Rick Schreurs

119 Left and right ventricular response to CRT Simulation-derived mechanistic insights The ability of CircAdapt to realistically simulate both LV and RV hemodynamics during pacing allowed us to further study the impact of differences between both chambers on cardiac output. In particular, the model showed a difference in response between dP/dt max and stroke volume/cardiac output. Similar differences have been observed in a clinical study where LV dP/dt max responses differed from responses in stroke work [36]. Our simulations provided a plausible explanation for this paradoxical observation. While LV and RV stroke volume of the first beat after a change of pacing delay can differ substantially, a common steady-state stroke volume and thereby cardiac output is reached due to balancing ventricular preload conditions during the next few beats. There are three main determinants of the newly achieved steady-state preload condition. First, the changes in stroke volumes directly change the end-systolic volumes and thereby the end-diastolic volumes in the next beat. Second, the pacing-induced changes in effective left and right AV-delays change the efficiency of atrial and ventricular diastolic filling and subsequent systolic contraction. Thirdly, changes in stroke volume affect the filling of the other ventricle through the systemic and pulmonary circulations. While dP/dt max changes congruently with first-beat stroke volume in the same ventricle, it is less affected by changes in preload. As a result, LV and RV dP/dt max are much more sensitive to changes of VV-delay and, hence, asynchrony of electrical activation than to changes of AV-delay. On the other hand, changes of AV-delay affect cardiac output more than LV and RV dP/dt max . Computer modeling in therapy optimization Other cardiac computer modelling studies have been conducted to investigate other factors in the optimization of CRT therapy. For example, in a cohort of 648 virtual patients it was found that the location of the LV pacing site is an important factor in response to CRT [37]. Electrophysical cardiac computer modeling studies also demonstrated the importance of LV pacing site and the potential of simulations to predict the electrical optimal pacing location and setting [38–40]. Our study shows, however, that an electrical optimum (lowest electrical dyssynchrony) might not necessarily be optimal for overall pump function. The CircAdapt simulations performed for this study can run on a single core in real time. CircAdapt requires activation time as input and lacks the cardiac electrophysiological model necessary to extract this information from standard clinical data. This input could, however, potentially be generated by other models, for example the ones referred to in the previous paragraph. This would also allow for testing of alternative pacing sites, which would result in different activation patterns, which subsequently can be used as input for CircAdapt simulations. A workflow where fast and anatomically realistic cardiac electrophysiological simulations are combined with cardiac mechanical and circulatory CircAdapt simulations might further increase the clinical applicability of cardiac computer models. 6