- Email: [email protected]
Albumin Dialysis for Liver Failure: A Systematic Review
Albumin Dialysis for Liver Failure: A Systematic Review Evangelos Tsipotis, Asim Shuja, and Bertrand L. Jaber Albumin dialysis is the best-studied extracorporeal nonbiologic liver support system as a bridge or destination therapy for patients with liver failure awaiting liver transplantation or recovery of liver function. We performed a systematic review to examine the efficacy and safety of 3 albumin dialysis systems (molecular adsorbent recirculating system [MARS], fractionated plasma separation, adsorption and hemodialysis [Prometheus system], and single-pass albumin dialysis) in randomized trials for supportive treatment of liver failure. PubMed, Ovid, EMBASE, Cochrane’s Library, and ClinicalTrials.gov were searched. Two authors independently screened citations and extracted data on patient characteristics, quality of reports, efficacy, and safety end points. Ten trials (7 of MARS and 3 of Prometheus) were identified (620 patients). By meta-analysis, albumin dialysis achieved a net decrease in serum total bilirubin level relative to standard medical therapy of 8.0 mg/dL (95% confidence interval [CI], 210.6 to 25.4) but not in serum ammonia or bile acids. Albumin dialysis achieved an improvement in hepatic encephalopathy relative to standard medical therapy with a risk ratio of 1.55 (95% CI, 1.16-2.08) but had no effect survival with a risk ratio of 0.95 (95% CI, 0.84-1.07). Because of inconsistency in the reporting of adverse events, the safety analysis was limited but did not demonstrate major safety concerns. Use of albumin dialysis as supportive treatment for liver failure is successful at removing albumin-bound molecules, such as bilirubin and at improving hepatic encephalopathy. Additional experience is required to guide its optimal use and address safety concerns. Q 2015 by the National Kidney Foundation, Inc. All rights reserved. Key Words: Liver failure, Albumin dialysis, MARS, Prometheus, First-pass albumin dialysis
INTRODUCTION Over the past several decades, a number of extracorporeal liver support therapies have been conceived to either bridge patients with liver failure to transplantation or serve as potential destination therapies for patients with acute or fulminant hepatic failure while awaiting tissue regeneration and recovery of liver function. These extracorporeal therapies are divided in 2 major categories, biologic and nonbiologic support systems. Biologic systems incorporate liver cells or tissues that simulate the excretory, synthetic, and metabolic functions of the liver,1,2 whereas nonbiologic systems use artificial membranes and adsorbents to detoxify the blood in patients with liver failure. Albumin dialysis is the beststudied and most promising nonbiologic system and is based on the removal of unwanted albumin-bound and water-soluble substances, such as bilirubin, bile acids, ammonia, nitrotyrosine, and fatty acids. There are 3 available albumin dialysis systems (Fig 1)3: the molecular adsorbent reticulating system or MARS (Teraklin, Rostock, Germany), single-pass albumin dialysis (SPAD), and the Prometheus system (Fresenius, Bad Homburg, Germany). MARS is a commercially available albumin dialysis system that removes protein-bound and water-soluble From Division of Nephrology, Department of Medicine, St. Elizabeth’s Medical Center, Boston, MA; and Department of Medicine, Tufts University School of Medicine, Boston, MA. Financial Disclosure: The authors declare that they have no relevant financial interests. Address correspondence to Bertrand L. Jaber, MD, MS, Division of Nephrology, Department of Medicine, St. Elizabeth’s Medical Center, 736 Cambridge Street, Boston, MA 02135. E-mail: [email protected] Ó 2015 by the National Kidney Foundation, Inc. All rights reserved. 1548-5595/$36.00 http://dx.doi.org/10.1053/j.ackd.2015.05.004
382
