA and their morphology, amplitude, and timing were compared. The impact of distance from the array to the endocardial surface and electrogram amplitude were analyzed.
Results A total of 22 patients age 52 ± 9 (mean ± SD) years were studied. During sinus rhythm, the median (range) morphology correlation and timing difference between contact and virtual atrial electrograms were 0.81 (0.27 to 0.98) and 4.2 (0 to 18.3) ms, respectively. These results were significantly worse than the corresponding far field individual ventricular electrograms; 0.91 (0.53 to 1.0) and 1.7 (0 to 18.3) ms (p < 0.001). For endocardial sites >40 mm from the array, the correlation was significantly worse than sites <40 mm: 0.73 (0.48 to 0.95) versus 0.87 (0.27 to 0.98) (p < 0.001). The correlation during AF was 0.72 (0.24 to 0.98), which deteriorated with increasing distance from the array. In the presence of adenosine induced atrioventricular block the correlation deteriorated 0.67 ± 0.16 versus 0.79 ± 0.11 (p < 0.001).
Conclusions Noncontact mapping can be performed in human LA; however, the accuracy of reconstructed electrograms is poor >40 mm from the center of the array, particularly during AF. Care must be taken interpreting isopotential maps if the entire endocardial surface of the LA is not close to the array.
Understanding the mechanisms of human atrial fibrillation (AF) is crucial to the development of effective catheter or surgical ablation strategies. Contact mapping, even with multipolar catheters, has limitations as a result of the chaotic and rapidly changing activation characteristic of AF. Much has been learned from animal models using multielectrode arrays attached to the epicardial surface of the heart (1) and optical mapping (2); however, these techniques cannot be easily applied in humans. Mapping of human AF has been limited to multielectrode arrays applied to small areas of the atrium during the nonphysiological conditions of cardiac surgery (3–5). Theoretically, noncontact mapping (NCM) is ideal for this purpose because it records simultaneously the endocardial activation of the entire chamber in which it is positioned; however, there has been limited validation of this technique for the mapping of atrial arrhythmias (6). The purpose of this study was to validate the use of NCM to map permanent AF in humans.
Data were collected from patients who were undergoing catheter ablation of permanent AF. All patients had been anticoagulated with warfarin for at least 6 weeks and underwent a transthoracic and transesophageal echocardiogram before the procedure to measure left atrial (LA) size and ensure that it was free of thrombus. All patients gave written consent before the procedure, and the study was approved by the City and North East London local research ethics committee.
Noncontact mapping previously has been described in detail (7). Two transseptal punctures were made to deliver a NCM array (Ensite Array, Endocardial Solutions Inc., St. Paul, Minnesota) and a mapping catheter (Navistar Thermocool, Biosense Webster, Diamond Bar, California) via sheaths into the LA. The NCM array was advanced toward the region of the LA appendage with the distal pigtail, which extends beyond the array, and aligned in the same sagittal plane as the left superior pulmonary vein (PV) ostium. Heparin was given by an intravenous bolus to maintain an activated clotting time >300 s throughout the procedure. The mapping catheter was moved around the LA under fluoroscopic guidance to allow reconstruction of the chamber on the NCM system, labeling the PV ostium, mitral annulus, septum, and appendage. Particular care was taken to ensure accurate geometry collection between the ipsilateral PVs and between the appendage and left PVs (Fig. 1).Before the ablation procedure and after restoration of sinus rhythm, the mapping/ablation catheter was moved to record electrograms around the right and left PVs the LA appendage, the roof, and the anterior and posterior LA walls. The PV electrograms were recorded at the sites of ablation, i.e., the ostium and vestibule of the vein.
Sinus rhythm was restored by catheter ablation and with additional internal cardioversion if required after ablation was completed. Ablation was delivered during the procedure in stages with radiofrequency energy applied to encircle the left and right PVs in ipsilateral pairs and create lines from the left inferior PV to mitral valve and along the roof of the LA between the PVs. Further ablation was performed at areas of persistent reentry identified by NCM.
Validation of reconstructed electrograms
A unipolar electrogram from the distal electrode of the mapping catheter was recorded using a ring electrode positioned on the proximal shaft of the array catheter in the inferior vena cava as a reference, identical to the reference used to create the virtual unipolar electrogram (VUE). The VUE at the point that the mapping catheter was in contact with the myocardium and the contact unipolar electrogram were both band filtered (1 to 100 Hz) and displayed. Segments of the simultaneous electrograms were exported as text files and processed using a specifically designed Excel (Microsoft, Redmond, Washington) spreadsheet macro. To validate the accuracy of the VUE, morphology was compared between the virtual and contact electrogram using a template matching algorithm, which produces a cross-correlation coefficient, C(k), indicating their similarity where X and Y are the corresponding amplitudes of all the points constituting the virtual and contact electrograms respectively. This formula has been used previously to compare electrograms recorded in the right atrium during AF (8) and in the ventricles (9,10).This algorithm determines the best match between the two electrograms across a time shift (k) of ±50 ms and the time shift at which correlation is highest is taken as the time difference between the electrograms. Time differences between electrograms were not measured during AF because of the chaotic and constantly varying electrograms.
