Introduction Atrial fibrillation (AF) can be effectively treated by catheter ablation. Typically, ablation targets the ostia of the pulmonary

Transcription

1 Initial experience with magnetic resonance imaging of atrial scar and co-registration with electroanatomic voltage mapping during atrial fibrillation: Success and limitations David D. Spragg, MD, FHRS, Irfan Khurram, MD, Stefan L. Zimmerman, MD, Hirad Yarmohammadi, MD, MPH, Bernie Barcelon, RT, Matthew Needleman, MD, David Edwards, MD, Joseph E. Marine, MD, FHRS, Hugh Calkins, MD, FHRS, Saman Nazarian, MD, FHRS From the Division of Cardiology, Johns Hopkins Hospital, Baltimore, Maryland. BACKGROUND Ablation for atrial fibrillation (AF) frequently requires multiple procedures to achieve durable restoration of sinus rhythm. Early studies have suggested that delayed enhancement magnetic resonance imaging (DE-MRI) of the left atrium (LA) can assist in performing repeat ablation procedures. OBJECTIVE The purpose of this study was to investigate the utility of DE-MRI in delineating regions of LA low voltage and PV reconnection sites in patients undergoing repeat PV isolation for recurrent AF. METHODS We enrolled 10 patients undergoing repeat ablation for AF recurrence to undergo preprocedural DE-MRI of the LA in conjunction with high-density voltage mapping (4100 sites) of the LA during the ablation procedure. LA wall regions with hyperenhancement were segmented from DE-MRI images and retrospectively co-registered with the electroanatomic LA map. The association between scar on DE-MRI images and low-voltage regions of the LA was assessed, as was the association between scar gaps and electrogram-determined PV reconnection sites. RESULTS Ten patients underwent successful DE-MRI imaging and repeat AF ablation without complication. In all 10 patients, the majority of PVs were found to have regained electrical continuity with the LA (30/37 PVs electrically active); all patients underwent successful reisolation of all PVs using standard ablation techniques. There was a significant association between scar identified by DE- MRI and low-voltage regions of the LA ( mv in scar regions; generalized estimating equations model clustered by patient, P o.001). However, there was no association between scar gaps and PV reconnection sites. CONCLUSION We demonstrate the co-registration of DE-MRI scar imaging and electroanatomic LA mapping, with agreement between regions of scar on DE-MRI and low voltage by mapping. However, at our center, this technique did not provide accurate information on the location of PV reconnection sites in patients undergoing repeat ablation for AF. KEYWORDS Ablation; Atrial fibrillation; Magnetic resonance imaging ABBREVIATIONS AF ¼ atrial fibrillation; CI ¼ confidence interval; DE-MRI ¼ delayed enhancement magnetic resonance imaging; LA ¼ left atrium; PV ¼ pulmonary vein (Heart Rhythm 2012;9: ) I 2012 The Heart Rhythm Society. All rights reserved. Introduction Atrial fibrillation (AF) can be effectively treated by catheter ablation. Typically, ablation targets the ostia of the pulmonary Dr. Nazarian has received honoraria for lectures (not Speakers Bureau) from St. Jude Medical, Biotronik, and Boston Scientific; is on the MRI advisory panel (unpaid) for Medtronic; is a scientific advisor to Biosense Webster; and is funded by a National Heart, Lung, and Blood Institute Career Development Award (K23HL089333). Dr. Spragg has received honoraria for lectures (not Speakers Bureau) from Medtronic. Dr. Yarmohammadi was supported by the Norbert and Louise Grunwald Arrhythmia Research fund. Dr. Khurram was supported by the Dr. Francis Chiaramonte, MD Foundation. Address reprint requests and correspondence: Dr. David Spragg, Division of Cardiology, Johns Hopkins and Johns Hopkins Bayview Medical Center, 4940 Eastern Avenue, Baltimore, MD address: dspragg1@jhmi.edu. veins (PVs), with the goal of durable electrical PV isolation. 1 Frequently, several ablation attempts are required to achieve lasting PV isolation and long-term restoration of normal sinus rhythm. Although ablation for AF has a safety profile that is improving, considerable issues with complications and patient safety persist, and strategies to maximize ablation efficacy and minimize the need for further ablation attempts are of obvious importance. 2 Recently, ablation strategies using delayed enhancement magnetic resonance imaging (DE-MRI) to delineate scar in the left atrium (LA) have been described. 3 Scar burden has been used to predict ablation efficacy, 3,4 stroke risk, 5 and, more recently, the site of PV reconnection in patients who have undergone initial AF ablation but suffered arrhythmia recurrence. 6,7 Thus, this form of imaging holds promise as a /$-see front matter B 2012 The Heart Rhythm Society. All rights reserved.

