- Email: [email protected]
Comparison of computed tomography imaging with intraprocedural contrast esophagram: Implications for catheter ablation of atrial fibrillation
Comparison of computed tomography imaging with intraprocedural contrast esophagram: Implications for catheter ablation of atrial fibrillation Emile G. Daoud, MD, John D. Hummel, MD, Mahmoud Houmsse, MD, David T. Hart, MD, Raul Weiss, MD, Zhenguo Liu, MD, Ralph Augostini, MD, Steven Kalbfleisch, MD, Macy C. Smith, MD, Rohit Mehta, MD, Ashish Gangasani, MD, Subha V. Raman, MD From the Department of Medicine, Division of Cardiology, Richard M. Ross Heart Hospital, Ohio State University Medical Center, Columbus, Ohio. BACKGROUND Computed tomography (CT) has been used to localize the esophagus before radiofrequency ablation (RFA) of atrial fibrillation (AF). OBJECTIVE The purpose of this study was to compare esophageal imaging by CT versus esophagram. METHODS CT imaging of the left atrium was performed in 57 patients 1 week before RFA and was imported into the CARTO mapping system. The electrophysiologist created a virtual shell of the left atrium and pulmonary veins (PVs) that was merged with the CT image; however, the CT-defined location of the esophagus was not displayed. The patient was then given 10 mL of oral contrast. Using fluoroscopy, an electroanatomic catheter tagged the esophageal borders outlined by esophagram. The CT-defined esophagus was then imported, and the borders were tagged on the merged map. In this manner, the esophagus borders by esophagram versus those by CT were compared.
Introduction A common strategy for radiofrequency ablation (RFA) of atrial fibrillation (AF) is to deliver radiofrequency lesions in a circumferential manner around the pulmonary veins (PVs). A rare but often fatal complication of this strategy is an atrioesophageal fistula.1,2 This complication is due to delivery of RF current in a region of the left atrium that is in close proximity to the esophagus. Accurate identification of the course of the esophagus during RFA of AF is an important tool to minimize the risk of thermal injury. Previous studies3– 8 have suggested that computed tomography (CT) imaging of the esophagus performed before the procedure accurately predicts the location of the esophagus at the time of the procedure. Other studies have defined the esophagus position by use of an electroanatomic catheter.9,10 In addition, studies have noted movement of the
RESULTS The maximum diameter of the esophagus by esophagram versus CT was not different (16.3 ⫾ 3.4 vs. 16.5 ⫾ 3.1 mm; P ⫽ .7). The esophagus was near the left PVs in 34 (62%), center in 13 (24%), and near the right PVs in eight (15%) patients. There was concordance between CT and esophagram in 48 of 55 patients (87%; P ⫽ .2). Ye, in 21 (44%) of 48 patients with concordant location, the CT-defined esophageal borders were separated from the esophagram-defined borders by ⱖ50% of the esophagus diameter. CONCLUSIONS Reliance on remotely acquired CT images does not ensure adequate intraprocedural localization of the esophagus or enhance recognition of esophageal motility. KEYWORDS Esophagus; Atrial fibrillation; Catheter ablation; Pulmonary vein; Computed tomography (Heart Rhythm 2008;5:975–980) © 2008 Heart Rhythm Society. All rights reserved.
esophagus.11 The purpose of this study was to assess the accuracy of a preprocedural CT scan in predicting the location of the esophagus at the time of the RFA procedure and to assess the extent of esophageal mobility during the procedure.
Methods Study population The subjects of this study were 57 patients undergoing curative RFA of symptomatic paroxysmal, persistent, or long-standing persistent AF. Inclusion criteria were completion of CT imaging of the chest within 1 week of planned RFA of AF and adequate imaging of the esophagus by contrast at the time of RFA procedure. Clinical characteristics of the study population are summarized in Table 1. The Institutional Review Board approved the study protocol.
CT imaging Address reprint requests and correspondence: Emile Daoud, M.D., DHLRI, 473 West 12th Avenue, Suite 200, Columbus, OH 43210-1252. E-mail address: [email protected]. (Received February 1, 2008; Accepted March 30, 2008.)
All patients underwent electrocardiographically gated CT angiography of the heart within 1 week before the ablation procedure. CT exams were performed on a 64-slice (SOMATOM Sensation64, Siemens Medical Solutions, Inc.,
1547-5271/$ -see front matter © 2008 Heart Rhythm Society. All rights reserved.
doi:10.1016/j.hrthm.2008.03.058
976 Table 1
Heart Rhythm, Vol 5, No 7, July 2008 Clinical characteristics of patient population (n ⫽ 57)
Parameter
Value
Male gender Age, years Type of AF (%): Paroxysmal Persistent Long-standing persistent Antiarrhythmic drug trials Left ventricular ejection fraction Left atrial diameter, mm Structural heart disease (%): None Hypertensive Ischemic
38 57 ⫾ 8 21 (37) 23 (40) 13 (23) 2.1 ⫾ 0.5 0.42 ⫾ 0.11 48.5 ⫾ 6.4 30 (53) 13 (23) 9 (16)
Forchheim, Germany) scanner with about 5 mL of oral barium paste administered just before imaging. Gantry rotation speed was 330 –370 ms per rotation based on heart rate, and tube voltage was 120 Kv; electrocardiogram dose modulation was applied in all studies to reduce tube current by 80% during the systolic phases of the cardiac cycle. Image data were acquired with inspiratory breath hold during injection of 80 mL of contrast agent at 5 mL/s followed by injection of 20 mL saline flush at 5 mL/s, with timing optimized using bolus tracking and a Hounsfield unit threshold of 100 units set in the left atrium. Overlapping 1.25-mm axial images stored in DICOM format were reconstructed at end-diastole. Motion artifact precluded adequate definition of the esophagus in two (3.5%) of 57 patients. The results, therefore, include the remaining 55 patients.
