Empirical and Computational Evaluation of Hemolysis in a Microfluidic Extracorporeal Membrane Oxygenator (ECMO) Prototype

Abstract

Microfluidic devices promise to overcome the limitations of conventional hemodialysis and oxygenation technologies by incorporating novel membranes with ultrahigh permeability into portable devices with low blood volume. However, the characteristically small dimensions of these devices contribute to both non-physiologic shear that could damage blood components and laminar flow that inhibits transport. While many studies have applied empirical and computational models of hemolysis to medical devices, such as valves and blood pumps, little is known about blood damage in the microfluidic flow regime. In this study, four design variants of a microfluidic membrane-based oxygenator and two controls (positive and negative) are introduced and modeled using a Computational Fluid Dynamics (CFD) model to predict hemolysis. The simulations were performed in ANSYS Fluent for nine shear stress-based Power Law hemolysis model variants. Empirical testing of the devices in a recirculating loop revealed levels of hemolysis significantly lower (3 ppm hemolysis for pump, tube, and device combined) than the hemolysis ranges (>10 ppm) observed in conventional oxygenators. We found that most of the nine tested hemolysis models overpredict (5x to 10x) hemolysis compared to empirical experiments. However, two models demonstrated higher predictive accuracy for hemolysis values in devices characterized by low shear conditions, while another set of three models exhibited better performance for devices operating under higher shear conditions. Our study highlights the limitations of combining hemolysis models with computational fluid dynamics models for a priori in silico device-induced hemolysis. Nevertheless, with a judicious selection of hemolysis models based on the shear ranges of the test device, we propose that computational modeling can complement empirical testing in the development of novel micro-dialyzers or oxygenators, allowing for a more efficient iterative design process. Furthermore, the low device-induced hemolysis (< 2 ppm) measured in our study at physiologically relevant flow rates is promising for the future development of microfluidic dialyzers and oxygenators.