Microfluidics and Nanofluidics

, Volume 2, Issue 6, pp 525–535 | Cite as

Microchannel bioreactors for bioartificial liver support

  • Jaesung Park
  • Mehmet Toner
  • Martin L. Yarmush
  • Arno W. Tilles
Research Paper

Abstract

An extracorporeal bioartificial liver (BAL) device containing viable hepatocytes has the potential to provide temporary hepatic support to liver failure patients, serving as a bridge to transplantation while awaiting a suitable donor. In some patients, providing temporary hepatic support may be sufficient to allow adequate regeneration of the host liver, thereby eliminating the need for a liver transplant. Although the BAL device is a promising technology for the treatment of liver failure, there are several technical challenges that must be overcome in order to develop systems with sufficient processing capacity and of manageable size. In this study, the authors describe the critical issues involved in developing a BAL device. They also discuss their experiences in hepatocyte culture optimization within the context of a microchannel flat-plate BAL device.

Keywords

Oxygen Hepatocyte Microgrooves Bioreactor Shear stress 

List of symbols

H or h

Channel height (μm)

Co

Inlet oxygen concentration (nmol/cm3)

Pe

Peclèt number (no unit)

Q

Volumetric flow rate (mL/min)

W

Channel width (cm)

D

Oxygen diffusion coefficient (cm2/s)

γ

Cell seeding density (cells/cm2)

Da

Damköhler number (no unit)

OUR (Vmax)

Oxygen uptake rate (nmol/s/106 cells)

τ

Shear stress (dyn/cm2)

1 Introduction

The liver is the largest solid organ in the body and performs numerous functions which are vitally important to maintaining metabolic homeostasis. These functions include synthesis of serum proteins, regulation of nutrients, production of bile, and metabolism and conjugation of compounds for excretion in the bile or urine. Hepatocytes, which are the predominant cell type within the liver, account for two-thirds of the liver mass. The liver is normally able to regenerate after acute injury and regain its function under appropriate physiological stimuli (Arias et al. 2001). However, liver failure occurs when the normal regenerative process is compromised and the residual functional capacity of the damaged liver is unable to sustain life.

Over 30,000 patients die annually in the United States from liver disease. The two principal causes of liver failure are cirrhosis and fulminant hepatic failure. Liver cirrhosis, whose etiology includes alcoholism and chronic hepatitis, is an irreversible process that occurs when fibrotic tissue replaces normal liver tissue as a result of chronic injury. Fulminant hepatic failure is a clinical syndrome defined by impaired mental and neuromuscular function whose etiology includes chemical and viral hepatitis. It occurs as a result of massive hepatocyte necrosis and is the most severe manifestation of end-stage liver disease with mortality rates greater then 80%. Although there is currently no cure for liver failure, orthotopic liver transplantation is the only clinically proven effective treatment for patients with end-stage liver disease (Arias et al. 2001; Smithson and Neuberger 1999). In the year 2004, there were 6,168 liver transplants performed in the United States and 1,856 patients died while waiting for a liver transplant (Based on UNOS OPTN data as of October 7, 2005). As can be seen from these statistics, a major limitation of this treatment is the scarcity of donor organs which results in patients dying while on the waiting list. There are currently over 17,930 patients on the waiting list to receive a donor liver.

In this article, we address the critical issues in bioartificial liver (BAL) development, and summarize our experience using a microchannel flat-plate BAL device.

2 Replacing liver functions

Over the past four decades, many attempts have been made to develop artificial liver systems to support patients with liver failure. The goal of these systems is to provide temporary hepatic support for patients with fulminant hepatic failure who are awaiting orthotopic liver transplantation. Since the liver has tremendous regenerative capacity, some liver failure patients may spontaneously recover if provided temporary hepatic support, thereby averting a liver transplant and the associated life-long immunosuppressive therapy. Duplicating the liver’s complex metabolic functions that are essential for survival has been a significant challenge. There are two general categories of extracorporeal liver support systems that have been developed, nonbiological and biological. Nonbiological methods, including hemodialysis and hemoperfusion, focus on the removal of toxins accumulating in the patient’s blood. In a randomized clinical trial, charcoal hemoperfusion, the most characterized nonbiological method, failed to show a survival benefit in patients with fulminant hepatic failure (O’Grady et al. 1988). Recently, there has been renewed interest in further refining these approaches, with three different systems, the BioLogic-DT System by HemoCleanse, Inc., West Lafayette, IN, USA (Ash 2001), the Molecular Adsorbent Recirculating System (MARS) by Teraklin AG, Rostock, Germany (Stange et al. 2002), and the Prometheus system by Fresenius Medical Care AG, Bad Homburg, Germany (Rifai et al. 2003), at various stages of clinical evaluation. Although biological-based approaches such as cross-circulation, extracorporeal liver perfusion with crosshemodialysis, and liver tissue hemoperfusion were all shown to provide limited functional support on a short-term basis, extracorporeal whole liver perfusion has experienced renewed interest in recent years, and it has successfully been used as a bridge to transplantation (Pascher et al. 2002).

The hybrid BAL device, in which functional hepatocytes are housed within a man-made synthetic device, can overcome some of the problems seen in other forms of liver support. These devices, with their metabolically active hepatocytes, can provide a broader range of liver-specific functions compared to non-biological or other biological-based systems (Hu et al. 1997). For these devices to function optimally, novel designs are needed which allow the maintenance of the high cell densities required of a clinical device, with minimal mass transfer limitations to the hepatocytes.

3 Design issues for a bioartificial liver device

For a BAL device to function optimally, it must maintain the hepatocytes in an environment that mimics the in vivo environment as close as possible. In order to do this, there are certain design criteria that must be met including: (1) to use a sufficient number of well-differentiated hepatocytes that can maintain long-term function, (2) to reduce mass transfer resistances and eliminate substrate limitations so that the device can function at maximum efficiency, and (3) to minimize the dead volume within the device thereby reducing plasma dilution effects in the patient. Various device configurations have been utilized in an effort to achieve these design criteria. In general, a bioreactor is inoculated with hepatocytes and the patient’s blood or plasma circulates through the device. The ideal bioreactor design would maximize mass transfer to the hepatocytes thereby allowing nutrients, including oxygen, and toxins from the patient’s blood or plasma to reach the hepatocytes. The treated blood or plasma, including metabolites and synthetic products, would then be returned to the patient’s circulation. Achieving this task requires a large surface area for cell attachment with uniform cell distribution and flow.