2 2004 Heart Rhythm, Vol 9, No 12, December 2012 tool for maximizing ablation effectiveness, particularly in patients who have undergone prior ablation and have an incomplete lesion set already present in the LA. Although scar imaging of the LA in patients with AF appears promising, the number of centers reporting successful application of this technique to guide clinical strategies remains low. We sought to determine whether, given our experience with imaging both native and ablation-induced scar in the left ventricle, 8 we could use DE-MRI images of the LA to assist in repeat catheter ablation for recurrent AF. In the current study, we report our initial experience with coregistration of DE-MRI images and electroanatomic mapping in a series of 10 patients referred for repeat catheter ablation of AF. Methods Patients Ten patients were enrolled in the current study. All patients had undergone initial ablation for AF at our institution and, because of recurrent AF, were scheduled to undergo repeat AF ablation. All patients consented to the investigation, and the ablation and imaging protocols were approved by the Johns Hopkins Institutional Review Board. MRI acquisition and analysis DE-MRI imaging was performed prior to repeat catheter ablation (5 3 days preprocedure). Images were acquired using a 1.5-Tesla Avanto scanner (Siemens, Erlangen, Germany). A 6-channel phased-array torso coil was used in combination with the spine matrix coil for a total of 12 independent receiver channels. A contrast enhanced 3-dimensional fast low angle shot MR angiography sequence [repetition time (TR) 2.5 ms, echo time (TE) 0.97 ms, field of view 370 mm, flip angle 251, in-plane resolution mm; slice thickness 1.2 mm)] was obtained immediately following administration of 0.1 mmol/kg of intravenous contrast (gadopentetate dimeglumine; Bayer Healthcare Pharmaceuticals, Montville, NJ) to define LA and PV anatomy and enable accurate image electroanatomic map co-registration. An additional 0.1 mmol/kg of contrast was injected for a total of 0.2 mmol/kg, and DE-MRI scans were acquired 15 minutes postinjection. DE-MRI images were obtained using a 3-dimensional inversion recovery prepared fast spoiled gradient recalled, respiratory navigated, ECG gated, and fat suppressed sequence (TR ms, TE 1.5 ms, field of view 340 mm, flip angle 101, in-plane resolution mm, slice thickness 2.0 mm). The respiratory navigator was positioned on the right hemidiaphragm to limit respiratory gating artifacts. Trigger time for images was selected to correspond with LA diastole based on 4-chamber cine images. The inversion time was identified for optimal nulling of the myocardium with a look-locker inversion recovery (TI scout) scan and ranged between 280 and 300 ms. Magnetic resonance angiography and DE-MRI images were analyzed and segmented using ITK-SNAP software ( 9 Areas with visual hyperenhancement in the LA wall were segmented from DE-MRI images and coregistered with the segmented angiogram image of the LA chamber to produce a single LA angiogram scar image file using previously validated custom software (Volley; Johns Hopkins, Baltimore, MD). 8 Electroanatomic mapping and catheter ablation At the time of ablation, patients underwent double transseptal puncture using standard techniques. A detailed voltage map of the LA was created using a 3.5-mm-tip, irrigated ablation catheter (2-mm interelectrode spacing; Thermo- Cool; Biosense Webster, Diamond Bar, CA) in conjunction with an electroanatomic mapping system (CARTO Merge; Biosense Webster). The entire LA, including the posterior wall, PV ostia, anterior regions, and LA appendage ostium, were mapped. During mapping the operator was blinded to scar images. Upon completion of the voltage data acquisition, the voltage map was registered to the magnetic resonance angiogram of the