Mechanisms of Arrhythmogenicity in Hypertrophic Cardiomyopathy: Insight from Noninvasive Electrocardiographic Imaging

Mechanisms of arrhythmogenicity in hypertrophic cardiomyopathy (HCM) are not well understood.To characterize an electrophysiological substrate of HCM in comparison to ischemic cardiomyopathy (ICM), or healthy individuals.We conducted a prospective case-control study. The study enrolled HCM patients at high risk for ventricular tachyarrhythmia (VT) (n=10; age 61±9 y; left ventricular ejection fraction (LVEF) 60±9%), and three comparison groups: healthy individuals (n=10; age 28±6 y; LVEF>70%), ICM patients with LV hypertrophy (LVH) and known VT (n=10; age 64±9 y; LVEF 31±15%), and ICM patients with LVH and no known VT (n=10; age 70±7y; LVEF 46±16%). All participants underwent 12-lead ECG, cardiac CT or MRI, and 128-electrode body surface mapping (BioSemi ActiveTwo, Netherlands). Non-invasive voltage and activation maps were reconstructed using the open-source SCIRun (University of Utah) inverse problem-solving environment.In the epicardial basal anterior segment, HCM patients had the greatest ventricular activation dispersion [16.4±5.5 vs. 13.1±2.7 (ICM with VT) vs. 13.8±4.3 (ICM no VT) vs. 8.1±2.4 ms (Healthy); P=0.0007], the largest unipolar voltage [1094±211 vs. 934±189 (ICM with VT) vs. 898±358 (ICM no VT) vs. 842±90 µV (Healthy); P=0.023], and the greatest voltage dispersion [median(interquartile range) 215(161-281) vs. 189(143-208) (ICM with VT) vs. 158(109-236) (ICM no VT) vs. 110(106-168)µV (Healthy); P=0.041]. Differences were also observed in other endo-and epicardial basal and apical segments.HCM is characterized by a greater activation dispersion in basal segments, a larger voltage, and a larger voltage dispersion through LV.


Introduction
Patients with hypertrophic cardiomyopathy (HCM) are at high risk of life-threatening ventricular arrhythmias and sudden cardiac death (SCD). 1 Mechanisms of arrhythmogenicity in HCM are complex and incompletely understood. It was previously shown that the late sodium current is increased in HCM, suggesting the importance of repolarization abnormalities. 2 At the same time, cardiac magnetic resonance (CMR) studies have shown that the myocardium in HCM is characterized by increased fibrosis burden, supporting an alternative mechanism for arrhythmogenesis -heterogeneity in electrical activation. The degree of late gadolinium enhancement in HCM is associated with SCD and appropriate implantable cardioverterdefibrillator (ICD) therapy. 3 Previously, a similar frequency of both reentrant monomorphic ventricular tachycardia (VT) and ventricular fibrillation (VF) was observed in HCM patients with ICD, 4 suggesting similarity of mechanisms with ischemic cardiomyopathy (ICM). While the presence of patchy scar in HCM suggests likely similarity in macro-reentrant VT mechanisms with post-myocardial infarction (MI) VT, VT ablation in HCM is less successful than in post-infarction VT. In HCM patients who underwent VT ablation, the incidence of VT recurrence, death, and cardiac transplantation at one year was one of the highest amongst all nonischemic cardiomyopathies (NICM), 5 even after adjusting for comorbidities. This may be due to designed this study with the goal to describe the EP substrate of HCM, in comparison to relatively well-understood EP substrate of post-infarction macro-reentrant VT.

Study population: inclusion and exclusion criteria
We conducted a single-center case-control study of high-risk HCM cases with three control groups (Clinical Trial Registration-URL: www.clinicaltrials.gov Unique identifier: NCT02806479). The study was approved by the Oregon Health & Science University (OHSU) Institutional Review Board (IRB), and all participants signed an informed consent form.
Enrollment was performed at OHSU in 2016-2018. Adult (age≥18y) non-pregnant participants were enrolled if the inclusion and exclusion criteria were met, as described below.
Inclusion criteria for HCM group were: (1) history of resuscitated sudden cardiac arrest, or documented sustained VT, or (2) maximal left ventricular (LV) wall thickness above 30 mm, or extensive fibrosis on CMR (above 10% of total myocardial volume), or (3) high risk of SCD (>7.5%/5y) as determined by HCM risk-SCD 8 score.
Healthy control group I (Healthy) was designed to include individuals who were free from structural heart disease and arrhythmogenic substrate in ventricles. The inclusion criterion required evaluation by a cardiac electrophysiologist for AV nodal reentrant tachycardia. used to define 17 segments of LV.
Additionally, data from the most recent echocardiogram was abstracted to provide additional information on baseline cardiac structure and function. For subjects who underwent CMR, left ventricular ejection fraction (LVEF) was calculated from ventricular volume measurements. For all other individuals, LVEF was calculated from the echocardiogram using the biplane Simpson method of discs. Regional LV function was evaluated by the echocardiographic wall motion score index. Motion and systolic thickening in each segment were scored as: normal or hyperkinesis = 1, hypokinesis = 2, akinesis = 3, and dyskinesis (or aneurysmal) = 4. Wall motion score index was calculated as the sum of all scores divided by the number of visualized segments. Resting peak LVOT gradient was calculated for all participants. In addition, HCM participants had peak LVOT gradient measured during Valsalva maneuver and at peak exertion.