Merging of CT and three-dimensional map Before the ablation procedure, importation into the mapping system and segmentation of the CT image data was performed using commercially available software (CARTO Merge, Biosense-Webster, Diamond Bar, CA).12–15 With this software, user-directed seed points followed by signal intensity-based segmentation allowed delineation of the following structures: esophagus, pulmonary artery, aorta, right atrium, right ventricle, left ventricle, left atrium, and PVs. The left atrium and PVs were exported into the real time mapping system. The electrophysiologist was blinded to the segmentation process, in particular, to the location of the esophagus on the CT. The ablation procedure was performed with conscious sedation and consisted of segmental PV isolation in 10 (18%) patients and circumferential PV isolation in 45 (82%) patients. Transseptal puncture was performed under ultrasound guidance, and catheters were positioned in the left atrium.16 A three-dimensional anatomical shell of the left atrium and PVs was constructed with an electroanatomic mapping system (CARTO, Biosense-Webster, Diamond Bar, CA). Using landmark and surface mapping points acquired during the creation of the anatomical shell, the imported CT image of the PVs and left atrium were merged
with the anatomical shell using surface registration.12,13 Once merged, software analysis was performed, in a pointby-point fashion, for closeness of fit. This analysis calculates the maximum, minimum, and mean distance of the acquired points from the imported CT image. Points that were greater than 2 standard deviations from the mean were deleted, and the image was reregistered. The number of points to create the anatomical shell of the left atrium was 64 ⫾ 11. The mean, maximum, and minimum distance of acquired points from the imported CT image were 1.5 ⫾ 0.3 mm, 4.8 ⫾ 0.5 mm, 0.6 ⫾ 0.3 mm, respectively.
Study protocol Once integration of the left atrium and PVs was complete, the patient, while awake and alert, was asked to swallow 5–10 mL of barium contrast (E-Z Paste, E-Z EM Canada, Westbury, NY) to outline the course of the esophagus on fluoroscopy.17 No sedation was administered until the barium had passed the region of the larynx on fluoroscopy. If there was insufficient contrast to outline the esophageal borders, then occasionally, usually toward the end of the procedure, an additional ⬇2 mL of barium contrast was administered. There were no complications with the esophagram. Starting with a left anterior oblique angle of 30° and a right anterior oblique angle of 15°, digital fluoroscopic images of the esophagus were obtained, but other angulations were used to optimize identification of the maximum diameter of the esophagus. In the large majority of patients, optimal imaging was with a left anterior oblique view. Once a specific fluoroscopic angle was used for the first measurement of the esophagus, the same angle was used when reassessing changes in esophageal dimensions/mobility. The electroanatomic catheter was then positioned along the posterior wall of the left atrium. With fluoroscopic guidance, the borders of the esophagus were tagged on the anatomical shell at the point of the maximum width. The segmented image of the esophagus was then imported into the integrated image of the left atrium and PVs. The ablation catheter tagged the borders of the imported esophagus at the same level as the points used to tag the esophagus by contrast (Figure 1). In this manner, the pair of electroanatomic tags defined the borders of the esophagus outlined by contrast swallowed during the procedure and the borders of the esophagus outlined by the CT performed before the procedure. The second part of the protocol was to assess mobility of the esophagus during the RFA procedure relative to its original position denoted by the electroanatomic tags. When the esophagus was noted to move, the electroanatomic catheter re-marked the borders of the esophagus. These tags were compared with the electroanatomic markers that defined the original position of the esophagus.
Study endpoints and statistical analysis The endpoint of the study was to assess the ability of CT imaging to predict the location of the esophagus during the RFA procedure, as outlined by the esophagram. Two tech-
Daoud et al
Esophagus during RFA AF
977
Figure 1 A: Reconstructed three-dimensional CT image of the left atrium and PVs (orange) that has been merged with the electroanatomic map. After swallowing E-Z Paste, using fluoroscopic guidance, the borders of the esophagram were tagged with yellow electroanatomical markers at the largest diameter of the esophagus. The electrophysiologist is blinded to the CT-defined esophagus position. B: The CT-defined esophagus (blue) is then imported into the merged CARTO image. The borders of the CT-defined esophagus are then marked with electroanatomical tags (green) at the same location as the tags for the esophagram. In this example, the esophagus locations by CT and esophagram were concordant.
niques were used to judge this accuracy. The course of the esophagus was categorized as right, center or left. Left or right position was defined as the esophagus overlying the ostia of the PVs, and center position was defined as the esophagus not overlying any PV (Figures 1 and 2). The second technique was to compare the medial and lateral borders of the esophagus. An inaccurate prediction of the location of the esophagus by CT was defined as the medial or lateral borders on CT being separated from the esophagram medial or lateral borders by ⱖ50% of the diameter of the esophagus. Other collected data included an assessment of the mobility of the esophagus by measuring the change in the esophageal borders from its original position. A large shift was defined as movement of ⱖ50% of the original esophagus diameter. In addition, the location of the esophagus on the preprocedural CT was compared with the location of the esophagus on the 6-month postablation CT. Continuous variables are expressed as mean ⫾ 1 standard deviation. Continuous variables were compared with a paired t-test, and categorical variables were compared by 2 analysis. P ⬍.05 was considered statistically significant.