LA using posterior PV ostia LA junctions and posterior wall points as landmarks. After completion of electroanatomic mapping, a 20-pole Lasso catheter was introduced into the LA. PVs were assayed with the Lasso catheter, and reconnection sites were identified by site of earliest ostial activation. These sites were localized on the CARTO map, and the position of the Lasso catheter was recorded for offline analysis. Table 1 Patient no. Patient characteristics Age (years) Interval period (month) Atrial fibrillation type CHADS 2 score Hypertension DM CVA Persistent.1 Yes No No Paroxysmal.0 No No No Paroxysmal.3 Yes No Yes Paroxysmal.2 Yes Yes No Paroxysmal.2 Yes Yes No Paroxysmal.1 Yes No No Persistent.1 Yes No No Paroxysmal.0 No No No Persistent.3 Yes No Yes Paroxysmal.0 No No No DM ¼ Diabetes CVA ¼ Stroke

3 Spragg et al Magnetic Resonance Imaging of Left Atrial Scar 2005 In all 10 patients, the PVs were reisolated in standard fashion using a wide circumferential approach followed by Lasso-targeted, segmental ablation along the circumferential ablation line. Upon reisolation of the PVs, catheters were removed and the patients underwent follow-up in the usual manner. Correlation of scar, voltage, and PV reconnection sites Correlation of scar location, as assessed by DE-MRI, and LA voltage was performed offline after the ablation procedure. Each acquired electrogram was determined to be either within or outside of regions of visual hyperenhancing LA scar (with determination of co-localization made while blinded to electrogram voltage). Localization of PV reconnection sites was compared to regions of scar during the same offline assessment. Statistical analysis Continuous variables are presented as mean SD. For analysis of pooled data, the overall mean atrial voltage amplitude in areas with delayed enhancement was compared with the overall mean atrial voltage in areas without delayed enhancement using the Student t-test. To account for intrapatient clustering of data and interpatient differences, a generalized estimating equations model for binary data with logit link function, clustered by patient, was used to examine the association of scar on MRI with atrial electrogram amplitude measures. A working autoregressive correlation structure was used because closely spaced voltage observations are more highly correlated than 2 observations spread further apart. Statistical analyses were performed using STATA (version 12; StataCorp, College Station, TX). Results Patients All 10 patients had undergone previous ablation at Johns Hopkins, with an average time between procedures of months. The clinical and procedural characteristics of the 10 patients are given in Table 1. All patients tolerated MRI and repeat PV isolation procedures without complication and have been followed up for 6 4 months post-redo procedure. Routine postprocedure evaluation has consisted of a follow-up ECG and office visit at 3 months. No patients have experienced recurrence of symptomatic AF since their second procedure. PV reconnection and repeat ablation All patients underwent a successful repeat PV isolation. In the 10 patients, 30 of 37 PVs (including 3 common left-sided PVs) were found to have electrically reconnected to the body of the LA; the 7 persistently isolated veins were all rightsided PVs. The sites of PV reconnection could be accurately determined in 18 of the 30 reconnected veins. Sites of reconnection were the left superior PV roof (1), the left superior PV posterior wall (4), the left inferior PV left Figure 1 A: Electroanatomic voltage map in the posteroanterior projection of a patient undergoing redo pulmonary vein isolation. Low voltage (o0.6 mv) is shown in red; healthy tissues is shown in purple.b: Delayed enhancement magnetic resonance imaging derived scar map of the same patient showing the same