Ablation of Myocardial Tissue With Nanosecond Pulsed Electric Fields

Background Ablation of cardiac tissue is an essential tool for the treatment of arrhythmias, particularly of atrial fibrillation, atrial flutter, and ventricular tachycardia. Current ablation technologies suffer from substantial recurrence rates, thermal side effects, and long procedure times. We demonstrate that ablation with nanosecond pulsed electric fields (nsPEFs) can potentially overcome these limitations. Methods We used optical mapping to monitor electrical activity in Langendorff-perfused New Zealand rabbit hearts (n = 12). We repeatedly inserted two shock electrodes, spaced 2–4 mm apart, into the ventricles (through the entire wall) and applied nanosecond pulsed electric fields (nsPEF) (5–20 kV/cm, 350 ns duration, at varying pulse numbers and frequencies) to create linear lesions of 12–18 mm length. Hearts were stained either with tetrazolium chloride (TTC) or propidium iodide (PI) to determine the extent of ablation. Some stained lesions were sectioned to obtain the three-dimensional geometry of the ablated volume. Results In all animals (12/12), we were able to create nonconducting lesions with less than 2 seconds of nsPEF application per site and minimal heating (< 0.2°C) of the tissue. The geometry of the ablated volume was smoother and more uniform throughout the wall than typical for RF ablation. The width of the lesions could be controlled up to 6 mm via the electrode spacing and the shock parameters. Conclusions Ablation with nsPEFs is a promising alternative to radiofrequency (RF) ablation of AF. It may dramatically reduce procedure times and produce more consistent lesion thickness than RF ablation.


Introduction
In contrast, nsPEF stimuli are too brief for capacitive charging, and displacement currents dominate over conduction currents [21,25]. Therefore intercellular electric connections do not affect (or just minimally affect) membrane charging, and every cell even in the syncytial tissue behaves as an independent entity. This situation is electrically equivalent to each cell having its own virtual cathode, and electroporation is not affected by the vicinity to the cathode. Instead, all cells within a high enough electric field area. For the same reason (not relying on conduction currents), nsPEF penetrates deeper and the E-field distribution in tissue is less affected by electrical inhomogeneities [21,25,26], resulting in more uniform ablation.
In this study, nsPEFs are applied via two thin parallel wire electrodes that penetrate the myocardial wall and ablate the volume between the electrodes. Lesions are created by repeated application (as in the case of RF ablation).

Optical mapping
The preparation was stained with the near-infrared dye DI-4-ANBDQBS. A stock solution was made by dissolving 10 mg dye in 1.2 ml of pure ethanol; for each experiment, 30 μL of the stock solution was diluted with 15 mL of Tyrode's solution and injected as a bolus. A 1000 mW, 671 nm diode laser (Shanghai Dream Lasers) was directed through a 5 degree conical diffusor and then through a dichroic mirror (λ crit = 690 nm) onto the heart to achieve uniform illumination. Fluorescence light passed the dichroic mirror and a 715 nm long pass filter and was recorded with a CCD camera ("Little Joe", SciMeasure, Decatur, GA) at 1000 frames per second.

Ablation electrodes and choice of ablation sites
Ablation electrodes were made of two parallel 250 μm tungsten needles (see Fig 1A). The electrode spacing was adjustable from 2 to 6 mm, and the electrodes were uninsulated over the terminal 4 mm and sharpened at the tip. Electrodes were dipped into surgical ink to mark the insertion sites and then inserted into the right or left epicardium (see Fig 1B), so that they penetrated the entire ventricular wall. Ablation sites were chosen at least 1 cm away from the septum. When linear lesions were desired, 4-5 consecutive ablations were performed next to each other to create a linear lesion of 12-18 mm length. transmission line and the load. In theory, this setup should charge the transmission line until the breakdown voltage of the spark gap is reached and apply rectangular pulses of duration t = 2l/v to the load, where l is the length of the transmission line and v the speed of light in the transmission line. In our case, l = 35 m, v = 0.66 c (c is the speed of light), and consequently, t350 ns [27] (see Fig 2B). The actual pulse shape was recorded with an oscilloscope (Tektronix 1001B, Beaverton, OR) and followed the theoretical prediction with good accuracy (see Fig 2C).