Results Location of esophagus The course of the esophagus was categorized as left, center, or right relative to the ostia of the PVs. By esophagography,
the esophagus was located near the left PVs in 34 patients (62%), in the center in 13 patients (24%), and near the right PVs in eight patients (15%). The CT prediction of the location of the esophagus was concordant with the esophagram in 48 (87%) of 55 patients. This did not represent a statistical difference (P ⫽ .2). However, even among the 48 patients in whom the CT and the esophagram were concordant, the esophagus borders by CT were separated from the esophagram borders by ⱖ50% of the diameter of the esophagus in 21 patients (44%). The mean separation between the esophageal border by esophagram compared with CT was 14 ⫾ 5 mm.
Esophageal mobility during RFA (Figures 3 and 4) The esophagus shifted by ⱖ50% of its original diameter during the procedure in 47 (85%) of 55 patients. During the procedure, the diameter of the esophagus changed, with a mean difference of the maximum and minimum diameter of the esophagus being 12 ⫾ 8 mm. By esophagram, the mean minimum and maximum diameter of the esophagus was 11 ⫾ 6 mm and 22 ⫾ 7 mm, respectively, and the range was 9 –28 mm. The esophagus shifted a mean of 16 ⫾ 9 mm (range 4 –27 mm). There was no significant difference in the diameter of the esophagus as measured by esophagram versus by CT (16.3 ⫾ 3.4 mm vs. 16.5 ⫾ 3.1, P ⫽ .7).
978
Figure 2 This figure represents a similar display as Figure 1. The borders of the esophagram are defined by the blue electroanatomic tags, and the borders of the esophagus, as defined by CT, are white. In this example, the esophagram and CT were not concordant: the esophagus was right (overlying the right PVs) by the esophagram but center by CT.
Esophageal position on post-RFA CT At the 6-month postablation CT image of the left atrium and PVs, the esophagus was located near the left PVs in 31 patients (56%), in the center in 14 patients (25%), and near the right PVs in 10 patients (18%). Relative to the PV ostia, the CT image of the esophagus at 6 months differed from the preprocedural CT in six of 55 patients (11%, P ⫽ .4). There were no atrioesophageal fistulae.
Discussion Main findings There are several main findings of this study. First, by statistical analysis, a CT performed 1 week before RFA
Heart Rhythm, Vol 5, No 7, July 2008 predicted the location of the esophagus at the time of the procedure. However, even though there was no statistical difference, the prediction of the esophagus location relative to the PVs by CT was inaccurate in 13% of the patients. This degree of error seems unacceptable when trying to carefully avoid delivering RF lesions adjacent to the esophagus. It is notable that a similar degree of variance in the location of the esophagus (11%) was demonstrated when comparing the preprocedural CT to the 6-month postprocedural CT, confirming the findings that the position of the esophagus changes over time. The second main finding is that, among the 48 patients in whom there was concordance between the CT and esophagram in predicting esophageal location, there were subtle but important shifts in the esophagus in 21 (44%) patients. A third finding is that the esophagus can considerably change dimensions during the RFA procedure, which is not predicted by CT imaging. This is entirely consistent with swallowing and esophageal peristalsis, events that can be expected to occur frequently between the time of CT image acquisition and the ablation procedure. In sum, these findings favor routine use of esophagram to locate and monitor the position of the esophagus rather than static CT imaging with esophageal contrast.
Previous studies Numerous studies have compared the position of the esophagus during the ablation procedure with the position of the esophagus by CT.6 – 8 However, the technique of assessing the position of the esophagus during the RFA procedure was to pass an electroanatomical catheter in and out of the esophagus.9,10 Consistent with the current study, these studies found no statistical difference in the location of the esophagus relative to the PVs but also reported a difference between CT and the virtual esophagus in a similar percentage of patients, ⬇10%. Key differences in the technique of prior studies compared with the current study are that the CT image was not
Figure 3 This figure is a left anterior oblique image of the esophagus outlined by barium contrast. There is a subtle shift in the position of the esophagus from image A to image B. When comparing images A and C, the esophagus position does not move, but the diameter increases in image C.
Daoud et al
Esophagus during RFA AF
979
Figure 4 This figure is a left anterior oblique image of the esophagus. There is no significant shift in the esophagus position; however, there is a dramatic increase in the esophagus diameter to the point that the esophagus is underfilled by barium.
merged with the three-dimensional CARTO map and the virtual esophagus tube generated by the electroanatomical catheter was stagnant and did not reflect variations in esophageal diameter or peristalsis. The important incremental value of the current study, therefore, is the assessment of a real time dynamic image of the esophagus relative to the CT image. The esophagram not only identified a discrepancy in the location of the esophagus relative to the PVs in 13% of patients, but even among those patients in whom the CT and esophagram correlated, subtle but important shifts (up to about 2.5 cm) in the esophageal borders were easily visualized on fluoroscopy. Movement of the esophagus has been reported during catheter ablation, with a mean distance of 20 mm.11 Measurements, however, were made relative to the spine, and there was no correlation made to CT imaging of the esophagus. In another study in which patients were under general anesthesia, a virtual esophagus and a diagnostic electrophysiology catheter were used to denote the esophagus. The esophagus was reported not to move; however, the current study demonstrates that esophageal movement can be subtle, cannot be adequately defined by a one-time creation of a virtual image, and is unlikely to be detected through use of a small-caliber, rigid electrophysiology catheter.
