Comprehensive Triple Scar Characterization With MDCT to Guide VT Ablations in ICD Patients
Ventricular tachycardia (VT) ablations are one of the emerging frontiers in clinical electrophysiology. Multiple clinical trials have extended the indications for life-saving defibrillator therapy to an increasing number of patients. Apart from those with previous life-threatening arrhythmias (secondary prevention), a large number of patients with a structurally abnormal heart benefit from placement of an ICD without previous episodes of ventricular tachycardia or fibrillation (primary prevention).
A large number of those patients present in the follow-up period with frequent and appropriate defibrillator shocks for ventricular arrhythmias. Indeed, VT storm (≥3 shocks per 24 hours) has been observed in ~4.5% of patients with primary and up to 40% of patients with secondary prevention. While pharmacological therapy is usually the first line therapy in this patient population, its long-term usage is often limited by efficacy and side effects, making VT ablation the next appropriate step. Moreover, at least two multi-center trials have also demonstrated a significant decrease of future ICD shocks with “prophylactic” VT ablation at the time of ICD implantation in patients with ischemic heart disease.
Therefore, a clinical need exists to make VT ablations faster, safer and more efficient to meet the growing clinical need.
However, procedure times frequently still surpass 5 hours, and a recurrence rate of 50+% in long-term follow-up is common. If structural heart disease is present, VT ablations target myocardial scar, which is the substrate for most reentrant VTs. In cases in which the ventricular arrhythmias are hemodynamically stable, standard entrainment criteria are used to determine the electrical circuit within the scar. Ablation within the critical pathway or at the exit site of the VT often results in acute elimination of the arrhythmia. However, up to 90% of VTs are either non-sustained or hemodynamically unstable and require a substrate-guided approach. In this case, the amplitude of bipolar voltage recordings is used to classify a location point as healthy myocardium, scar or border zone (“voltage mapping”). Using a clinical mapping system, up to several hundred of those mapping points are combined to build a voltage map of the heart. It provides information about the anatomic location and extent of the scar and border zone. An ablation catheter is used to pace within the scar and along its border zone trying to match the 12-lead morphology of the VT (“pace mapping”). An identical morphology of the pace map and VT suggests a location close to the VT exit site from the scar and ablation lesions at that site have a good chance to eliminate the VT.
However, voltage mapping has several important limitations. Low voltage recordings can represent true myocardial scar, but can also be the result of poor catheter contact with the cardiac muscle. Voltage mapping can take >1-2 hours and prolong the procedure times. This also limits the average spatial resolution of voltage maps, and small areas of scar between two mapping points may be overlooked. Importantly, a single endocardial voltage recording is a poor surrogate of a complex intramural scar with variable degrees of scar transmurality or patchy myocardial fibrosis. Therefore, a detailed assessment of myocardial scar using a different approach than voltage mapping alone has the potential of providing helpful information and improve VT ablations.
Cardiac imaging has the ability to non-invasively differentiate normal from abnormal myocardium. Of special interest is multi-detector computed tomography (MDCT), which has recently undergone rapid technical advances improving imaging quality and reducing radiation/contrast use. MDCT does not have the relative contraindication in ICD patients that MRI has, and offers a spatial resolution that surpasses the 6mm resolution, which can be achieved with PET imaging. Indeed, modern CT imaging protocols achieve a sub-millimeter resolution, which allows the differentiation of the heart wall into the endocardium, epicardium and mid-myocardium.
A recent study by Tian et al1 examined the potential of 3D CT imaging to facilitate VT ablations. In this novel approach, the authors performed triple assessment of anatomic, dynamic and perfusion characteristics obtained from a single CT scan to achieve a detailed scar characterization.
Eleven patients with ischemic cardiomyopathy that were scheduled to undergo a clinically indicated VT ablation at the University of Maryland were enrolled in the study. Patients were on average 65 years old and had advance heart failure with an ejection fraction of 22%. A single CT scan was performed using a 64-slice Philips Brilliance CT scanner. For the anatomic evaluation, regional end-systole wall thickness (ESWT) and end-diastole wall thickness (EDWS) was determined at 40% and 80% of the cardiac cycle. Dynamic parameters including segmental wall thickening (WT) and wall motion (WM) were defined on short axis slices and extracted for further processing. Perfusion imaging was obtained after IV contrast injection and perfusion defects were hand-planimetered on 2D images by an experienced radiologist. Average radiation exposures were 12.35±3.67 mSv.
To assess if those parameters could be used to predict the voltage-map defined myocardial scar, a 17-segment model of the bipolar voltage map was compared to the CT parameters. First, the authors1 assessed the extracted anatomic data. Areas of normal voltage demonstrated a preserved left ventricular wall thickness of 10.1±1.8 mm and 12.9±1.7 mm during end-systole and end-diastole. In segments of decreased voltage, the myocardial wall was significantly thinner with 6.5±0.75 mm and 6.6±0.94 mm (p
When the investigators assessed the predictive value of the dynamic parameters, they found that areas of preserved voltage demonstrated normal dynamic parameters with wall thickening of 32.7±5.5% and wall motion of 3.5±0.92 mm. With abnormal voltage the myocardium lost most of its wall thickening (0±4.4%) as well as its wall motion (1.1±0.4 mm), which again was highly statistically significant.