Several bioreactor designs incorporate membranes of different selectivity’s to prevent direct blood or plasma contact with the hepatocytes. In these designs, mass transfer is determined by the molecular weight cutoff (i.e., pore diameter) of the membrane for a given pressure and flow rate, which can influence the performance of the bioreactor. The idea is to select a molecular weight cutoff which allows transport of proteins (e.g., albumin) to the patient’s circulation and toxins out of the patient’s circulation while preventing immune-mediated injury to the hepatocytes. The properly selected membrane can also exclude the transport of xenogeneic substances, as well as cells, to the patient’s circulation. Some designs use membranes with low molecular weight cutoff (70–100 kD) which allow passage of serum albumin but excludes proteins of higher molecular weights, such as immunoglobulins, thereby providing immunoprotection to the hepatocytes. Other designs use microporous membranes with large pore diameters (0.2 μm) which allow free passage of plasma proteins as well as large molecular weight proteins (e.g., clotting factors) and toxins (soluble or protein bound) between blood or plasma and the hepatocytes. These microporous membranes, however, do exclude passage of cells (e.g., blood cells and hepatocytes). The rational for using membranes with large pore diameters is that fluid convection is enhanced, thereby improving transfer of substrates and products to and from the hepatocytes.

Another important issue in the design of a BAL device is the maintenance of sufficient oxygen supply to the hepatocytes. Hepatocytes are highly metabolic with high oxygen uptake rates (Balis et al. 1999; Foy et al. 1994; Rotem et al. 1994, 1992). Therefore, they require adequate oxygenation to maintain viability and function. In order to oxygenate the circulating blood or plasma, some designs incorporate an oxygenator within the bioreactor itself, while other designs use an inline oxygenator within the extracorporeal perfusion circuit. Physiological temperatures are maintained via heat exchangers placed in the perfusion circuit.

4 Bioartificial liver devices undergoing clinical trials

Most devices undergoing clinical trials are designed based on hollow fiber technology. The Extracorporeal Liver Assist Device (ELAD) by Vital Therapies, Inc. (San Diego, CA, USA), is a hollow fiber device that uses human hepatoblastoma (C3A) cells loaded into the extracapillary space, with the patient’s blood flowing through the capillary lumina. Although initial studies showed that the device was safe for clinical use and provided metabolic support to patients with late-stage liver failure, demonstrating efficacy has been more difficult. In a clinical trial of 24 patients with acute liver failure, half of whom were assigned to treatment with the device and the other half to the control group, no clinical survival benefit was seen (Ellis et al. 1996). This finding was partly explained by the fact that the control group had a higher than expected survival.

The HepatAssist by Arbios Systems, Inc. (Los Angeles, CA, USA) is a hollow fiber device that uses cryopreserved porcine hepatocytes attached to collagen-coated dextran microcarriers. The hepatocytes are loaded into the extracapillary space, and patient plasma then flows through the capillary lumina. Within the flow loop is a charcoal adsorption column, which removes certain organic compounds from the plasma prior to contacting the hepatocytes. A phase I clinical trial showed that this device could serve as a bridge to liver transplantation in patients with acute liver failure (Watanabe et al. 1997). Recently, a clinical, multicenter phase II/III randomized trial of a hollow fiber BAL device was conducted at several United States and European sites. The results showed a trend toward improved survival in fulminant hepatic failure patients who received treatment with this BAL device (Demetriou et al. 2004).

The Bioartificial Liver Support System (BLSS), developed by Excorp Medical (Oakdale, MN, USA), is a hollow fiber device that uses porcine hepatocytes embedded in a collagen matrix. The patient’s blood is perfused through the capillary lumina. A phase I study using this device to treat four patients with acute liver failure and acute-on-chronic liver failure indicated that the device was well tolerated by the patients, although no conclusions regarding efficacy could be drawn (Mazariegos et al. 2001; Patzer et al. 2002). Results of the phase I/II trail are pending.

It has been suggested that these hollow fiber devices are subject to substrate limitations due to the relatively large diameter of the fibers, and the transport resistances associated with the fiber wall (Catapano 1996; Hay et al. 2001, 2000). Given the high oxygen utilization of hepatocytes and low solubility of oxygen in plasma, the adequate delivery of oxygen in hollow fiber BAL devices has been problematic. To improve oxygenation, some designs utilize hollow fibers as conduits for oxygen delivery. The Modular Extracorporeal Liver System (MELS) of the Charite Virchow-Klinikum in Berlin, Germany consists of a CellModule, DetoxModule, and DialysisModule. The CellModule is a bioreactor in which discrete bundles of woven capillary membranes enter and leave the bioreactor, forming a three-dimensional structure (Gerlach et al. 1994). The hepatocytes are distributed in a collagen matrix on the membrane framework, and the extracapillary space is perfused with plasma. The capillary bundles allow independent oxygen supply and plasma inflow and outflow. The DetoxModule is for albumin dialysis to remove albumin-bound toxins, and the DialysisModule is for hemofiltration. A phase I clinical trial using only the CellModule bioreactor component, containing porcine hepatocytes, in eight patients with acute liver failure revealed that the extracorporeal liver support with this bioreactor was safe and well tolerated (Sauer et al. 2003). All treated patients were successfully bridged to transplantation.

The Academic Medical Center Bioartificial Liver (AMC-BAL) developed by Flendrig and colleagues uses a three-dimensional, spirally wound, non-woven polyester matrix for hepatocyte attachment with integrated hollow fibers for oxygen delivery to the cells (Flendrig et al. 1997). In contrast to other designs, this system uses direct contact between the patient’s plasma and the matrix attached hepatocytes to improve bidirectional mass transfer. In Italy, this device seeded with primary porcine hepatocytes underwent a phase I trial in 12 patients with acute liver failure. Eleven of the 12 patients were successfully bridged to liver transplantation, and one patient treated with the BAL recovered without needing a liver transplant (van de Kerkhove et al. 2002, 2005). No adverse events were noted.