A further comparison between the contact electrogram and VUE was made in SR by measuring their amplitude. A derivation of the amplitude was made by measuring the root mean squared (RMS) of all positive and negative points away from the isoelectric line that constituted the atrial electrogram. Both the absolute RMS and the ratio of VUE to contact electrogram RMS were analyzed with respect to anatomical location and distance from the array.
Thirty-second recordings of contact electrograms and VUEs were taken before and after ablation with the contact catheter located in the sites described in the previous text. Seven-second segments of these recordings were downloaded to calculate the morphology correlation coefficient at these sites. After cardioversion to sinus rhythm, individual atrial electrograms and the corresponding far field ventricular electrograms were recorded around the LA at the same sites recorded during AF.
The results were analyzed further to determine whether the amplitude of the electrogram, anatomical location in the LA, and the distance from or orientation to the array impacted on the correlation coefficient. There is a theoretical possibility that, because of to the oval shape of the array, electrogram reconstruction may be less reliable at polar compared with equatorial locations. To evaluate whether the ventricular component of the unipolar electrogram influenced correlation during AF, paired segments were analyzed before and during atrioventricular (AV) nodal block with adenosine, recorded at the same location and stage of the procedure (Fig. 2).
Normal distribution of continuous data were tested using a Kolmogorov-Smirnov test. The electrogram morphology correlation scores during AF were normally distributed but not during sinus rhythm. Continuous and normally distributed data are expressed as mean ± SD and an independent ttest used to test differences between two groups. Non-normally distributed data are expressed as median (range) and a Mann-Whitney Utest was used to test differences between two groups.
Data were collected from 22 patients (19 men, 3 women) ages 52 ± 9 years. They had AF of long duration (median 36 months, range 12 to 288 months) and had taken 3.5 ± 1.1 antiarrhythmic drugs. Their left atria (LA) were dilated with parasternal long axis diameter of 46 ± 7 mm.
Validation of reconstructed electrograms in sinus rhythm
A total of 154 sinus rhythm electrograms were compared, and the results are summarized in Table 1.The atrial electrogram morphology cross-correlation and absolute timing difference were 0.81 (0.27 to 0.98) and 4.2 (0 to 18.3) ms, respectively. These were significantly worse than found for the individual far field ventricular electrograms 0.91 (0.53 to 1.0) and 1.7 (0 to 18.3) ms (p < 0.001). Both the correlation and timing of the atrial electrogram deteriorated with increasing distance from the center of the array (Fig. 3).For endocardial sites >40 mm compared with <40 mm, the morphology correlation but not timing difference of the atrial electrogram was significantly worse: morphology correlation 0.73 (0.48 to 0.95) versus 0.87 (0.27 to 0.98) (p < 0.001) and timing 5.0 (0 to 18.3) ms versus 4.2 (0 to 15.8) ms (p = 0.09). This deterioration was not observed for the ventricular electrogram (Fig. 4).
The amplitude (RMS) of the contact electrogram was not dependent on the distance of the catheter from the array: ≤40 mm, 0.26 (0.09 to 0.73) versus >40 mm, 0.29 (0.09 to 0.77) (p = 0.45). However, a significant reduction in the amplitude of the VUE was observed: ≤40 mm, 0.24 (0.09 to 0.56) versus >40 mm, 0.17 (0.06 to 0.41) (p < 0.001) (Fig. 3) and, therefore, a reduction in the ratio of VUE to contact electrogram ratio ≤40 mm, 0.91 (0.27 to 2.3) versus >40 mm, 0.57 (0.13 to 2.4) (p < 0.001). Endocardial sites in which the amplitudes were closely matched, for example, VUE: contact amplitude ratio >0.9, were closer to the array (30 [15 to 54] mm vs. 34 [14 to 73] mm, p = 0.01) and had a better morphology correlation (0.86 [0.32 to 0.98] vs. 0.74 [0.21 to 0.96], p < 0.01). A low-amplitude VUE was associated with a poorer correlation; however, the lowest amplitude signals were those found at sites furthest from the center of the array (Fig. 3). The anatomical sites with the worst electrogram morphology correlation during SR were the LA appendage and left PVs, which again were those furthest from the array.
