The Gradiflow is a novel membrane-based preparative electrophoresis technology that separates macromolecules based on a dual strategy targeting size and charge.1 The selective removal of molecules from biological fluids is achieved by the application of an electrical potential across a set of polyacrylamide membranes with defined pore sizes.
The technology is currently used in the in vitro purification or separation of a range of moieties such as proteins, nucleotides, and complex sugars from biological fluids.2,3 However, there is the potential for this technology to be implemented in various in vivo processes such as hemodialysis, on-line removal of specific biological molecules, and selective protein or pathogen removal from biological fluids. Previous in vitro studies have demonstrated the effective removal of beta-2 microglobulin and phosphate from plasma and blood using the Gradiflow device,4 indicating its potential use as an adjunct to current dialytic therapies. In order for this technology to be applied in an in vivo or ex vivo setting, the biocompatibility and physiologic effects of the Gradiflow process must be investigated. The aim of this initial ex vivo study was to investigate the safety of the small-scaled Gradiflow prototype by assessing its physiologic, hematologic, and biochemical effects in an ovine model. The potential biocompatibility of this system was investigated with the treatment of both whole blood and plasma using an extracorporeal circuit linked to the Gradiflow system.
Materials and Methods
The Gradiflow utilizes three thin sheets of polyacrylamide membranes to contain two fluid streams with a separation membrane separating the two recirculating streams from each other. The restriction membranes are the two outer membranes, and these act to separate the two fluid streams from the electrodes and the buffer reservoir (Figure 1). Polypropylene grid supports between the separation and restriction membranes aid in maintaining the structural integrity of the cartridge and increasing the turbulent flow of the two fluid streams, thereby maximizing exposure time to the membranes and decreasing bio-fouling. The electrodes are used to apply an electrical potential across the membranes, resulting in the migration of solutes toward the electrode carrying the opposite charge. The restriction membranes act as barriers that prevent the movement of solutes into the buffer reservoir, while the pore size of the separation membrane can be varied depending on the molecular size of the target solute. Removal strategies for specific solutes can be accomplished on the basis of charge and molecular size.
The Gradiflow instrument and tubing was primed to drain with sterile water (Baxter Health Care, Sydney, Australia), followed by 1 l of a 1% Etoxa clean solution (Sigma Chemical Co, Castle Hill, Australia) to reduce the risk of endotoxin contamination. After this procedure, the system was rinsed with sterile water, and then saline, and primed with a 10,000 IU/l heparin saline solution. Separation cartridges were constructed using 5 kDa restriction membranes and 16 cm2 separation polyacrylamide membranes of defined pore sizes (Life Therapeutics, Australia). The buffer was a bicarbonate based dialysate solution (Fresenius Medical Care, Sydney, Australia) with 12units/ml heparin added. The flow rate of blood or plasma through the Gradiflow device was set at 20 ml/min, and the recirculating buffer reservoir at 2 l/min. The recirculating dialysate solution was changed at regular intervals during the procedure. The Gradiflow process consisted of the Gradiflow separation cartridge together with a stainless steel heat exchanger in line before the cartridge. The temperature of blood or plasma was monitored at both entry and exit points from the heat exchanger and separation cartridge. Blood temperature was regulated by control of the flow rate of chilled water through the heat exchanger.
Medical grade PVC dialysis tubing (Gambro, Baulkham Hills, Australia) was primed to drain with sterile water, then saline and primed with a 10,000 IU/l heparin saline solution.
Healthy cross-bred wethers were supplied through the Biological Resources Centre, The University of New South Wales. All animals were 50–70 kg in weight, approximately 2 years of age, and were easily identified through ear-tagging. Animals used for each procedure were randomly assigned and kept nil by mouth for 24 hours before the procedure. The study complied with all relevant animal handling requirements and was granted appropriate ethics committee approval.
Measurements and Records
Procedure sheets and anesthetic records were maintained for all procedures. Pulse, respiration rate, oxygen partial pressure (SpO2), and temperature were recorded at 10-minute intervals, and blood pressure was monitored during all procedures. The hematologic and biochemical parameters were analyzed at an external veterinary laboratory (IDEXX Laboratories Pty Ltd, Rydalmere, NSW Australia). A list of parameters assessed is shown in Table 1.
In total, 12 sheep were studied: 6 were exposed to the Gradiflow and 6 were used as controls. The effects of the procedure using both whole blood (Group BG) and plasma (Group PG) were analyzed. The Gradiflow procedure was performed with and without electrical activation (63.3V) in an ABAB format with 1-hour periods of activation (A) followed by no activation (B). The six control sheep were tested without the Gradiflow component: three sheep with whole blood (Group B) and three with the plasma separator (Group P).
