Patient-Specific Simulations in Interventional Cardiology Practice: Early Results From a Clinical/Engineering Center

Abstract

Patient-specific models have been recently applied to investigate a wide range of cardiovascular problems, including cardiac mechanics, hemodynamic conditions, and structural interaction with devices [1]. The development of dedicated computational tools, which combined the advances in the field of image elaboration, finite element (FE), and computational fluid dynamic (CFD) analyses, has greatly supported not only the understanding of human physiology and pathology but also the improvement of specific interventions by taking into account realistic conditions [2,3]. However, the translation of these technologies into clinical applications is still a major challenge for the engineering modeling community [4].Over the last decade, our engineering group, which is part of the Cardiorespiratory Unit in the main UK children hospital, has been involved in the development of patient-specific models to support the development of the percutaneous pulmonary valve implantation (PPVI) procedure [5,6]. This experience allowed the creation of a library with over 600 anatomical models of cardiovascular structures and 24 devices for a variety of cardiovascular interventions. In this context, there was an increasing clinical demand to extend the use of such models in planning interventions, especially in complex cases of congenital heart disease.In this study, we report the early results of the translations of patient-specific computational simulations into clinical practice to predict outcomes of percutaneous interventions.Six patients who were referred to our Center for PPVI (n = 4) and stenting of aortic coarctation (CoA, n = 2) were included in this study. Image data conventionally acquired for patients' clinical assessment, including computed tomography and magnetic resonance imaging, were postprocessed to reconstruct the three-dimensional anatomy of the site of cardiovascular structures. For each case, patient-specific computational analyses were set up in order to simulate peri- and postoperative scenarios using different devices, which vary in terms of size and shape, from our library of available device models. In particular, following morphological measurements, FE analyses simulated the implantation phases of selected devices and their interaction with the individual implantation site, while CFD analyses were performed to investigate the hemodynamics before and after the intervention. Results of patient-specific models were presented and discussed during the multidisciplinary cardiac meeting in our unit to assess the feasibility of the intervention. Finally, clinical outcomes were compared with the model predictions.The simulations concerning the patients referred for PPVI indicated feasibility in all cases and outlined the best-fitting device for each individual following analyses of the interaction between the available stents and the implantation sites (Figs. 1(a)–1(c)). Clinical procedures provided successful outcomes and a good agreement with the predictions in three cases. One PPVI was not performed following balloon sizing because, during the catheterism, a highly distensible implantation site was felt not suitable for the stability of the stent.The simulations of stenting of the first CoA case provided information on the optimal size of a covered stent, which was identified as best solution to restore physiological flow conditions (Fig. 1(d)). This prediction was in excellent agreement with the clinical outcomes. In the second CoA case, a thoracic endovascular graft was suggested. This intervention is currently planned to avoid surgery in this patient.This study reports the early results of a prospective study, which aims to translate patient-specific models into clinical practice. The first results are promising in terms of reliability of the simulations, response time, and usefulness in clinical practice. However, future studies will need to include more realistic data about mechanical characterization of specific patients and possibly further reduce the computational costs. The translation of these technologies is crucial, as it can limit the procedural risks of currently available devices and support the development of new devices that are morphologically and functionally suitable for each patient.