Body surface potentials recording and ECG electrodes localization
A routine clinical resting 12-lead electrocardiogram (ECG) was recorded during the study visit, and ECG metrics were measured by the 12 SL algorithm (GE Marquette Electronics, Milwaukee, WI).
Unipolar ECG potentials were recorded on the body surface using the ActiveTwo biopotential measurement system (BioSemi, Amsterdam, the Netherlands) with 128 Ag/AgCL electrodes (4 panels of 32 electrodes; each panel is arranged as four strips of 8-electrodes; diameter of the ECG electrodes 5 mm), as previously described. 11 The sampling rate of the signal was 16,384 Hz; bandwidth DC-3,200 Hz. ECG electrodes were localized by three-dimensional (3D) photography approach, using a Kinect camera (Microsoft, Redmond, WA, USA), as All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint previously reported. 11 In addition, for co-registration of torso images, five CMR-or CT-specific markers were placed on each patient's chest during scanning, to mark ECG electrodes locations.

Reconstruction of torso and heart meshes
We constructed 3D meshes of a continuous surface of the endocardium (excluding papillary muscles) and epicardium of both ventricular chambers. The 3D heart and torso meshes were reconstructed using a semi-automatic approachimage growing method of a continuous surface 11, 12 from CMR/CT images using ITK-snap software (PICSL, USA). 13 Each cardiac mesh was manually reviewed to ensure a continuous segmentation of epicardium and endocardium of both ventricular chambers and exclude atria chambers and papillary muscles. Both torsi meshes segmented by the 3D photography method, and DICOM images were matched using the coregistered CMR/CT markers and electrode position, as previously described. 11 The resolution of the cardiac mesh was 3.6 ± 0.5 mm with 3992 ± 735 nodes.

Inverse solution and reconstruction of the cardiac activation map
The workflow is shown in Figure 1. One clean normal sinus beat was selected for analysis; an absence of extrasystole before and after selected beat was verified. We used the open-source SCIRun problem-solving environment developed at the Center for Integrative Biomedical Computing (University of Utah, UT), 14,15 which was previously used to compute forward and inverse solutions 16 and reconstruct unipolar epicardial and endocardial electrograms (EGMs).
The inverse problem was solved as the potential-based formulation (boundary element method), as a weighted minimum norm problem by applying a Tikhonov L2-norm regularization.
The steepest downslope of each unipolar EGM was determined automatically, using MATLAB (The MathWorks Inc, Natick, MA) software application. Then, each pair of neighboring unipolar EGMs together with resulting bipolar EGM (calculated as their difference) 9 was reviewed by at least two investigators (AW, KY, NMR, EAPA) blinded to the study groups assignment, to verify the consistency of morphology and the steepest downslope detection

Unipolar voltage potential map
Unipolar voltage was measured in each reconstructed EGM, and a unipolar voltage potential maps were constructed. The peak-to-peak voltage on each unipolar EGM was automatically measured using MATLAB (The MathWorks Inc, Natick, MA) software application. An accuracy of unipolar EGM peaks detection was validated by investigators (AW, KY, NMR, EAPA) blinded to groups assignment. We used the standardized myocardial segmentation and nomenclature 10 to define 17 segments of LV. Mean unipolar voltage was calculated for each segment. RV endocardial surface in 5 segments (basal anteroseptal and inferoseptal, mid-cavity anteroseptal and inferoseptal, and apical septal) served as an "epicardial" surface of LV.
Standard deviation (SD) of unipolar voltage distribution in each segment served as a measure of voltage dispersion within each segment.
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint

Ventricular Conduction Velocity
Ventricular conduction velocity between each pair of neighboring nodes was calculated as a distance between nodes divided by the difference in LATs between corresponding nodes. Mean conduction velocity was calculated for each LV segment. 10 RV endocardial surface in 5 segments (basal anteroseptal and inferoseptal, mid-cavity anteroseptal and inferoseptal, and apical septal) served as an "epicardial" surface of LV. Dispersion of conduction velocity was measured as SD of ventricular conduction velocity in each segment.

Statistical analysis
Statistics of normally distributed variables are summarized as mean ± SD. Distributions of all variables were reviewed. Outliers of ventricular conduction velocity representing nonphysiological values (> 100cm/s) were removed from further analyses. After verifying the normality of distribution, we tested the hypothesis that the means (mean voltage and mean conduction velocity) are the same across four study groups while removing the assumption of equal covariance matrices. The Wald chi-squared statistic with James's approximation 17 was used to calculate P-values.
We used a Kruskal-Wallis test of the hypothesis that four study groups are from the same population, to compare voltage and velocity dispersions (measured as an SD of velocity and voltage in each of 17 segments 10 ), which have a non-normal distribution. Non normally distributed variables are summarized as the median and interquartile range (IQR).
A P-value of less than 0.05 was considered significant. Statistical analyses were performed using STATA MP 15.1 (StataCorp LLC, College Station, TX).
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint

Study population
Clinical characteristics of the study population are shown in Table 1. Most of HCM patients (80%) had previously undergone genetic testing. The definitive, disease-causing MYHBPC3 mutation was found in 2 patients. Half of the HCM participants had survived a sudden cardiac arrest, and another half had documented the history of sustained VT/VF. While half of the HCM patients had a history of severe LVOT obstruction (up to 153 mmHg at peak exertion), they had already undergone surgical myectomy, resulting in vastly improved LVOT gradients (provoked peak LVOT gradient 20±20 mmHg), at the time of enrollment.
In VT-free post-MI group, the scar was located in the anteroseptal region in 90% of participants. In post-MI VT group, the scar was located in the inferoposterior region in 40% and anteroseptal in 60%. Single-chamber ICD was implanted in approximately 50% of groups II-III controls, and HCM. The other half of the population had a dual-chamber ICD implanted.
LV systolic function was normal in healthy controls and HCM participants, whereas ischemic cardiomyopathy (ICM) with reduced LVEF was confirmed for both post-MI groups (Table 1).

Mean unipolar voltage and unipolar voltage dispersion
A representative example of a voltage map is shown in Figure 3A. Mean unipolar voltage (Supplemental Table 1 and Figure 4) was significantly different across all 4 study groups in basal anteroseptal and apical septal segments, on both sides of septum -LV and RV endocardium.
Also, a significant difference in voltage across all four groups was observed on both endocardial and epicardial surface of basal anterior and anterolateral segments, the endocardial surface of anterior apical segment, and the epicardial surface of basal inferior and mid-inferolateral All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint segments. Healthy individuals had the smallest mean unipolar voltage, whereas HCM was characterized by the largest voltage ( Figure 4). Unipolar voltage in two post-MI groups was similar, and had intermediate values, as compared to healthy and HCM participants.
Voltage dispersion was significantly smaller in healthy controls, as compared to the other three groups (Supplemental Table 2 and Figure 5). Remarkably, in several segments, voltage dispersion in HCM was the highest amongst all four groups, significantly exceeding voltage dispersion in both ICM groups. The unipolar voltage dispersion was significantly different across study groups in both endocardial and epicardial segments in basal anterior, basal anterolateral and inferolateral, apical inferior, and the epicardial surface of the apex.

Mean ventricular conduction velocity and velocity dispersion
A representative example of the activation map is shown in Figure 3B. In Healthy controls, we observed a normal activation pattern, which initiated in the septal region and propagated from endocardium to epicardium, with several breakthroughsnear the RV apex and anterior paraseptal aspects of the epicardium in regions adjacent to the left anterior descendent coronary artery. Activation proceeded from apex to the inferior basal area in both RV and LV, with the inferolateral LV base and the region near right ventricular outflow tract (RVOT) being the latest to activate.
Overall, mean conduction velocity was mostly similar in all 4 study groups (Supplemental Table 4 and Figure 6). In most LV segments, there was no difference in mean conduction velocity in healthy as compared to both ICM groups. Mean conduction velocity was significantly slower in HCM participants as compared to other groups in endocardial basal anterior, and epicardial basal inferior, basal anterolateral, apical anterior, and apical septal segments. All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint Overall, velocity dispersion was largely similar across all four study groups. Velocity dispersion in HCM participants was significantly greater than in the other three groups only in one segment: RV endocardial surface corresponding to the basal anteroseptal segment (Supplemental Table 4).

Discussion
Our study revealed important features of EP substrate in HCM, which differentiate HCM  author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

EP substrate and mechanisms of arrhythmogenesis in HCM
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint correlation between low voltage areas on the map, co-localized with a post-infarction scar in post-MI patients. However, scattered intramural fibrosis in HCM did not manifest by the low voltage on endocardial, nor epicardial maps. Reported findings included local conduction delay or conduction block, fractionated electrograms, and reduced voltage. 23  author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the

Noninvasive mapping of ventricular activation
In this study, we used the Forward/Inverse problem toolkit from the SCIRun problem-solving environment, which is used by many investigators in the field. 14,15,31,32 However, knowing the limitations of ECGi method, 33 we intentionally limited our analysis by averaged "per segment" data. Duchateau et al 33 observed mean activation time error of approximately 20 ms. In this All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint study, we observed ventricular activation dispersion of approximately 20 cm/s in all study participants, including healthy controls, with no meaningful differences in ventricular activation dispersion between groups. Therefore, we interpret that ventricular activation dispersion in our study quantifies an error of local ventricular conduction velocity measurement between two nods. Thus, we accept that actual ventricular conduction velocity can be on average either higher or lower by 20 cm/s. This study error precludes interpretations of specific ventricular conduction phenomena, yet our findings still provide insight into HCM mechanisms.

Limitations
A case-control study is susceptible to bias. We selected only high-risk HCM patients, and our All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint       LV endocardial and 17 LV epicardial segments. Segments nomenclature as described in Figure 5 legend.
All rights reserved. No reuse allowed without permission.
author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/19002782 doi: medRxiv preprint Table 1