Author + information
- Received April 28, 2017
- Accepted June 1, 2017
- Published online December 18, 2017.
- Nele Vandersickel, PhDa,∗ (, )
- Alexandre Bossu, Msb,
- Jan De Neve, PhDc,
- Albert Dunnink, PhDd,
- Veronique M.F. Meijborg, PhDd,
- Marcel A.G. van der Heyden, PhDb,
- Jet D.M. Beekmanb,
- Jacques M.T. De Bakker, PhDb,
- Marc A. Vos, PhDb and
- Alexander V. Panfilov, PhDa
- aDepartment of Physics and Astronomy, Ghent University, Ghent, Belgium
- bDepartment of Medical Physiology, University Medical Center Utrecht, Utrecht, the Netherlands
- cDepartment of Data Analysis, Ghent University, Ghent, Belgium
- dDepartment of Clinical and Experimental Cardiology, Academic Medical Center, Amsterdam, the Netherlands
- ↵∗Address for correspondence:
Dr. Nele Vandersickel, Department of Physics and Astronomy, Ghent University, Krijgslaan 281, S9, Ghent, Oost-Vlaanderen 9000, Belgium.
Objectives This study investigated the arrhythmogenic mechanisms responsible for torsade de pointes (TdP) in the chronic atrioventricular block dog model, known for its high susceptibility for TdP.
Background The mechanism of TdP arrhythmias has been under debate for many years. Focal activity as well as re-entry have both been mentioned in the initiation and the perpetuation of TdP.
Methods In 5 TdP-sensitive chronic atrioventricular block dogs, 56 needle electrodes were evenly distributed transmurally to record 240 unipolar local electrograms simultaneously. Nonterminating (NT) episodes were defibrillated after 10 s. Software was developed to automatically detect activation times and to create 3-dimensional visualizations of the arrhythmia. For each episode of ectopic activity (ranging from 2 beats to NT episodes), a novel methodology was created to construct directed graphs of the wave propagation and detect re-entry loops by using an iterative depth-first-search algorithm.
Results Depending on the TdP definition (number of consecutive ectopic beats), we analyzed 29 to 54 TdP: 29 were longer than 5 beats. In the total group, 9 were NT and 45 were self-terminating. Initiation and termination were always based on focal activity. Re-entry becomes more important in the longer-lasting episodes (>14 beats), whereas in all NT TdP, re-entry was the last active mechanism. During re-entry, excitation fronts were constantly present in the heart, while during focal TdP, there was always a silent interval between 2 consecutive waves (142 ms) during which excitation fronts were absent. Interbeat intervals were significantly smaller for re-entry episodes—220 versus 310 ms in focal. Electrograms recorded in particular areas during NT TdP episodes had significantly smaller amplitude (0.38) than during focal episodes (0.59).
Conclusions TdP can be driven by focal activity as well as by re-entry depending on the duration of the episode. NT episodes are always maintained by re-entry, which can be identified in local unipolar electrograms by shorter interbeat intervals and smaller deflection amplitude.
Torsade de pointes (TdP) is a specific type of abnormal heart rhythm that can lead to sudden cardiac death and is responsible for 20% of the sudden unexplained deaths (1). Two definitions are being used for TdP, first, as a polymorphic ventricular tachycardia (PVT) that exhibits distinct characteristics on the electrocardiogram (ECG) such as varying amplitude or typical twisting, “twisting of the Pointes” (2). Second, in clinics, any arrhythmia in the setting of long QT (LQT) syndrome ranging from (multiple) premature ectopic beats to monomorphic ventricular tachycardia to PVT that is either self-terminating or can degenerate in ventricular fibrillation (VF) is often considered a TdP. In this paper, we used the latter definition and investigated the corresponding ECG characteristics of each episode. Prolongation of the QT interval is associated with an increased risk of developing TdP. The prolongation can be inherited, leading to different LQT syndromes (LQT1 to LQT15) (3). These arise from mutation of 1 of the genes, which via altered channel dynamics, cause prolongation of the duration of the ventricular action potential (APD) and thus lengthening the QT interval. In addition, prolongation can be a dangerous side effect of drugs (acquired LQT syndrome). TdP as a prescription drug side effect has been a major liability and reason for withdrawal of medications from the marketplace (4); examples include terfenadine, sertindole, astemizole, grepafloxacin, cisapride (5). In many cases, this effect can be directly linked to QT prolongation mediated predominantly by inhibition of the hERG channel. Whether re-entry or abnormal impulse formation (focal mechanism) or both are involved in TdP is still unclear, but of paramount importance to understand the mechanism of this arrhythmia and possible therapies.
