Compartmentalized Structure of the Moderator Band Provides a Unique Substrate for Macroreentrant Ventricular Tachycardia

Supplemental Digital Content is available in the text. Background Papillary muscles are an important source of ventricular tachycardia (VT). Yet little is known about the role of the right ventricular (RV) endocavity structure, the moderator band (MB). The aim of this study was to determine the characteristics of the MB that may predispose to arrhythmia substrates. Methods Ventricular wedge preparations with intact MBs were studied from humans (n=2) and sheep (n=15; 40–50 kg). RV endocardium was optically mapped, and electrical recordings were measured along the MB and septum. S1S2 pacing of the RV free wall, MB, or combined S1-RV S2-MB sites were assessed. Human (n=2) and sheep (n=4) MB tissue constituents were assessed histologically. Results The MB structure was remarkably organized as 2 excitable, yet uncoupled compartments of myocardium and Purkinje. In humans, action potential duration heterogeneity between MB and RV myocardium was found (324.6±12.0 versus 364.0±8.4 ms; P<0.0001). S1S2-MB pacing induced unidirectional propagation via MB myocardium, permitting sustained macroreentrant VT. In sheep, the incidence of VT for RV, MB, and S1-RV S2-MB pacing was 1.3%, 5.1%, and 10.3%. Severing the MB led to VT termination, confirming a primary arrhythmic role. Inducible preparations had shorter action potential duration in the MB than RV (259.3±45.2 versus 300.7±38.5 ms; P<0.05), whereas noninducible preparations showed no difference (312.0±30.3 versus 310.0±24.6 ms, respectively). Conclusions The MB presents anatomic and electrical compartmentalization between myocardium and Purkinje fibers, providing a substrate for macroreentry. The vulnerability to sustain VT via this mechanism is dependent on MB structure and action potential duration gradients between the RV free wall and MB.

Myocardial and Purkinje staining were quantified from histological sections using ImageJ. Purkinje and myocardium were masked and an RGB color threshold was applied to isolate myocardial-specific or Purkinje-specific staining from collagen. The quantity of each tissue component was quantified from the integral of frequency histograms of pixel intensity.

Optical and electrical mapping of the sheep myocardium
Novel dual coronary-perfused left anterior and right ventricular wedge preparations were imaged using optical mapping of the right ventricular endocardial surface. Preparations were loaded with the voltage-sensitive dye Di-4-ANBDQBS that was excited by illumination of the endocardial surface using monochromatic LEDs at 627 nm (Cairn Research Ltd, Kent, UK). Optical images (100x100 pixels) of signals passed through a 715nm long-pass filter were acquired using a Micam Ultima CMOS camera (SciMedia USA Ltd, CA, USA) at 2 kHz with a spatial resolution of 0.7x0.7 mm. Optical signals were filtered using a low-pass frequency filter at 120 Hz followed by spatial averaging (kernel 2.1 mm) and temporal averaging (kernel 1.5 ms).
Pseudo-ECG recordings across tissue preparations were recorded throughout experiments. In addition, to record activity along the moderator band and septum, unipolar electrograms were recorded from 5 locations simultaneously: 1) septum, close to the MB attachment; 2) proximal (septal) end of the MB; 3) mid-MB; 4) distal (RV) end of the MB and 5) the anterio-lateral papillary muscle. A reference electrode was positioned far from recording electrodes. All electrical recordings were acquired at 10 kHz and signals were treated by a forward-backward Butterworth filter with a 120 Hz low-pass cut-off.

Pacing protocols
Two stimuli locations were used: endocardial surface of the RV mid-wall and the MB at approximately half of its length from the septum. Measurements of activation latency at varying stimulation currents were recorded at a basic cycle length of 2 Hz from the stimulation threshold. S1S2 pacing was assessed for each pacing site, and a combined S1RV-S2MB protocol at twice the stimulation threshold. The stimulation sequence consisted of a train of 15 S1 stimuli at a basic cycle length of 2 Hz followed by S2. The S1S2 coupling interval was augmented until capture of an S2 beat to identify the effective refractory period (ERP). Coupling intervals were adjusted incrementally from the ERP six times by 5 ms, two times by 10 ms, two times by 25 ms and two increments of 50 ms.

Analyses and statistics
Optical activation time (AT) was the time of the maximal derivative of the fluorescent signal.
RT (RT) was taken from a fixed level of 80% of repolarization. AT and RT of unipolar signals were the time of the minimum and maximum derivatives, respectively. VT involving the MB was defined as 3 or more re-entry cycles with uni-directional propagation along the MB with activation of the RV free wall originating from the insertion of the MB. The incidence of VT was determined from the frequency of occurrences throughout the S1S2 pacing protocol. Statistical differences were assessed by the Wilcoxon matched-pairs signed rank test and defined by P<0.05.