Limitations The limitations of this study are that the findings are limited to patients who undergo RFA with conscious sedation. Movement of the esophagus may not be as significant in patients who receive general anesthesia. A second limitation is the possibility that the mere administration of barium contrast in this setting may act as a stimulant for esophageal motility, although this phenomenon is not noted when esophagography is performed to evaluate esophageal motility.18 A third limitation is that this study measured only luminal dimensions; however, this technique was consistent when assessing the esophagus by esophagram and by CT. A final limitation is that the dimensions of the esophagus were not measured in a continuous manner; however, the esoph-
agus position was monitored during the duration of the procedure by fluoroscopy.
Conclusions A common practice of identifying the esophagus during curative RFA of AF is a CT image performed preprocedure or creation of a virtual esophagus with use of an electroanatomical catheter during the procedure. However, the findings of the current study highlight that these imaging techniques may offer false assurance that RFA lesions are being applied in a safe position. Esophagography not only defines the general location of the esophagus relative to the PVs but also identifies subtle but important changes in the position of the esophagus that may occur during the procedure, which is especially relevant when ablating with catheters that can create large lesions (⬇8 mm). A real time esophagram appears to be a more reliable method to monitor these shifts rather than static CT imaging or a virtual esophagram.
References 1. Pappone C, Oral H, Santinelli V, et al. Atrio-esophageal fistula as a complication of percutaneous transcatheter ablation of atrial fibrillation. Circulation 2004;22: 2724 –2726. 2. Scanavacca MI, D’avila A, Parga J, et al. Left atrial-esophageal fistula following radiofrequency catheter ablation of atrial fibrillation. J Cardiovasc Electrophysiol 2004;15:960 –962. 3. Ren JF, Lin D, Marchlinski FE, et al. Esophageal imaging and strategies for avoiding injury during left atrial ablation for atrial fibrillation. Heart Rhythm 2006;10:1156 –1161. 4. Pollak SJ, Monir G, Chernoby MS, et al. Novel imaging techniques of the esophagus enhancing safety of left atrial ablation. J Cardiovasc Electrophysiol 2005;16:245–248. 5. Wang SL, Ooi CG, Siu CW, et al. Endocardial visualization of esophageal-left atrial anatomic relationship by three-dimensional multidetector computed tomography “navigator imaging.” PACE 2006;29:502–508. 6. Kobza R, Auf der Maur C, Kurtz C, et al. Esophagus imaging for radiofrequency ablation of atrial fibrillation using a dual-source computed tomography system: preliminary observations. J Interv Card Electrophysiol 2007;3:167–170. 7. Piorkowski C, Hindricks G, Schreiber D, et al. Electroanatomic reconstruction of the left atrium, pulmonary veins, and esophagus compared with the “true anatomy” on multislice computed tomography in patients undergoing catheter ablation of atrial fibrillation. Heart Rhythm 2006;3:317–327. 8. Kottkamp H, Piorkowski C, Tanner H, et al. Topographic variability of the
980
9.
10. 11. 12.
13.
esophageal left atrial relation influencing ablation lines in patients with atrial fibrillation. J Cardiovasc Electrophysiol 2005;2:146 –150. Sherzer AI, Feigenblum DY, Kulkarni S, et al. Continuous nonfluoroscopic localization of the esophagus during radiofrequency catheter ablation of atrial fibrillation. J Cardiovasc Electrophysiol 2007;2:157–160. Shah BK, Yarlagadda RK, Tan V, et al. Three-dimensional identification of the esophagus during ablation of atrial fibrillation. Heart Rhythm 2005;5:565. Good E, Oral H, Lemola K, et al. Movement of the esophagus during left atrial catheter ablation for atrial fibrillation. J Am Coll Cardiol 2005;11:2107–2110. Cristoforetti A, Masè M, Faes L, et al. A stochastic approach for automatic registration and fusion of left atrial electroanatomic maps with 3D CT anatomical images. Phys Med Biol 2007;52:6323– 6337. Malchano ZJ, Neuzil P, Cury RC, et al. Integration of cardiac CT/MR imaging with three-dimensional electroanatomical mapping to guide catheter manipulation in the left atrium: implications for catheter ablation of atrial fibrillation. J Cardiovasc Electrophysiol 2006;17:1221–1229.
