fax: 49 30 450 559909
Liver Failure and Chronic Liver Disease
In vitro comparison of the molecular adsorbent recirculation system (MARS) and single-pass albumin dialysis (SPAD)
Article first published online: 26 APR 2004
Copyright © 2004 American Association for the Study of Liver Diseases
Volume 39, Issue 5, pages 1408–1414, May 2004
How to Cite
Sauer, I. M., Goetz, M., Steffen, I., Walter, G., Kehr, D. C., Schwartlander, R., Hwang, Y. J., Pascher, A., Gerlach, J. C. and Neuhaus, P. (2004), In vitro comparison of the molecular adsorbent recirculation system (MARS) and single-pass albumin dialysis (SPAD). Hepatology, 39: 1408–1414. doi: 10.1002/hep.20195
- Issue published online: 26 APR 2004
- Article first published online: 26 APR 2004
- Manuscript Accepted: 2 FEB 2004
- Manuscript Received: 25 JUL 2003
The detoxification capacities of single-pass albumin dialysis (SPAD), the molecular adsorbents recirculation system, (MARS) and continuous veno-venous hemodiafiltration (CVVHDF) were compared in vitro. In each experiment 4,100 mL of toxin-loaded human plasma was processed for 6.5 hours. MARS treatment (n = 6) was undertaken in combination with CVVHDF. For SPAD (n = 6) and CVVHDF (n = 6) a high-flux hollow fiber hemodiafilter (identical to the MARS filter) was used. Levels of ammonia, urea, creatinine, bilirubin, and bile acids were determined. Concentrations before and after application of detoxification procedures were expressed as differences and were compared using the Kruskal-Wallis test. Post hoc comparisons for pairs of groups were adjusted according to Bonferroni-Holm. Time, group, and interaction effects were tested using the nonparametric ANOVA model for repeated measurements. SPAD and CVVHDF induced a significantly greater reduction of ammonia levels than MARS. No significant differences were found among SPAD, MARS, and CVVHDF with respect to other water-soluble substances. SPAD induced a significantly greater reduction in bilirubin levels than MARS. Reductions in bile acid levels were similar for SPAD and MARS. When operating MARS in continuous veno-venous hemodialysis mode, as recommended by the manufacturer, no significant differences in the removal of bilirubin, bile acids, urea, and creatinine were found. However, MARS in continuous veno-venous hemodialysis mode was significantly less efficient in removing ammonia than MARS in CVVHDF mode. In conclusion, the detoxification capacity of SPAD is similar to or even greater than that of MARS. (HEPATOLOGY 2004;39:1408–1414.)
Liver failure, whether fulminant or acute-on-chronic, is a life-threatening syndrome. When conservatively managed, acute hepatic failure is associated with a poor prognosis; mortality rates are 60% to 80%.1 Urgent liver transplantation (LTx) is the standard therapy for acute liver failure (ALF) in individuals who, according to clinical and laboratory criteria, have less than a 20% chance of survival.2, 3 However, ALF is a potentially reversible condition. Early liver transplantation eliminates the chance of spontaneous liver regeneration. Following transplantation, patients require lifelong immunosuppression; they have increased risks of infectious complications and malignancy. Some patients are considered to be unsuitable for transplantation due to social or medical reasons, such as active drug abuse or a morbid psychological profile; the presence of multiorgan failure might contraindicate LTx.
There is a growing disparity between the numbers of suitable donor organs and patients waiting for transplantation. Accordingly, efforts have been made to optimize the allocation of donor organs, to find alternatives to cadaveric liver transplantation, and to develop extracorporeal methods to support or replace the functions of the failing liver. Moreover, in hepatic failure, patients with a capacity for liver regeneration may be provided with liver support until liver regeneration takes place; such patients would not require transplantation. In cases of chronic liver failure, temporary extracorporeal liver support therapy may be beneficial during episodes of an acute exacerbation of liver failure (acute-on-chronic liver failure).