In recent years, it has been argued that the start of TdP is due to a focal beat, which is most probably related to early afterdepolarizations (EAD), which occur due to reduced repolarization reserve (6–13). However, the mechanism underlying TdP perpetuation is still under debate. In animal experiments, some investigators claim that a focal mechanism is the basis of the perpetuation of a TdP (7,11–16), and others claim that nonstationary re-entry is responsible (9,10,17). In other studies, both mechanisms are observed (6,8,14). Also in theoretical studies, both mechanisms were proposed. These studies show that for re-entrant activity, the typical twisting ECG pattern may be due to the drift of a re-entrant circuit (18–21) or meandering of a re-entry spiral wave (22). In cases of focal mechanism, twisting of the ECG may be caused by multiple shifting foci generated by EAD (23,24), or a competition between focal beats generated by fixed heterogeneities with reduced repolarization reserve (25).
Therefore, important questions are how the mechanism of TdP could affect its dynamics and whether there are any specific markers that could determine the mechanism of TdP. In this study, these questions were approached by the application of novel directed-graph algorithms to these unique experimental data obtained in a highly susceptible arrhythmogenic model: the anesthetized chronic atrioventricular block (CAVB) dog model, which is known for its contractile, structural, and electrical adaptations (26). Eventually, the ventricular remodeling (27) reduces repolarization reserve and subsequently predisposes about 70% of canine hearts to develop TdP as a result of a dofetilide challenge.
Detailed electrical mapping experiments
All animal experiments were performed in accordance with the guidelines formulated by the European Community for the use of experimental animals (EU Directive 2010/63/EU) and with approval from the Committee for Experiments on Animals of Utrecht University (the Netherlands). After pre-medication including atropine, methadone, and acepromazine, anesthesia was induced in 5 CAVB dogs by pentobarbital and maintained by isoflurane 1.5% in a N2O:O2 mixture (2:1). After exposing the heart, 56 needles were inserted in the left ventricle (LV) and right ventricle (RV) and in the septal wall as previously described (28): 30 needles in the LV; 18 in the RV; and 8 in the septum, see also Online Figures 1 and 2. All needles were composed of 4 electrode terminals, each recording unipolar electrograms (ActiveTwo system, Biosemi, Amsterdam, the Netherlands). Dofetilide (25 μg/kg/5 min) was then administered in an effort to induce TdP. If arrhythmia episodes lasted over 10 s, the heart was defibrillated.
In the previous studies performed in CAVB dogs, a TdP episode was defined as an arrhythmia of at least 5 ectopic beats (not including the first paced beat). Because this definition is based on a rather arbitrary number of ectopic beats on one hand, and that this study investigates the association between the mechanism(s) involved with duration of episodes on the other hand, we have chosen to also take into account episodes of <5 beats. In addition, as the definition of TdP is not clearly stated, and very often is interpreted as any arrhythmia in the setting of the LQT syndrome, in this study we also took into account all observed arrhythmias with or without polymorphism or this typical twisting. The twisting was defined as a QRS peak that is pointing upward in one beat and downward in the next beat or vice versa.
All the information about the analysis of the signals of the electrodes can be found in the Online Appendix.
Detection of re-entry loops
For a given activation sequence, re-entry loops were automatically detected in the following way. First, for each electrode all the neighbors were found by Delaunay triangulation (29,30), with respect to their 3-dimensional (3D) coordinates. In addition, we connected all the points in the layer closest to the apex, as it is possible for the electrical signal to travel any path via the apex. Second, for a certain time tl, activation times of all electrodes were selected that had activation times >tl, whereby the first activation of an electrode after tl is called ti. Then, a graph was created, see Online Figure 3, which connected all neighbors that had the following difference in activation times:(1)whereby t1, t2 are the activation times of 2 neighboring electrodes, d1,2 is the distance between these 2 electrodes and CVmin is the minimal conduction velocity in the tissue. Notice that these connections are unidirectional: t1→t2. As for this particular graph, only 1 activation time per electrode was chosen, we created a second graph in exactly the same way for time (ti + 1 ms) and merged these 2 graphs. Finally, loops were detected by using an iterative depth-first-search algorithm as described by Johnson (31). If no re-entry loops were found, the episode was classified as focal. An important parameter is the CVmin. The conduction velocity was measured in CAVB dogs with an epicardial multielectrode (14). The conduction velocity parallel to the fiber direction was 0.56 and 0.51 mm/ms for the RV and LV and 0.34 and 0.23 mm/ms, respectively, for the conduction perpendicular to the fiber direction (14). These values are higher than the control values due to the hypertrophy of the CAVB hearts. In the current experiments, we measured 0.4 mm/ms as the smallest conduction velocity. Therefore, we have chosen 0.2, 0.3, and 0.4 mm/ms for determining loops and to assess the effect of conduction velocity on the number of loops. As 0.2 mm/ms is rather small, and distances between needles can be sometimes larger than 3 cm (needles in the septum connected with the neighboring needles), giving rise to possible accepted time differences of 150 ms between certain needles, we also repeated all the analyses by setting a maximum distance of 20 mm between the needles and a maximal time difference of 50 ms between the needles. For all episodes, loops were continuously detected each 100 ms.
This novel method is an enhancement of the activation maps approach. As the first step, we always made activation maps, however, we added connectivity graphs to the activation maps and applied automatic algorithms for the identification of the re-entrant loops.
Additional criteria for the identification of focal activity versus re-entry
1. Silence time: the interbeat interval at a given time minus the total activation time; in other words, this number represents the time between the latest activation of 1 beat and earliest activation of the next beat.
2. Interbeat interval: the arrhythmia period.
3. Amplitude: the amplitude of the local electrograms for each beat.
In all comparisons (see Online Tables 1 and 2), the parameters are expressed for the following: 1) all the ectopic beats of the focal mechanism marked in Online Table 2 (red in Figure 1); 2) all the re-entry beats of the mixed focal and re-entry cases also marked in Online Table 2 (green color in Figure 1); and 3) all the re-entry beats of the nonterminating (NT) cases of Online Table 1 (blue color in Figure 1).
A statistical model was developed to predict the mechanism (re-entry versus focal) based on the interbeat interval, a quantity that can be assessed solely by the ECG. As a first step in building the prediction model, we randomly split the data in 2 parts: the training data and the test data. The training data consisted of two-thirds of the original dataset; the test data consisted of the remaining one-third. A logistic regression model with the interbeat interval as a predictor was fitted to the training data. Please refer to the Online Appendix for more details.
We analyzed recordings of 54 arrhythmic episodes in 5 dogs: 29 were longer than 5 beats and 9 TdP episodes (in 4 animals) were NT (longer than 10 s).
Re-entry loops were identified and the robustness of the loop procedure was tested. Because minimal conduction velocity and electrode distances are crucial parameters for loop estimation, several values were used (see the Methods section).
Data from application of the re-entry-loop algorithm for the 9 NT episodes of TdP are shown in the Online Table 1. In short, re-entry loops were found for all 9 episodes starting with beats 6 → 7, 5 → 6, 3 → 4, 11 → 12, 3 → 4, 3 → 4, 7 → 8, 6 → 7, and 4 → 5 from the loop-finding algorithm with CV = 0.4 mm/ms (first paced beat included). In comparison, the 3D visualizations show re-entry with beats 4 → 5, 3 → 4, 3 → 4, 5 → 6, 2 → 3, 2 → 3, 6 → 7, 6 → 7, and 4 → 5, and lie therefore close to each other.
In Figure 2, an example of a clear re-entry path is shown in the 3D interpolated heart. In this case, 1 rotation of the re-entry wave exactly spans 1 QRS complex. In Online Figure 4, we showed the corresponding activation map of this re-entry path (see Online Figures 5–7 for some corresponding ECG traces).
For the 45 self-terminating episodes (see Online Figures 8–11), we summarized the results in Online Table 2. We found contribution of re-entry loops in only 5 cases intermingled between focal activation before (respectively, 1, 7, 3, 1, 2 focal beats) and after (respectively, 4, 4, 3, 3, 6 focal beats), see also Online Figure 12.
In Figure 3, we present an example of an ectopic beat triggered by focal activity found in dog 5 for the episode t = 570 to 575 s after the start of the measurements. At time t = 572.835 s, the beginning of the focal beat is indicated by a black arrow (apex of the LV anterior wall) (Figure 3A). From here activation moves toward the LV lateral wall (Figure 3B) and via the posterior wall (Figure 3C) to the anterior wall of RV (Figure 2D) where activation ends and tissue repolarizes again (Figure 3E). At time t = 573.11, a new focal activation emerges from the same LV anterior apical location (black arrow) (Figure 3F). In Online Figure 8, we have given the corresponding activation maps of these 2 beats, which again show that these beats are focal.
From the loop-finding algorithm, it is clear that the larger the number of beats, the greater the chance to find re-entry. In Figure 1, we summarized the results of the loop-finding algorithm: we used the results with the time filter and velocity (0.3 to 0.4 mm/ms), which mostly coincided with 2D and 3D visualizations. In Online Figure 12, we presented for each episode how many beats were focal versus how many beats were re-entry beats.
Silent time in between beats
In cases of re-entrant activity, an excitation wave front is present at every time instant. However, in cases of focal activity, we may have a silent time during which wave fronts are absent. In Figure 4, the boxplots of the silent times are shown: First, for the focal beats (group 1), the median silent time is 142.42 ms (interquartile range [IQR]: 95 to 239 ms), for the terminating re-entry beats (group 2), it is 0.48 ms (IQR: −20 to 31 ms), and for the NT re-entry beats (group 3), it is −33.08 ms (IQR: −63 to −14 ms). All pairwise comparisons of these medians are significantly different (all p values <0.0001). Comparisons within group 1 are described in the Online Appendix.
In rare cases in group 1 the duration of the beat was larger than the interbeat interval, meaning that the next beat started before the end of the previous beat, giving rise to a negative number in Figure 4 (left). In most cases, the interbeat interval is a positive number. In contrast, the silent time is usually around 0 or is negative for the re-entry cases, groups 2 and 3, meaning that there is no quiescent time between the different beats when re-entry is present. In other words, not all electrodes are activated within a certain interbeat interval. This difference reflects itself (Online Videos 1 and 2), showing that there is time deprived of activation in between the different focal beats versus continuous activity in the cases for which re-entry loops were present.
Alternative parameters to discriminate between the arrhythmogenic mechanisms
It is normally assumed that re-entrant arrhythmias are fastest: lower interbeat interval (Figure 5). In Figure 5 (left), a histogram of all the interbeat intervals per dog is shown. The right column depicts the value of the interbeat interval after 5 s (or the last value available) of the start of re-entry and the evolution of the interbeat interval once the re-entry started. These data reveal that the interbeat interval during re-entry gradually decreased, whereas after 5 s, it is lower than the shortest interbeat interval from a focally activated episode. The median interbeat interval of group 1 is 310 ms (IQR: 257 to 376 ms), of group 2 is 221.65 ms (IQR: 215 to 231 ms), and of group 3 is 220 ms (IQR: 207 to 237 ms). The median of groups 1 and 2 is statistically different (p < 0.0001). Groups 1 and 3 are also statistically different (p < 0.0001), and there is no significant difference between groups 2 and 3 (p = 0.89). Therefore, we find that re-entry beats have a smaller interbeat interval than do the focal beats.
Amplitudes of the signals
We noticed that in the cases that re-entry loops are exhibited, often some signals with lower amplitudes are found as depicted in Figure 6, whereas for pure focal episodes, the amplitudes are usually larger, as in Online Figure 1. Comparing the 5% lowest amplitudes for each beat for the 3 different groups revealed the following: for group 1, 0.59 (IQR: 0.52 to 0.66); for group 2, 0.52 (IQR: 0.45 to 0.64); and for group 3, 0.38 (IQR: 0.34 to 0.41). For groups 1 and 3, this difference is significant (p < 0.001), but it is not significant for groups 1 and 2, (p = 0.45) or groups 2 and 3 (p = 0.03). These data are shown in Figure 7.
The lower amplitudes of some of the electrodes in cases of re-entry could be explained by the fact that we are measuring inside the core of a spiral wave (34). We have demonstrated this in Online Figure 13, where we simulated a spiral wave and recorded a unipolar signal inside (green and red arrows) and outside (blue arrow) the core of the spiral wave. We see indeed that the amplitude of the signal decreases and become wider closer to the center of the spiral wave, similarly as in Figure 6. In comparison, during focal episodes, the amplitude does not change, or even becomes larger (see the red arrow and blue arrow in the bottom of Online Figure 13). In addition, we checked indeed, that for the case of Figure 2, where the center of the spiral wave was easily detected manually, the signal of the amplitude was indeed lower in the core of the spiral wave and larger outside the core of the spiral wave. However, with our current software, this could not be checked automatically and is beyond the scope of the paper. Another possible explanation is that in cases of multiple re-entry spirals, or more chaotic waves, in some electrodes, the signals come from different sides, which automatically reduces the amplitude of the signals on the electrodes.
Based on the interbeat interval, the prediction model identifies the beats as from re-entrant or focal origins. The quality of these predictions was evaluated on the test data. The model exhibits good predictive power: the positive predicted value is 76% with a negative predicted value of 90%. This means that of all the beats that are predicted as focal, 76% of them were truly focal. For all beats labeled as re-entry, 90% were genuine re-entries.
These results illustrate that the mechanism can be predicted based on the information given by the interbeat interval. As a consequence, the uncertain data can now be predicted with this model, see the Online Appendix, Online Figure 14, and the Online Table 3.
ECG and TdP
As a quantitative definition of a typical TdP is usually unclear in the published reports, we shall add a qualitative comment on the ECG recordings with respect to the mechanism.
First, for the NT cases, in 7 of 9 of the cases, we found the typical varying amplitude, as shown in Online Figure 5. Qualitatively, smaller amplitudes are usually accompanied by more chaotic wave patterns. Also notice that in this case, smaller amplitudes were observed simultaneously in all the leads; however, this is not always the case. In 2 of 9 cases, we found a different type of ECG without varying amplitude. First, for 1 episode, we found a more monomorphic ventricular tachycardia, as shown in Online Figure 6. In this case, the waves were less chaotic, and there was anatomical re-entry around the LV. Second, as shown in Online Figure 7, we also found 1 episode where the amplitude did not change, but the signal drifted. In this case, we observed clearly a single spiral that was slightly meandering in around the LV posterior and the septum posterior. This behavior is similar to the pseudo-ECG computed in Abildskov and Lux (19), where a spiral wave that drifts due to a heterogeneous medium in 2D.
Second, for the terminating cases (see Online Table 2, the certain cases), we found twisting in 14 of 35 cases, as shown in Online Figure 9. This twisting can be related to the different origins of the focal beats; however, in some cases, we observed twisting even when the focal beat came from the same region. In 21 of 35 cases, the ECG signature is more polymorphic, as shown in the example in Online Figure 10. For the 4 terminating cases, where re-entry loops were detected, we found twisting of ECG signal (Online Figure 11).