5 Microchannel flat-plate hepatocyte bioreactor design

In an effort to maximize oxygen availability to the hepatocytes and to reduce mass transport limitations, we developed a microchannel flat-plate bioreactor with an internal gas permeable membrane through which oxygen was supplied (Tilles et al. 2001a, b) (Fig. 1a). The hepatocytes were attached to a collagen-coated glass substrate (25 × 75 mm2) and were in direct contact with the perfusing medium. A gas permeable membrane separated the liquid compartment from the oxygenating gas compartment. This design allowed oxygen delivery to the hepatocytes to be decoupled from the medium flow, thereby allowing oxygen delivery and flow to be studied independently. In these studies, the bioreactor channel heights ranged between 50 and 500 μm and medium flow rates ranged between 0.06 and 4.18 mL/min. Our initial studies modeled oxygen concentrations in microchannel flat-plate bioreactors with and without the internal membrane oxygenator (Tilles et al. 2001a).
Fig. 1

Schematic representation of microchannel flat-plat bioreactors a with internal membrane oxygenator, and b without internal membrane oxygenator

6 Oxygen modeling in bioreactors with and without internal membrane oxygenation

6.1 Flat-plate bioreactor without internal membrane oxygenation

In the case for a microchannel flat-plate bioreactor with a high aspect ratio between longitudinal length and height under fully-developed laminar flow (typically Reynolds number < 2,000), one can neglect the axial diffusion. In this bioreactor, without the internal oxygenating membrane (Fig. 1b), oxygen supply to the hepatocytes is dependent on convective transfer only. The oxygen concentration can be expressed as
$$ \frac{{\partial \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}}} {{\partial \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x}}} = \frac{1} {{{\text{Pe}}}}\frac{{\partial ^{2} \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}}} {{\partial \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y}^{2} }} $$
(1)
$$ \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x} = \frac{x} {h}; \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y} = \frac{y} {h}; \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C} = \frac{C} {{C_{0} }}, $$
where h is the height of the bioreactor compartment, C(x,y) is the oxygen concentration, C0 is the uniform inlet oxygen concentration (nmol/cm3), Pe is Peclèt number (= Q/wD), Q is the volumetric flow rate, w is the channel width, and D is oxygen diffusion coefficient (2.0 × 10−5 cm2/s). Boundary conditions for the bioreactor without internal membrane oxygenation are
$$ \begin{aligned}{} \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C} & = 1\,{\text{at}}\, \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x} = 0 \\ \frac{{\partial \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}}} {{\partial \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y}}} & = - \frac{{\gamma ({\text{OUR}})h}} {{C_{0} D}} = - {\text{Da}}\,{\text{at}}\, \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y} = 1, \\ \frac{{\partial \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}}} {{\partial \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y}}} & = 0\,{\text{at}}\, \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y} = 0, \\ \end{aligned} $$
(2)
where γ is the cell seeding density (i.e., number of cells per unit area), Da is the Damköhler number, and OUR is the oxygen uptake rate of hepatocytes (nmol/s/106 hepatocytes). Assuming OUR of hepatocytes is constant and equal to the maximum oxygen uptake rate, Vmax, Damköhler number reduces to
$$ {\text{Da}} = \frac{{\gamma \,V_{{{\text{max}}}} \,h}} {{C_{0} \;D}} $$
(3)
and Eq. 1 can be solved to yield (Carslaw and Jaeger 1959):
$$ \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}( \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x}, \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y}) = \frac{C} {{C_{o} }} = 1 - \left[ {\frac{{{\text{Da}}}} {{{\text{Pe}}}} \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x} + {\text{Da}}} \right[\frac{{3 \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y}^{2} - 1}} {6} - \frac{2} {{\pi ^{2} }}{\sum\limits_{n = 1}^\infty {\frac{{( - 1)^{n} }} {{n^{2} }}\cos (n\pi \ifmmode\expandafter\tilde\else\expandafter\sim \fi{y})} }\left. {\left. {{\text{e}}^{{ - (n\pi )^{2} \frac{{ \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x}}} {{{\text{Pe}}}}}} } \right]} \right], $$
(4)
where n is an integer. The non-dimensional average concentration can be obtained as
$$ \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{ave}}}} ( \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x}) = 1 - \frac{{{\text{Da}}}} {{{\text{Pe}}}} \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x}. $$
(5)

6.2 Flat-plate bioreactor with internal membrane oxygenation

For very small channel heights (∼ 100 μm) and low flow rates (< ∼1 mL/min), typical conditions used in small-scale bioreactors (Stefanovich et al. 1996), the time constant for convection, ttransit = L/U, is much larger than the time constant for diffusion, tdiffusion = h2/D. Thus, delivery of oxygen to the bioreactor surface is primarily dominated by the diffusion characteristics of the system. Under these conditions, for a Michaelis–Menten-behaving system, as already reported by Peng and Palsson (1996) and Yarmush et al. (1992b), the dimensionless cell surface oxygen concentration, \( \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{cs}}}} ,\) is given by the following equation:
$$ \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{cs}}}} = - \frac{{({\text{Da}} + \beta - 1) \pm {\sqrt {({\text{Da}} + \beta - 1)^{2} + 4\beta } }}} {2}, $$
(6)
where \( \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{cs}}}} = \frac{{C_{{{\text{cs}}}} }} {{C*}},\)\( \beta = \frac{{K_{{0.5}} }} {{C*}}, \)C* is the oxygen concentration at the aqueous membrane surface, and K0.5 is a constant value of oxygen concentration at the cell surface for an OUR of Vmax/2. The average oxygen concentration at the location (x) from the inlet can be described as
$$ \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{ave}}}} = \frac{{1 + \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{cs}}}} }} {2}. $$
(7)

7 Outlet oxygen measurements in bioreactors with and without internal membrane oxygenation

Studies using porcine hepatocytes compared outlet oxygen concentration in bioreactors with (Fig. 1a) and without (Fig. 1b) the internal membrane oxygenator (Tilles et al. 2001a). Figure 2 summarizes the effects of medium flow rates on outlet oxygen partial pressure in bioreactors, with and without the internal membrane oxygenator. In the bioreactor without the internal membrane oxygenator, the outlet oxygen tension decreased gradually from 154 mmHg at a flow rate of 3.5 mL/min to 114 mmHg at a flow rate of 0.5 mL/min, with an approximate overall decrease of 26%. As the flow rate was further decreased to 0.1 mL/min, output oxygen tension precipitously decreased to 2.0 mmHg, which corresponded to a decrease of about 99% from the concentration measured at the highest medium flow rate. This suggested significant oxygen limitations were occurring at volumetric flow rates of 0.1 mL/min, and lower, in the bioreactor without the internal membrane oxygenator. Figure 2 also contains the model fit (Eq. 5) to the experimental data.
Fig. 2

