Re-Interpreting Complex Atrial Tachycardia Maps Using Global Atrial Vectors

The mapping and ablation of Atrial Tachycardias (AT) can be challenging, particularly in patients with prior ablation or structural atrial disease. Entrainment is the foundation to confirm anatomically-based reentry but, in patients with structural disease, may transform or terminate AT. Activation mapping is thus often used in parallel. In recent years, high resolution catheters have increasingly been used to reveal micro-reentry at sites that appeared focal¹, to improve definition of gaps in scar or ablation lines² and to improve mechanistic definition³. In principle, it should be straightforward to use electroanatomical systems to mark electrograms and create isochrones for stable AT to guide ablation. In practice, it can be difficult to mark activation in electrograms that are fractionated or have low amplitude, which are common at sites of scar or prior ablation where AT often arise. It can thus be difficult to determine which activation sequence best represents arrhythmia and should be targeted for ablation. This remains a major clinical challenge, and there is a real need for objective tools to improve this interpretative process.

Using Global Atrial Patterns to Interpret Atrial Tachycardia Maps

In this issue of the Journal, Kuroda et al. [Kuroda JCE 2021 – Editor please insert] systematically evaluated the novel “Coherent™” algorithm that uses global patterns of propagation to resolve ambiguities in electrogram marking and create a “best-fit” map during stable arrhythmias. In N=77 patients with scar-related AT, the team retrospectively compared this novel algorithm with standard activation mapping, referenced to the gold standard of full clinical assessment (entrainment and acute ablation outcome). The novel algorithm more accurately determined AT location and mechanism (macro-reentry, localized reentry and focal tachycardia) in single (67.2% vs 44.8%, P=0.009) and dual-loop circuits than standard mapping, accurately reclassified mechanisms in cases when standard mapping failed, and reduced inter-observer variability.

Technical Approach

This global vector-based algorithm was first reported by Anter et al. ⁴ and has several components. First, the algorithm quantifies AT activation globally within the chamber, and uses these data to modify annotation of individual electrograms to produce a physiologically plausible final map. Second, the algorithm presents maps of vector propagation, that can differentiate focal or reentrant propagations from dead-end pathways, and of conduction velocity, that can identify slow conducting isthmuses of reentry. Finally, the authors write that the resulting activation map is independent of the mapping window, and displays color coded isochrones without predefined early and late activation. This algorithm has been shown to identify rotational activity at sites of complex electrograms where traditional mapping suggests focal activity (figure, from Anter et al. ⁴).

Figure. Atrial tachycardia by standard activation mapping (left) and algorithmically reinterpreted maps (middle) focused on a site of complex electrograms (right).

Electrograms at this site (right atrium near the coronary sinus) are difficult to interpret in isolation. The novel algorithm found that marking these electrograms to produce highly curved vectors of propagation i.e. localized reentry best resolved the global vector patterns. Ablation here terminated AT. Standard mapping less accurately portrayed this map as a focal mechanism. Reproduced with permission from figure 4 of Anter et al ⁴.

Critical Appraisal of Electrogram Re-Annotation by Global Vectorial Analysis

To solve the problem of assigning activation times in multi-component electrograms, the algorithm identifies all possible markings then selects that which best reconciles global activation (i.e. provides the lowest spatial and temporal errors). This is an elegant approach. However, greater detail would be useful on how this selection is performed. If the algorithm uses rules (deterministic equations), their accuracy should be reported for different types of rhythm and patient profiles. If the algorithm uses more complex classification tools such as machine learning, the labeling and generalizability of training data are key⁵. Global vectors are sensitive to regions that are over- or under-sampled, and so signal acquisition should be as spatially complete and uniform as possible. In theory, algorithmic selection could assign a low rank to correct but ‘presumed unlikely’ maps, although such diagnostic errors were not reported in this study. By optimizing maps globally, localized patterns with a small global impact such as micro-reentry with a small organized domain could be overlooked. The micro-reentry shown in figure 1⁴ controls relatively large atrial areas, and it would be fascinating to study the sensitivity of this approach to smaller micro-reentrant domains. Extending this logic, it remains to be studied if this global vectorial approach could identify organized driver sites in AF, which may have local control yet a small impact on global activation vectors, for which several algorithms are in clinical use and evaluation^6–8.

How much information is needed to map AT?

A fundamental question in arrhythmia mapping is how much data should be integrated. Kuroda et al. should be credited not only for performing a rigorous study, but also for acknowledging that the best results were achieved by methods “combining activation mapping, entrainment and termination as determinants of true arrhythmia mechanisms” [Kuroda JCE 2021 – editor to fill in]. One could imagine future algorithms using machine learning or rule-based logic to integrate additional parameters such as electrogram amplitude to reflect vectorial direction, fractionated electrograms, beat-to-beat variability in cycle length or other characteristics. It is also important to compare this algorithm to other reported global vectorial approaches, such as those reflecting conduction velocity near scar⁹ or those currently applied to panoramic contact baskets^{10, 11}. These comparisons should ultimately be performed prospectively, and may need to be tuned to patients with or without prior ablation, with and without atrial structural remodeling, and so on.

In summary, the authors should be congratulated for this clinical study of a very innovative and physiologically plausible approach to provide automatic re-interpretation of complex atrial arrhythmia maps. We eagerly anticipate future developments in this field.