Heart Rhythm, Vol 5, No 7, July 2008 14. Piorkowski C, Hindricks G, Schreiber D, et al. Electroanatomic reconstruction of the left atrium, pulmonary veins, and esophagus compared with the “true anatomy” on multislice computed tomography in patients undergoing catheter ablation of atrial fibrillation. Heart Rhythm 2006;3:317–327. 15. Tops LF, Bax JJ, Zeppenfeld K, et al. Fusion of multislice computed tomography imaging with three-dimensional electroanatomic mapping to guide radiofrequency catheter ablation procedures. Heart Rhythm 2005;10: 1076 –1081. 16. Daoud EG, Kalbfleisch SJ, Hummel JD. Intracardiac echocardiography to guide transseptal left heart catheterization for radiofrequency catheter ablation. J Cardiovasc Electrophysiol 1999;10:358 –363. 17. Yamane T, Matsuo S, Date T, et al. Visualization of the esophagus throughout left atrial catheter ablation for atrial fibrillation. J Cardiovasc Electrophysiol 2006;17:105. 18. Summerton SL. Radiographic evaluation of esophageal function. Gastrointest Endosc Clin N Am 2005;15:231–242.
Introduction A common strategy for radiofrequency ablation (RFA) of atrial fibrillation (AF) is to deliver radiofrequency lesions in a circumferential manner around the pulmonary veins (PVs). A rare but often fatal complication of this strategy is an atrioesophageal fistula.1,2 This complication is due to delivery of RF current in a region of the left atrium that is in close proximity to the esophagus. Accurate identification of the course of the esophagus during RFA of AF is an important tool to minimize the risk of thermal injury. Previous studies3– 8 have suggested that computed tomography (CT) imaging of the esophagus performed before the procedure accurately predicts the location of the esophagus at the time of the procedure. Other studies have defined the esophagus position by use of an electroanatomic catheter.9,10 In addition, studies have noted movement of the
RESULTS The maximum diameter of the esophagus by esophagram versus CT was not different (16.3 ⫾ 3.4 vs. 16.5 ⫾ 3.1 mm; P ⫽ .7). The esophagus was near the left PVs in 34 (62%), center in 13 (24%), and near the right PVs in eight (15%) patients. There was concordance between CT and esophagram in 48 of 55 patients (87%; P ⫽ .2). Ye, in 21 (44%) of 48 patients with concordant location, the CT-defined esophageal borders were separated from the esophagram-defined borders by ⱖ50% of the esophagus diameter. CONCLUSIONS Reliance on remotely acquired CT images does not ensure adequate intraprocedural localization of the esophagus or enhance recognition of esophageal motility. KEYWORDS Esophagus; Atrial fibrillation; Catheter ablation; Pulmonary vein; Computed tomography (Heart Rhythm 2008;5:975–980) © 2008 Heart Rhythm Society. All rights reserved.
esophagus.11 The purpose of this study was to assess the accuracy of a preprocedural CT scan in predicting the location of the esophagus at the time of the RFA procedure and to assess the extent of esophageal mobility during the procedure.
Methods Study population The subjects of this study were 57 patients undergoing curative RFA of symptomatic paroxysmal, persistent, or long-standing persistent AF. Inclusion criteria were completion of CT imaging of the chest within 1 week of planned RFA of AF and adequate imaging of the esophagus by contrast at the time of RFA procedure. Clinical characteristics of the study population are summarized in Table 1. The Institutional Review Board approved the study protocol.
CT imaging Address reprint requests and correspondence: Emile Daoud, M.D., DHLRI, 473 West 12th Avenue, Suite 200, Columbus, OH 43210-1252. E-mail address: [email protected]. (Received February 1, 2008; Accepted March 30, 2008.)
All patients underwent electrocardiographically gated CT angiography of the heart within 1 week before the ablation procedure. CT exams were performed on a 64-slice (SOMATOM Sensation64, Siemens Medical Solutions, Inc.,
1547-5271/$ -see front matter © 2008 Heart Rhythm Society. All rights reserved.
doi:10.1016/j.hrthm.2008.03.058
976 Table 1
Heart Rhythm, Vol 5, No 7, July 2008 Clinical characteristics of patient population (n ⫽ 57)
Parameter
Value
Male gender Age, years Type of AF (%): Paroxysmal Persistent Long-standing persistent Antiarrhythmic drug trials Left ventricular ejection fraction Left atrial diameter, mm Structural heart disease (%): None Hypertensive Ischemic
38 57 ⫾ 8 21 (37) 23 (40) 13 (23) 2.1 ⫾ 0.5 0.42 ⫾ 0.11 48.5 ⫾ 6.4 30 (53) 13 (23) 9 (16)
Forchheim, Germany) scanner with about 5 mL of oral barium paste administered just before imaging. Gantry rotation speed was 330 –370 ms per rotation based on heart rate, and tube voltage was 120 Kv; electrocardiogram dose modulation was applied in all studies to reduce tube current by 80% during the systolic phases of the cardiac cycle. Image data were acquired with inspiratory breath hold during injection of 80 mL of contrast agent at 5 mL/s followed by injection of 20 mL saline flush at 5 mL/s, with timing optimized using bolus tracking and a Hounsfield unit threshold of 100 units set in the left atrium. Overlapping 1.25-mm axial images stored in DICOM format were reconstructed at end-diastole. Motion artifact precluded adequate definition of the esophagus in two (3.5%) of 57 patients. The results, therefore, include the remaining 55 patients.
