Multi-scale modeling of the systemic effects of VA-ECMO system

Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) is employed in patients with cardiogenic shock (CS), a condition marked by impaired left ventricular (LV) contractility. Despite technological advances, mortality remains high, and complications such as LV distension, impaired gas exchange, and ischemia at the return site are frequent. Outcomes appear to be strongly related to the etiology of CS, resulting in a wide and heterogeneous spectrum of pathological state. However, no clear classification of CS based on etiology currently exists, and the hemodynamic parameters assessed are still limited. Numerical simulations can help guide clinical decisions; however, to fully exploit potentialities of these tools standardised protocols for data collection are required. In this thesis, different numerical approaches are employed to analyse the hemodynamics of subjects supported by VA-ECMO: the lumped parameter approach for global evaluation and computational fluid dynamics (CFD) for local hemodynamic assessment. The lumped parameter approach is first applied to characterize CS and VA-ECMO support. Two systemic circulation models of an adult male subject—a condensed (CM) and an extended (EM) configuration—are compared to evaluate how structural detail influences clinical predictions. The CS condition is modeled by introducing in the lumped characterisation of the LV an impairment factor k, which regulates the description of the pathological scenario. The results show that, even though starting from the same physiological conditions, the CS scenario produced by the two models diverge progressively. This highlights the importance of calibration tailored to the specific pathological scenario. Indeed, the need for standardised protocols for hemodynamic measurements is of utmost importance in order to ensure improved quality and comparable outcomes between two different models. Moreover, inconsistencies among the diagnostic criteria are noticed, as each one is associated to a different value of k. The VA-ECMO setting is defined in accordance with the clinical requirements (i.e., based on MAP (mean arterial pressure) and VA-ECMO flow), yielding coherent predictions in both models. Interestingly, the model predictions suggest a redundancy in the flow rate delivered by the VA-ECMO device. An evaluation of the local hemodynamics in the return site is also performed. By means of CFD approach, the standard cannula is compared to two innovative designs featuring one (single-hole cannula, SH) or four (multi-hole cannula, MH) holes, respectively. To comprehensively assess performance, both laminar flow (consistent with VA-ECMO weaning) and turbulent flow (consistent with target support) regimes are analysed. The standard cannula confirms the inadequacy in preventing limb ischemia, while the new designs improve distal perfusion, though some limitations are evident: distal thrombus-prone region in the vessel in the MH cannula at low support, and intra-cannula stagnation in SH at high support. Areas at risk of wall abnormal response are also identified, especially during weaning. To address some limitations inherent to each numerical approach, a coupling procedure is applied to simulate an appropriate VA-ECMO flow distribution in the 0D cardiovascular model and to prescribe physiologically consistent time-dependent boundary conditions at the return site. Simulations of the weaning phase with the standard configuration reveal an evolution of the jet during the cardiac cycle and show how distal and proximal flow patterns change with the different phases of the heartbeat. These findings highlight the potential of numerical modeling to complement clinical assessment of cardiogenic shock and call for standardised hemodynamic protocols to improve patient management.