The accumulation of water-soluble toxins, such as ammonia and mercaptans, and albumin-bound water insoluble compounds, such as bilirubin, bile acids, short chain fatty acids, and aromatic amino acids, has been implicated in the pathogenesis of hepatic encephalopathy4–6 and dysfunction of organs other than the brain in patients with liver failure.7, 8
Conventional continuous veno-venous hemodiafiltration (CVVHDF) has been shown to be effective in the removal of water-soluble toxins. To clear the blood of albumin-bound, hydrophobic substances, additional “adsorber” or “acceptor” substances are necessary to enhance exchange. Albumin is one of the potential acceptor substances. Stange, Mitzner, and coworkers introduced a detoxification system that involved albumin dialysis—the molecular adsorbents recirculation system (MARS). An albumin solution is circulated in a closed circuit, which is separated from the patient's blood by a high flux hemodialysis filter. The albumin acts as an acceptor for potential toxins and is partially regenerated by including an anion exchanger and a charcoal adsorber in a closed circuit; the albumin is dialyzed using a standard dialysis solution in continuous veno-venous hemodialysis (CVVHD) or CVVHDF mode. To operate the system, additional hardware (MARS monitor), including a recirculation pump, is necessary.9, 10
Single-pass albumin dialysis (SPAD) is a simple method of albumin dialysis, which uses standard renal replacement therapy machines without an additional perfusion pump system. The patient's blood flows through a circuit containing a high-flux hollow fiber hemodiafilter, identical to that used in the MARS system. The other side of this membrane is cleansed by an albumin solution flowing in the counter-direction; this solution is discarded after passing the filter. CVVHDF may be undertaken in the first circuit using the same high-flux hollow fibers. Waiving of sorbent columns for regeneration of adsorber-albumin, initial clinical application of SPAD was undertaken using higher concentrations of albumin in the dialysis fluid than in MARS.11
Material and Methods
In Vitro Test System
The clearance of bilirubin, bile acids, ammonia, creatinine, urea, and uric acid was evaluated using an in vitro test system that consisted of a conventional device for renal replacement therapy (B. Braun Trio, B. Braun Melsungen AG, Melsungen, Germany) in combination with a MARS monitor and the MARS treatment kit, which consisted of a MARS Flux Dialyzer, the diaMARS AC250 charcoal unit, and the diaMARS IE250 ion exchanger (Teraklin AG, Rostock, Germany) for MARS (in CVVHDF mode, n = 6; in CVVHD mode, n = 4) and a Fresenius HdF 100S polysulfone high-flux hemodiafilter (Fresenius Medical Care AG, Bad Homburg, Germany), identical to the MARS Flux Dialyzer, for SPAD (n = 6) and CVVHDF (n = 6).
The Trio continuous renal replacement therapy (CRRT) device consists of a blood module and three fluid balance monitors operated by interfacing roller pumps. A disposable measuring chamber with an ultrasound sensor controls fluid balance; the pumps are recalibrated after each measurement cycle. Both high-flux dialysis filters, the Fresenius HdF 100S and the Teraklin MARS Flux Dialyser, are identical; both are based on Fresenius polysulfone hollow fibers.
To compare the different methods of dialysis with MARS, including an adsorber and an ion-exchanger having a specific capacity, a smaller scale of test system, incorporating smaller volumes of plasma and dialysis fluid, is not possible. Therefore, all systems were operated for 6.5 hours; the amount of plasma, loaded with potential toxins, was similar to that in an adult human.
A roller pump enabled circulation of the prepared plasma (4,100 mL) at a flow rate of 130 mL/min. The temperature was kept at 37.0 °C using a water bath. The primary circuit included the plasma reservoir, tubing, and a roller pump, operating at 130 mL/min; the circuit was identical for all assessments. To form a reservoir, a glass bottle, having a volume of 5000 mL, was connected to two tubes, one for the inflow and one for the outflow of plasma. The distance between inflow and outflow was 20 cm, to avoid circulation shunts. The outflow of the bottle was connected to the dialysis cartridge; the outflow of the dialyzer was connected to the inflow of the bottle (Fig. 1).
Human plasma served as a substitute for the patients' blood. Having obtained informed consent and the approval of the local ethics committee, the plasma was obtained from patients undergoing plasmapheresis treatment. The material was heparinized (5 IU/mL), stored at −20 °C, and depleted of fibrin clots before use. The plasma was spiked with the following substances to achieve concentrations equivalent to 3 × normal: 82.9 mg chenodeoxycholic acid, 51.6 mg cholic acid, 50.9 mg dehydrocholic acid, 15.7 mg deoxycholic acid, 7.5 mg lithocholic acid, 701.6 mg unconjugated bilirubin, 4,804.8 mg urea, 537.9 mg uric acid (Sigma-Aldrich Chemie, Munich, Germany), 32.1 mg ammonium chloride (Merck KGaA, Darmstadt, Germany), and 158.4 mg creatinine (Merck KGaA, Darmstadt, Germany) were dissolved in 200 mL 0.1 M sodium hydroxide solution. The pH value was adjusted to 7.4 by titration with acetic acid (20%). This solution (200 mL) was mixed with 3,500 mL pooled human plasma and 400 mL of 20% human albumin solution.