Ablation protocol
At each ablation site, we applied trains of 350 ns pulses (either 6 pulses at 3 Hz of 20 pulses at 1 Hz) of different amplitudes. We used an adjustable spark gap to allow the adjustment of the pulse amplitude, the pulse frequency was adjusted by adjusting the supply voltage, provided by a 0-20 kV power supply (EH Series, Glassman, Highbridge, NJ). When we generated trains of Electrode configuration for nsPEF ablation. A: Electrode geometry: Two parallel 250 μm tungsten needles uninsulated at the tip (yellow segment). B: Electrode placement in the tissue. A slab of cardiac tissue (pink) is penetrated by two needle electrodes. Upon shock delivery, an area between and around the electrodes is ablated (yellow). C: Electrode placement in a rabbit heart. Electrodes were inserted from the epicardium, penetrating all the way through the myocardial wall.  pulses, the pulse amplitude was reproducible within ± 0.2 kV over the amplitude range used in this study.

Lesion analysis
For each element of the lesion, the two surface insertion points A and B of the electrodes were identified. At the midpoint of the line AB, the extension of the lesion perpendicular to the line AB was considered the width of the lesion segment. The width of the lesion was defined as the average of the widths of the segments.

PI/TTC staining and sectioning
After the creation and electrophysiological evaluation, preparations were stained either with propidium iodide (PI, 30 mM/30 min) or tetrazolium chloride (TTC, 30 mM/20 min), for further study of the geometry of the ablated volume. For PI stains, we subsequently washed out the PI using our coronary perfusion for 40 minutes, leaving only the cells with compromised membrane stained with PI [28].
For both TTC and PI stains we cut lesions out of the tissue preparation and mounted each lesion separately in a block of agar. We then sectioned the lesions (section thickness: 300 μm). For PI stains, we recorded the fluorescence (illuminated at λ = 532 nm) for each section. By defining a fluorescence threshold that corresponds to dead tissue, the ablated area was identified.

Modeling of field distribution
We used the COMSOL Multiphysics package (COMSOL, Palo Alto, California) for finite element modeling of field distributions. We modeled the tissue as a circular, homogeneous sheet with uniform conductance and a radius of 4 cm. At the electrode locations, the sheet had two circular holes of 250 μm diameter (the electrode diameter), spaced 2 mm apart. The Laplace equation, ΔV = 0, was solved on the tissue domain for the transmembrane voltage. The boundary conditions were that at the interior (electrode) holes, V equaled the respective applied electrode potentials, and at the outer boundary, V was zero.

