Author + information
- Received December 29, 2009
- Revision received March 15, 2010
- Accepted April 13, 2010
- Published online October 12, 2010.
- Benoit Desjardins, MD, PhD,
- Fred Morady, MD and
- Frank Bogun, MD⁎ ()
- ↵⁎Reprint requests and correspondence
: Dr. Frank Bogun, Division of Cardiology, CVC Cardiovascular Medicine, 1500 East Medical Center Drive SPC 5853, Ann Arbor, Michigan 48109-5853
Objectives The purpose of this study was to correlate 3-dimensional distribution of epicardial fat on computed tomography (CT) with electroanatomical (EA) voltage maps obtained during percutaneous epicardial mapping in order to determine the fat thickness cut-off that results in voltage attenuation and to establish normal ventricular epicardial voltage criteria in the absence of fat.
Background Epicardial fat can mimic scar tissue when epicardial voltage mapping is performed, as both result in low epicardial voltage. Cardiac CT can differentiate epicardial fat from scar or muscle on the basis of their distinct attenuations.
Methods Transcutaneous epicardial mapping was performed in a consecutive series of 14 patients. A cardiac CT was performed before the procedure and a 3-dimensional image of the epicardial fat was generated and registered with the epicardial EA voltage map.
Results In patients without cardiomyopathy (n = 8), a voltage ≥1.5 mV best correlated with the absence of epicardial fat. A fat thickness ≥2.8 mm resulted in voltage attenuation and best separated low voltage (<1.5 mV) from normal voltage (≥1.5 mV; sensitivity 81%, specificity 81%, area under the curve 0.85). In patients without cardiomyopathy, the low-voltage area matched well with the area of epicardial fat. In the 6 patients with nonischemic cardiomyopathy, the low-voltage area by far exceeded the area accounted for by epicardial fat; this corresponded with the presence of scar tissue. Epicardial ablations at sites with >10 mm of fat were ineffective.
Conclusions Cardiac CT identifies epicardial fat that can mimic scar tissue during epicardial EA voltage mapping, which is important during epicardial mapping and ablation.
Ventricular arrhythmias are often generated by scar tissue that is characterized by low-voltage bipolar electrograms on an electroanatomical map. Scar tissue can be located on the epicardial surface of the heart, especially in patients with nonischemic cardiomyopathy. Fat frequently is present on the epicardium of the heart. As with scar tissue, fat tissue is characterized by low voltage (1). If an arrhythmia originates in an area covered by epicardial fat, conventional ablation may be inadequate, and irrigated catheters may be required to reach the arrhythmogenic focus. Epicardial fat tissue can readily be identified by cardiac computed tomography (CT) based on the distinct attenuation of fat tissue (2). The objective of this study was to correlate a 3-dimensional (3D) display of epicardial fat with the electroanatomical (EA) map obtained during a percutaneous epicardial mapping procedure. The aim was to determine cut-off values of fat thickness resulting in voltage attenuation, to establish voltage criteria for normal ventricular epicardium in the absence of fat tissue, and to assess whether the ablation outcome was affected by the presence of epicardial fat.
Fourteen consecutive patients (11 men, mean age 42 ± 14 years; ejection fraction 49 ± 20%) underwent transcutaneous epicardial mapping for treatment of drug-refractory ventricular or supraventricular arrhythmias. All patients had undergone unsuccessful endocardial mapping and ablation before the epicardial mapping procedure. Six of the 14 patients had nonischemic cardiomyopathy (dilated cardiomyopathy, n = 5; arrhythmogenic right ventricular dysplasia, n =1; mean ejection fraction: 0.33 ± 0.19%) and ventricular tachycardia (VT). The other 8 patients had normal left ventricular function and no evidence of structural heart disease (ejection fraction: 0.61 ± 0.06%). One of these 8 patients had ventricular tachycardia, 2 had frequent premature ventricular complexes (PVCs), 4 had Wolff-Parkinson-White syndrome, and 1 had atrial tachycardia (Table 1).