The MARS was set up and rinsed according to the manufacturer's instructions. Six hundred milliliters of human albumin solution (20%) were added to the loop circuit. Standard dialysis solution (SH-BIC 35 basis solution combined with SH-EL 02; B. Braun Shiwa GmbH KG, Glandorf, Germany) was used in the dialysis circuit. The albumin solution was circulated at 130 mL/min; the dialysis circuit was operated at 1000 mL/h without fluid removal. MARS was operated in combination with CVVHDF (n = 6) and, in a substudy, in combination with CVVHD (n = 4).
To obtain the albumin dialysate solution for use in SPAD (n = 6), 1000 mL human albumin (20%) were mixed with 3500 mL dialysis solution (SH-BIC 35 basis solution combined with SH-EL 02, B. Braun Shiwa GmbH KG, Glandorf, Germany); the result was an albumin solution of 4.4%. The same standard dialysis solution was used in the dialysis circuit. The albumin solution passed the filter at a rate of 700 mL/h; the dialysis circuit was operated at 1000 mL/h without fluid removal.
SH-BIC 35 basis dialysis solution combined with SH-EL 02 (B. Braun Shiwa GmbH KG, Glandorf, Germany) was used in the CVVHDF circuit (n = 6).
Comparison of Different Modes of Operation of MARS
To compare MARS with SPAD and CVVHDF, we operated MARS in the CVVHDF mode. However, the manufacturers of MARS suggest operating their system in the CVVHD mode (Fig. 1a). Accordingly, we undertook MARS in the CVVHD mode (n = 4). The detoxification capacity of the two operation modes of MARS were compared. The dialysis circuit was operated at 1000 mL/h without fluid removal.
Analysis of Biochemical Variables
Plasma samples (5 mL) were obtained before the start of detoxification, and, during detoxification, every 10 minutes during the first hour, and every 30 minutes thereafter.
The levels of bilirubin, bile acid, ammonia, creatinine, urea, and albumin were determined. All samples were analyzed twice using standard techniques. Urea was measured using a kinetic ultraviolet assay based on Talke and Schubert's method (Roche Diagnostics GmbH, Mannheim, Germany). Quantitative determination of creatinine was undertaken using a kinetic calorimetric assay (Jaffé method, Roche Diagnostics GmbH, Mannheim, Germany). Ammonia was quantified using a standard enzyme-linked glutamate dehydrogenase-based method (Roche Diagnostics GmbH, Mannheim, Germany). Bile acids were determined using an enzymatic color test (Merckotest Bile Acids, E. Merck, Darmstadt, Germany). Bilirubin was quantified using a colorimetric assay (Roche Diagnostics GmbH, Mannheim, Germany).
Statistical analyses were undertaken using SAS for Windows V8 (SAS Institute Inc., Cary, NC) and SPSS 11 for Windows (SPSS Inc., Chicago, IL). Evaluation and graphical presentation of the data were undertaken using Microsoft Excel (version X for Apple Macintosh OS X, Microsoft, Redmond, WA) and Deltagraph 5 (version 5.0.1 for Apple Macintosh OS X, SPSS Inc./Red Rock Software, Salt Lake City, UT).
Due to small sample sizes, nonparametric tests were applied. Concentrations before and after procedures were calculated as differences and compared using the Kruskal-Wallis test. Post hoc comparisons for testing pairs of groups were adjusted according to Bonferroni-Holm. The nonparametric one- and two-way ANOVA model for repeated measurements, according to Brunner et al., was applied to test the effects of time, group (CVVHDF, MARS, and SPAD), and interactions between time and group. A significant time factor indicates that the concentration changes significantly over time. Differences among groups over the whole treatment period are characterized as group effects. A significant time/group interaction indicates that there is a significant difference in the time course of the concentration between the groups CVVHDF, MARS, and SPAD. A P value less than or equal to .05 was considered statistically significant.