Outlet oxygen partial pressure (pO2) as a function of flow rate (Q) for the bioreactor without (filled circle) and with (open circle) internal membrane oxygenation. Error bars represent standard deviation. The solid line represents the mathematical model fit (Da = 0.08) to the experimental data for the bioreactor without the internal membrane oxygenator. The dashed line represents the model prediction (Da = 0.08, β = 0.052) for the bioreactor with internal membrane oxygenation (Eq. 7) (Tilles et al. 2001a)

We used the mathematical model to predict the cell surface oxygen concentration, which can be estimated from Eq. 4. Figure 3 shows non-dimensionalized cell surface oxygen concentration (Ccs/C0) in the axial direction of the bioreactor without the internal membrane oxygenator at various Pe (with corresponding volumetric flow rates). The inlet pO2 for the reactor was fixed at 159 mmHg, based upon the constraint that the medium was oxygenated with 21% oxygen prior to its entrance into the bioreactor. The model predictions were based on a Da of 0.08, which was obtained from the experimental data fit. The average oxygen concentration decreased in a linear fashion along the length of the bioreactor as more cells consumed oxygen. Assuming that porcine hepatocytes were oxygen-limited when the oxygen concentration at the cell surface was below the normalized K0.5/C0 (∼ 0.05, Balis et al. 1999), one can predict the extent of oxygen metabolism in the bioreactor. For Pe = 33.3 (corresponding to Q = 0.1 mL/min), the average concentration fell below K0.5/C0 at \( \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x} \sim 400, \) which predicted that approximately 50% of the cells in the bioreactor of \( \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x} = 750, \) would be exposed to oxygen partial pressures of less than K0.5 and thus would be oxygen limited. At Pe = 167 (corresponding to Q = 0.5 mL/min), no hepatocytes were exposed to oxygen rate-limiting conditions. Although higher Pe would theoretically result in less oxygen limitations due to increased medium velocity, there would be a practical upper limit due to the potential deleterious effects of increased shear stresses on hepatocyte viability and function. Additionally, in the clinical setting, the flow rate would be governed by the cardiac output of the patient.
Fig. 3

Non-dimensional cell surface oxygen concentration \( {\left( { \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{cs}}}} = C_{{{\text{cs}}}} /C_{0} } \right)} \) as a function of axial direction \( {\left( { \ifmmode\expandafter\tilde\else\expandafter\sim \fi{x}} \right)} \) for the bioreactor without internal membrane oxygenation at various Pe (with corresponding volume flow rates). Inlet oxygen partial pressure (pO2) was fixed at 159 mmHg (21%). The bioreactor channel length was 7.5 cm, and the channel height was 100 μm (Tilles et al. 2001a)

8 Beneficial effects of internal membrane oxygenation

In the bioreactor with the internal membrane oxygenator (Fig. 1a), the variability of outlet pO2 tension across all flow rates was minimal when compared to that exhibited by the bioreactor without the internal membrane oxygenator, with average values ranging from 148 mmHg at 1.0 mL/min to 160 mmHg at 0.1 mL/min (Fig. 2). Flow rates between 0.1 and 1.0 mL/min were selected because of the low oxygen tensions encountered in the bioreactor without internal membrane oxygenation at these flow rates. At the lowest flow rate, the resulting mean outlet oxygen tension of the bioreactor with internal membrane oxygenation demonstrated a 75-fold increase over that seen in the bioreactor without the internal membrane oxygenator. In this bioreactor configuration, air entering countercurrent to the medium flow supplies oxygen to downstream hepatocytes which otherwise would be oxygen depleted at lower medium flow rates. This leads to an increased outlet pO2 in the bioreactor with internal membrane oxygenation compared to the bioreactor without internal membrane oxygenation. Utilizing the expression for \( \ifmmode\expandafter\tilde\else\expandafter\sim \fi{C}_{{{\text{ave}}}} \) developed for the bioreactor with the internal membrane oxygenator, the average concentration of oxygen in the channel was found to be 153 mmHg and shows as a dashed line in Fig. 2. This simple model was able to predict the oxygen concentration with reasonable accuracy.

It is important to predict the channel height at which hepatocytes become oxygen-limited in the bioreactor with internal membrane oxygenation. Computing the oxygen concentration at the cell surface (Eq. 6) as a function of both Da and β (i.e., K0.5/C*) reveals that as Da increases to 2, the outlet oxygen concentration decreases to the K0.5 value. This corresponds to a channel height of 865 μm, above which the cellular oxygen metabolism is rate limited. Since the channel height in the bioreactor was maintained below this value, there were no oxygen limitations, thereby resulting in increased outlet oxygen concentration in the bioreactor with the internal membrane oxygenator under the flow conditions of 0.1–1.0 mL/min.

9 Hepaotcyte function in bioreactors with and without internal membrane oxygenator

To assess the beneficial effect of internal membrane oxygenation on the function of hepatocytes, rat hepatocytes co-cultured with 3T3-J2 murine fibroblasts were used in the bioreactors with and without internal membrane oxygenation. The rat hepatocyte/3T3-J2 fibroblast co-culture combination was used because this culture system has shown long-term stability and has been extensively characterized and published in the literature (Bhatia et al. 1999). Volumetric flow rates within the two bioreactor configurations were maintained at 0.06 mL/min. The channel height for the bioreactor without the internal membrane oxygenator was 130 μm with a corresponding wall shear stress of 0.14 dyn/cm2 and the channel height of the bioreactor with the internal membrane oxygenator was 115 μm with a corresponding wall shear stress of 0.18 dyn/cm2. Figure 4 shows the albumin and urea synthesis rates on day 3 of perfusion within the two bioreactors. The albumin synthesis rate for the bioreactor without the internal membrane oxygenator was 4.8 μg/day/106 hepatocytes and was 65.9 μg/day/106 for the bioreactor with the internal membrane oxygenator. The urea synthesis rate was 38.7 μg/day/106 in the bioreactor without the internal membrane oxygenator and was 347.2 μg/day/106 in the bioreactor with the internal membrane oxygenator. This corresponded to greater than a 1,300% increase in the albumin synthesis rate and greater than a 500% increase in the urea synthesis rate within the bioreactor with the internal membrane oxygenator compared to the bioreactor without the internal membrane oxygenator, clearly indicating the significance of oxygenation in the bioreactor.
Fig. 4