The Gradiflow procedure was performed in an ex vivo ovine model (Figure 2). Vascular access was gained via direct cutdown of the animal's carotid artery and external jugular vein. A shunt was inserted and the extracorporeal circuit established at 200 ml/min. A baseline blood sample was collected from the shunt once blood access was achieved. For the plasma procedures, a membrane plasma separator (Fresenius Medical Care) was incorporated into the extracorporeal circuit (Figure 3).
Anticoagulation was achieved by an initial bolus of 5,000 IU heparin followed by additional 5,000 IU boluses every hour thereafter for the duration of the procedure. In some procedures using the Gradiflow component, additional anticoagulation was achieved with the continuous infusion of heparin into the Gradiflow line.
Treatment of Whole Blood with the Gradiflow (Group BG)
After blood flow through the extracorporeal circuit was stabilized at 200 ml/min, the parallel Gradiflow blood circuit was established at 20 ml/min. Blood samples were taken simultaneously from both pre- and postsampling ports after 5, 10, 30, and 60 minutes, after which the electrical potential of the Gradiflow component (63.3 V) was activated, and samples collected after 5, 10, 30, and 60 minutes. The electrical potential was then switched off and blood samples were collected as previously described. After the final blood samples were collected, the sheep were euthanized.
Treatment of Plasma with the Gradiflow (Group PG)
After plasma flow through the extracorporeal circuit was stabilized at 200 ml/min, the parallel Gradiflow plasma circuit was established at 20 ml/min. Thereafter, procedures were identical to those of the whole blood experiments.
Treatment of Whole Blood without the Gradiflow (Group B)
The extracorporeal blood circuit for Group B is illustrated by Figure 2 (minus the Gradiflow component). During the control procedures, blood flow through the extracorporeal circuit was maintained at 200 ml/min and the Gradiflow circuit was not in use. Blood samples were taken simultaneously from both pre- and postsampling ports after 5, 10, 30, 60, 90, 120, 180, and 240 minutes. After the final blood samples were collected, the sheep were euthanized.
Treatment of plasma without the Gradiflow (Group P)
The extracorporeal blood circuit for Group P is illustrated in Figure 3 (minus the Gradiflow component). After plasma flow through the extracorporeal circuit was stabilized at 200 ml/min, a parallel bypass circuit was established at 20 ml/min. The Gradiflow circuit was not in use during the control procedures. Except for the incorporation of the plasma filter, procedures were identical to those of Group B.
A weighted mean, equivalent to the “area under the curve” was determined for each 1-hour period of the procedure for both the pre- and postsample, with error bars representing the standard error of the mean. This was calculated by using a trapezoidal method, and it should be noted that the baseline or preprocedure sample was not included in this calculation.
To adjust for variation in baseline levels between different sheep, analysis of covariance on the means, calculated over the eight observations (four time periods × two sampling sites) was performed.5 Overall tests on the effect of the time period and sampling site and their interaction with the plasma membrane and Gradiflow component were determined using repeated-measures analysis of variance with the Huynh-Feldt correction factor.6 In addition, test of specific contrasts7 were performed to explore any changes in the pattern of difference.
Changes in Physiologic Parameters
The plasma control procedures were well tolerated in all three animals, and pulse, respiration rate, blood pressure, and SpO2 remained stable throughout the procedure.
Similarly, all three animals in the blood control group showed no gross adverse effects from the surgery or extracorporeal circuit. Transient perturbations were observed in the SpO2 and pulse rate of one animal; on postmortem, the left lung was collapsed, most probably due to poor position and ventilation of the sheep during the procedure and exacerbated by fluid overload.
Although clotting was observed in all the Gradiflow blood procedures, the animals did not show any gross adverse effects to the procedure. To minimize the clotting, a continuous heparin infusion was introduced directly into the Gradiflow circuit. In one animal, a further reduction in clotting was achieved by the replacement of the stainless steel heat exchanger with a PVC blood cooling coil.
The Gradiflow plasma procedures were uneventful in two out of three animals. The third animal experienced difficulty with decreased respiration rate, SpO2, and heart rate within 10 minutes of initial exposure to the Gradiflow circuit. Artificial ventilation and adrenalin were administered for the remainder of the procedure. Biopsy indicated accumulation of red blood cells and nonspecific protein material in the septae and interstitial spaces of the lung, with no significant increase in neutrophil numbers. The remaining membranes and the storage solution from this procedure were tested and found to be contaminated with a gram positive, coagulase-negative Staphylococcus, probably Staphylococcus epidermidis.