projection of the left atrium. Scar is shown in purple; healthy tissue is shown in red. A white border has been manually traced around the dominant region of scar on the posterior wall of the left atrium. C: Combined projection of scar and voltage data, in the same posteroanterior orientation. The scar border (white tracing) has been superimposed on the image to show full extent of the scar (rather than showing multiple projections of the same image with different voltage and scar transparencies). The site of the left inferior pulmonary vein reconnection is shown (yellow dot; see Figure 4). superior isthmus (4), the left inferior PV inferior border (1), the right superior PV roof (1), the right superior PV right inferior PV isthmus (2), the right superior PV posterior wall

4 2006 Heart Rhythm, Vol 9, No 12, December 2012 (2), and the right inferior PV inferior border (3). In the remaining PVs, there was either no record of the Lasso position that could be reliably correlated with PV electrograms or there was durable isolation from the previous procedure. PV isolation was confirmed by entrance block following a 60-minute waiting period, and the patients underwent follow-up in the usual fashion. DE-MRI scar imaging and voltage correlation Patients underwent detailed voltage mapping before image registration with the preacquired magnetic resonance angiogram (90 24 points sampled per patient; total of 893 points). An example of a co-registered scar and voltage electroanatomic map from one of the patients in our study is shown in Figure 1. Voltage mapping shows a broad region of low voltage across the posterior wall of the LA (extending into the ostia of the PVs), with a region of tissue with normal voltage at the posterior roof (Figure 1A). DE-MRI imaging of LA scar localizes predominantly to regions of low LA voltage (Figure 1B and 1C). DE-MRI imaging did not reveal scarring within the PV ostia, suggesting that the low voltages seen in these regions were due to mapping within the PVs. The overall mean atrial voltage amplitude for the 893 measures in 10 patients was mv (0.40 mv between-subject standard deviation and 0.94 mv withinsubject standard deviation). Differences in LA voltage in regions of scar vs healthy tissue, as assessed by DE-MRI, are shown for each of the 10 patients (Figure 2A) and for the group (Figure 2B). In pooled analysis, the mean voltage in 565 points identified as scar by DE-MRI [ mv (0.26 mv between-subject standard deviation and 0.56 mv withinsubject standard deviation)] was lower than the mean voltage in 328 points identified as normal by DE-MRI [ mv (0.47 mv between-subject standard deviation and 1.17 mv within-subject standard deviation), P o.001]. In a generalized estimating equations model, accounting for clustering of data by individual patients, identification of scar by DE-MRI was significantly associated with reduced local bipolar voltage ( mv, P o.001). We categorized LA bipolar electrogram voltage r0.5 mv as low and 40.5 mv as normal. In regions of normal voltage (40.5 mv), there were essentially no occurrences of fractionated, high-frequency, multipeak electrograms. Such electrograms were frequently seen, however, in regions displaying low (o0.5 mv) bipolar electrogram amplitude. The sensitivity and positive predictive value of MRI delayed enhancement regions identifying areas of low voltage were 0.84 [95% confidence interval (CI) ] and 0.80 (95% CI ), respectively (Figure 3). Figure 2 Box-plots of voltage amplitude in regions of normal and scar myocardium identified by delayed enhancement magnetic resonance imaging (MRI). A: Data by patient. B: Pooled data across all patients. Box-plots display the median and the 25th to 75th percentile range (center line and solid box), the lower and upper adjacent values (whiskers), and outlier data points (dots).