Merging of CT and three-dimensional map Before the ablation procedure, importation into the mapping system and segmentation of the CT image data was performed using commercially available software (CARTO Merge, Biosense-Webster, Diamond Bar, CA).12–15 With this software, user-directed seed points followed by signal intensity-based segmentation allowed delineation of the following structures: esophagus, pulmonary artery, aorta, right atrium, right ventricle, left ventricle, left atrium, and PVs. The left atrium and PVs were exported into the real time mapping system. The electrophysiologist was blinded to the segmentation process, in particular, to the location of the esophagus on the CT. The ablation procedure was performed with conscious sedation and consisted of segmental PV isolation in 10 (18%) patients and circumferential PV isolation in 45 (82%) patients. Transseptal puncture was performed under ultrasound guidance, and catheters were positioned in the left atrium.16 A three-dimensional anatomical shell of the left atrium and PVs was constructed with an electroanatomic mapping system (CARTO, Biosense-Webster, Diamond Bar, CA). Using landmark and surface mapping points acquired during the creation of the anatomical shell, the imported CT image of the PVs and left atrium were merged
with the anatomical shell using surface registration.12,13 Once merged, software analysis was performed, in a pointby-point fashion, for closeness of fit. This analysis calculates the maximum, minimum, and mean distance of the acquired points from the imported CT image. Points that were greater than 2 standard deviations from the mean were deleted, and the image was reregistered. The number of points to create the anatomical shell of the left atrium was 64 ⫾ 11. The mean, maximum, and minimum distance of acquired points from the imported CT image were 1.5 ⫾ 0.3 mm, 4.8 ⫾ 0.5 mm, 0.6 ⫾ 0.3 mm, respectively.
Study protocol Once integration of the left atrium and PVs was complete, the patient, while awake and alert, was asked to swallow 5–10 mL of barium contrast (E-Z Paste, E-Z EM Canada, Westbury, NY) to outline the course of the esophagus on fluoroscopy.17 No sedation was administered until the barium had passed the region of the larynx on fluoroscopy. If there was insufficient contrast to outline the esophageal borders, then occasionally, usually toward the end of the procedure, an additional ⬇2 mL of barium contrast was administered. There were no complications with the esophagram. Starting with a left anterior oblique angle of 30° and a right anterior oblique angle of 15°, digital fluoroscopic images of the esophagus were obtained, but other angulations were used to optimize identification of the maximum diameter of the esophagus. In the large majority of patients, optimal imaging was with a left anterior oblique view. Once a specific fluoroscopic angle was used for the first measurement of the esophagus, the same angle was used when reassessing changes in esophageal dimensions/mobility. The electroanatomic catheter was then positioned along the posterior wall of the left atrium. With fluoroscopic guidance, the borders of the esophagus were tagged on the anatomical shell at the point of the maximum width. The segmented image of the esophagus was then imported into the integrated image of the left atrium and PVs. The ablation catheter tagged the borders of the imported esophagus at the same level as the points used to tag the esophagus by contrast (Figure 1). In this manner, the pair of electroanatomic tags defined the borders of the esophagus outlined by contrast swallowed during the procedure and the borders of the esophagus outlined by the CT performed before the procedure. The second part of the protocol was to assess mobility of the esophagus during the RFA procedure relative to its original position denoted by the electroanatomic tags. When the esophagus was noted to move, the electroanatomic catheter re-marked the borders of the esophagus. These tags were compared with the electroanatomic markers that defined the original position of the esophagus.
Study endpoints and statistical analysis The endpoint of the study was to assess the ability of CT imaging to predict the location of the esophagus during the RFA procedure, as outlined by the esophagram. Two tech-
Daoud et al
Esophagus during RFA AF
977
Figure 1 A: Reconstructed three-dimensional CT image of the left atrium and PVs (orange) that has been merged with the electroanatomic map. After swallowing E-Z Paste, using fluoroscopic guidance, the borders of the esophagram were tagged with yellow electroanatomical markers at the largest diameter of the esophagus. The electrophysiologist is blinded to the CT-defined esophagus position. B: The CT-defined esophagus (blue) is then imported into the merged CARTO image. The borders of the CT-defined esophagus are then marked with electroanatomical tags (green) at the same location as the tags for the esophagram. In this example, the esophagus locations by CT and esophagram were concordant.
niques were used to judge this accuracy. The course of the esophagus was categorized as right, center or left. Left or right position was defined as the esophagus overlying the ostia of the PVs, and center position was defined as the esophagus not overlying any PV (Figures 1 and 2). The second technique was to compare the medial and lateral borders of the esophagus. An inaccurate prediction of the location of the esophagus by CT was defined as the medial or lateral borders on CT being separated from the esophagram medial or lateral borders by ⱖ50% of the diameter of the esophagus. Other collected data included an assessment of the mobility of the esophagus by measuring the change in the esophageal borders from its original position. A large shift was defined as movement of ⱖ50% of the original esophagus diameter. In addition, the location of the esophagus on the preprocedural CT was compared with the location of the esophagus on the 6-month postablation CT. Continuous variables are expressed as mean ⫾ 1 standard deviation. Continuous variables were compared with a paired t-test, and categorical variables were compared by 2 analysis. P ⬍.05 was considered statistically significant.