Albumin and urea synthesis on Day 3 of culture for rat hepatocytes co-cultured with 3T3-J3 fibroblasts in bioreactors with (130 μm channel height) and without (115 μm channel height) the internal membrane oxygenator. Volumetric flow rate was set at 0.06 mL/min. Corresponding shear stresses were 0.14 dyn/cm2 in the bioreactor without the internal membrane oxygenator and 0.18 dyn/cm2 in the bioreactor with the internal membrane oxygenator. Results are expressed as mean of three experiments ± SD (Tilles et al. 2001b)

10 Effect of shear stress on hepatocyte function

Since hepatocyte function was shown to be significantly decreased in the bioreactor without the internal membrane oxygenator, all experiments to determine the effects of various flow conditions were only conducted in the bioreactor with the internal membrane oxygenator. The flow rate (mL/min) − channel height (μm) combinations used to asses the effect of flow conditions on the function of hepatocytes were: 0.06–500; 0.06–180; 0.06–115; 0.06–85; 3.8–180; 3.8–115; and 3.8–85, which corresponded to wall shear stresses of 0.01, 0.07, 0.18, 0.33, 5, 10, and 21 dyn/cm2, respectively. These combinations were sorted into a low shear stress group (0.01–0.33 dyn/cm2) and a high shear stress group (5–21 dyn/cm2). Figure 5a shows the results of daily albumin synthesis rates, presented as percentages of the corresponding daily static controls, at both the low (upper panel) and high (lower panel) shear stresses. For flow conditions that resulted in low shear stresses, the normalized daily albumin synthesis rates were not significantly different throughout the 3 days in the bioreactors across the four shear stresses tested (ANOVA, P = 0.12). There also were no statistically significant differences between the normalized daily albumin synthesis rates for day 3 and day 0 (non-perfused, prior to placement into bioreactor) for any of the four shear stresses. For flow conditions that resulted in high wall shear stresses, the normalized daily albumin synthesis rates decreased throughout the three days in the bioreactor for the three wall shear stresses tested (ANOVA, P < 0.01). Across the group, the day 3 normalized daily albumin synthesis rates were statistically lower than those on day 0 (Tukey’s test, P < 0.05).
Fig. 5

Normalized albumin (a) and urea (b) synthesis at low (upper panels) and high (lower panels) shear stresses for rat hepatocytes co-cultured with 3T3-J2 fibroblasts for 3 days of continuous perfusion in the microchannel flat-plate bioreactor with internal membrane oxygenator. H channel height (μm); τ shear stress (dyn/cm2) (Tilles et al. 2001b)

Urea synthesis rates were noted to decrease throughout the 3 days of perfusion in both the low and high shear stress groups (ANOVA, P < 0.05) (Fig. 5b). In the high shear stress group, there was also a statistically significant decrease in the normalized daily urea synthesis rates on day 3 compared to day 0 (lower panel), whereas in the low shear stress group, there were no statistically significant differences between day 3 and day 0 urea synthesis rates (upper panel) (Tukey’s test, P < 0.05). Comparison of day 3 results between low and high shear stress groups showed that albumin and urea production rates were 2.6 and 1.9 times greater, respectively, than that at high shear stress (t-test, P < 0.01).

These results are of great benefit in the selection of proper bioreactor operating conditions. For example, increasing medium flow rate is beneficial in delivering oxygen to the cell surface. However, our results indicate there is a critical flow range above which synthetic function of the hepatocytes can be greatly diminished by the increased shear stress. Therefore, in a BAL device where recirculation is used, our results suggest that there is an upper limit on the recirculation rate for optimal functioning of the hepatocytes, unless the hepatocytes are protected from the detrimental effects of high shear stresses caused by the flowing medium (Park et al. 2005). Also, in the design of a BAL, minimizing the dead volume of the bioreactor by reducing the channel height, for the same medium flow rate, causes an increase in wall shear stress, thereby placing a lower limit on the dead volume within the bioreactor.

11 Microgrooved substate bioreactor for shear stress protection

Although the studies above demonstrated the beneficial effects of the internal membrane oxygenator on hepatocyte viability and function in the bioreactor, the incorporation of the membrane into the bioreactor adds a level of difficulty in terms of scale-up, especially given that the channel height is only 50–100 μm. When considering a clinical BAL device, it is estimated that 10% of the liver cell mass, or approximately 1010 hepatocytes would be required to support a patient in liver failure (Yarmush et al. 1992a). This translates into a device requiring a surface area of approximately 10 m2. In a microfabricated bioreactor with a planar geometry, obtaining such a large surface area would require a novel design, such as stacking plates of seeded cells (Taguchi et al. 1996; Uchino et al. 1988).

A means of shielding hepatocytes from the detrimental effects of shear stress would thus allow increased medium flow rates for convective delivery of oxygen, thereby eliminating the need for the internal oxygenation membrane between each stacked plate of cells. This would potentially reduce the complexity of the device and allow for easier scale-up. To test this concept, we used photolithographic techniques to fabricate microgrooves (55 μm deep, 100 μm wide, 50 μm groove spacing) onto the underlying glass substrate of a flat-plate bioreactor as a means of reducing the shear stresses associated with the high medium flow rates required for adequate oxygen delivery (Fig. 6) (Park et al. 2005). The microfabricated grooves were designed to reduce the shear stress imposed on the hepatocytes while maintaining direct contact of the hepatocytes with the perfusing medium.
Fig. 6

Schematic representation of the bioreactor with microgrooved substrate to protect the hepatocytes from shear stress effects. The width and the depth of each groove were 100 and 55 μm, respectively (Park et al. 2005)

12 Shear stress modeling in microgrooved substrate bioreactor

Since the complex geometry of the microfabricated grooved substrate was not easily amenable to an analytical solution, we used finite element analysis (FEA) to predict the shear stresses imposed on the hepatocyte/fibroblast cocultures in the microfabricated grooved substrate and the flat-substrate bioreactors. In the microgrooved-substrate bioreactor, for a channel height (H) of 100 μm and a volumetric flow rate of 0.7 mL/min, the velocity vector change at the cell surface was smaller than that near the upper wall, indicating that the shear stress was lower at the cell surface compared to that at the upper wall. Also, the shear stress varied along the cell surface within the groove, i.e., the shear stress at the cell surface decreased closer to the groove side wall. In this microgrooved-substrate bioreactor, at a medium flow rate of 4.0 mL/min, the shear stresses at the cell surface ranged between 0.02 and 0.5 dyn/cm2, with an average of 0.25 dyn/cm2. In the flat-substrate bioreactor, the shear stress at the cell surface was 15 dyn/cm2 at a flow rate of 4.0 mL/min. At this medium flow rate, the shear stresses at the cell surface in the microgrooved-substrate bioreactor could be reduced as much as 30 times compared to that in the flat-substrate bioreactor.

