Multiscale modelling of single-ventricle hearts for clinical decision support: a Leducq Transatlantic Network of Excellence.

Abstract

Keywords: Congenital heart surgery † Modelling† HaemodynamicsDeveloped in the 1930s and 1940s to study hydraulic problems,computational fluid dynamics (CFD) has become a practical mod-elling tool to solve and analyse physical phenomena that involvefluid flows. In fact, the Food and Drug Administration is now inte-grating computational modelling into the evaluation and testingprocesses with increasing frequency and mandate [1]. In congeni-tal cardiac surgery, where the objective of the operative recon-struction is to reshape the blood flow within the heart and greatvessels to achieve the best dynamics and tissue oxygen delivery,CFD is a natural tool to uncover suboptimal circulations andimprove surgical techniques. However, a lack of a common lan-guage and mutual understanding of each other’s expertise haveoften stymied this logical collaboration between cardiac surgeonsand engineers. The breakthrough came in 1996 when Marc deLeval at Great Ormond Street Hospital in London and his collea-gues at Milan’s Politecnico collaborated to show that CFD model-ling can be used to improve surgical procedures in the totalcavopulmonary connection (TCPC) [2].Advances in computationalmethods have led to numerous contributions in the field of con-genitalheartdiseasesandsurgery, includingassistdevicedevelop-ment, studying of valvularand aortic pathologies, modifications tothe Fontan operation and the continuing effortsto understand themodified physiologies in single ventricularcirculations [3–8].While these advances have shed light into some of the alteredflow dynamic phenomena that are unique in congenital heartsurgery, there has been an increased recognition that modellingapproachesthat only focus on the local or the surgical domain willmiss or underrepresent the overall effects on the entire cardiovas-cular and pulmonary physiology. For example, an isolated flowmodel of the TCPC intended to estimate power loss cannotpredict the Fontan pressure, nor could an isolated model of a3.5-mm modified Blalock–Taussig (mBT) shunt used to calculateshear stress reveal the systemic oxygen delivery. In effect, thehaemodynamics of the operative reconstruction site are dynamic-ally coupled to the rest of the cardiovascular system. New multi-scale modelling methods have been developed to provide acomputationally efficient approach to correctly model both localandsystems-levelbehaviour.Withoutgoingintothemathematicalbackground, a multiscale model of the cardiovascular systemcombines the detailed 3D, anatomically accurate CFD model ofthe desired surgical reconstruction with a zero-dimensional (0D)hydraulic lumped-parameter network (LPN) representation of therest of the cardiovascular circulation system. Computationally, theflow and pressure output values from the 3D CFD model becomethe input pressure and flow values to the 0D LPN and, in turn,these same outputs from the LPN model become the input valuesto the CFD model of the surgical domain. This multiscale ap-proach, such as shown for a TCPC model (Fig. 1), allows forclosed-loop circulatory modelling. The initiating conditions set bythe user put into motion a set of calculations that iteratively arriveat flow and pressure solutions anywhere in the circulation. In amultiscale TCPC model, not only would shear stress and powerloss within the TCPC be calculated, but also would a host of clinic-ally relevant physiological variables, such as Fontan pressures,pressure–volume relationship of the single ventricle and cerebralperfusion. And when combined with fundamental oxygen equa-tions, systemic and end organ, such as cerebral and myocardial,oxygen delivery can be assessed. Further adjustments to thesemodels allow for simulations underexercise and growth effects.In 2010, Fondation Leducq, a private foundation in France,awarded a 5-year Transatlantic Networkof Excellence grant to ourgroup to continue collaborations that began in 1996. Comprisingfour American and three European clinical and engineering insti-tutions, the Modeling of Congenital Heart Alliance (MOCHA)investigators were tasked to apply the state of the art in computa-tional and engineering to all three stages of the surgical palliativepathway of single-ventricle physiology to provide a novel clinicaldecisionsupportsystem.Ouraimistoestablishanewinvestigativeparadigm in which patient-specific anatomy and physiology areused in an engineering model to predict surgical outcomes andsupplement patient management, as shown in our research cycle(Fig. 2). Such a process involves virtual surgery and computation-al/experimental simulations using clinical data acquired fromechocardiography, computed tomography, magnetic resonance