Results Location of esophagus The course of the esophagus was categorized as left, center, or right relative to the ostia of the PVs. By esophagography,
the esophagus was located near the left PVs in 34 patients (62%), in the center in 13 patients (24%), and near the right PVs in eight patients (15%). The CT prediction of the location of the esophagus was concordant with the esophagram in 48 (87%) of 55 patients. This did not represent a statistical difference (P ⫽ .2). However, even among the 48 patients in whom the CT and the esophagram were concordant, the esophagus borders by CT were separated from the esophagram borders by ⱖ50% of the diameter of the esophagus in 21 patients (44%). The mean separation between the esophageal border by esophagram compared with CT was 14 ⫾ 5 mm.
Esophageal mobility during RFA (Figures 3 and 4) The esophagus shifted by ⱖ50% of its original diameter during the procedure in 47 (85%) of 55 patients. During the procedure, the diameter of the esophagus changed, with a mean difference of the maximum and minimum diameter of the esophagus being 12 ⫾ 8 mm. By esophagram, the mean minimum and maximum diameter of the esophagus was 11 ⫾ 6 mm and 22 ⫾ 7 mm, respectively, and the range was 9 –28 mm. The esophagus shifted a mean of 16 ⫾ 9 mm (range 4 –27 mm). There was no significant difference in the diameter of the esophagus as measured by esophagram versus by CT (16.3 ⫾ 3.4 mm vs. 16.5 ⫾ 3.1, P ⫽ .7).
978
Figure 2 This figure represents a similar display as Figure 1. The borders of the esophagram are defined by the blue electroanatomic tags, and the borders of the esophagus, as defined by CT, are white. In this example, the esophagram and CT were not concordant: the esophagus was right (overlying the right PVs) by the esophagram but center by CT.
Esophageal position on post-RFA CT At the 6-month postablation CT image of the left atrium and PVs, the esophagus was located near the left PVs in 31 patients (56%), in the center in 14 patients (25%), and near the right PVs in 10 patients (18%). Relative to the PV ostia, the CT image of the esophagus at 6 months differed from the preprocedural CT in six of 55 patients (11%, P ⫽ .4). There were no atrioesophageal fistulae.
Discussion Main findings There are several main findings of this study. First, by statistical analysis, a CT performed 1 week before RFA
Heart Rhythm, Vol 5, No 7, July 2008 predicted the location of the esophagus at the time of the procedure. However, even though there was no statistical difference, the prediction of the esophagus location relative to the PVs by CT was inaccurate in 13% of the patients. This degree of error seems unacceptable when trying to carefully avoid delivering RF lesions adjacent to the esophagus. It is notable that a similar degree of variance in the location of the esophagus (11%) was demonstrated when comparing the preprocedural CT to the 6-month postprocedural CT, confirming the findings that the position of the esophagus changes over time. The second main finding is that, among the 48 patients in whom there was concordance between the CT and esophagram in predicting esophageal location, there were subtle but important shifts in the esophagus in 21 (44%) patients. A third finding is that the esophagus can considerably change dimensions during the RFA procedure, which is not predicted by CT imaging. This is entirely consistent with swallowing and esophageal peristalsis, events that can be expected to occur frequently between the time of CT image acquisition and the ablation procedure. In sum, these findings favor routine use of esophagram to locate and monitor the position of the esophagus rather than static CT imaging with esophageal contrast.
Previous studies Numerous studies have compared the position of the esophagus during the ablation procedure with the position of the esophagus by CT.6 – 8 However, the technique of assessing the position of the esophagus during the RFA procedure was to pass an electroanatomical catheter in and out of the esophagus.9,10 Consistent with the current study, these studies found no statistical difference in the location of the esophagus relative to the PVs but also reported a difference between CT and the virtual esophagus in a similar percentage of patients, ⬇10%. Key differences in the technique of prior studies compared with the current study are that the CT image was not
Figure 3 This figure is a left anterior oblique image of the esophagus outlined by barium contrast. There is a subtle shift in the position of the esophagus from image A to image B. When comparing images A and C, the esophagus position does not move, but the diameter increases in image C.
Daoud et al
Esophagus during RFA AF
979
Figure 4 This figure is a left anterior oblique image of the esophagus. There is no significant shift in the esophagus position; however, there is a dramatic increase in the esophagus diameter to the point that the esophagus is underfilled by barium.
merged with the three-dimensional CARTO map and the virtual esophagus tube generated by the electroanatomical catheter was stagnant and did not reflect variations in esophageal diameter or peristalsis. The important incremental value of the current study, therefore, is the assessment of a real time dynamic image of the esophagus relative to the CT image. The esophagram not only identified a discrepancy in the location of the esophagus relative to the PVs in 13% of patients, but even among those patients in whom the CT and esophagram correlated, subtle but important shifts (up to about 2.5 cm) in the esophageal borders were easily visualized on fluoroscopy. Movement of the esophagus has been reported during catheter ablation, with a mean distance of 20 mm.11 Measurements, however, were made relative to the spine, and there was no correlation made to CT imaging of the esophagus. In another study in which patients were under general anesthesia, a virtual esophagus and a diagnostic electrophysiology catheter were used to denote the esophagus. The esophagus was reported not to move; however, the current study demonstrates that esophageal movement can be subtle, cannot be adequately defined by a one-time creation of a virtual image, and is unlikely to be detected through use of a small-caliber, rigid electrophysiology catheter.