13 Oxygen concentration in microgrooved substrate bioreactor

The outlet oxygen partial pressure for the microgrooved-substrate bioreactor seeded with cocultures of hepatocytes (0.5 × 106) and fibroblasts (1.5 × 106) was measured using a ruthenium-coated optical probe (Park et al. 2005). The outlet oxygen partial pressure decreased from 144 mmHg (i.e., 90% of inlet oxygen concentration) at a medium flow rate of 2.2 mL/min to 106 mmHg (i.e., 67% of inlet oxygen concentration) at a flow rate of 0.45 mL/min. As the flow rate decreased below 0.45 mL/min, the outlet oxygen tension decreased dramatically to 0 mmHg at a flow rate of 0.06 mL/min, indicating that the oxygen in the bioreactor was completely depleted.

14 Comparison of bioreactors with and without the microgrooved substrate

A series of experiments were conducted to evaluate the role of the microfabricated grooved substrates on the function of hepatocytes in the bioreactor. Albumin (Fig. 7) and urea (Fig. 8) synthesis rates were measured for the hepatocyte/fibroblast cocultures in the flat-substrate and microgrooved-substrate bioreactors. The substrates remained in static conditions (days -2, -1, and 0) prior to placing them into the bioreactor on day 0, and continuing under perfusion conditions for 5 days (days 1 through 5). Substrates used as controls remained in static conditions throughout the experiment. The results are presented as percentages of the average synthesis rates of the last 3 days in static conditions (days -2, -1, and 0). The average daily synthesis rates, under static conditions, for hepatocytes seeded on the microgrooved substrates, flat substrates, and controls were 262 μg/10hepatocytes (SD = 84 μg/10hepatocytes) for albumin and 793 μg/10hepatocytes (SD = 250 μg/10hepatocytes) for urea. Medium flow rates tested were 0.06, 0.7, and 4.0 mL/min. For the microgrooved substrates, these medium flow rates corresponded to shear stresses of 0.01, 0.1, and 0.5 dyn/cm2, respectively, and for the flat substrates, these medium flow rates corresponded to shear stresses of 0.23, 2.7, and 15.2 dyn/cm2, respectively. At the medium flow rate of 0.06 mL/min, the albumin (Fig. 7a) and urea (Fig. 8a) synthesis rates progressively decreased throughout the 5 days of perfusion in both, the flat-substrate and microgrooved-substrate bioreactors (ANOVA, P ≦ 0.0011). These findings are consistent with the results of the outlet oxygen concentration measurements described above, indicating that oxygen limitations were occurring at this medium flow rate for both the flat-substrate and microgrooved-substrate bioreactors.
Fig. 7

Albumin synthesis rates of rat hepatocytes (0.5 × 106) cocultured with 3T3-J2 fibroblasts (1.5 × 106) in the microgrooved-substrate and flat-substrate bioreactors at flow rates of 0.06 mL/min (a), 0.7 mL/min (b), and 4.0 mL/min (c). See text for corresponding shear stresses. The average daily albumin synthesis rate, under static conditions, was 262 μg/106 hepatocytes (SD = 84 μg/106 hepatocytes). Albumin synthesis rates are normalized to the average of the last 3 days in static culture. Results are expressed as mean of three experiments ± SD (Park et al. 2005)

Fig. 8

Urea synthesis rates of rat hepatocytes (0.5 × 106) cocultured with 3T3-J2 fibroblasts (1.5 × 106) in the microgrooved-substrate and flat-substrate bioreactors at flow rates of 0.06 mL/min (a), 0.7 mL/min (b), and 4.0 mL/min (c). See text for corresponding shear stresses. The average daily urea synthesis rate, under static conditions, was 793 μg/106 hepatocytes (SD = 250 μg/106 hepatocytes). Urea synthesis rates are normalized to the average of the last 3 days in static culture. Results are expressed as mean of three experiments ± SD (Park et al. 2005)

In the microgrooved-substrate bioreactor, at medium flow rates of 0.7 mL/min (shear stress = 0.1 dyn/cm2) and 4.0 mL/min (shear stress = 0.5 dyn/cm2), the normalized albumin (Fig. 7b, c) and urea (Fig. 8b, c) synthesis rates remained relatively stable throughout the 5 days of perfusion (ANOVA, P ≧ 0.074). There were no significant differences between the day 5 and day 0 albumin or urea synthesis rates for either of these two flow rates (Tukey's test, P ≧ 0.05). For the microgrooved-substrate bioreactor, the normalized albumin and urea synthesis rates were not significantly different compared to the day 5 static controls (t-test, P > 0.09). In the flat-substrate bioreactor, the normalized albumin (Fig. 7b, c) and urea (Fig. 8b, c) synthesis rates both progressively decreased over the 5 days of perfusion at medium flow rates of 0.7 mL/min (shear stress = 2.7 dyn/cm2) and 4.0 mL/min (shear stress = 15.2 dyn/cm2) (ANOVA, P ≦ 0.002). There were also statistically significant decreases in the normalized albumin synthesis rate on day 1 compared to day 0, and the urea synthesis rate on day 3 compared to day 0 (Tukey’s test, P < 0.05). For the flat-substrate bioreactor, the normalized albumin and urea synthesis rate on day 5 was also significantly decreased compared to the day 5 static controls (t-test, P < 0.01). These results indicate that the microgrooved-substrates were successful at reducing the detrimental effects of shear stress on the cultured hepatocytes induced by high medium flow rates while allowing the maintenance of stable hepatocyte function.