This study was approved by the University of Michigan Institutional Review Board. After informed consent was obtained, vascular access was obtained in a femoral vein. Two multielectrode catheters were positioned in the His bundle position and the right ventricular apex. In the patients with structural heart disease, programmed ventricular stimulation at 2 right ventricular sites was performed using up to 4 extrastimuli before and after the ablation procedure to assess for inducible VTs (3). Pericardial access was obtained as described by Sosa et al. (4) Thereafter, 3,000 U of heparin were administered followed by an additional 1,000 U/h; if left ventricular endocardial mapping was performed, a bolus of 5,000 U of heparin was administered, with additional heparin as needed to maintain an activated clotting time >250 s.
Mapping and ablation procedure
An 8-F multipolar catheter (QwikStar, Biosense Webster, Diamond Bar, California) was used for mapping and ablation within the pericardial space. Epicardial voltage mapping was performed with a mean of 226 ± 212 points for tip reconstruction and a mean of 1,044 ± 564 points for combined tip and shaft reconstruction. Intracardiac and epicardial electrograms and the surface electrocardiograms were displayed at a speed of 100 mm/s. All electrograms were filtered at bandpass settings of 50 to 500 Hz. The recordings were stored on optical disc (EP Med Systems Inc., West Berlin, New Jersey). The voltage maps were performed during sinus rhythm and were displayed as maps, including the tip reconstruction and the combined reconstruction map (Fig. 1).Analysis was carried out at the epicardial sites that were acquired by the catheter tip only. In patients with structural heart disease, pace mapping was performed at sites with a voltage <1.5 mV to map ventricular arrhythmias. In patients without structural heart disease, PVCs and supraventricular tachycardias were mapped by activation mapping. The maps were compared with the actual locations of fat in 3 patients who underwent heart transplantation (Fig. 1).
When radiofrequency energy was delivered in the epicardial space, this was preceded by coronary angiography to document a distance of >5 mm from a major epicardial coronary artery (5). Ablation was first attempted with the Qwikmap catheter at a power of up to 50 W and a target temperature of 60°C. If there was insufficient heating or if the power was <20 W, the catheter was changed to an open-irrigated catheter (Thermocool, Biosense Webster). Radiofrequency energy was delivered with the irrigated-tip catheter at an initial power of 20 W. The power was titrated to achieve a 10-ohm impedance drop. The irrigant was removed continuously via the pericardial sheath.
The low-voltage bipolar electrograms in patients with and without cardiomyopathy were analyzed and correlated with the overlying epicardial fat thickness. The electrogram readers were blinded to the CT data. The electrogram width and the timing from the beginning of the QRS complex to the end of the recorded electrogram were measured. The number of distinct spikes within the electrogram was assessed. The presence of isolated potentials was documented. An isolated potential was defined as an electrogram that was separated by an isoelectric segment of >20 ms from the ventricular electrogram.
Processing of CT images
Each patient underwent an electrocardiogram-gated cardiac CT on a 64-slice scanner (General Electric, Milwaukee, Wisconsin) with intravenous iodinated contrast agent before the epicardial mapping procedure. A stack of CT images (0.5 × 0.5 mm resolution, 1.25-mm thickness) from a single cardiac phase at 70% of the RR-wave interval was then post-processed (Osirix, Geneva, Switzerland). First, the pericardial membrane was identified (Fig. 2,left panel) and traced in a single contour around the heart on sample images. Although on some images the location of the pericardial membrane was difficult to identify, the 3D nature of the CT data and the location of the epicardial vessels made it possible to follow the 3D course of the pericardial membrane in all cases. The contour was made continuous behind the heart for simplicity. The contours on the other images were then interpolated by the software and manually corrected for exact fit. If fluid was present in the pericardial sac, the visceral layer of the pericardium was included in the contour. The outer contour of the myocardium then was identified by thresholding with manual adjustments. The area between the 2 contours represented the fat and epicardial vessels. The fat outside the pericardium was not included because it would not be interposed between myocardium and the mapping catheter. The reproducibility of the technique for semiautomatic measurements of the epicardial fat contours was assessed. Intraobserver variability was 0.65 ± 0.64 mm (range per subject: 0.41 to 1.06 mm). Interobserver variability was 0.81 ± 0.78 mm (range per subject: 0.51 to 1.20 mm).
