1st supervisor: Adelaide De Vecchi, King’s College London
2nd supervisor: Ronak Rajani, Guy’s and St Thomas’ NHS Foundation Trust
In the UK, 1.5 million individuals over the age of 65 are affected by heart valve disease. Mitral regurgitation is one of the most commonly occurring conditions and its prevalence is projected to double by 2050, carrying serious implications for healthcare providers who will be addressing the needs of a growing number of patients too old and frail for open-chest surgery. TMVR is the treatment of choice in these cohorts of patients. Specific concerns of this challenging procedure include difficulty in anchoring the device to the deformed mitral annulus, with risk of valve migration or leak, and potential for left ventricular outflow tract obstruction (LVOTO) by the implanted valve, with risk of refractory heart failure and late mortality. This project will develop image-based personalised computer models to simulate TMVR outcomes, with the goal to predict potential LVOTO and valve migration/leaks in the preprocedural planning phase.
The project will be divided in two parts. In the first stage, personalised models of ventricular blood flow will be combined with Computer-Aided Design (CAD) models of a commercial bioprosthesis using the commercial software Star-CCM+. Data will be extracted from multi-phase CT images (anatomy, wall motion) and Doppler echocardiography (velocity profile at valve planes) acquired pre- and post-TMVR in ~20 patients who have undergone the procedure at St Thomas’ Hospital. Two pre-TMVR models with and without implanted device (baseline and predictive models) and one post-TMVR model with implanted device (validation model) will be created and validated based on metrics extracted from the corresponding imaging datasets. 3D simulations of the patient-specific blood flow will be performed in each case using Star-CCM+. The baseline and validation models will be independently validated against the pre- and post-TMVR data, respectively, while the predictive model based on pre-TMVR data will be compared to the post-TMVR images.
The second part will involve the creation of an in-silico database for performing sensitivity analysis. The following metrics will be quantified from each simulation: systolic and diastolic pressure gradients (as a surrogate measure for afterload and preload, respectively), degree of LVOTO, global longitudinal strain and blood residence time inside the ventricle (as surrogate measures for ejection efficiency) and fluid-dynamic forces on the device (for device stability). Additional models will be created for each patient where the degree of LVOTO is changed by altering the depth of placement of the device inside the ventricle. This data will be used to produce response maps of the changes in these haemodynamic metrics for varying degrees of LVOTO. Regression analysis will be performed to test the effects of variations in the degree of LVOTO between pre- and post-TMVR on ventricular pressure gradients, fluid-dynamic forces on the device and mean blood residence time and strains. The outcome of this study will provide proof-of-concept data to characterise the mechanistic links between pre-TMVR metrics and risks of LVOTO and device migration given a specific device implant. Risk predictions will be prospectively evaluated against the follow-up echocardiographic exams after 6 and 12 months by blinding the researcher to the patient outcome with respect to these complications.
Figure 1: Image-based predictive modelling of transcatehter mitral valve replacement in patients using interventional imaging, device modelling and personalised blood flow simulations