15 Implications for scale-up

A benefit of the microgrooved-substrate used in this bioreactor design lies in the scaling-up of this device to a clinically relevant surface area available for hepatocyte culture. The modular design of this BAL device makes scaling-up a matter of connecting individual modules in parallel as repeating units (Chan et al. 2004). The flow characteristics in each individual module remain identical, i.e., channel height, volumetric flow rate per module, and oxygen delivery to the hepatocyte surface are unchanged. Depending on the clinical requirements of an individual patient both before and during the course of treatment with the BAL device (i.e., child vs. adult, etiology and degree of liver failure), modules could either be added or removed thereby allowing an easy method of increasing or decreasing, respectively, the total hepatocyte mass of the device. One configuration of a module is depicted in Fig. 9 as a radial flow design consisting of a stack of glass substrates (50 mm diameter, 1.1 mm thickness), with microgrooves on one surface of each substrate to protect the seeded hepatocytes from flow-induced shear stress. Adjacent substrates are separated using 100 μm spacers, thereby forming a flow channel between each substrate. The stacked glass substrates are contained in a polycarbonate housing. A report describing the in vitro testing of this device will be published as a separate article.
Fig. 9

Schematic representation of a stacked radial flow bioreactor design. The glass substrates (50 mm diameter, 1.1 mm thickness) have micropatterned grooves on one surface to protect the seeded hepatocytes from flow-induced shear stress. Channel height between each substrate is maintained at 100 μm

16 Conclusion

An extracorporeal BAL device is a promising technology for the treatment of liver failure. Our studies using the various configurations of our microchannel hepatocyte bioreactor provide information that is useful in designing efficient BAL devices for clinical applications.

Notes

Acknowledgments

This work was partially supported by grants from the National Institutes of Health (K08 DK66040) and (R01 DK43371). The microfabrication was performed at the BioMicroElectroMechanical Systems (BioMEMS) Resource Center at the Massachusetts General Hospital funded by the National Institutes of Health (P41 EB02503).