Both a 3D epicardial fat image and a 3D myocardial image were exported by the software for further processing. The thickness of epicardial fat was computed from the 3D epicardial fat map using 3D mathematical morphological operators and displayed in a color-coded map (ORS Visual, Montreal, Quebec, Canada) (Fig. 3).The 3D epicardial fat and myocardial images were registered with the 3D epicardial voltage map using customized software on Matlab (Mathworks, Natick, Massachusetts) (Fig. 1). Landmark registration initially was performed using 3 reference points: the epicardial apex and the most lateral tricuspid and mitral annular points. This was followed by surface registration using a Gauss-Newton optimization approach. The positional error of the registered map was 2.97 ± 0.85 mm using surface registration statistics. For each epicardial mapping point, the amount of underlying epicardial fat tissue was determined at the corresponding specific location on the registered cardiac CT (Figs. 1 and 4).
In the 8 patients without structural heart disease, an optimal threshold value of fat thickness resulting in voltage attenuation due to underlying epicardial fat was determined by comparing sites with ≥1.5 and <1.5 mV and constructing receiver-operator characteristic (ROC) curves (Fig. 5,left panel). In addition, the optimal epicardial voltage separating sites with and without epicardial fat was also determined by constructing ROC curves at multiple fixed-fat thicknesses (Fig. 5, right panel).
The distribution of ventricular epicardial fat was displayed on a 3D map and was correlated with low-voltage points on the electroanatomical maps (Fig. 1). Low epicardial voltage in this study was defined as a voltage <1.5 mV based on prior studies using endocardial mapping data (6). Within the ventricular epicardium, the percentages of low- and normal-voltage points projecting on fat and nonfat tissue were determined. The size of fat tissue covering the left and right ventricles was determined. Because a voltage cutoff of <1.0 mV also was used in prior studies (7), the epicardial fat thickness differentiating sites with a voltage ≥1.0 and <1.0 mV was also determined.
The area of epicardial fat was compared with the size of ventricular low-voltage areas as determined by the epicardial electroanatomical voltage map. Scar was defined as low-voltage sites in areas devoid of epicardial fat; this area was measured (Fig. 6)and was correlated with the presence of scar on delayed enhancement magnetic resonance imaging (MRI) in 3 patients. The regions of scar were compared with the areas of epicardial fat. An MRI was not obtained in 5 of 6 patients with cardiomyopathy who had an implanted cardioverter defibrillator.
At effective ablation sites and at other epicardial regions of interest where mapping criteria were met, the voltage during sinus rhythm and the thickness of the epicardial fat was determined (Fig. 6).
An MRI was performed to assess for ventricular scar in 3 patients with ventricular arrhythmias who did not have a contraindications for MRI. The MRI was performed with a 1.5-T MRI scanner (Signa Excite CV/i, General Electric) with a 4- or 8-element phased array coil placed over the chest of patients in the supine position. Images were acquired with electrocardiography gating during breath-holds. Dynamic short- and long-axis images of the heart were acquired using a segmented, k-space, steady-state, free-precession pulse sequence (repetition time 4.2 ms, echo time 1.8 ms, 1.4 × 1.4-mm in-plane spatial resolution, slice thickness 8 mm). Fifteen minutes after administration of 0.20 mmol/kg intravenous gadolinium DTPA (Magnevist, Berlex Pharmaceuticals, Wayne, New Jersey), 2-dimensional delayed enhancement imaging was performed using an inversion-recovery sequence (8) (repetition time 6.7 ms, echo time 3.2 ms, in-plane spatial resolution 1.4 × 2.2 mm, slice thickness 8 mm) in the short axis and long axis of the left ventricle at matching cine-image slice locations. The inversion time (250 to 350 ms) was optimized to null the normal myocardium.