Validation of reconstructed electrograms in AF
A total of 125 AF electrogram segments were analyzed. The morphology correlation between contact and virtual electrograms was 0.72 (0.24 to 0.98); however, there was a deterioration in correlation with increasing distance from the center of the array (Fig. 5).For VUEs >40 mm from the array, the morphology correlation was significantly worse compared with those <40 mm (0.67 [0.22 to 0.96] vs. 0.75 [0.22 to 0.96], p = 0.01) (Fig. 5). Electrograms at polar sites had a significantly worse correlation than at equatorial sites (0.69 [0.23 to 0.98] vs. 0.77 [0.25 to 0.97], p = 0.009); however, polar sites were significantly further from the center of the array (36 [14 to 76] mm vs. 22 [14 to 58] mm, p < 0.001). Anatomical sites that had the worst correlation were the left PVs at 0.65 (0.25 to 0.98) and LA appendage at 0.66 (0.23 to 0.91); however, these sites also were the positions furthest from the array at 42 (14 to 63) mm and 36 (28 to 76) mm, respectively (Fig. 6).
Effect of adenosine on correlation during AF
For 21 separate 30 s episodes of AF, paired 7 s segments were analyzed during and after AV block induced by adenosine bolus. The number of ventricular complexes in the adenosine segments was 5 ± 2 compared with 12 ± 3 without adenosine. A significantly worse correlation was noted during AV block (0.67 ± 0.16 vs. 0.79 ± 0.11, p < 0.001). There also was a small reduction in the amplitude (RMS) of the contact (0.34 ± 0.19 vs. 0.4 ± 0.0, p = 0.001) and virtual (0.34 ± 0.15 vs. 0.29 ± 0.13, p = 0.06) unipolar electrogram. This small difference is likely to be attributable to the loss of the ventricular electrograms. The atrial cycle length, measured using the bipolar electrogram from the ablation catheter, did not change significantly (from 158 ± 23 ms to 151 ± 21 ms, p = 0.07) during an adenosine bolus.
During sinus rhythm morphology, correlation and timing between atrial contact and VUEs were determined strongly by the distance of the catheter from the center of the array. At distances greater than 40 mm from the center of the array (typically the LA appendage or left PVs), both the correlation and timing were worse and the amplitude of the atrial VUE was lower. Far field ventricular electrograms, however, correlated and timed much more closely and remained accurate even at distances >40 mm from the array suggesting that the amplitude of the signal is important. During AF, electrograms did not correlate as well, and AV block induced with adenosine caused a reduction in this correlation, confirming that the ventricular component of the unipolar electrogram exerts a larger and more consistent influence on morphology correlation than the low-amplitude fibrillating atrial electrogram. During AF at endocardial sites <40 mm from the array, the correlation was only moderate (median, 0.75) and caution must be applied when interpreting AF electrograms recorded even further from the array.
Previous validation studies of electrogram morphology
Noncontact mapping has been validated predominantly in the left ventricle (LV). Thiagalingam et al. (11) demonstrated a close correlation between contact and noncontact electrograms (0.87 to 0.89) in the ovine LV during sinus rhythm and pacing. These findings were almost identical to those of Schilling et al. (7) who found a cross correlation of 0.87 during sinus rhythm in the human LV. However, both groups recognized a rapid deterioration of morphology correlation for endocardial points >40 mm (11) or >34 mm (7) from the center of the array. A similar morphology correlation was demonstrated during ventricular tachycardia (12). However, it should be noted that in all these studies, validation was performed on ventricular electrograms with a consistent morphology and that such an approach is not feasible for atrial or ventricular fibrillation. In a recent canine LV study, Everett et al. (13) demonstrated highly correlated electrograms in sinus rhythm (0.94) and ventricular fibrillation (0.90); however, the authors did not state over what period of time the electrograms were analyzed.
Considerably less validation has been done in the atrium, and it has been limited to the right atrium (RA). Kadish et al. (6) showed in the canine RA that atrial electrograms from randomly selected 2 s windows during sinus rhythm, pacing or atrial flutter all correlated very well (≥0.8). During AF, they also demonstrated good correlation (0.81); however, again this deteriorated if electrograms were analyzed further than 40 mm from the array center. Schilling et al. (7) demonstrated a mean cross correlation of 0.72 for AF electrograms in the human RA analyzing segments of varying length.