The initial concentrations of the toxins added to the plasma (determined prior to the start of detoxification, t = 0 min) were similar in all experiments. Thus, there was a satisfactory basis for comparing the systems (bilirubin, mean: 18.3 ± 1.0 mg/dL; bile acids: 78.1 ± 5.8 μmol/L; urea: 170.1 ± 39.6 mg/dL; ammonia: 223.4 ± 31.6 μmol/L; and creatinine: 5.35 ± 1.6 mg/dL).
Bilirubin, as a marker substance for albumin-bound toxins, was reduced by SPAD from 18.5 ± 1.1 mg/dL to 12.9 ± 2.0 mg/dL. MARS in CVVHDF mode reduced bilirubin from 18.4 ± 0.8 mg/dL to 15.6 ± 1.0 mg/dL (Fig. 2); using the CVVHDF mode, no removal occurred (pre: 18.5 ± 1.5 mg/dL; post: 18.2 ± 1.2 mg/dL). Similar effects were found for bile acids (Fig. 3). Using SPAD, the initial bile acid concentration was 77.2 ± 4.8 μmol/L; posttreatment it was 22.1 ± 6.3 μmol/L. For MARS, the initial bile acid concentration was 75.9 ± 5.0 μmol/L; posttreatment it was 21.5 ± 3.1 μmol/L. For CVVHDF, the pretreatment value was 80.7 ± 7.4 μmol/L; posttreatment it was 65.5 ± 6.2 μmol/L.
The decrease in concentrations over the complete time course of experiments was significant for the albumin dialysis-based techniques (MARS: bilirubin, P = .001; bile acids, P < .001; and SPAD: bilirubin, P = .001; bile acids, P < .01; time effect by nonparametric ANOVA for repeated measurements). When CVVHDF was applied, no significant decrease in bilirubin levels occurred (P = .056). However, levels of bile acids decreased significantly (P < .001). Comparisons of the time course of the different groups revealed a significant difference between MARS in the CVVHDF mode and SPAD; SPAD was associated with a significantly greater reduction in bilirubin than was induced by MARS in the CVVHDF mode (P < .01, relating to delta [0-390 min], using the Kruskal-Wallis test and the Bonferroni-Holm adjustment). CVVHDF alone, however, was associated with no significant reduction in bilirubin levels (Fig. 2). No significant differences between SPAD and MARS in CVVHDF mode were found with respect to reduced bile acid levels. However, both systems were associated with significantly greater reductions of bile acids than were induced by CVVHDF (P < .05, relating to delta [0-390 min] using the Kruskal-Wallis test and Bonferroni-Holm adjustment).
The concentration of urea was reduced by SPAD from 180.7 ± 27.8 mg/dL to 13.5 ± 1.8 mg/dL. Using MARS in the CVVHDF mode, concentrations of urea were 118.3 ± 25.9 mg/dL before and 16.3 ± 1.63 mg/dL after applying the procedure. Similar results were obtained for ammonia (SPAD: pre, 238.4 ± 30.6 μmol/L; post, 46.7 ± 12.8 μmol/L; MARS in CVVHDF mode: pre, 198 ± 18.2 μmol/L; post, 78.6 ± 11.5 μmol/L), and creatinine (SPAD: pre, 5.6 ± 1.5 mg/dL; post, 0.3 ± 0.05 mg/dL; MARS: pre, 3.5 ± 0.6 mg/dL; post, 0.1 ± 0.05 mg/dL). Maximum removal of water-soluble substances (urea, ammonia) occurred after CVVHDF (Fig. 4). Ammonia levels decreased from 217.4 ± 35.3 μmol/L to 24.7 ± 6.3 μmol/L. Urea levels were 188.5 ± 19.5 mg/dL before treatment and 13.5 ± 1.6 mg/dL after 6.5 hours of CVVHDF; corresponding figures for creatinine were 6.1 ± 1.3 mg/dL before, and 0.3 ± 0.05 mg/dL after. The decrease in concentration over the complete time course of experiments was significant for each of the techniques (P < .001, time effect by nonparametric ANOVA for repeated measurements). SPAD (P < .05) and CVVHDF (P < .05) were associated with significantly greater reductions of ammonia levels than those induced by MARS in CVVHDF mode (relative to delta [0-390 min] using the Kruskal-Wallis test and Bonferroni-Holm adjustment). Testing for an interaction effect (time*group) yielded similar results (P < .001). There was no significant difference between CVVHDF and SPAD with respect to reduction of ammonia levels. All of the techniques reduced creatinine levels to <0.5 mg/dL (0.1 to 0.3 mg/dL) after 390 minutes. Urea was reduced to levels <16.5 mg/dL by all three techniques; there were no significant differences among SPAD, MARS in CVVHDF mode, and CVVHDF.
