Background and aims:  The effect of veno-arterial extracorporeal membrane oxygenation (VA-ECMO) on left atrial pressure (LAP) in the presence of interventricular interaction and the Frank–Starling mechanism is unknown. We developed and validated a mock circulatory loop (MCL) incorporating a novel, 3D-printed biventricular heart model and Frank–Starling algorithm, and used this model to assess the determinants of LAP during VA-ECMO support.

Methods: The MCL was designed to allow a separate ventricle or biventricular configuration, with or without an active Frank–Starling mechanism. The biventricular model with Frank–Starling mechanism was validated in terms of (1) the presence and degree of ventricular interactions; (2) its ability to simulate Frank–Starling physiology; and (3) its capacity to simulate normal and pathological cardiac states. In the separate ventricle and biventricular with Frank–Starling models, we assessed the effect on LAP of changes in mean aortic pressure (mAoP), ECMO pump speed, LV contractility and ECMO return flow direction.

Results: In the biventricular configuration, clamping RA inflow decreased RAP, with a concurrent decrease in LAP, consistent with direct ventricular interaction. With a programmed Frank–Starling mechanism, decreasing RAP was associated with a significant reduction in both LV outflow and LV end-systolic pressure. In the biventricular model with a Frank–Starling algorithm, the MCL was able to reproduce pre-defined normal and pathological cardiac output, and arterial and ventricular pressures. Increasing aortic pressure caused a linear increase in LAP in the separate ventricle model, which was attenuated in the biventricular model with Frank–Starling mechanism. Increasing ECMO pump speed caused no change in LAP in the separate ventricle model (p = 0.75), but significantly decreased LAP in the biventricular model with Frank–Starling mechanism (p = 0.039), with stabilization of LAP at the highest pump speeds. Changing the direction of VA-ECMO return flow did not affect LAP in either the separate ventricle (p = 0.91) or biventricular model with Frank–Starling mechanism (p = 0.76).

Conclusions: Interventricular interactions and the Frank–Starling mechanism can be simulated in a physical, biventricular MCL. In their presence, the effects of VA-ECMO on LAP are mitigated, with LAP reduction and stabilization at maximal VA-ECMO speeds.