Continuous variables are expressed as mean ± 1 SD. Comparisons were performed using Student ttest, paired when appropriate. Discrete variables were compared using chi-square test or the Fisher exact test, as appropriate. The Pearson correlation coefficient was used to compare the size of epicardial ventricular fat as determined by CT and the size of the ventricular low-voltage area as assessed by EA mapping. ROC curves were constructed to determine the best combined sensitivity/specificity to differentiate low voltage due to epicardial fat from normal voltage. A p value <0.05 indicated statistical significance.
Epicardial voltage mapping and fat thickness in patients without cardiomyopathy
Low-voltage areas were identified in all 8 patients in the atrioventricular groove and interventricular groove and at the inferolateral right ventricle in 7 patients (Table 2).The cut-off value of epicardial fat thickness that best separated epicardial low-voltage sites (<1.5 mV) from sites with normal voltage (≥1.5 mV) was 2.8 mm (sensitivity 81%, specificity 81%, area under the curve 0.85) (Figs. 4 and 5). Using <1.0 mV as a cut-off value for low voltage, the epicardial fat thickness that best separated sites with epicardial low-voltage from sites with normal voltage was 3.0 mm (sensitivity 83%, specificity 78%, area under the curve 0.85).
For fixed-fat thickness threshold in the 0- to 5-mm range, an epicardial voltage of 1.5 mV was located at the inflexion point of each respective ROC curve, maximizing the sum of sensitivity and specificity, and best separated sites with low voltage and normal voltage (Fig. 5).
Ninety-seven percent of all sites devoid of fat had a voltage ≥1.5 mV. Including sites with a thin layer of epicardial fat (<2.8 mm), 92% of all sites had a voltage ≥1.5 mV. Only 4% of low-voltage points projected on areas devoid of fat. The mean voltage at sites devoid of epicardial fat was 3.5 mV, and the mean voltage of epicardial mapping points where ≥2.8 mm of fat was present was 1.7 mV (p < 0.0001). The thickness of epicardial fat was correlated inversely with the epicardial voltage (R = −0.40, p < 0.0001) (Fig. 4).
Epicardial fat, voltage mapping, and scar
In the patients without structural heart disease, there was a strong correlation between the surface area of low voltage and the surface area of epicardial fat covering the ventricular surface (R = 0.99; p < 0.0001). In these patients, the low-voltage area was 70 ± 43 cm2. This matched with the area of epicardial fat overlying the ventricular myocardium (76 ± 50 cm2; p = 0.2).
In patients with nonischemic cardiomyopathy, the area of low voltage was larger (201 ± 86 cm2; p = 0.003), and the area of epicardial fat was similar (113 ± 42 cm2; p = 0.4) compared with patients without structural heart disease. The discrepancy between the area of fat and the area of low voltage in patients with cardiomyopathy was a mean of 88 ± 71 cm2, which was significantly larger than in patients without cardiomyopathy (−5 ± 11 cm2; p = 0.003). Epicardial scar accounted for this discrepancy (Fig. 6).