Single nsPEF ablations create conduction block
Ablation with nsPEFs consistently created conduction block if the field strength was sufficient. Panel B shows the same cardiac surface after nsPEF application, with two spots of surgical ink indicating the positions of the ablation electrodes. Panels C and D compare the action potential amplitudes (obtained via optical mapping) before and after nsPEF application, showing that the action potential amplitude goes to zero in the region surrounding the nsPEF electrodes. Panels E and F illustrate the propagation of excitation before and after nsPEF application. Before the application, excitation propagates in all directions, with elliptical isochrones whose long axis corresponds to the local fiber direction. After the shock, propagation in the direction of the nsPEF application is blocked, so that excitation propagates around this electrically inactive region, as shown by the smoothly changing activation times along this path, to eventually excite the tissue on the other side of the nsPEF ablation. Note that the data in Panel F do not exclude the possibility that the observed block is really just very slow conduction (which would excite the tissue on the other side of the lesion after the wave that went around the lesion arrived). This interpretation appears, however, contrived in light of the absence of electrical activity in the treated region (Panel D) and the evidence of cell death in this region presented below.
The field necessary for ablation depends on the shock parameters and electrode spacing (see Fig 4). For our default parameters (2.3 mm spacing, 6 shocks at 3 Hz), field strengths above 2.3 kV consistently caused block (n = 19). When we changed to 20 shocks at 1 Hz, a lower pulse amplitude of 2.0 kV was consistently sufficient (n = 5). For an electrode separation of 4 mm (20 pulses at 1 Hz), a pulse amplitude of 4 kV was consistently sufficient (n = 3).
In general, we found that pulse amplitudes above half the thresholds mentioned above were still sometimes successful in creating block (2/5), while shock amplitudes below 50% of the Photograph of the cardiac surface after ablation. The electrode insertion points are marked by black dots. C: Action potential amplitude map before shock application. Black corresponds to zero action potential amplitude, white to maximal action potential amplitude. D: Action potential amplitude map after shock application. E: Activation map before shock application. Colors code the time after stimulus application at which a surface element is activated. Black areas are activated first, white areas last (see scale). Small red "x" marks the stimulation site, arrow indicate the direction of propagation. F: Activation map after shock application. Activation is blocked at the site of shock application (small red dots indicate shock electrode positions. doi:10.1371/journal.pone.0144833.g003 thresholds stated above never succeeded in creating a lesion (n = 7). All these data are from single ablations; they are complemented by the data for longer lesions below (a blocking lesion implies that all individual ablations are blocking).

Effect of shock amplitude on geometry of ablated volume
When we varied the shock amplitude while keeping all other shock parameters fixed (4 mm electrodes spacing, 20 pulses at 1 Hz) and subsequently stained the tissue with PI, we found a characteristic dependency of the geometry of the ablated volume on shock amplitude, shown in Fig 4. For a small shock amplitude (1 kV/cm), we observe ablation only in limited areas around the ablation electrodes (see Fig 5A). For a larger shock amplitude (2 kV/cm), the areas around the electrodes merge so that a contiguous lesion is created (see Fig 5B).
Computed field distributions for both 1 and 2 kV/cm are shown in Fig 5C and 5D. They support the idea that tissue dies whenever the local field strength is above a critical threshold. Comparing Fig 5A and 5C and Fig 5B and 5D, we estimate that this critical threshold should be in the range of 3-5 kV/cm.  ~16 mm. Based on our experience with single application, we chose set our standard protocol to 6 pulses at 3 Hz. The field strength was set to 2.3 kV for 2.3 mm electrode separation and to 4.3 kV for 4 mm electrode separation (i.e.~10 kV/cm if calculated simply as "voltage over distance"). A representative example with 2.3 mm electrode separation is shown in Fig 6. Panel B shows that electrical activity has ceased where nsPEF was applied. Panel C shows that activation propagates around the treated region (as indicated by the arrows), and the smooth change in activation time indicates that this propagation happens with approximately constant speed. While activation also propagates from the stimulation site towards the lesion,  the abrupt jump in activation time at the lesion is strong evidence that conduction block has indeed been achieved.

Lesion statistics
We created a total of 16 lesions in 12 rabbit hearts. Based on our single ablations, we expected consistent ablation success with low individual ablation times (< 2 s), and indeed, all lesions created were nonconducting (16/16). For 2.3 mm electrode separation, we assessed the lesion width both with PI stains and TTC stains (see Fig 7). For PI stains (n = 7) we found an average width of 2.3+/-0.2 mm, while with TTC stains (n = 7), the average was 3.1+/-0.56 mm. For 4 mm electrode separation (n = 2), the evaluation of the lesion with TTC staining gave a thickness between 5 and 6 mm.

3D Geometry of lesions
We also investigated the 3D geometry of selected lesions we created. Fig 8 shows TTC staining in a series of sections from epi-to endocardium (Panel A) and a 3D reconstruction of the lesion geometry (Panel B). The section geometry is far more consistent from epi-to endocardium than is typical for RF or cryoablation [10]. Also, the boundary of the ablated region is very sharp; the zoomed picture (Panel C) shows that even at the level of individual cardiac fibers there is a very abrupt transition from unstained to almost fully stained tissue.