5 Spragg et al Magnetic Resonance Imaging of Left Atrial Scar 2007 Normal Voltage (>0.5 mv) Low Voltage ( 0.5 mv) Total No Scar 239 (27%) 89 (10%) 328 Scar 114 (13%) 451 (50%) 565 Total Figure 3 Data from contingency table analysis of scar/normal tissue by delayed enhancement magnetic resonance imaging (MRI) vs low/normal voltage by electroanatomic mapping demonstrating significant association between scar presence or absence with low or normal voltages, respectively. The specificity and negative predictive value of absence of MRI delayed enhancement identifying regions of normal voltage were 0.68 (95% CI ) and 0.73 (95% CI ), respectively. DE-MRI scar imaging and PV reconnection sites There was no appreciable identification of PV connections by gaps in the scar images determined by DE-MRI. Completely circumferential lesions were only seen in 2 of 37 PVs. PV reconnection sites were frequently seen in regions of scar (Figure 4). Of the 18 PV reconnection sites identified, 13 (72%) were found to be within imaged scar, whereas 5 (28%) were away from imaged scar. Discussion In the current report, we present our initial experience with co-registration of electroanatomic data and DE-MRI scar mapping in a series of 10 patients. In each patient we performed preprocedural DE-MRI assessment of LA scar as well as detailed LA voltage mapping prior to ablation. Electroanatomic and scar data were co-registered, and the associations between (1) scar distribution and areas of low LA voltage and (2) scar distribution and PV reconnection sites were determined. We believe that several important conclusions can be drawn from these early data. First, there appears to be a reliable association between LA scar imaged by MRI and low-voltage areas of the LA by endocardial mapping. Second, the resolution of scar delineation by current DE- MRI techniques at our center was not sufficient to predict where PV reconnection occurred. Finally, we describe what we believe to be a novel technique for displaying scar, anatomic imaging, and LA voltage mapping in the same image. Scar, voltage, and gaps We chose patients undergoing repeat catheter ablation for a number of reasons. These patients are anticipated to have a significant burden of scar, both from gradual native fibrosis and from the application of lesion sets during first attempt at ablation. Our study of the scar burden and distribution in these patients is interesting in several respects. We could not distinguish between native scar and scar created by application of radiofrequency energy. In our imaging series, we saw fibrosis in patterns that appeared to correlate with previous lesion application, but this scar was contiguous with other large areas of scar not anticipated to be caused by radiofrequency energy, based on distribution in the LA in comparison with the previously applied lesion set. The implications for this are significant. If DE-MRI is to be used as a tool to assess the complete circumferential isolation of PV ostia, it is likely important whether that circumferential scar is secondary to aging and gradual fibrosis or to ablation. It is not intuitive that atrial myocardium scarred by gradual fibrosis (which may enhance on DE-MRI) is a reliable barrier to electrical propagation. Indeed, scarring and attendant slowing (but not block) of conduction throughout the LA are postulated to be central components of AF maintenance. 10,11 Second, we anticipated in this patient population that there would be electrical reconnection between the PVs and the LA body, reflecting gaps in the initial lesion set and resulting in recurrent AF. Although other centers have recently described cases of DE-MRI demonstrating ablation line gaps correlating with PV reconnection sites, 6,7 we did not find this to be reliable using our current imaging protocols. Rather, we found that in many PVs, gap sites were found in regions of scar (as delineated by MRI). This likely reflects the presence of small bundles of viable myocardium, too thin to detect with current DE-MRI resolution. This is in contrast with our results in MRI-based targeting of ventricular tachycardia circuits in the left ventricle and demonstrates the difficulties of scar imaging in the thin-walled LA. 8 Although the positive and negative predictive values for DE-MRI in predicting regions of low and normal LA voltage were good in the current study, they were nevertheless imperfect. We suspect that several factors, including limited image resolution, motion or flow artifacts, volume averaging in DE-MRI imaging of fibrosis, imperfect LA voltage mapping due to variable catheter contact or angle, and imprecise anatomic and electrical map registration, may conspire to limit diagnostic accuracy. An additional contribution from the current study is the development of a method for co-registration and concurrent display of scar, voltage, and anatomy in the same projection. In future studies of MRI-assisted AF ablation, this sort of simultaneous display of voltages and scar should provide real-time assessment of whether that scar (as determined by MRI) represents electrically inert tissue or whether redo ablation is needed within a region of scar because of residual, electrically active tissue.

