Poster Q9 - Computational modeling of myocardial contractility in mechanical dyssynchrony

Introduction: Heart failure (HF) is an emerging epidemic affecting approximately 1% to 2% of the adult population. Patients with HF often develop mechanical dyssynchrony (MD) that affects regional cardiac function and coronary blood flow in the septum and left ventricular free wall (LVFW), which are fed by the left anterior descending (LAD) and left circumflex (LCx) arteries, respectively. The effects of MD on global left ventricular (LV) functions, i.e., LV pressure and volume, and ejection fraction (EF), have been assessed from experimental/clinical studies. However, it remains unclear how MD affects regional contractile function and coronary perfusion in the LV as it is difficult to measure these quantities directly in experiments. The lack of this knowledge is an impediment to improving the current 30% non-responder rate of cardiac resynchronization therapy (CRT), which is frequently used to treat HF patients. Here we seek to address this issue by developing an inverse finite element (FE) modeling framework based on animal-specific 3D echocardiographic (ECHO) images and hemodynamics measurements acquired from a swine MD model to understand the impact of MD on regional contractile function and coronary blood flow distribution in the LV.

Materials and

Methods: All animal experiments were performed using Yorkshire swine (n = 9) in accordance with national and local ethical guidelines. All measurements were acquired under healthy and MD conditions by right atrium (RAP) and right ventricular pacing (RVP), respectively, at heart rates of 100 and 140 bpm. In each condition, 3D ECHO images were acquired while simultaneously recording LV volume. LV pressure and volume were measured using a pressure-volume (PV) catheter. The coronary flow in the LAD and LCx arteries were measured under regulated condition simultaneously and then under the passive condition from the fully dilated vessels by injecting adenosine. The 3D ECHO images were postprocessed using TomTec software to calculate systolic dyssynchrony index (SDI) to represent the degree of MD, reconstruct 3D LV geometry, and derive the regional longitudinal and circumferential strain in each AHA segment.
An inverse FE model was developed based on the reconstructed LV geometry and measurements of the LV PV loops, regional strains to estimate the regional contractility under the 2 pacing conditions (Fig. 1A1). Specifically, parameters associated with the passive myocardium were adjusted until the values of LV end-diastolic pressure (EDP) and LVEDV match the measurements at ED time point. Thereafter, the local time-dependent active stress parameter, which reflects the local contractility, is estimated by minimizing the mismatch between simulated and measured LV PV loops and regional strains. Global and regional contractility was optimized until the model predicted LV PV loops and regional strains match the measurements. Computational frameworks are implemented using FEniCS and dolfin-adjoint.

Results, Conclusions, and Discussions: The experimental measurements show that the SDI with RAP was 8.6±1.2% at 100 bpm (p < 0.05) and 8.9±2.0% at 140 bpm (p < 0.05), whereas was 14±4.8% at 100 bpm (p < 0.05) and 19.4±6.9% at 140 bpm (p < 0.01) with RVP, indicating that the degree of MD increases with RVP, which is more at higher heart rates. Fig. 1B shows LVEDP and LVESP were significantly decreased with RVP (-4.4±1.0 mmHg at 100 bpm, p< 0.01, and -6.5±1.5 mmHg at 140 bpm, p< 0.01; ˗25.6±4.0 mmHg at 100 bpm, p< 0.001, and -35.7±5.4 mmHg at 140 bpm, p< 0.001). Fig. 1C shows LVEDV and LVESV were significantly decreased with RVP (-14.7±1.8 mL at 100 bpm, p< 0.001, and ˗11.3±3.3 mL at 140 bpm, p< 0.05; -4.6±1.4 mL at 100 bpm, p< 0.05). Fig. 1D shows regulated and passive LAD flows were significantly decreased with RVP (˗7.6±2.3 mL/min, p< 0.01 and -33.2±8.9 mL/min, p< 0.05 at 100 bpm; -5.7±0.8 mL/min, p< 0.001 and -54.0±7.4 mL/min, p< 0.01 at 140 bpm). Regulated LCx flow (Fig. 1E) did not change (-0.2±1.3 mL/min at 100 bpm and -3.6±2.6 mL/min at 140 bpm). Passive flow rate, however, was significantly decreased with RVP (-17.5±7.8 mL/min at 100 bpm, p< 0.05 and -31.1±8.0 mL/min at 140 bpm, p< 0.01). In addition, the inverse FE model predicts the effects of MD on LV contractility. The error between experimental measurements and model predictions of the LV PV loops is less than 5% in each case. The model predicted global LV contractility reduction from RAP to RVP is 70±51, p< 0.002 mmHg at 100 bpm and 100±86, p< 0.02 mmHg at 140 bpm, respectively. The estimated septal contractility is reduced by 16±163 mmHg, p=0.77 at 100 bpm and 93±91 mmHg, p< 0.1 at 140 bpm with RVP (compared to RAP) (Fig. 1F). The estimated regional LVFW contractility is reduced by 120±187 mmHg, p< 0.05 at 100 bpm and 46±102 mmHg, p=0.24 at 140 bpm with RVP (compared to RAP) (Fig. 1F).
The adverse effects of MD on LV contractility and regional coronary hemodynamics may be interrelated based on the “perfusion-contraction matching”. Identification of these interrelated factors may have clinical implications in the identification of treatment of MD (i.e., CRT) and improvement in treatment.

Acknowledgements (Optional):