References

  1. Arias IM, Boyer JL, Chisari FV, Fausto N, Schacter D, Shafritz DA (2001) The liver: biology and pathobiology. Lippincott Williams & Wilkins, PhiladelphiaGoogle Scholar
  2. Ash SR (2001) Powdered sorbent liver dialysis and pheresis in treatment of hepatic failure. Ther Apher 5:404–416CrossRefPubMedGoogle Scholar
  3. Balis UJ, Behnia K, Bhatia SN, Sullivan SJ, Yarmush ML, Toner M (1999) Oxygen consumption characteristics of porcine hepatocytes. Metab Eng 1:49–62CrossRefPubMedGoogle Scholar
  4. Bhatia SN, Balis UJ, Yarmush ML, Toner M (1999) Effect of cell–cell interactions in preservation of cellular phenotype: cocultivation of hepatocytes and nonparenchymal cells. Faseb J 13:1883–1900PubMedGoogle Scholar
  5. Carslaw HS, Jaeger JC (1959) Conduction of heat in solids. Oxford University Press, LondonGoogle Scholar
  6. Catapano G (1996) Mass transfer limitations to the performance of membrane bioartificial liver support devices. Int J Artif Organs 19:18–35PubMedGoogle Scholar
  7. Chan C, Berthiaume F, Nath BD, Tilles AW, Toner M, Yarmush ML (2004) Hepatic tissue engineering for adjunct and temporary liver support: critical technologies. Liver Transpl 10:1331–1342CrossRefPubMedGoogle Scholar
  8. Demetriou AA, Brown RS Jr, Busuttil RW, Fair J, McGuire BM, Rosenthal P, Am Esch JS II, Lerut J, Nyberg SL, Salizzoni M, Fagan EA, de Hemptinne B, Broelsch CE, Muraca M, Salmeron JM, Rabkin JM, Metselaar HJ, Pratt D, De La Mata M, McChesney LP, Everson GT, Lavin PT, Stevens AC, Pitkin Z, Solomon BA (2004) Prospective, randomized, multicenter, controlled trial of a bioartificial liver in treating acute liver failure. Ann Surg 239:660–667; discussion 667–670Google Scholar
  9. Ellis AJ, Hughes RD, Wendon JA, Dunne J, Langley PG, Kelly JH, Gislason GT, Sussman NL, Williams R (1996) Pilot-controlled trial of the extracorporeal liver assist device in acute liver failure. Hepatology 24:1446–1451PubMedCrossRefGoogle Scholar
  10. Flendrig LM, la Soe JW, Jorning GG, Steenbeek A, Karlsen OT, Bovee WM, Ladiges NC, te Velde AA, Chamuleau RA (1997) In vitro evaluation of a novel bioreactor based on an integral oxygenator and a spirally wound nonwoven polyester matrix for hepatocyte culture as small aggregates. J Hepatol 26:1379–1392CrossRefPubMedGoogle Scholar
  11. Foy BD, Rotem A, Toner M, Tompkins RG, Yarmush ML (1994) A device to measure the oxygen-uptake rate of attached cells—importance in bioartificial organ design. Cell Transplantation 3:515–527PubMedGoogle Scholar
  12. Gerlach JC, Encke J, Hole O, Muller C, Ryan CJ, Neuhaus P (1994) Bioreactor for a larger scale hepatocyte in-vitro perfusion. Transplantation 58:984–988PubMedCrossRefGoogle Scholar
  13. Hay PD, Veitch AR, Gaylor JD (2001) Oxygen transfer in a convection-enhanced hollow fiber bioartificial liver. Artif Organs 25:119–130CrossRefPubMedGoogle Scholar
  14. Hay PD, Veitch AR, Smith MD, Cousins RB, Gaylor JD (2000) Oxygen transfer in a diffusion-limited hollow fiber bioartificial liver. Artif Organs 24:278–288CrossRefPubMedGoogle Scholar
  15. Hu WS, Friend JR, Wu FJ, Sielaff T, Peshwa MV, Lazar A, Nyberg SL, Remmel RP, Cerra FB (1997) Development of a bioartificial liver employing xenogeneic hepatocytes. Cytotechnology 23:29–38CrossRefGoogle Scholar
  16. van de Kerkhove MP, Di Florio E, Scuderi V, Mancini A, Belli A, Bracco A, Dauri M, Tisone G, Di Nicuolo G, Amoroso P, Spadari A, Lombardi G, Hoekstra R, Calise F, Chamuleau RA (2002) Phase I clinical trial with the AMC-bioartificial liver Academic Medical Center. Int J Artif Organs 25:950–959PubMedGoogle Scholar
  17. van de Kerkhove MP, Poyck PP, Deurholt T, Hoekstra R, Chamuleau RA, van Gulik TM (2005) Liver support therapy: an overview of the AMC-bioartificial liver research. Dig Surg 22:254–264. Epub 2005 Sep 2020Google Scholar
  18. Mazariegos GV, Kramer DJ, Lopez RC, Shakil AO, Rosenbloom AJ, DeVera M, Giraldo M, Grogan TA, Zhu Y, Fulmer ML, Amiot BP, Patzer JF (2001) Safety observations in phase I clinical evaluation of the Excorp Medical Bioartificial Liver Support System after the first four patients. ASAIO J 47:471-475CrossRefPubMedGoogle Scholar
  19. O’Grady JG, Gimson AE, O’Brien CJ, Pucknell A, Hughes RD, Williams R (1988) Controlled trials of charcoal hemoperfusion and prognostic factors in fulminant hepatic failure. Gastroenterology 94:1186–1192PubMedGoogle Scholar
  20. Park J, Berthiaume F, Toner M, Yarmush ML, Tilles AW (2005) Microfabricated grooved substrates as platforms for bioartificial liver reactors. Biotechnol Bioeng 90:632–644CrossRefPubMedGoogle Scholar
  21. Pascher A, Sauer IM, Hammer C, Gerlach JC, Neuhaus P (2002) Extracorporeal liver perfusion as hepatic assist in acute liver failure: a review of world experience. Xenotransplantation 9:309–324CrossRefPubMedGoogle Scholar
  22. Patzer IJ, Lopez RC, Zhu Y, Wang ZF, Mazariegos GV, Fung JJ (2002) Bioartificial liver assist devices in support of patients with liver failure. Hepatobiliary Pancreat Dis Int 1:18–25PubMedGoogle Scholar
  23. Peng CA, Palsson BO (1996) Determination of specific oxygen uptake rates in human hematopoietic cultures and implications for bioreactor design. Ann Biomed Eng 24:373–381PubMedGoogle Scholar
  24. Rifai K, Ernst T, Kretschmer U, Bahr MJ, Schneider A, Hafer C, Haller H, Manns MP, Fliser D (2003) Prometheus–a new extracorporeal system for the treatment of liver failure. J Hepatol 39:984–990CrossRefPubMedGoogle Scholar
  25. Rotem A, Toner M, Bhatia S, Foy BD, Tompkins RG, Yarmush ML (1994) Oxygen is a factor determining in-vitro tissue assembly—effects on attachment and spreading of hepatocytes. Biotechnol Bioeng 43:654–660CrossRefGoogle Scholar
  26. Rotem A, Toner M, Tompkins RG, Yarmush ML (1992) Oxygen-uptake rates in cultured rat hepatocytes. Biotechnol Bioeng 40:1286–1291CrossRefGoogle Scholar
  27. Sauer IM, Kardassis D, Zeillinger K, Pascher A, Gruenwald A, Pless G, Irgang M, Kraemer M, Puhl G, Frank J, Muller AR, Steinmuller T, Denner J, Neuhaus P, Gerlach JC (2003) Clinical extracorporeal hybrid liver support-phase I study with primary porcine liver cells. Xenotransplantation 10:460–469CrossRefPubMedGoogle Scholar
  28. Smithson JE, Neuberger JM (1999) Acute liver failure. Overview. Eur J Gastroenterol Hepatol 11:943–947PubMedCrossRefGoogle Scholar
  29. Stange J, Hassanein TI, Mehta R, Mitzner SR, Bartlett RH (2002) The molecular adsorbents recycling system as a liver support system based on albumin dialysis: a summary of preclinical investigations, prospective, randomized, controlled clinical trial, and clinical experience from 19 centers. Artif Organs 26:103–110CrossRefPubMedGoogle Scholar
  30. Stefanovich P, Matthew HWT, Toner M, Tompkins RG, Yarmush ML (1996) Extracorporeal plasma perfusion of cultured hepatocytes: effect of intermittent perfusion on hepatocyte function and morphology. J Surg Res 66:57–63CrossRefPubMedGoogle Scholar
  31. Taguchi K, Matsushita M, Takahashi M, Uchino J (1996) Development of a bioartificial liver with sandwiched-cultured hepatocytes between two collagen gel layers. Artif Organs 20:178–185PubMedCrossRefGoogle Scholar
  32. Tilles AW, Balis UJ, Baskaran H, Yarmush ML, Toner M (2001a) Internal membrane oxygenation removes substrate oxygen limitations in a small-scale flat-plate hepatocyte bioreactor. In: Ohshima N (eds) International symposium on tissue engineering for therapeutic use 5. Elsevier, Amsterdam, pp 59–71Google Scholar
  33. Tilles AW, Baskaran H, Roy P, Yarmush ML, Toner M (2001b) Effects of oxygenation and flow on the viability and function of rat hepatocytes cocultured in a microchannel flat-plate bioreactor. Biotechnol Bioeng 73:379–389CrossRefGoogle Scholar
  34. Uchino J, Tsuburaya T, Kumagai F, Hase T, Hamada T, Komai T, Funatsu A, Hashimura E, Nakamura K, Kon T (1988) A hybrid bioartificial liver composed of multiplated hepatocyte monolayers. ASAIO Trans 34:972–977PubMedGoogle Scholar
  35. Watanabe FD, Mullon CJ, Hewitt WR, Arkadopoulos N, Kahaku E, Eguchi S, Khalili T, Arnaout W, Shackleton CR, Rozga J, Solomon B, Demetriou AA (1997) Clinical experience with a bioartificial liver in the treatment of severe liver failure A phase I clinical trial. Ann Surg 225:484–494CrossRefPubMedGoogle Scholar
  36. Yarmush ML, Dunn JCY, Tompkins RG (1992a) Assessment of artificial liver support technology. Cell Transplant 1:323–341Google Scholar
  37. Yarmush ML, Toner M, Dunn JCY, Rotem A, Hubel A, Tompkins RG (1992b) Hepatic tissue engineering - Development of critical technologies. Ann N Y Acad Sci 665:238–252Google Scholar

Copyright information

© Springer-Verlag 2006

Authors and Affiliations

  • Jaesung Park
    • 1
  • Mehmet Toner
    • 1
  • Martin L. Yarmush
    • 1
  • Arno W. Tilles
    • 1
  1. 1.Center for Engineering in Medicine and Surgical ServicesMassachusetts General Hospital, Shriners Hospitals for Children and Harvard Medical SchoolBostonUSA

Personalised recommendations