6 2008 Heart Rhythm, Vol 9, No 12, December 2012 Figure 4 A: Catheter-projection view of the Lasso in the ostium of the left inferior pulmonary vein, with superimposition of subsequently applied repeat ablation lesions. Site of pulmonary vein reconnection is shown (yellow dot) and is located in the center of the scar region (see Figures 1B and 1C). B: Lasso electrograms showing earliest activation at Ls19,20 and Ls5,6 (adjacent poles in this particular deployment of the Lasso). Study limitations This is a small study, and greater numbers of patients are needed to verify our results. Results may be limited by the possibility of positional errors when registering electroanatomic maps to corresponding DE-MRI images, although we believe that these inaccuracies are of limited magnitude ( mm). 12 Additionally, results may be limited by lower density of sampled electroanatomic map points in the anterior LA, the size of the electrode tip (3.5 mm), interelectrode distance, variable angle and force of contact, and inherent mismatches in the resolution of DE-MRI vs electrogram mapping. Finally, scar extent was determined by visual assessment of areas of hyperenhancement. Although less objective than computational methods, we believe that this methodology more accurately reflects the clinical utility of scar imaging in this scenario. Conclusion DE-MRI imaging can be used to assess scar distribution in the LA, with reasonable correlation between scar and lowvoltage areas delineated by endocardial mapping. However,

7 Spragg et al Magnetic Resonance Imaging of Left Atrial Scar 2009 the resolution of scar mapping in our hands was not sufficient to reliably predict ablation gaps and PV reconnection sites. Changes in image processing may help with increasing image definition. Finally, we provide an example of simultaneously displaying voltage and scar distributions on the same anatomic projection. This sort of display may be useful in future, real-time MRI-based ablation. References 1. HRS/EHRA/ECAS Expert Consensus Statement on Catheter and Surgical Ablation of Atrial Fibrillation: Recommendations for Personnel, Policy, Procedures and Follow-Up A report of the Heart Rhythm Society (HRS) Task Force on Catheter and Surgical Ablation of Atrial Fibrillation Developed in partnership with the European Heart Rhythm Association (EHRA) and the European Cardiac Arrhythmia Society (ECAS); in collaboration with the American College of Cardiology (ACC), American Heart Association (AHA), and the Society of Thoracic Surgeons (STS). Endorsed and Approved by the governing bodies of the American College of Cardiology, the American Heart Association, the European Cardiac Arrhythmia Society, the European Heart Rhythm Association, the Society of Thoracic Surgeons, and the Heart Rhythm Society. Europace 2007;9: Hoyt H, Bhonsale A, Chilukuri K, et al. Complications arising from catheter ablation of atrial fibrillation: temporal trends and predictors. Heart Rhythm 2011;8: Oakes RS, Badger TJ, Kholmovski E, et al. Detection and quantification of low voltage left atrial tissue using delayed enhancement MRI in patients with atrial fibrillation. Circulation 2009;119: McGann CJ, Kholmovski EG, Oakes RS, et al. New magnetic resonance imaging-based method for defining the extent of left atrial wall injury after the ablation of atrial fibrillation. J Am Coll Cardiol 2008;52: Daccarett M, Badger TJ, Akoum N, et al. Association of left atrial fibrosis detected by delayed- enhancement magnetic resonance imaging and the risk of stroke in patients with atrial fibrillation. J Am Coll Cardiol 2011;57: Reddy VY, Schmidt EJ, Holmvang G, et al. Arrhythmia recurrence after atrial fibrillation ablation: can magnetic resonance imaging identify gaps in atrial ablation lines? J Cardiovasc Electrophysiol 2008;19: Badger TJ, Daccarett M, Akoum NW, et al. Evaluation of left atrial lesions after initial and repeat atrial fibrillation ablation: lessons learned from delayedenhancement MRI in repeat ablation procedures. Circ Arrhythm Electrophysiol 2010;3: Estner HL, Zviman MM, Herzka D, et al. The critical isthmus sites of ischemic ventricular tachycardia are in zones of tissue heterogeneity, visualized by magnetic resonance imaging. Heart Rhythm 2011;8: Yushkevich PA, Piven J, Hazlett HC, et al. User-guided 3D active contour segmentation of anatomical structures: significantly improved efficiency and reliability. Neuroimage 2006;31: Benjamin EJ, Chen PS, Bild DE, et al. Prevention of atrial fibrillation: report from a national heart, lung, and blood institute workshop. Circulation 2009;119: Kirchhof P, Lip GY, Van Gelder IC, et al. Comprehensive risk reduction in patients with atrial fibrillation: emerging diagnostic and therapeutic options. Executive summary of the report from the 3rd AFNET/EHRA consensus conference. Thromb Haemost 2011;106: Dong J, Calkins H, Solomon SB, et al. Integrated electroanatomic mapping with three- dimensional computed tomographic images for real-time guided ablations. Circulation 2006;113: