Multiscale Computational Model Predicts Ventricular Remodeling After Surgical Mitral Valve Repair

Introduction: The most successful treatment for primary mitral valve (MV) regurgitation is the surgical repair of the valve. Mitral valve reconstructive surgery (MVr) is associated with 95% freedom from reoperation at 15 years after surgery. However, the clinical benefits of surgical repair are reduced for patients with severe heart failure secondary to the valvular disease. Although there are recommendations for early intervention in patients with severe mitral regurgitation (MR), there is no prediction model for the degree of LV remodeling after repair. We aimed to calibrate and validate a predictive computational tool to calculate the patient-specific path to recovery after MVr based on multiscale modeling of ventricular mechanics and cellular signaling.

Methods: We evaluated retrospectively pre- and post-operative echocardiography data from five patients with degenerative MR that underwent complex MVr at a tertiary referral center. Patient-specific data was grouped by the pre-operative EF (EF<60% and EF≥60%). We developed a multiscale model of cardiovascular mechanics that was calibrated to pre-clinical data from animal models and clinical data from randomized trials on the outcomes of MVr. Systolic dysfunction was modeled by reducing simulated adrenergic compensation of LV contractility. Herein, we show simulation predictions for a patient with preoperative MR2+ and normal LV function, and a patient with MR3+ and systolic dysfunction. Validation consists of paired comparison of the model predictions of LV mass, LV end diastolic volume (LVEDV), LV end systolic volume (LVESV), and LV ejection fraction (LVEF) to patient-specific echocardiographic measurements.

Results: The model successfully reproduces the main differences observed on the path to recovery between patients with mild MR and normal systolic function and patients with severe MR and signs of systolic dysfunction as reported in clinical trials. Patients with severe MR show more predicted LVEDV recovery after surgery compared to less severe cases (27% vs 15% decrease). However, these patients showed no predicted recovery of the EF, and no significant improvement of LVESV because the model assumes that systolic dysfunction is irreversible post-repair.

Conclusion: The model reproduced trends in mass and volume adaptation after MVr when compared to previous clinical trials. The model also showed agreement with patient-specific data when stratified by EF. In future work, we will improve the patient-specific predictions by customizing the model parameters to consider individual vital signs, echocardiography, electrocardiogram, and the impact of pharmaceuticals. In addition, the model will be validated by longitudinal follow-up of patient-specific echocardiographic data.