Changes in Hemoglobin Levels
Hemoglobin levels initially tended to decrease in all treatment and control groups, a trend that could be attributed to the hemodilution effect of the extracorporeal circuit (Figure 4). Hemoglobin levels then remained stable for the remainder of the procedure. There were no statistically significant changes in hemoglobin levels from time period A and B (p = 0.26), nor was there any significant effect of the plasma filter (p = 0.32), the Gradiflow component (p = 0.44), or an interaction of the two (p = 0.18). There were no significant changes in other red cell or platelet indices between the treatment or control groups.
Changes in White Cell Count
An initial drop in white cell count (WCC) was observed in all groups (Figure 4). In the plasma groups, which were exposed to the additional polypropylene plasma filter membrane, this decrease was more dramatic and persisted for the entire duration of the procedure, resulting in a significantly lower mean WCC (p < 0.05). After the initial leukopenia, cell counts increased steadily throughout the procedure for all groups. This trend was significant (p < 0.05) for all time periods, but was not influenced by the presence of the Gradiflow (p > 0.62) or the plasma filter (p > 0.53).
Changes in Neutrophil Levels
The early decrease in WCC could be attributed to a corresponding decrease in neutrophil levels during the first hour of the procedure (Figure 4). The decrease in mean neutrophil levels was more marked in the plasma groups. After the initial neutropenia, cell counts increased steadily for the remainder of the procedure in all treatment and control groups. This trend was significant (p < 0.05) for all time periods, but was not influenced by the presence of the Gradiflow (p > 0.20) or the plasma filter (p > 0.56). There were no significant changes in any other hematologic parameters assessed between control and treatment groups.
Changes Aspartate Serum Transaminase Levels
There was a slow decline in aspartate serum transaminase (AST) over the course of the procedure (p < 0.05); however, this trend was independent of the Gradiflow or the plasma filter (Figure 5).
Changes in Unconjugated Bilirubin Levels
Total bilirubin and conjugated bilirubin levels were measured, and the difference between the two used to determine the level of unconjugated bilirubin present (Figure 5). Total bilirubin levels gradually increased throughout the duration of the procedure in all treatment and control groups (p < 0.05), which was explained by a corresponding increase in conjugated bilirubin levels (p < 0.05). Both trends were unrelated to the presence of the Gradiflow. As a result, unconjugated bilirubin levels remained constant and within normal levels.
Changes in Creatine Kinase Levels and Other Biochemical Parameters
There was a slow increase in creatine kinase levels seen over the course of the procedure (p < 0.05) (Figure 5). Mean creatine kinase levels in the treatment and control groups increased significantly between time periods A and B, but this increase was independent of the Gradiflow or the plasma filter. The Gradiflow device did not exert an influence on any other enzyme or biochemical parameters assessed, with no significant changes observed during the procedure for either treatment or control groups.
Changes in Hematologic and Biochemical Parameters of the Gradiflow Plasma Treatment (PG) Animal with Sepsis
The hematologic and biochemical results for this animal indicated a significant insult to the animal and the possibility of an acute infection. Against the general trend, hemoglobin levels increased during the procedure in this animal. Total white cell counts, after an initial decrease, increased drastically during the first hour of the procedure attributed to a pronounced increase in neutrophil levels perhaps indicating a physical stress or infection. The overall results were unchanged when data from this animal were included or not.
The Gradiflow was originally designed for in vitro use. As such, the Gradiflow device poses several issues pertaining to biocompatibility. Previous in vitro investigations of the Gradiflow device indicated that hemolysis levels of anticoagulated whole human blood with and without an electrical potential were within internationally acceptable standards of 5%.8 Approximately 1.5% hemolysis was observed in 100 ml of recirculating blood after 60 minutes with an electrical potential.9 The Gradiflow process did not alter production of thrombin/antithrombin III complex, activated partial thromboplastin times, or complement C3a during a 60-minute period, with or without an electrical potential.9 Although the in vitro treatment of blood and plasma through the Gradiflow suggested that it was a biocompatible process, the effect of the design, material, and electrical potential on whole blood and plasma was yet to be determined in an in vivo or ex vivo setting.
The current study demonstrated that the Gradiflow procedure was well tolerated in an ex vivo ovine model. Although one animal in the plasma treatment group experienced shock requiring inotropic support, the procedure was uneventful for the other animals. The response in this animal is believed to be caused by Staphylococcus contamination of the membranes and storage solution and not exposure to the Gradiflow system. After this initial insult, the surgical position and fluid overload in the animal may have further exacerbated these effects.