Ablation Speed
We were able to create non-conducting lesion with 6 pulses delivered at 3 Hz in every attempt. This corresponds to a treatment time of 1.67 s per location. We also tried to achieve ablation with single pulses, but even at field strengths of 5-6 kV, the maximum that our generator could supply, we were not able to create lesions consistently.

Thermal effects
The amount energy deposited per pulse can be computed from the shock parameters and the generator capacitance (3.1 nF), and it is less than 8 mJ (for 2.3 kV, a typical pulse amplitude). Even with 20 pulses the total energy is below 200 mJ (compared to typically~1,000 J per site for RF ablation). In direct experimental measurements at the midpoint between the electrodes, we were not able to detect any temperature increase after the shock, indicating that the increase was smaller than~0.1°C.

Discussion
We have demonstrated that using nsPEFs, we can reliably ablate myocardial tissue with more consistent lesion cross section than RF/cryoablation, in a fraction of the time, and without thermal side effects. Lesion width was consistent and could be adjusted between 2 and 5.5 mm by choosing different electrode distances and pulse amplitudes.

Implications for recurrence
Atrial fibrillation usually recurs when pulmonary vein (PV) isolation lesions become conductive again [9]. While the precise mechanism for the loss of lesion integrity is not known, it is reasonable to assume that lesions most easily become conductive at locations at which they are particularly narrow (or even have a gap). It appears that a promising approach to prolong lesion integrity is to make lesions more uniform in width, specifically avoiding that the width is below some minimum width anywhere along the lesion. Our data suggest that nsPEF ablation may give electrophysiologists the tools to achieve such greater lesion uniformity.

Absence of thermal effects
The complete absence of thermal effects and the associated complications side effects is a major advantage of nsPEF ablation over RF and cryoablation. RF ablation suffers from significant rates of major complications [12,13]. Esophageal injuries are common [29] and can in extreme, rare [30] cases lead to fistulae, which are associated with high mortality [31]. PV stenosis is another important complication that was reported to occur in 0 to 19% (mean, 2%; median, 3.1%) after 2004 [32]. It is possible that the nsPEF rates for PV stenosis will be lower (PV stenosis is a consequence of the inflammatory response [33,34], which may be aggravated by heating). Compared to cryoablation, the elimination of phrenic nerve palsy [35] would be a major improvement. For stroke, another important complication of thermal ablation, it is less clear how often it is a direct consequence of thermal effects, but a since nsPEF ablation excludes the risks of charring and thrombus formation, it should provide some benefit here as well.

Speed of applications
Since nsPEF ablation requires only requires less than 2 seconds per ablation site, it is realistic to assume that overall procedure times, e.g. for AF ablation, can be reduced substantially. Such a reduction will reduce the overall surgical risk, reduce the stress on patients, and free up resources of the hospitals that are expected to perform more and more ablation procedures.

Effect of shock amplitude on lesion geometry
Our experimental finding that smaller shock amplitudes lead to ablation only in the vicinity of the electrodes, while larger shock amplitudes ablate the whole region between the electrode is consistent with similar experiments performed in 3D in-vitro tumor models [36] as well as computations of field distributions [37]. For 3D in-vitro tumor models, a critical field that predicted cell death was also identified [36].
While we get good qualitative agreement between our experiments and model, we recognize that our model excludes important electrical properties of the myocardium, in particular its anisotropic conductivity and the fact that the direction of highest conductivity changes across the myocardial wall ("twisted anisotropy"). We are in the process of developing a model that includes these features of cardiac tissue.

