Mapping Ventricular Tachycardia with Electrocardiographic Imaging

The study and treatment of ventricular tachycardia (VT) should be an excellent opportunity for electrocardiographic imaging (ECGI) to make a significant clinical impact. Scar-related VT is an important problem with a high burden of disease that has effective therapeutic options.¹ VT has a relatively large electrocardiographic signal, and a high signal-to-noise ratio (SNR), with a lesser impact of the cardiac conduction system on myocardial activation. There is now a substantial body of literature regarding inverse calculation of cardiac electrical activity from body surface potentials and heart-torso geometry information, often known as electrocardiographic (ECGI).² Although ECGI was used for localization of target sites for stereotactic radioablation, permitting an entirely noninvasive procedure,³ its use as a standalone means for clinical arrhythmia localization has otherwise been limited. Quantitative clinical validation of the accuracy ECG has been challenging. Graham and colleagues are to be congratulated for the assembly of a relatively large series of patients who underwent simultaneous invasive electrophysiologic study with catheter ablation, while ECGI electrodes were in place in order to quantitate the accuracy of ECGI in localizing sites of VT exit from myocardial scar.⁴ These data are valuable and difficult to acquire: real estate on the thorax during a catheter ablation procedure is at a premium, with electroanatomic mapping (EAM) system localization patches, defibrillation pads, and procedural 12-lead ECG electrodes competing for space with electrodes for body surface potential mapping (BSPM).

One might expect that ECGI would have its greatest accuracy in identifying the location of epicardial sources of origin of ventricular activation. As such, epicardial contact catheter mapping of ventricular arrhythmias should be an ideal circumstance in which to quantitate the accuracy of ECGI. In a series of 4 patients undergoing epicardial contact mapping, we identified a mean error of 13±9 mm (in the absence of myocardial scar) for localizing pacing sites using Tikhonov regularization,⁵ which improved to 8±6 mm using a Bayesian method.⁶ VT-exit site localization errors for 2 of these patients with epicardial exit VTs were 24 mm and 28 mm using the former method, and 13 mm and 9 mm using the latter method.⁶ These data are similar to the findings of Graham et al in this larger group of patients undergoing endocardial mapping and ablation, and to their prior findings in 8 patients undergoing epicardial mapping, in whom a mean error of 13.2 mm for localizing pacing sites was identified.⁷

ECGI is a refinement of standard electrocardiography. The increase from 9 electrodes to a range of 60+ to 252 electrodes and the addition of imaging-derived patient-specific anatomy and lead location increases the information content of the recording. It does not specifically include assumptions about myocardial scar function or anatomy. For the purposes of localizing VT-exit sites, does this provide a meaningful clinical advantage over the standard ECG? Most 12-lead ECG localization algorithms identify a culprit segment of activation rather than a location on the patient’s personal anatomy, providing an approximate region of interest.⁸ When comparing the accuracy of a continuous localization method, such as ECGI to a segmental localization technique, we should not be surprised that continuous localization yields better accuracy metrics. Regionalizing algorithms inherently discretize the anatomic continuity of the cardiac geometry, creating templates intended to represent a mean for each segment; the margin of one segment will inevitably have an ECG signature that is more alike to an adjacent segment margin than to its own far margin. Inclusion of personalized patient anatomy, however, along with recorded evoked potentials from known pacing sites during an invasive procedure, can achieve relatively accurate localization of sites of ventricular activation,⁹ similar to ECGI.¹⁰

In the study by Graham and colleagues, some of the challenges intrinsic to quantitating localization accuracy are evident. The identification of a reliable ground truth is difficult. It is quite reasonable to use an EAM system with reported sub-millimeter accuracy, but it must be recalled that there is likely incomplete adjustment for motion due to the cardiac and respiratory cycles, and that registration with cardiac imaging is imperfect. Furthermore, clinicians who perform catheter ablation procedures for VT will be aware that the exact scar exit site is infrequently perfectly identifiable, and indeed does not form the ideal ablation target in many instances. Utilization of entrainment mapping and including sites with stimulation-QRS intervals as long as 50% of the cycle length will include sites significantly proximal to the exit site. Identification of exit sites with the use of pace-mapping using quantitative correlation to ventricular arrhythmia ECG patterns also has limitations, as there are anatomic regional variations in the relationship between 12-lead correlation coefficient mapping and excitation source location.^{11, 12} Furthermore, the potential accuracy of localization will be limited by the resolution of the anatomic mesh utilized. There are further complexities attributable to limitations on the ability of ECGI to localize septal sites of activation, and the potential discrepancy between endocardial sources/sites identified during a clinical procedure and sites identified by ECGI which map to the epicardium. These limitations should not deter the continued pursuit of accurate methods for mapping VT circuits.

Ultimately, the most appropriate role for ECGI in this field is yet to be determined. As noted by Graham et al, ECGI, as it is available currently, lacks adequate accuracy and precision to guide catheter ablation without detailed contact mapping. It is possible, however, that it will be sufficient to guide more regional ablative therapies, such as stereotactic beam radiotherapy. Yet even for this application, it would be extremely useful if it could accurately identify the anatomy and functional characteristics of myocardial scar, to derive noninvasive arrhythmogenic substrate maps. It is not yet clear that this can be reliably accomplished.¹³ Furthermore, it would be useful if it could reliably distinguish right ventricular from septal activation, and if it could differentiate between endocardial, mid-myocardial and epicardial events.^{14, 15}

Mapping myocardial activation sites does not take advantage of the full breadth of information available from ECGI. For the purposes of catheter ablation of VT using current techniques, the most valuable information is that which can be gained within myocardial scar or at its margins. Identification of a scar exit site is useful to direct further contact mapping within or adjacent to scar, which may be used to design an appropriate ablation strategy. Details of myocardial function, activation and physiology outside of scar are typically of secondary importance, although they could be of critical importance for the purposes of risk stratification, delivery of cardiac resynchronization therapy, or assessment of need for revascularization.

Has this technology reached a mature plateau, or are there likely to be further advancements which may overcome some of the above limitations? It is possible that the incorporation of the location and morphology of myocardial scar, derived from cardiac imaging studies could improve its performance for localization of VT substrate. Furthermore, it is conceivable that signal averaging could permit greater insight into activation within or at the margins of scar, or that mapping could be combined with functional/provocative testing to gain greater insight into individual patient anatomic and functional substrate. There remains considerable reason for optimism that ECGI will provide meaningful contributions to our understanding of normal and abnormal cardiac function, as well as to clinical care.