Epicardial fat and outcome of ablation
Eleven targeted epicardial VTs (cycle length: 397 ± 90 ms) in 6 patients (all with nonischemic cardiomyopathy) and 1 accessory epicardial atrioventricular pathway were successfully ablated on the epicardial surface. In the 3 patients without structural heart disease in whom epicardial PVCs or VT were targeted, despite the earliest activation being detected in the basal epicardial left ventricle, ablation failed. In 1 patient with an epicardial posterior pathway, epicardial ablation also failed in the epicardium. The voltage did not differ significantly at effective and ineffective epicardial target sites (0.65 ± 0.52 mV vs. 0.37 ± 0.04 mV; p = 0.6). However, the epicardial fat layer was significantly thinner at effective than at ineffective ablation sites (1.1 ± 1.3 mm vs. 13.2 ± 2.1 mm; p < 0.0001). In all patients with effective ablation of VT, the target sites were located at low-voltage sites devoid of epicardial fat (Fig. 2, right panel) (mean voltage: 0.73 ± 1.1 mV). In the 3 patients with idiopathic PVCs or VTs, despite the earliest activation time being located in the epicardium, ablation was ineffective, and the site of origin was located beneath an epicardial fat layer of 10.8 to 14.6 mm. The voltage at these sites was 0.4 ± 0.08 mV. In the patient with effective ablation of an epicardial accessory pathway, the pathway was located at a posteroseptal site with a voltage of 1.8 mV beneath an epicardial fat layer of 1.8 mm. In a patient with unsuccessful ablation of an epicardial posterior accessory pathway, the thickness of epicardial fat overlying the target site was 14.3 mm.
In 3 patients (2 patients with Wolff-Parkinson-White syndrome and 1 patient with atrial tachycardia), ablation was effective at an endocardial site.
In the 3 hearts that were explanted due to heart transplantation 2 weeks to 2 months after the ablation procedure, no discrete scar was detected epicardially, and there was diffuse fibrosis throughout the entire ventricular myocardium. In all 3 hearts, the epicardial fat distribution macroscopically closely matched with the epicardial fat distribution on the CT images (Fig. 1).
Isolated potentials were only recorded in patients with cardiomyopathy and were not identified in patients without structural heart disease (p < 0.05). Electrogram duration and number of spikes within the electrogram did not differ when sites from patients with overlaying epicardial fat ≥2.8 mm were compared with sites from patients with less epicardial fat (114 ± 21 ms vs. 101 ± 11 ms, p = 0.1; and 2.9 ± 1.1 spikes vs. 2.6 ± 0.6 spikes, p = 0.3, respectively). Also, there was no difference in the percentage of electrograms extending beyond the end of the QRS complex comparing sites from patients with ≥2.8 mm epicardial fat with electrograms from patients with less epicardial fat (0.34 ± 0.22 vs. 0.33 ± 0.17; p = 0.8).
The mean follow-up period was 23 ±14 months. In the 3 patients with idiopathic PVCs or VT, arrhythmia control was achieved by treatment with Vaughan Williams class IC antiarrhythmics, amiodarone, or beta-blockers. Arrhythmias in 3 patients with cardiomyopathy who did not undergo transplantation were controlled by therapy with dofetilide, amiodarone, or beta-blockers. In the patients with supraventricular arrhythmias, no arrhythmias recurred in 3 of 6 patients in whom all medications were discontinued. The other 3 patients remained on a beta-blocker or Vaughan Williams class IC antiarrhythmic medication.
This report is unique in that CT technology was used to characterize the extent of fat tissue by extracting and integrating epicardial fat information into EA voltage maps, thereby helping to distinguish epicardial fat from scar tissue. Radiofrequency energy applications were ineffective in part due to a substantial layer of epicardial fat measuring ≥10 mm in thickness. An epicardial fat layer ≥2.8 mm resulted in epicardial voltage attenuation. In patients without cardiomyopathy, an epicardial voltage of 1.5 mV best separated sites with and without overlying epicardial fat.
Identification and registration of epicardial fat
Low-voltage endocardial electrograms are used to differentiate normal endocardium from scar tissue. Although this technique is valid for the endocardium, this is not true for the epicardium, where low voltage may be due to fat tissue. This limitation of epicardial voltage mapping can be overcome by extracting and registering epicardial fat into the electroanatomical voltage map. Using CT for identification of epicardial fat is highly reproducible and accurate, as demonstrated in this study.
