Mass and momentum transfer in membrane-based bioartificial liver systems

Abstract

Liver failure, caused by acute or chronic end-stage liver disease (ESLD) imposes a significant disease burden worldwide. Chronic liver disease and cirrhosis is ranked as 12th cause of death in the United States and 4th in middle-aged adults. Researchers in Mayo Clinic report liver-related mortality as 8th by using a more comprehensive definition accounting for other aspects of liver disease as well. Currently, liver transplantation remains the conventional treatment for ESLD as the only medically proven method to promote patient’s health. To avoid the problem of inadequate donor organs and yet provide a comprehensive range of liver functions, cell-based therapies have been actively under investigation to potentially provide a substitute for transplantation, or a temporary support for liver failure patients. Studies on the latter aim has led to development of extracorporeal bioartificial liver (BAL) devices. Hepatic cell cultures are exploited for different applications in liver disease studies, drug toxicity testing, and bioartificial liver (BAL) devices. However, development of such systems is often hindered by the peculiar characteristics and intricate requirements of primary hepatocytes, challenging their prolonged functionality and viability in vitro. Despite the development of various 3D cell culture systems using perfused bioreactors, mass transfer properties still remain a critical and controversial topic, especially oxygen supply as the limiting and modulating factor The aim of this work is to enhance and optimize a prototype hollow fiber membrane bioreactor (HFMBR) providing efficient mass transfer for nutrient provision and catabolite removal, promoting prolonged viability and functionality of hepatocytes. In this bioreactor, two bundles of hollow fibers are employed in a crossed configuration: one bundle for supplying the oxygenated medium, and the other for removing the medium from the extra-capillary space. Optimization of the operational culture conditions to enforce an in vivo-like microenvironment is an intrinsic part of the process that requires a clear understanding of the in vitro cellular microenvironment. Oxygen transport in a convection-enhanced, crossed-configuration HFMBR hosting hepatocyte spheroids was investigated through mass transfer modelling using COMSOL Multiphysics®, a specialized, commercial finite-element software. The permeability of hollow fibers (hydraulic, albumin solution) was evaluated experimentally, showing significant, irreversible decrease in the permeance of the membranes due to protein absorption during culture period. Bioreactor’s hydrodynamics was investigated through residence time distribution analysis, by which a portion of the bioreactor was diagnosed as stagnant region. Finally, oxygen diffusion through the medium and the effect of different conditionings on the oxygen sensor’s readings in multi-well plates were studied. Mass transfer in static culture systems – both as a monolayer and as spheroids – was evaluated using a diffusion-reaction model numerically solved for oxygen (steady-state study) and urea (time-dependent study). In addition to the size and number of spheroids, sufficiency of oxygen supply to cells also depended on their distribution (the distance between them) and the amount of culture medium in each well. A convection-diffusion-reaction time distribution analysis, by which a portion of the bioreactor was diagnosed as stagnant region. Finally, oxygen diffusion through the medium and the effect of different conditionings on the oxygen sensor’s readings in multi-well plates were studied. Mass transfer in static culture systems – both as a monolayer and as spheroids – was evaluated using a diffusion-reaction model numerically solved for oxygen (steady-state study) and urea (time-dependent study). In addition to the size and number of spheroids, sufficiency of oxygen supply to cells also depended on their distribution (the distance between them) and the amount of culture medium in each well. A convection-diffusion-reaction time distribution analysis, by which a portion of the bioreactor was diagnosed as stagnant region. Finally, oxygen diffusion through the medium and the effect of different conditionings on the oxygen sensor’s readings in multi-well plates were studied. Mass transfer in static culture systems – both as a monolayer and as spheroids – was evaluated using a diffusion-reaction model numerically solved for oxygen (steady-state study) and urea (time-dependent study). In addition to the size and number of spheroids, sufficiency of oxygen supply to cells also depended on their distribution (the distance between them) and the amount of culture medium in each well. A convection-diffusion-reaction time distribution analysis, by which a portion of the bioreactor was diagnosed as stagnant region. Finally, oxygen diffusion through the medium and the effect of different conditionings on the oxygen sensor’s readings in multi-well plates were studied. Mass transfer in static culture systems – both as a monolayer and as spheroids – was evaluated using a diffusion-reaction model numerically solved for oxygen (steady-state study) and urea (time-dependent study). In addition to the size and number of spheroids, sufficiency of oxygen supply to cells also depended on their distribution (the distance between them) and the amount of culture medium in each well. A convection-diffusion-reaction model was developed to describe momentum and mass transfer in the bioreactor, in which the influential parameters were parametrized through implementation of applicable correlations. The model was numerically solved for two different types of geometries: (i) single-spheroid model using a periodic/symmetric unit cell within the bioreactor to locally represent the system decreasing the computational complexity of the model, (ii) miniaturized bioreactor model. The single-spheroid model was used to carry out a systematic parametric study to evaluate the effect of different parameters – oxygen tension (Co,sat), perfusion rate (QBR), hollow fiber spacing (δHF), spheroid diameter (Dsph), Michaelis-Menten kinetics for oxygen uptake (Vmax, Km) and porosities of the spheroid (εcc) and the membrane (εm) – on dissolved oxygen concentration (DOC) profile. Dimensionless numbers were defined for in-depth analysis of oxygen transfer and how each parameter can affect that. Among the operational conditions, Co,sat was found much more influential than QBR. Due to the mild advection added, the extra-spheroid resistances to diffusive mass transfer, i.e. the membrane (governed by εm) remains an important factor. However, εcc was found as a key intrinsic property strongly affecting intra-spheroid DOC profile. Maintaining physiological DOC range in large spheroids (Dsph=400μm) with different porosities was investigated in the single-spheroid model. Regulation of DOC profile was more manageable in spheroids with higher εcc, which lead to feasibility of achieving physiological oxygen concentrations. Low-porosity spheroids demonstrated a sharper concentration gradient, challenging sufficient oxygen supply to cells. Temporal shrinkage of spheroids due to rearrangement of cells changes the microstructure of the spheroid, the effect of which was numerically studied and proved to adversely affect the transport properties and consequently the DOC profile inside the spheroid. In the end, values from an experimental study were incorporated into the model to analyze the cellular microenvironment during the experiment, and the predictive capacity of the model regarding the outcome. Miniaturized bioreactor model was developed to reduce the computational cost while providing a more realistic model for the bioreactor. Another major advantage of this approach is capacitating investigation of the fluid dynamics inside the bioreactor. Notable DOC drop along the lumina of the supplying bundle was observed, consistent with the DOC gradient in the extra-capillary space along the supplying bundle. Having retentate flow in the hollow fibers significantly reduced these gradients and improved oxygen supply to the cells. Oxygen transfer was not noticeably affected by different flow patterns realized through using both bundles supplying or both removing the medium. However, minimization of the stagnant region had in fact a negative influence on oxygen supply. The miniaturized bioreactor model was also modified based on the experimental results for comparison with the single-spheroid model and the actual bioreactor, showing better compatibility with the real case.