Comparison of Different Modes of Operation of MARS
When operating MARS in the recommended CVVHD mode (Fig. 1a), the concentration of urea was reduced from 204.8 ± 9.7 mg/dL to 50.0 ± 4.0 mg/dL, creatinine was reduced from 6.7 ± 1.2 mg/dL to 0.2 ± 0.1 mg/dL, ammonia was reduced from 247.0 ± 16.5 to 102.6 ± 8.3 μmol/L, bilirubin was reduced from 17.7 ± 0.4 mg/dL to 14.4 ± 1.1 mg/dL, and bile acids were reduced from 78.7 ± 6.5 μmol/L to 17.7 ± 1.0 μmol/L. When MARS in CVVHD and CVVHDF modes were compared, no significant differences were found with respect to the removal of bilirubin, bile acids, urea, and creatinine. However, MARS in CVVHDF mode was significantly more efficient in removing ammonia than MARS in CVVHD mode (P < .05, with respect to delta [0-390 min] using the Kruskal-Wallis test and Bonferroni-Holm adjustment). Comparison of SPAD and MARS in CVVHD mode indicated that they had similar capacities for detoxifying bilirubin.
The concept that a critical issue in liver failure is the accumulation of toxins not cleared by the failing liver led to the development of artificial filtration and adsorption devices. A hypothesis was proposed that the removal of lipophilic, albumin-bound substances, such as bilirubin, bile acids, metabolites of aromatic amino acids, medium-chain fatty acids and cytokines, may have a beneficial effect on the clinical course of patients in liver failure.
SPAD is a simple detoxification technique based on continuous veno-venous hemodiafiltration and solute bound dialysis. Human albumin, as a binding agent, is added to the dialysis solution, to enable solute transfer from the patient's blood to the dialysis solution. MARS is a commercially available product based on the principle of albumin dialysis. In MARS, however, albumin is regenerated by passage through a charcoal and anion exchange column. The aim of this study was to compare the detoxification capacity of SPAD with that of MARS. Our study showed that the efficiencies of SPAD and the more sophisticated, more complex and, hence, more expensive MARS technique were similar. This finding may lead to a wider clinical application of the SPAD detoxification technique; the combination of detoxification and dialysis techniques with bioreactor technology for cell-based extracorporeal liver support would be facilitated.
Conceptual and procedural aspects of SPAD and MARS merit comparison and discussion. Both methods are based on a CRRT device. To undertake SPAD, no additional dialysis hardware is necessary. When MARS therapy is undertaken, an operation device (MARS monitor) is required and has to be combined with the CRRT unit. Accordingly, disposables for the MARS monitor (MARS treatment kit) and for the dialysis unit are necessary, whereas only standard equipment for undertaking continuous veno-venous hemodiafiltration and a high-flux hemofilter are needed for SPAD. SPAD is cheaper and much simpler than MARS (Fig. 1); SPAD may be set up in an intensive care unit more quickly than MARS, which requires more disposables and involves the time-consuming process of priming and charging the MARS circuit.
The application time of MARS is limited by its design. As the circulating albumin solution is regenerated via the charcoal and ion exchanger columns, depending on the toxin load, the system eventually becomes saturated. The SPAD system, however, is not limited in a similar way; its capacity can be increased by simply adding human albumin solution. SPAD was more efficient in removing some compounds than MARS. There were marked differences in the removal of ammonia and bilirubin. The higher efficiency of SPAD may be attributed to the fact that conventional dialysis takes place directly in the primary circuit, where plasma is dialyzed. In MARS, dialysis takes place in the secondary circuit; the recirculating albumin solution is dialyzed (Fig. 1). However, if MARS, as suggested by the manufacturer, is applied in the CVVHD mode, the difference in the capacity of SPAD and MARS to remove bilirubin is not significant.
