Electrocardiogram phenotypes in hypertrophic cardiomyopathy caused by distinct mechanisms: apico-basal repolarization gradients vs. Purkinje-myocardial coupling abnormalities

Abstract Aims To identify key structural and electrophysiological features explaining distinct electrocardiogram (ECG) phenotypes in hypertrophic cardiomyopathy (HCM). Methods and results Human heart–torso anatomical models were constructed from cardiac magnetic resonance (CMR) images of HCM patients, representative of ECG phenotypes identified previously. High performance computing simulations using bidomain models were conducted to dissect key features explaining the ECG phenotypes with increased HCM Risk-SCD scores, namely Group 1A, characterized by normal QRS but inverted T waves laterally and coexistence of apical and septal hypertrophy; and Group 3 with marked QRS abnormalities (deep and wide S waves laterally) and septal hypertrophy. Hypertrophic cardiomyopathy abnormalities characterized from CMR, such as hypertrophy, tissue microstructure alterations, abnormal conduction system, and ionic remodelling, were selectively included to assess their influence on ECG morphology. Electrocardiogram abnormalities could not be explained by increased wall thickness nor by local conduction abnormalities associated with fibre disarray or fibrosis. Inverted T wave with normal QRS (Group 1A) was obtained with increased apico-basal repolarization gradient caused by ionic remodelling in septum and apex. Lateral QRS abnormalities (Group 3) were only recovered with abnormal Purkinje-myocardium coupling. Conclusion Two ECG-based HCM phenotypes are explained by distinct mechanisms: ionic remodelling and action potential prolongation in hypertrophied apical and septal areas lead to T wave inversion with normal QRS complexes, whereas abnormal Purkinje-myocardial coupling causes abnormal QRS morphology in V4–V6. These findings have potential implications for patients’ management as they point towards different arrhythmia mechanisms in different phenotypes.

The first step of the computational pipeline consisted in constructing personalized 3D biventricular surfaces of the heart, torso, lungs and ribs from CMR images. MRI slices of the heart (long and short axis) were acquired, and pre-processing steps such as alignment (1), and manual contour delineation were applied to generate 3D surfaces of the bi-ventricular geometry from the patient's CMR images. An iterative process consisting in remeshing, smoothing, and fitting of a surface towards the input contours is performed to generate the heart surfaces following (2). Torso surface geometries were derived from combining manual contouring with a statistical model of the torso. Ribs and lungs were personalized from a template to the patient's anatomy (3). The next step was to generate tetrahedral volumetric meshes from these surfaces using the software Tetgen (4). Each tetrahedron was defined by four nodes and allowed the propagation of the electrical signal throughout the mesh. The meshing process required the definition of a tetrahedral edge length and multiple optimization parameters. In our case, a fine and detailed mesh was required in the heart in order to obtain numerical convergence in the simulations of electrical propagation with the Chaste software. Torso, lungs and ribs were designed as passive tissues with different fixed conductivities, allowing a coarser mesh to be defined in these regions (following (5)'s work). The edge length of each element was defined as 0.4mm for the heart, 3mm for the lungs, 2mm for the ribs, and 10mm for the torso based on (5) and (6)'s benchmarking work. An adaptive meshing, providing smooth transitions between different edge lengths, was used by controlling the volume of the tetrahedra in order to ensure a smooth propagation of the electrical signal in the whole domain. In order to simulate the 12-lead ECG, ten nodes were defined to model electrodes on the torso surface, adapting the location defined in (5) to each personalized torso by geometrical deformation.

Electrophysiological and activation model
The 3D biventricular volumetric meshes were therefore composed of a set of nodes (vertices of the tetrahedra, defined by their coordinates) and elements (tetrahedra, defined by four nodes). The electrophysiological simulation described below followed the framework published in (5). The modified version of the electrophysiological O'Hara-Rudy model of ventricular cells (7) (8)  In order to simulate the ECG on the torso surface, ten nodes were defined to model electrodes placed on the torso, adapting the location defined in (5) to each personalized torso by geometrical deformation. Fibre orientations were computed for each geometry following a rule-based method replicating findings in (14).

Action potential duration heterogeneities
Several configurations of heterogeneities representing the dispersion of action potential duration were considered. Apex-to-base heterogeneities were modelled by introducing a gradient of IKs conductance from base to apex to reach an APD difference of 50ms. Similarly, interventricular heterogeneities were modelled by modifying the gradient of gKs between left and right ventricles (25ms APD RV to LV difference). Finally, transmural heterogeneities were defined using layers of endocardial, midmyocardial and epicardial cells with different APD properties as described in the O'Hara-Rudy model (7). The combination of apex-to-base and transmural heterogeneities with 45% endocardial cells, 25% mid-myocardial cells and 30% epicardial cells yielded the best simulations compared to the clinical ECGs and was therefore chosen in the rest of the paper.
Simulations were run using Chaste (15) on the Archer UK national supercomputer using parallelization on 960 cores.  As a sensitivity analysis, we analysed the influence of varying both the intracellular and extracellular conductivities on the shape of the QRS, to explore whether these changes could explain the QRS abnormalities observed in Group C patients. Results are shown on Figure S2 below. Reducing the conductivities increased the amplitude and width of the QRS complexes, especially the R wave in V1-3, but did not change the morphology and polarity of the QRS waveforms, especially in leads V5-6. Doubling the conductivities did not yield severe changes in QRS morphology. It slightly reduced the width of the QRS in all leads, overall decreased the amplitude of the QRS (especially the R wave in leads V1-4), and slightly increased the amplitude of the S wave in V5-6 while still exhibiting a tall R wave in V5-6. In all cases, the QRS in V4-6 still exhibited a large R amplitude with small S wave, different from the real ECG showing a small R amplitude with deep and wide S wave. These results suggest that simple homogeneous conductivity changes do not lead to severe morphological abnormalities in the lateral leads of Group 3 patients. Figure S2: Analysis of conductivity changes on the ECG. Real ECG (black) is represented along with a normal simulation (blue), a simulation with intracellular and extracellular conductivities divided by 2 (red), and a simulation with intracellular and extracellular conductivities multiplied by 2 (green).

Fibre disarray modelled as isotropic propagation
Diffusion tensor imaging (DTI) provides information on the regions that may be affected by fibre disarray. Low fractional anisotropy (FA) may be the indicator of disarray, fibrosis, cell enlargement, or a combination of all. Low FA was observed in the basal and mid slices, in the anterior RV/LV insertion point at the epicardial border ( Figure S3A). Patchy low FA was also noticeable all around the septum. In the apical slice, low FA was seen in the endocardium of the anterior and posterior LV/RV insertion point. As observed in Figure S3B, isotropic propagation hardly affected the QRS complex in leads V1-V6. isotropic orientation of the myocyte fibres (red) along with normal simulation (blue) and real ECG (black) in leads V1 to V6. Figure S4. Activation maps (ms) in the case of slow conduction in the mid-septum (left) and slow conduction in the entire septum from base to apex (right). Large activation delays are observed in the anterior of the ventricles caused by the region of slow conduction in the mid septum, which extended to the anterior, but these are not present in the entire septum case.