In summary, different behaviors are possible, as are more polymorphic ECG and typical ECG with greatly varying amplitude and twisting, although in the case of re-entry loops, the latter behavior was observed more often. Therefore, looking qualitatively at the ECG, it is not possible to conclude the mechanism. Also, there is a clear need to quantitatively analyze ECG and try to connect it to the mechanism. However, this is beyond the scope current study.
In summary, we found that the initiation of every episode is generated by a focal mechanism. This is in accordance with the published reports, which claim that a TdP always starts with an EAD-depended focal beat (6–13). The underlying mechanism of the perpetuation consists of different types: 1) NT TdP (>10 s), which are all perpetuated by re-entry; 2) spontaneously terminating TdP that were perpetuated by purely focal activity; and 3) spontaneously terminating TdP that had mixed perpetuation: focal–re-entry–focal. The duration of the TdP indicates the mechanism (Figure 1): longer-lasting TdP (>14 beats) usually have re-entry as the underlying mechanism, and re-entry could be distinguished from focal activity in 4 different ways: 1) using a novel algorithm based on directed graphs that can detect loops when fed with activation sequences; 2) the interbeat interval is smaller in case of re-entry; 3) during re-entry, we found that there is always activation present, whereas for a focal episode, there is a mean quiescence of 142 ms in between beats; and 4) in cases of NT re-entry, the amplitudes of certain signals are substantially smaller. Interestingly, these different mechanisms could appear in the same dog, for example, in dog 1, self-terminating focal episodes as well as nonsustained re-entry and sustained re-entry were possible. Finally, we also created a prediction model, which allows us to accurately predict the mechanism (focal vs. re-entry): the positive predictive value is 76% for a negative predictive value of 90%.
This result explains the long-lasting debate on whether re-entry or focal origin is the underlying mechanism of TdP, as we found that both mechanisms are possible and they can both give rise to varying amplitudes in the ECG, which is considered as the signature of a “typical” TdP. Part of the confusion in published reports may be because the signature of TdP is usually based on the interpretation of the observer, as there exists no clear quantitative general definition. Many papers use TdP and PVT as different arrhythmias, but it is clear from this study, that it is difficult to separate these 2. From this study, it also follows that having re-entry in an episode is dangerous as it is possible that it does not stop spontaneously.
Literature: differences based on definitions?
Therefore, in the light of our results, it is interesting to review the published reports regarding the mechanism of TdP. In the canine model in El-Sherif et al. (8), the sodium channel inactivation is slowed, creating a sustained inward current during the plateau and prolongation of APD. Their study also proposed 2 possible mechanisms of TdP. First, the typical twisting of a TdP was assigned to varying orientations of the re-entry wave. Interestingly, the sustained episodes always had re-entry activity, thereby agreeing with our results, but also nonsustained episodes were possible. Second, they also found focal activity as a possible mechanism for TdP; however, they named it a PVT. This may add to the confusion whether a TdP has re-entry as basis or focal origin. In a later study also by El-Sherif et al. (9), a surrogate canine model of the LQT3 syndrome using the neurotoxin anthopleurin-A, was investigated and here only re-entry was found as a possible mechanism of TdP. In Schreiner et al. (12), the CAVB dog model was also investigated, hereby inducing TdP with almokalant, a potassium channel (IKr) blocker. They recorded 10 episodes of arrhythmias whereby 9 of 10 were self-terminating. In 2 self-terminating cases, re-entry was found. In the NT case, the episodes were purely focal, however, this episode deteriorated into VF, and it was not mentioned whether re-entry was present during VF (or maybe just before the start of the VF). It should be mentioned that purely focal episodes could last up to 60 beats, which is much higher than in the present study. Third, in Murakawa et al. (11) and Senges et al. (13), TdP was induced by cesium chloride in dogs, and all episodes were purely focal according to their data. However, in these studies, episodes were recorded that were NT, which is not in agreement with the data from the current study, where all NT episodes show re-entry. This discrepancy can be related to the cesium chloride injection, which is maybe not a good model for TdP (35). Fourth, a study by Kim et al. (15) showed that in rabbit hearts challenged with isoproterenol, the basis of PVT was of focal origin. In their study, no “typical” TdP ECG were shown with the varying amplitude and twisting of the axis (therefore the term TdP was never mentioned). In their study, it was hypothesized that the foci had no preferred location and varied randomly in the LV and RV. The mechanism of perpetuation was shown to be related to chaotic synchronization of EAD, which generated triggered activity. It would be interesting to investigate whether the foci in our study have preferred locations connected to heterogeneities, or if they are truly random. However, it was already suggested by Dunnink et al. (16) that in the current CAVB dog model, heterogeneities might play an important role. Also in the study by Asano et al. (6), TdP was investigated in the Langendorff perfused rabbit heart by applying quinidine and E-4031. Again, the definition of TdP plays a pivotal role, as in their study a TdP was defined as a gradual undulating change on the ECG. It was found that only re-entry could cause this typical twisting, whereas there were also other purely focal episodes, but they were not categorized as TdP, rather as PVT.