Peszynski et al. compared SPAD and MARS in vitro using procedures similar to those presented here.12 However, they applied SPAD without hemodiafiltration. In addition, only 1000 mL of plasma were used in their experiments. Using their approach, a potential saturation of the regenerating charcoal and anion exchange column is unlikely. The technique has been tailored to the needs of adult patients with a plasma volume of 3000–4000 mL. Overall efficiency may be crucial; it can be addressed adequately only by using an amount of toxin-loaded plasma similar to that of an adult patient with liver failure. The experiments undertaken by Peszynski et al. do not permit statistical analysis. It appears that each experiment was undertaken only once. The concentrations of the substances measured in the plasma are not mentioned, and it is not possible to judge whether experimental conditions were similar for each technique. Comparisons of overall total clearances, without documentation of the toxin load in plasma or the duration of each experiment, make it impossible to evaluate the efficiency of the detoxification techniques.
Our experiments demonstrated that SPAD and MARS have similar efficiencies in clearing substances that may be relevant in liver failure. As albumin dialysis is a costly procedure, the financial aspects of these procedures are important: For a seven-hour treatment with MARS, the costs include approximately $370 for 600 mL human albumin solution (20%), $2,120 for a MARS treatment kit and $150 for disposables used by the dialysis machine. The total cost of a treatment with this therapy is approximately $2,641. The costs of undertaking SPAD, using the approach applied in our experiments, include $610 for 1000 mL of human albumin solution (20%), approximately $50 for a high-flux dialyzer, and $150 for tubing. The total cost of a SPAD treatment is approximately $810, or about 30% of the cost of an equally efficient MARS treatment. The expenditure for the MARS monitor, which is necessary to undertake MARS, is not included in this comparison.
Our study has compared the efficiency of two procedures of albumin dialysis under controlled conditions. Whether or not the results of our in vitro studies can be extrapolated to patients with liver failure, and whether or not these procedures are of beneficial for patients with liver failure are uncertain; further investigations are required to resolve these issues. So far, mainly case reports or descriptive studies of the application of albumin dialysis in subpopulations of patients with liver failure have been published. Heemann et al. recently demonstrated, in a prospective controlled study, that treatment with the MARS is most beneficial for patients with alcohol-induced liver disease and superimposed alcoholic hepatitis.9 However, the findings in that study cannot necessarily be extrapolated to other patient groups.14 Data on the effects of MARS in acute liver failure and in primary graft dysfunction are encouraging, but limited.15 Albumin dialysis, using MARS, has been shown to be effective in the treatment of hepatic encephalopathy16 and intractable pruritus.17 Furthermore, there are indications that its application improves circulatory and renal function in patients with cirrhosis and ascites.18, 19
The detoxification capacities of SPAD and MARS are similar. The SPAD procedure is less complex and less expensive; it is currently being evaluated clinically at our center. In our in vitro experiments we used a 4.4% solution of human albumin. This is the concentration we are using in our clinical applications of albumin dialysis; it is based on case reports using this approach for extracorporeal detoxification11, 20 and on previous in vitro studies by Awad et al.21 Albumin is the most expensive component of albumin dialysis; dose-determining studies should be undertaken to find the ideal concentration of albumin and the ideal flow-rate of the albumin solution. Such in vitro studies are currently in progress.
The authors thank the chief hepato-technician of our team, Ulrich Stumborg, and the nurses and physicians at the Charité Nephrology Department for procuring the human plasma. We also thank Brigitte Wegner, Institute for Medical Biometrics, Charité, for her advice concerning the statistics, and Anne Carney and Wolfgang Mudra for help in preparing the manuscript and the artwork.
- 2Selection criteria and results of orthotopicliver transplantation in the UK. In: WilliamsR, HughesRD, eds. Acute Liver Failure: Improved Understanding and Better Therapy. London: Miter Press, 1991; 77–80., , .
- 8Intensive care management of patients with acute liver failure with emphasis on systemic hemodynamic instability and cerebral edema: a critical appraisal of pathophysiology. Can J Gastroenterol 2000; 14(Suppl): 105–111., , .
- 10The 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 2002; 26: 103–110, , , , .
- 13Characteristics of an albumin dialysate hemodiafiltration system for the clearance of unconjugated bilirubin. ASAIO J 1997; 43: 745–749., , , , , , .
- 21Characteristics of an albumin dialysate hemodiafiltration system for the clearance of unconjugated bilirubin. ASAIO J 1997; 43: 745-–749., , , , , , et al.