To assess which parameters might be the most clinically useful, Tian et al1 performed further analysis using receiver operating curves. Area under the curves for ESWT, EDWT, WT, and WM were 0.83±0.05, 0.75±0.06, 0.79±0.06, and 0.68±0.06, respectively. The best prediction could be achieved by combining ESWT and WT, which resulted in an area under the curve of 0.85±0.05 and predicted areas of abnormal voltage correctly in 81.7% of cases.
Of special interest in this study1 were the results of the third imaging modality, as perfusion imaging additionally allows one to determine the transmural extent and intramural location of myocardial scar. All patients were found to have areas of hypoperfusion. Thirty-eight percent of the analyzed segments demonstrated first-pass hypoperfusion, which matched in 78% with areas of abnormal voltage. As expected, in a cohort of ischemic cardiomyopathy cases, non-transmural scar was mostly seen in endocardial location (Figure 2A) and could be clearly differentiated from fully transmural scar.
Using custom-made software, three-dimensional models of abnormal anatomic, dynamic and perfusion characteristics were created, which could be integrated into the clinical mapping system (CartoMerge, Biosense Webster, Inc.). Very good registration results could be achieved with all of the registration methods tested. A combination of landmark point registration using apical, mitral valve and RV septal insertion sites and surface registration resulted in an accuracy of 3.31±0.52 mm. This allowed the accurate display of any chosen abnormal characteristics embedded into a three-dimensional model in the clinical mapping system even prior to the actual VT ablation. Location, extent and even transmurality of scar extracted from perfusion could be displayed during the ablation (Figures 2A and B).
Additional analysis was performed after registration of the reconstructed 3D CT data with the voltage map. In quantitative analysis, the area of abnormal perfusion correlated best with the area of abnormal voltage measurements including scar and border zone. Voltage mapping demonstrated an abnormal voltage area of 97.7±41.3 cm2, representing 37.4±11.4% of the left ventricle. The 3D CE-CT hypoperfusion map demonstrated a myocardial hypoperfusion area of 88.2±36.3 cm2, with an abnormal LV myocardium burden of 33.3±8.5%. Voltage measurements within the CE-CT hypoperfusion reconstruction area demonstrated a voltage amplitude of 1.14±0.23 mV compared with 8.4±5.1 mV in the area with normal myocardium perfusion (P
An average of 2.2±1.5 VTs per patient were inducible, of which 16 were deemed to represent the clinical VT. Fifteen of the 16 targeted VTs were successfully ablated and were no longer inducible at the end of the procedure. Successful ablation sites were within the area of CT hypoperfusion in 82%, and were within 10mm of the perfusion-defined border zone in >90% of patients (Figure 2C: CT scar indicated by black mesh, white arrow indicates successful ablation site). No procedure-related complications were observed. In this population, a 36% mortality was observed during 12-month follow-up and was due to chronic heart failure. Forty-five percent of patients had recurrent VT and 27% underwent a second VT ablation.
This is the first study to utilize the comprehensive information available from a single CT scan to potentially provide a roadmap for a novel image-guided approach for VT ablations. Cardiac imaging has been introduced for electrophysiological procedures over the last decade. However, pre-procedural imaging has been mostly adopted for the ablation of atrial fibrillation to assess anatomic abnormalities such as pulmonary vein morphology. Accordingly, the clinical display is limited to a solely anatomical reconstruction of the endocardial left atrial shell. Adoption of cardiac imaging for similarly complex ventricular procedures has been slow, as a simple anatomic display of the endocardial surface does not provide the information of the VT substrate that is clinically desirable. An extraction of scar and display embedded into the healthy myocardium would fulfill this need, but is not easily possible with the currently available commercial systems.
The current study by Tian et al1 demonstrates the ability to perform a triple evaluation of normal and scarred LV myocardium using a single MDCT scan and provide this information in a clinical mapping system. The triple modality evaluation of scar does not require additional procedures, and provides immediate confirmation and comprehensive characterization of the scar tissue. Among those parameters tested, anatomic information regarding phase-specific wall thickness combined with dynamic systolic and diastolic data was able to achieve the most accurate delineation of the VT scar. Additionally, first-pass perfusion imaging provided supplementary information about the scar thickness and location within the LV wall. Scar morphology such as the presence of preserved epicardial myocardium behind endocardial scar could identify potential channels of surviving heart muscle. Scar located exclusively in the midmyocardium or the epicardium may go unnoticed by endocardial voltage mapping, and CT imaging could direct mapping and possible ablation approaches.
Most importantly, all the extracted information from the MDCT was successfully reconstructed as a three-dimensional shell and could be registered with the clinical mapping system. All scar information from the triple modality imaging based on anatomic, dynamic and perfusion data can therefore be available to the electrophysiologist during the procedure. By sequentially selecting anatomic, dynamic and perfusion reconstructions, a detailed characterization is available for any chosen mapping or ablation point. Such a display can work as immediate feedback for the electrophysiologist during the various phases of the VT ablation procedure. If a mapping point displays low voltage but none of the imaging modalities suggest the presence of scar, poor catheter contact with the heart wall is likely and should trigger further mapping. The high spatial resolution of CT is superior to the standard mapping density during voltage mapping and may avoid overlooking small areas of scar. This approach may also allow reducing unnecessary mapping in CT defined normal areas and direct further mapping to myocardial scar. The direct display of corresponding scar thickness at the mapping site allows a better understanding of scar geometry and mapping signals. Nearly all of the successful ablation sites in this study were located within the border zone of the CT-defined myocardial scar. This also suggests that such a detailed imaging approach may in the future enable a rather imaging guided ablation approach, which will be tested in future prospective treatment trials.