Comparison to DC and pulsed electric fields of longer duration
There is a long history of DC ablation, in fact DC used to be the standard modality of thermal ablation before it was established that RF has lower complication rates [38,39]. Also, pulsed electric fields have been used before for ablation of cardiac tissue [40][41][42], although these studies used longer pulses (in the micro-and millisecond range) and much higher ablation energies.
Ablation with millisecond pulses has been attempted by placing circular electrodes on pig ventricular myocardium and applying defibrillator shocks of 100-200J [41]. While lesion depths up to 7 mm were achieved, the lesions were not uniformly transmural. Microsecond pulses were successfully used to create nonconducting lesions in pig atrial appendages, although with energies approximately 1,000 times higher than those presented here [40].
The fact that we demonstrate nsPEF ablation here with a fraction of the applied energy suggests that nsPEF ablation is more efficient for ablation than longer pulses. This result is consistent with theoretical analyses that suggest that nsPEF ablation electroporates cells in a tissue more uniformly. It has been shown in modeling studies that the strong fields used for nanosecond pulses electroporate all cells indiscriminately, while the weaker fields associated with longer pulses lead to spatially heterogeneous, incomplete electroporation [26,43]. Recent experiments in rat embryonic cardiomyocytes likewise suggest that nsPEF electroporates cells more uniformly that millisecond pulsed electric fields [44]. While it would be in principle possible to generate pulses that are both strong and long, such pulses would deposit large energies, which would lead to unwanted thermal effects. Nanosecond pulses also have the benefit that it has already been shown that besides ablating cells via necrosis, they can also ablate them via apoptosis [45,46].

Clinical application of nsPEFs ablation
In this paper, we are demonstrating the principle of nsPEF ablation in isolated hearts. In clinical practice, nsPEF ablation could either be performed via surgical clamps during open heart surgery, or using a catheter (just as RF-or cryoablation). For nsPEF ablation during open heart surgery, surgical clamps similar to those used in RF ablation [47] could be equipped with arrays of penetrating electrodes to allow the creation of extended lesions with a single nsPEF application. The development of this technology is straightforward and we plan to move it towards clinical practice first. For nsPEF catheter ablation, retractable penetrating electrodes would be placed in a catheter similar to those used in RF ablation. The thin, sharpened electrodes can be inserted into the myocardium with small contact force. Note that the wall thickness of the rabbit ventricles (2-4 mm) closely matches that of the human atria, which would be one important target for nsPEF ablation. Ventricular ablation, while performed less commonly, is also a very important target due to the grave risk of ventricular tachyarrhythmias that patients receiving ventricular typically carry. The substantially thicker human ventricles would not pose a special challenge to our nsPEF ablation approach; the electrode length would be increased accordingly. It is reasonable to expect that the consistent cross section of lesions that we observe, and that is due to the translational symmetry of our electrode configuration across the wall, would also be observed in thicker tissue (because the translational symmetry would still be given).

Arrhythmia induction
An important concern regarding shock application in a clinical setting is that shocks may induce fibrillation in the patient. In our rabbit model, nsPEFs never induced fibrillation that lasted long enough for us to observe it (several seconds would be detectable for us), but it is well-known that larger species such as human, dog, and pig are more prone to the induction of fibrillation [48]. In fact, application of microsecond shocks to pig heart frequently does induce fibrillation. This problem can, however, be addressed by timing the shock application so that it occurs right after the QRS complex (i.e. when all of the myocardium is depolarized) [49].

Limitations
This study has several limitations. Even though the consistent width of our lesions suggests that recurrence will be low, this has not been tested experimentally. Long-term survival studies that are necessary to address this limitation, and we are currently planning such studies. Also, it would be desirable to have an electrode configuration in which the electrodes don't penetrate the tissue multiple times. We are currently testing an alternative electrode configuration in which one electrode is places on the epicardium and the other electrode on the endocardium. This new configuration is straightforward to implement for open heart surgery, but for a catheter-based approach, there will be challenges related to the alignment of catheters.

Conclusions
Ablation with nsPEF is a promising alternative to RF and cryoablation that may overcome limitations of current clinical practice. Chronic animal studies and studies in large animal hearts are needed to evaluate the clinical potential of nsPEF ablation.