In the study by Kozhevnikov et al. (10), which also investigated the CAVB dog model, 14 dofetilide-induced episodes were investigated whereby 2 were sustained. In their study also, the first 1,2 beats were focal, but then all episodes showed re-entry. When the re-entry terminated, the TdP episode stopped, indicating that it did not end by a focal mechanism as we found in the current study. However, re-entry was detected manually, so it might have been overlooked. Also in the study by Fadi et al. (17), re-entry was hypothesized as the mechanism for TdP in a canine model and related to the existence of M-cells. However, they only investigated wedges instead of the whole heart.
In a previous study by Boulaksil et al. (14), it was found that TdP were mainly perpetuated by focal activity, however, NT TdP were not investigated. Here we intensified this study by giving 4 different methods to distinguish re-entry from focal mechanism. Also, this is the first time that in mapping experiments, loops were found by using automatic algorithms that detect loops by creating directed graphs (31) based on the activation times. Moreover, this method could be very powerful and used for any mapping experiment to detect loops, also for atrial fibrillation and VF. Directed graphs can be made depending on the conduction velocity, which would add extra information in comparison with phase-mapping technologies.
The fact that re-entrant patterns organize sustained cardiac arrhythmias was also shown by Krummen et al. (36). They found that sustained VF in patients with and without structural heart disease is organized by functional re-entry patterns (rotors) and stability of VF correlated with rotor stability.
Mechanism in humans
With the help of these methods, we could potentially uncover the mechanism of TdP in humans, which has never been studied before. From the ECG trace alone, we could determine the interbeat interval, which coincides with the distance between the different peaks in the ECG. If we find a gradual decrease of the interbeat interval as indicated in Figure 5, this could indicate re-entry. Also, the prediction model we created showed that the interbeat interval can be a good indicator of the mechanism in TdP. In addition, if it were possible to do inverse mapping experiments in patients with, for example, inverse ECG (ECGi) (37), we could also trace the amplitudes on the epicardium and find whether there are certain locations in the heart where the local ECG shows smaller amplitudes as in Figure 6. This could provide a breakthrough in our understanding on the mechanism of TdP in humans and possible therapy.
Another interesting observation is that we find the typical twisting of the points in the ECG in focal as well as in re-entry episodes. Although it is beyond the scope of this paper, it would be interesting to uncover ECG characteristics depending on the mechanism, which could again help to determine the mechanism in humans.
Many questions still remain unsolved. First, how does a spiral wave start? Our hypothesis would be that a correctly placed focus can cause re-entry due to conduction block (like an S1-S2 pacing protocol). This is because in some cases, we do get re-entry (9 NT + 5 terminating episodes), but in most cases, we only found focal activity (40 cases). Even more importantly, once re-entry is established, why does it sometimes stop spontaneously? Understanding this mechanism could be of paramount importance for clinical applications, as it could help developing new therapeutic methods for LQT syndromes or other patients with risk of developing TdP. Now, as a therapeutic treatment, if medication alone is not successful, implantable cardioverter-defibrillators are implanted, which can be painful when delivering a shock. Interestingly, we can now understand why defibrillation of this arrhythmia is successful, as it is basic knowledge that spiral waves are terminated after synchronization of the tissue.