Limitations The limitations of this study are that the findings are limited to patients who undergo RFA with conscious sedation. Movement of the esophagus may not be as significant in patients who receive general anesthesia. A second limitation is the possibility that the mere administration of barium contrast in this setting may act as a stimulant for esophageal motility, although this phenomenon is not noted when esophagography is performed to evaluate esophageal motility.18 A third limitation is that this study measured only luminal dimensions; however, this technique was consistent when assessing the esophagus by esophagram and by CT. A final limitation is that the dimensions of the esophagus were not measured in a continuous manner; however, the esoph-
agus position was monitored during the duration of the procedure by fluoroscopy.
Conclusions A common practice of identifying the esophagus during curative RFA of AF is a CT image performed preprocedure or creation of a virtual esophagus with use of an electroanatomical catheter during the procedure. However, the findings of the current study highlight that these imaging techniques may offer false assurance that RFA lesions are being applied in a safe position. Esophagography not only defines the general location of the esophagus relative to the PVs but also identifies subtle but important changes in the position of the esophagus that may occur during the procedure, which is especially relevant when ablating with catheters that can create large lesions (⬇8 mm). A real time esophagram appears to be a more reliable method to monitor these shifts rather than static CT imaging or a virtual esophagram.
References 1. Pappone C, Oral H, Santinelli V, et al. Atrio-esophageal fistula as a complication of percutaneous transcatheter ablation of atrial fibrillation. Circulation 2004;22: 2724 –2726. 2. Scanavacca MI, D’avila A, Parga J, et al. Left atrial-esophageal fistula following radiofrequency catheter ablation of atrial fibrillation. J Cardiovasc Electrophysiol 2004;15:960 –962. 3. Ren JF, Lin D, Marchlinski FE, et al. Esophageal imaging and strategies for avoiding injury during left atrial ablation for atrial fibrillation. Heart Rhythm 2006;10:1156 –1161. 4. Pollak SJ, Monir G, Chernoby MS, et al. Novel imaging techniques of the esophagus enhancing safety of left atrial ablation. J Cardiovasc Electrophysiol 2005;16:245–248. 5. Wang SL, Ooi CG, Siu CW, et al. Endocardial visualization of esophageal-left atrial anatomic relationship by three-dimensional multidetector computed tomography “navigator imaging.” PACE 2006;29:502–508. 6. Kobza R, Auf der Maur C, Kurtz C, et al. Esophagus imaging for radiofrequency ablation of atrial fibrillation using a dual-source computed tomography system: preliminary observations. J Interv Card Electrophysiol 2007;3:167–170. 7. Piorkowski C, Hindricks G, Schreiber D, et al. Electroanatomic reconstruction of the left atrium, pulmonary veins, and esophagus compared with the “true anatomy” on multislice computed tomography in patients undergoing catheter ablation of atrial fibrillation. Heart Rhythm 2006;3:317–327. 8. Kottkamp H, Piorkowski C, Tanner H, et al. Topographic variability of the
980
9.
10. 11. 12.
13.
esophageal left atrial relation influencing ablation lines in patients with atrial fibrillation. J Cardiovasc Electrophysiol 2005;2:146 –150. Sherzer AI, Feigenblum DY, Kulkarni S, et al. Continuous nonfluoroscopic localization of the esophagus during radiofrequency catheter ablation of atrial fibrillation. J Cardiovasc Electrophysiol 2007;2:157–160. Shah BK, Yarlagadda RK, Tan V, et al. Three-dimensional identification of the esophagus during ablation of atrial fibrillation. Heart Rhythm 2005;5:565. Good E, Oral H, Lemola K, et al. Movement of the esophagus during left atrial catheter ablation for atrial fibrillation. J Am Coll Cardiol 2005;11:2107–2110. Cristoforetti A, Masè M, Faes L, et al. A stochastic approach for automatic registration and fusion of left atrial electroanatomic maps with 3D CT anatomical images. Phys Med Biol 2007;52:6323– 6337. Malchano ZJ, Neuzil P, Cury RC, et al. Integration of cardiac CT/MR imaging with three-dimensional electroanatomical mapping to guide catheter manipulation in the left atrium: implications for catheter ablation of atrial fibrillation. J Cardiovasc Electrophysiol 2006;17:1221–1229.
Heart Rhythm, Vol 5, No 7, July 2008 14. Piorkowski C, Hindricks G, Schreiber D, et al. Electroanatomic reconstruction of the left atrium, pulmonary veins, and esophagus compared with the “true anatomy” on multislice computed tomography in patients undergoing catheter ablation of atrial fibrillation. Heart Rhythm 2006;3:317–327. 15. Tops LF, Bax JJ, Zeppenfeld K, et al. Fusion of multislice computed tomography imaging with three-dimensional electroanatomic mapping to guide radiofrequency catheter ablation procedures. Heart Rhythm 2005;10: 1076 –1081. 16. Daoud EG, Kalbfleisch SJ, Hummel JD. Intracardiac echocardiography to guide transseptal left heart catheterization for radiofrequency catheter ablation. J Cardiovasc Electrophysiol 1999;10:358 –363. 17. Yamane T, Matsuo S, Date T, et al. Visualization of the esophagus throughout left atrial catheter ablation for atrial fibrillation. J Cardiovasc Electrophysiol 2006;17:105. 18. Summerton SL. Radiographic evaluation of esophageal function. Gastrointest Endosc Clin N Am 2005;15:231–242.