Ionic models
Membrane electrical activity (Im) for myocardium was based on the Mahajan-Shiferaw cell model 1 and the Aslanidi rabbit Purkinje cell model 2 . This ionic model is well suited for the study of arrhythmias and arrhythmia therapy 3 . The supplement details the changes and lists modifications for each model parameter (supplemental table 2) that were applied to adjust for modeling the effective heart size (I) of sheep within the rabbit ventricular geometry for better correspondence with experiments 4 . Here, we scale the wavelength of reentry in the sheep heart to the model, preserving the patterns of ventricular fibrillation in the smaller heart geometry 4 . Parameters for calculating the effective heart size are listed in supplement table 3. I is defined as: Here, F is the maximum dominant frequency (Hz) and m is heart mass (g). For sheep, we obtain I=51.8  Cell models were subjected to 320 s of pacing at a basic cycle length of 320 ms for stabilization and the last AP was used for comparison. Initial states of ionic variables were captured 1 ms prior to stimulation of the last beat. Supplemental

Ventricles and moderator band computer model
We used a realistic 3D rabbit ventricle finite element mesh with anisotropic fibers and Purkinje system that we have previously developed 6 . The model contained >3,000,000 tetrahedral elements with edge lengths on the order of 300 μm (supplemental figure 5A). A MB was manually added to the geometry (supplemental figure 5B). MB computational elements merged with the myocardial elements, thus making the MB electrically continuous with the adjoining myocardium. An additional branch of the Purkinje system was inserted that extended from the His bundle and followed the central axis of the MB to form junctions at the mid RV free wall with four intramural terminal points. PMJs were absent along the MB. To mimic experimental preparations, clipped wedge geometry was also used to compare with whole heart simulations (supplemental figures 5C and 5D). BCLexpt is the basic cycle length of S1 pulses in experiments.

In silico reentry study
An S1S2 stimulus protocol was applied in the model by point stimulation of S1 pulses followed by a single S2 test pulse. S1S2 coupling intervals ranging from 100 ms were tested. Simulations were repeated while incrementing S1S2 coupling intervals by 1 ms for each iteration until a maximum of 180 ms. Re-entry was confirmed if nodes in the muscular compartment at the midpoint of the MB depolarized above a threshold of -30 mV three or more times following S2 stimuli.

Structural substrate of the MB
The MB structure is composed of two excitable, yet uncoupled compartments in sheep.
Purkinje fibers were separated from the myocardium in a coaxial configuration by lipid deposits and extensive extracellular collagen surrounding individual bundles, as seen in insertions (supplemental figure 3) with no apparent coupling. The Purkinje fiber/muscle ratios were highly variable, ranging from 0.04 to 0.63 (supplemental figure 4). This may be explained by a high intra-species variability of the MB thicknesses (3.3±0.9 mm, minimum=2.1 mm, maximum=5.0 mm, in sheep).

MB conduction behaviour
Conduction delays between the RV free wall and MB myocardium in sheep were dependent upon the direction of propagation across the MB insertion and the coupling interval. This was determined through S1S2 pacing of either the RV free wall or the myocardial compartment of the MB. Across all experiments, activation latency of the MB following RV stimulation at the S1 basic cycle length was 12.7±6.1 ms. This was increased to 16.0±8.3 ms at the ERP of the stimulation site. For MB stimulation, activation latency of the RV was 14.4±13.6 ms at the basic cycle length but significantly prolonged to 32.5±16.7 ms at the ERP (P<0.05). An example of short coupled propagation along the MB is shown for S1S2RV and S1S2MB pacing in supplemental figure 5.
Supplemental figure 5. Short coupled S1S2 pacing intervals. The last of a train of S1 pulses and the S2 beat from the ERP were captured by unipolar recordings and simultaneous optical mapping of the RV free wall for RV (A) and MB (B) pacing (black arrows). Directions of propagation (red arrows) are shown.

Macroreentry involving the human MB
Conduction behavior and repolarization heterogeneity were examined from two human wedge preparations with intact MBs. MB thicknesses from human preparations were (4.0 and 8.5 mm in donor #1 and donor #2, respectively). Pacing at a basic cycle length of 500 ms on the MB triggered activation propagating bi-directionally towards both septal and RV free wall insertions of the MB.
The activation latency of the RV when pacing MB was 10.8 ms and 17.9 ms, respectively (see supplemental figure 6). In donor #2, a sufficiently short coupled stimulation at the same site resulted in uni-directional propagation towards the RV free wall to excite the septum from the RV free wall instead of the MB directly.
Supplemental figure 6. Short-coupled stimulation of the MB in humans. Activation patterns during pacing human donor #1 MB at A) basic cycle lengths of 500 ms and B) a short coupled beat of 325 ms. C) Unipolar electrical and optical AP traces extracted from the septum, along the MB and the RV site of early activation following the final S1 and S2 pulses. D-F) Same as for A-C for donor #2.

In silico reentry
Simulations of S1S2RV, S1S2MB and S1RV-S2MB pacing protocols are shown in supplemental videos 4-6. In each case, VT could be induced in homogeneous and heterogeneous models, but windows of vulnerability for VT were increased in the heterogeneous model (supplemental figure 7A-B). Furthermore, the direction of the reentrant circuit was reversed for S1S2RV pacing as a result of reversal of the activation and repolarization sequences of the S2 beat relative to S1S2MB and S1RV-S2MB simulations. A further model was implemented whereby the Purkinje network was omitted (supplemental figure 7C and supplemental video 7). In this case, neither S1S2RV nor S1S2MB could induce VT, indicating a predominant role of Purkinje in the induction of the VT as opposed to the muscular compartment of the MB. However, inducibility of sustained stable reentrant arrhythmias was unchanged in S1RV-S2MB simulations. To examine the impact of the wedge preparation in experiments, the whole ventricle mesh was clipped to a geometry corresponding closely to experiments and zero-boundary conditions applied to clipped surfaces. In a model with homogeneous ionic properties, windows of vulnerability for each pacing protocol were largely unchanged, particularly for S1RV-S2MB pacing (supplemental figure 7D and supplemental video 8).