Multiscale Modeling of VA-ECMO to Guide Clinical Management:Insights into Hemodynamics, Oxygen Delivery, and Ventricular Load

Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) is a life-saving therapy used in patients with severe cardiac failure or cardiogenic shock. By providing extracorporeal circulatory and respiratory support, VA-ECMO can rapidly restore systemic perfusion and oxygen delivery. However, the retrograde arterial flow generated by the ECMO circuit profoundly alters cardiovascular physiology, increasing left ventricular afterload and modifying the distribution of oxygenated blood. These effects may lead to complications such as ventricular distension, pulmonary congestion, and differential oxygenation of the upper and lower body. Despite widespread clinical use, the physiological mechanisms underlying these interactions remain incompletely understood. This thesis aimed to improve the mechanistic understanding of VA-ECMO by developing and applying a computational modeling framework spanning multiple physiological scales. Three-dimensional computational fluid dynamics (CFD) models and zero-dimensional closed-loop cardiovascular simulations were used to investigate how ECMO flow, native cardiac function, and vascular anatomy jointly influence hemodynamics and oxygen transport. Chapters 2–4 focus on vascular-scale flow and oxygenation using CFD simulations of the aorta and major branches. These studies show that ECMO flow rate strongly affects aortic flow competition and oxygen mixing between retrograde ECMO blood and native ventricular output. Low ECMO flows may result in insufficient oxygen delivery to the upper body, whereas excessive flows create pronounced mixing zones in the ascending aorta, providing a mechanistic explanation for Harlequin (North–South) syndrome. Anatomical variability of the aortic arch and supra-aortic branches was found to significantly influence regional oxygen delivery, highlighting the patient-specific nature of ECMO support. Pulsatile ECMO flow improved flow oscillation and shear patterns while maintaining comparable mean oxygen delivery, suggesting potential benefits for vascular physiology. Chapter 5 addresses cardiac-scale physiology using a lumped-parameter cardiovascular model incorporating a physiologically based ventricular formulation. The simulations demonstrate that increasing ECMO flow markedly elevates left ventricular afterload, leading to ventricular distension and a substantial reduction in native stroke volume and ejection fraction. While ECMO maintains total systemic perfusion, the heart’s mechanical contribution is progressively suppressed. These results quantitatively reproduce clinical observations of left ventricular overload during VA-ECMO and explain why excessive support may hinder myocardial recovery. Taken together, the findings reveal that VA-ECMO reshapes ventricular–arterial coupling across multiple scales. The suppression of native cardiac output provides a unifying interpretation of the aortic flow competition and oxygen mixing patterns observed in the CFD analyses. Thus, ECMO-induced changes in oxygen distribution depend not only on circuit settings and vascular anatomy, but also on altered ventricular mechanics. In conclusion, this thesis demonstrates that optimal VA-ECMO management requires balancing systemic support with ventricular unloading rather than maximizing flow. The computational framework developed here provides a mechanistic foundation for future patient-specific modeling and supports the development of more physiologically informed ECMO optimization strategies.