This study differs from previous validation studies in several ways. First, this study is the only one performed in vivo in the LA. Second, we examined both sinus rhythm and AF electrograms and applied the cross-correlation algorithm over a consistent 7 s window in AF. This method prevents selection bias of electrograms that may correlate well over short intervals but not overall. Our findings are consistent with the limited available human data (8) that AF electrograms reconstructed by NCM in the LA have an acceptable correlation if recorded within 40 mm of the center of the balloon but not as good as for regular rhythms recorded in the LV. A further concern regarding the accuracy of reconstructed electrograms is that we have shown that when fewer ventricular complexes are present that the correlation significantly deteriorates.
NCM of AF
To date, there have been no published studies of NCM to investigate persistent or permanent AF in the LA. Schilling et al. (8) demonstrated reentry circuits in the RA with a wide variety in the number and complexity of the wavefronts. Two studies have investigated the onset of episodes of paroxysmal AF in the LA using NCM. Weber et al. (14) identified that repetitive firing from trigger zones (principally PVs) maintained AF in some cases but also in others rapid break up and disorganization of wavefronts particularly on the posterior wall could maintain AF without the need for a repetitive firing focus. Markides et al. (15) showed the initiation of AF by wavefronts from focal triggers breaking up and forming reentry circuits over a line of functional conduction block found in the LA. Evidence exists from other studies that the posterior wall and LA appendage may support reentry circuits that could maintain AF even in models of established AF (2). Noncontact mapping should be ideally suited to investigate this hypothesis in humans; however, when the array is introduced via the interatrial septum in a large chamber, care would need to be taken to ensure the LA appendage is within 40 mm of the center of the array. In our study of patients with permanent AF, the transthoracic echocardiographic parasternal long axis LA diameter was only 46 ± 7 mm. However, because of the eccentric geometry of the LA, the septum to lateral wall diameter is much greater than this and therefore the tip of the LA appendage and left superior pulmonary vein may be >40 mm from the center of the array.
An ablation catheter with its relatively large distal electrode is not the ideal tool to record a unipolar electrogram on the atrial endocardium. Thiagalingam et al. (11) demonstrated that VUEs represent a sum of transmural activation; however, for contact electrograms, we cannot be sure of the exact degree of catheter contact, pressure applied at the tip, and orientation of the irrigated 3.5-mm tip to the endocardium, which may all alter the morphology of the contact but not the noncontact unipolar electrogram. There was no attempt in this study to stratify electrograms with regard to areas of endocardial scarring or degree of fractionation during AF, which may influence the accuracy of the reconstructed VUE.
Reconstructed VUEs recorded in the LA become less accurate when the distance from the center of the array is >40 mm. During AF, this correlation is worse then in sinus rhythm, and VUEs may not be sufficiently accurate to interpret unless recorded close to the array. The LA appendage and left PVs have been recognized as areas that might promote reentry maintaining AF; however, caution must be applied when analyzing isopotential maps of AF with NCM to ensure that the entire virtual endocardium is within an acceptable distance form the center of the array.
Drs. Earley and Abrams were funded by grants from the British Heart Foundation charity. Dr. Schilling is on the scientific advisory board of Biosense-Webster and the Speakers’ Bureau of St. Jude Medical (formerly Endocardial Solutions).
- Abbreviations and Acronyms
- atrial fibrillation
- left atrial/atrium
- left ventricle
- noncontact mapping
- pulmonary vein
- right atrium
- root mean square
- virtual unipolar electrogram
- Received November 6, 2005.
- Revision received March 20, 2006.
- Accepted April 4, 2006.
- American College of Cardiology Foundation
- Schuessler R.B.,
- Grayson T.M.,
- Bromberg B.I.,
- Cox J.L.,
- Boineau J.P.
- Jalife J.,
- Berenfeld O.,
- Mansour M.
- Konings K.T.,
- Kirchhof C.J.,
- Smeets J.R.,
- Wellens H.J.,
- Penn O.C.,
- Allessie M.A.
- Sahadevan J.,
- Ryu K.,
- Peltz L.,
- et al.
- Kadish A.,
- Hauck J.,
- Pederson B.,
- Beatty G.,
- Gornick C.
- Schilling R.J.,
- Peters N.S.,
- Davies D.W.
- Schilling R.J.,
- Kadish A.H.,
- Peters N.S.,
- Goldberger J.,
- Davies D.W.
- Gepstein L.,
- Hayam G.,
- Ben Haim S.A.
- Gornick C.C.,
- Adler S.W.,
- Pederson B.,
- Hauck J.,
- Budd J.,
- Schweitzer J.
- Schilling R.J.,
- Peters N.S.,
- Davies D.W.
- Everett T.H.,
- Wilson E.E.,
- Foreman S.,
- Olgin J.E.
- Markides V.,
- Schilling R.J.,
- Ho S.Y.,
- Chow A.W.,
- Davies D.W.,
- Peters N.S.