Another question is why the focal activity sustains up to even 14 beats, as in dog 2, 695 to 700 s after start of the measurements (Online Table 2). In this case, these 14 beats even originated from the same location. In other cases, the locations can switch during the episode. One possible explanation might be that there exist local heterogeneities, which have reduced repolarization reserve in comparison with the surrounding tissue. In the CAVB model, it was demonstrated that the first focal beat always originated from the area with the steepest gradient in APD (16). In computer simulations, it was also shown that heterogeneities with reduced repolarization reserve in comparison with the surrounding tissue can produce focal beats on a global reduction in repolarization (25). However, it was not shown why they would spontaneously stop. A possible argument might be that because of the faster heart rate, APD shortens, and EAD are less likely to be formed.
It should be noted that the spacing between the different needles is rather large, 1.2 cm, and the needles do not measure the activity in the Purkinje system. Potentially, many loops could be missed in the loop-finding algorithm due to these large distances or via the Purkinje system. However, as we have 3 different methods that show other differences between focal origin and re-entry, it strengthens the correctness of the loop-finding algorithm. In the mapping experiments, it is rather difficult to evaluate the transversal and longitudinal conduction velocity since we have no information about the fiber-cardiomyocytes orientation in the measured areas and its anisotropy. We have therefore always chosen a particular conduction velocity for all electrodes, although in principle, the conduction velocity between certain electrodes should depend on the fiber orientation.
COMPETENCY IN MEDICAL KNOWLEDGE: This study shows that re-entry is relevant for longer-lasting TdP (>14 beats) in the CAVB dog model that do not terminate spontaneously.
TRANSLATIONAL OUTLOOK: This research should be extended to humans to understand whether re-entry also is the basis for sustained TdP in humans.
Dr. Vandersickel gratefully acknowledges Pawel Kuklik for useful discussions.
Dr. Vandersickel is funded by Research Foundation–Flanders (FWO). All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors' institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- duration of the ventricular action potential
- chronic atrioventricular block
- early afterdepolarization(s)
- interquartile range
- long QT
- left ventricle
- polymorphic ventricular tachycardia
- right ventricle
- torsade de pointes
- ventricular fibrillation
- Received April 28, 2017.
- Accepted June 1, 2017.
- 2017 American College of Cardiology Foundation
- Tester D.J.,
- Ackerman M.J.
- Noble D.
- Asano Y.,
- Davidenko J.M.,
- Baxter W.T.,
- Gray R.A.,
- Jalife J.
- El-Sherif N.,
- Caref E.F.,
- Yin H.,
- Restivo M.
- El-Sherif N.,
- Chinushi M.,
- Caref E.B.,
- Restivo M.
- Kozhevnikov D.O.,
- Yamamoto K.,
- Robotis D.,
- Restivo M.,
- El-Sherif N.
- Murakawa Y.,
- Sezaki K.,
- Yamashita T.,
- Kanese Y.,
- Omata M.
- Boulaksil M.,
- Jungschleger J.G.,
- Antoons G.,
- et al.
- Dunnink A.,
- Stams T.R.G.,
- Bossu A.,
- et al.
- Akar F.G.,
- Yan G.X.,
- Antzelevitch C.,
- Rosenbaum D.S.
- Gray R.A.,
- Jalife J.,
- Panfilov A.,
- et al.
- Winfree A.T.
- Sato D.,
- Xie L.H.,
- Sovari A.A.,
- et al.
- Vandersickel N.,
- de Boer T.P.,
- Vos M.A.,
- Panfilov A.V.
- Fleuren M.A.,
- van Dommelen P.,
- Dunnink T.
- O’Rourke J.
- Koenker R.
- R Core Team
- Allessie M.A.,
- Bonke F.I.,
- Schopman F.J.
- Krummen D.E.,
- Hayase J.,
- Morris D.J.,
- et al.
- Ghanem R.N.,
- Burnes J.E.,
- Waldo A.L.,
- Rudy Y.