Clotting was experienced in all three of the Gradiflow whole blood procedures approximately 2 hours into the procedure. The extracorporeal bloodlines showed no signs of clotting and rinsed back cleanly at the end of the procedures. Similarly, the extracorporeal circuits in the plasma Gradiflow procedures showed no signs of clotting. A reduction in clotting in the Gradiflow blood circuit would be achieved by redesigning and changing the materials used in the heat exchanger and Gradiflow cartridge. The majority of the clotting occurred in the heat exchanger, and although stainless steel was used, the fluid path was disturbed in sections. Although difficult to visualize, the majority of the clotting appeared to initiate at the inlet and outlet of the device, where blood flow turbulence and restriction was present. The relevant scaling up and redesign of the heat exchanger should significantly reduce the risk of clotting. Current pediatric blood oxygenator devices that incorporate stainless steel heat exchangers could be used to aid in the redesign. Similarly, a redesign of the Gradiflow cartridge should reduce clotting, as the current lattice design induces flow turbulence at the entry and exit points of the fluid pathway. Clotting was seen at the exit point of the cartridge, probably caused by a reduction in flow rate experienced. Useful reference points for the redesign of the cartridge include flat plate dialyzers that have incorporated spacers and inlet and outlet designs that produce smooth consistent flows. The scaling up of the Gradiflow component and use of relevant tubing sizes, flow rates, and membrane surface area will further address this issue.
Clotting may have been exacerbated by the removal of some heparin during the procedure. The effect of using other anticoagulants on clotting factors and other components in the clotting cascade should be analyzed further to determine suitable anticoagulant regimens for future procedures.
Overall, there were no gross adverse reactions to the ex vivo Gradiflow procedure, and no other significant changes were observed in the physiologic, hematologic, or biochemical parameters assessed. A significant neutropenia was observed in the control and treatment plasma groups, a phenomenon that is observed in current hemodialysis and plasma exchange therapies using synthetic membranes.10 Similarly, the presence of the Gradiflow in the extracorporeal circuit had no significant effect on enzyme or electrolyte levels, with no statistically significant differences between the control and treatment groups for any biochemical or hematologic value.
The results of this study indicate the potential of the Gradiflow technology for the in vivo removal of a range of moieties as in hemodialysis or indeed the in vitro removal of substances in a blood bank setting. Further design and material changes, including relevant scaling up of the Gradiflow device should be performed before further testing in animals suffering from renal failure and subsequent ex vivo trials in humans.11,12
The authors thank Dr. Weiyun Yu, Ms. Kate Noble, Ms. Lynn Ferris, and Mr. John Klemes of the Graduate School of Biomedical Engineering, The University of New South Wales, Australia, for their assistance and hard work.
1. Horvath SZ, Corthals GL, Wrigley CW, Margolis J: Multifunctional apparatus for electrokinetic processing of proteins. Electrophoresis
15: 968–971, 1994.
2. Corthals GL, Margolis J, Williams KL, Gooley AA: The role of pH and membrane porosity in preparative electrophoresis. Electrophoresis
17: 771–775, 1996.
3. Rylatt DB, Napoli M, Ogle D, Gilbert A, Lim S, Nair CH: Electrophoretic transfer of proteins across polyacrylamide membranes. J Chromatogr A
865: 145–153, 1999.
4. Lee JI, Nair H, Harris DCH: In vitro
Characterization of phosphate and beta–2 microglobulin removal by the Gradiflow [abstract]. J Am Soc Nephrol
13: 418A, 2002.
5. Frison L, Pocock SJ: Repeated measures in clinical trials: Analysis using mean summary statistics and its implications for design. Stat Med
11: 1685–1704, 1992.
6. Frison L, Pocock SJ: Repeated measures in clinical trials: Analysis using mean summary statistics and its implications for design. Stat Med
11: 1685–1704, 1992.
7. StataCorp. Stata Statistical Software: Release 6.0.
College Station, TX: Stata Corporation, 1999.
8. Harris RJ: ANOVA: An Analysis of Variance Primer.
Itasca, IL, F.E. Peacock,1994.
9. ISO 10993-1: Biological Evaluation of Medical Devices—Part 1: Evaluation and Testing.
Geneva, Switzerland, International Organization for Standardization, 2003.
10. Tabor B, Geissler B, Odell R, Schmidt B, Blumenstein M, Schindhelm K: Dialysis neutropenia: The role of the cytoskeleton. Kidney Int
53: 783–789, 1998.
11. Bosch T, Schmidt B, Spencer PC, et al: Ex vivo biocompatibility evaluation of a new modified cellulose membrane. Artif Organs
11: 144–148, 1987.
12. Bosch T, Lennertz A, Schmidt B, et al: DALI apheresis in hyperlipidemic patients: Biocompatibility, efficacy and selectivity of adsorption of lipoproteins from whole blood. Artif Organs
24: 81–90, 2000.