Epicardial radiofrequency ablation is substantially limited by the presence of epicardial fat. Even when an irrigated-tip catheter was used, a fat thickness of ≥3.5 mm could not be penetrated by an ablation lesion in a prior report (9). Therefore, knowing the thickness of fat overlying the ventricular epicardium is important in the planning of an epicardial mapping and ablation procedure. In 4 patients, the presence of epicardial fat limited the efficacy of the ablation procedure despite the use of an irrigated-tip ablation catheter. Epicardial VTs were effectively ablated in areas of low voltage devoid of epicardial fat, indicative of scar. If an appropriate target site for VT ablation cannot be identified despite extensive endocardial and epicardial mapping, it is possible that an epicardial VT circuit is present and that the appropriate target site is sheltered by epicardial fat. However, another possible explanation would be an intramural circuit. If there are no contraindications, MRI could demonstrate intramural scarring suggestive of an intramural circuit. MRI technology can also be used to image epicardial fat; however, in patients with implanted defibrillators MRIs are often contraindicated, and a CT should be used for fat imaging.
Cano et al. (7) described electrogram characteristics in patients undergoing epicardial mapping and ablation procedures. In patients without cardiomyopathy, a voltage cut-off of 0.94 mV separated areas with normal voltage from areas with low voltage if electrograms within the regions of the atrioventricular groove and the epicardial vessels were excluded. If these areas were not excluded, the bipolar voltage comprising 95% of the recorded sites was >0.61 mV. Fat tissue was not quantified in this study, and therefore, the cut-off value of 0.94 mV separating low from normal voltage might not have been accurate. In the present study, a cut-off value of 1.5 mV was accurate for differentiating low-voltage from normal-voltage sites. Using a lower cut-off value might underestimate the size of the arrhythmogenic substrate, and sites critical to an epicardial VT might be missed.
This was a small series, and the findings need to be confirmed in a larger study. Furthermore, although it is possible to identify fat tissue, one cannot be certain about the absence of scar below the epicardial layer of fat. However, analysis of the electrogram characteristics suggests that scarring often was present under a layer of epicardial fat. A cardiac MRI with delayed enhancement may be helpful to identify scar that underlies fat.
Registration accuracy between the CT data and the epicardial mapping data is within 3 mm, which impacts assessment of fat thickness to a minor extent. Real-time coronary angiography is, however, required before ablation to maximize patient safety.
Lack of adequate contact of the catheter with tissue might also result in low voltage. However, in the case of epicardial mapping, this was less likely to occur because none of the patients had pericardial effusions, and if an irrigated-tip catheter was used, the irrigant was removed continuously.
CT epicardial fat imaging facilitates the correct interpretation of epicardial voltage maps. The integration of extracted CT fat images into 3D EA mapping systems can also help to identify areas that are difficult to penetrate because of extensive epicardial fat.
Dr. Desjardins is now affiliated with the University of Pennsylvania, Philadelphia, Pennsylvania. Dr. Desjardins was supported by National Institutes of Healthgrant no. K23 EB006481. All other authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- computed tomography
- magnetic resonance imaging
- premature ventricular complex
- receiver-operator characteristic
- ventricular tachycardia
- Received December 29, 2009.
- Revision received March 15, 2010.
- Accepted April 13, 2010.
- American College of Cardiology Foundation
- Hummel J.D.,
- Strickberger S.,
- Daoud E.,
- et al.
- D'Avila A.,
- Gutierrez P.,
- Scanavacca M.,
- et al.
- Marchlinski F.E.,
- Callans D.J.,
- Gottlieb C.D.,
- Zado E.
- Cano O.,
- Hutchinson M.,
- Lin D.,
- et al.
- d'Avila A.,
- Houghtaling C.,
- Gutierrez P.,
- et al.