toxins. It comprises 2 separate dialysis circuits (Fig 1A); the first circuit contains exogenous human serum albumin, which is in contact with the patient’s blood through a semipermeable membrane (molecular weight cutoff of 50-60 kDa). Water-soluble and protein-bound toxins in the blood pass through this membrane. The toxin-enriched albumin solution is then passed through another dialyzer to remove water-soluble toxins using a counter-current bicarbonate-based dialysate. Albuminbound toxins are removed by 2 adsorbent cartridges that contain activated charcoal and an anion exchanger. The regenerated albumin solution is then ready for new uptake of toxins from the blood. Although MARS has been used in Europe for the treatment of acute-onchronic liver failure, severe alcoholic hepatitis, severe pruritus because of cholestasis, and intoxication with protein-bound substances, in the United States, it has been cleared for use in the treatment of drug overdose, poisoning, and hepatic encephalopathy but is not indicated for the treatment of chronic liver disease or as a bridge to liver transplantation.4,5 Relative contraindications to the use of MARS include severe sepsis, coagulopathy, and bleeding. A previous metaanalysis of small, randomized, and quasi-randomized trials published in 2012 demonstrated a clinical benefit of MARS compared with standard medical therapy in terms of lowering serum total bilirubin levels (net change 27.0 mg/dL; 95% confidence interval [CI], 210.4 to 23.7; P , .001) and improving hepatic encephalopathy (odds ratio, 3.0; 95% CI, 1.9-5.0; P , .001); however, there was no observed mortality benefit (odds ratio, 0.91; 95% CI, 0.64-1.31; P ¼ .62).6 Two randomized controlled trials of MARS have since been completed.7,8 Single-pass albumin dialysis is an extracorporeal liver support system that uses a conventional dialysis circuit in which an exogenous albumin solution is passed once
Advances in Chronic Kidney Disease, Vol 22, No 5 (September), 2015: pp 382-390
Albumin Dialysis
383
of albumin dialysate used, blood flow rate, anticoagulation use, duration of treatment, and number of treatment sessions), and duration of follow-up. The efficacy end points of interest were changes in circulating levels of total bilirubin, ammonia, and bile acids; improvement in hepatic encephalopathy; and all-cause mortality. To assess these end points, we extracted the mean values of the solutes at baseline and at the end of the study period and the net change from baseline, the number of patients experiencing an improvement in the West-Haven grade of hepatic encephalopathy (as defined in individual trials), and the number of patients who died. Safety end points of interest included gastrointestinal bleeding, catheter-related bleeding, thrombocytopenia, infections and sepsis, pulmonary adverse events, and serious adverse events. Disagreements were resolved through consensus. The corresponding authors of 3 trials were contacted for data clarification. When indicated, values reported as medians with ranges were converted to estimates of means with standard deviations.13 The Cochrane’s Collaboration tool was used to assess risk of bias, which covers 6 domains of bias (selection, performance, detection, attrition, reporting, and other bias). Two authors (A.S. CLINICAL SUMMARY and E.T.) independently rated each domain’s risk Use of albumin dialysis, including MARS and the of bias as low, high, or unPrometheus system, as supportive treatment for liver clear, with the implication failure is successful at removing albumin-bound of a third author (B.J.) in molecules, such as bilirubin and at improving hepatic case of disagreement. encephalopathy.
through the dialysate compartment and then discarded.9 This liver support system has shown some promise in a case series of children with acute liver failure.10 The Prometheus system is a variant of albumin dialysis and combines fractionated plasma separation, adsorption, and hemodialysis.11 It uses a 250-kDa semipermeable membrane generating an albumin-containing plasma-like solution. The patient’s albumin-containing plasma solution is then adsorbed on 2 albumin-detoxifying columns before reuniting with blood cells. Diffusive hemodialysis is then performed to remove water-soluble solutes. This system is advantageous as it relies on endogenous rather than exogenous albumin. The HELIOS trial represents the largest multicenter clinical experience using the Prometheus system.12 To provide an update on MARS and review the scarce literature on the other 2 albumin dialysis modalities, SPAD and the Prometheus system, we conducted a meta-analysis of all randomized controlled trials published to date that compared the efficacy and safety of any of the 3 albumin dialysis systems, including head-tohead comparisons. METHODS
Literature Search and Study Selection The literature search and study selection were performed independently by 2 authors (A.S. and E.T.). With the assistance of a medical librarian, the following electronic data However, survival benefit has not been established yet. Data Synthesis and bases were searched for Additional experience is required to guide its optimal use relevant citations: PubMed, Analysis and address safety concerns. Ovid, EMBASE, and CoRandom-effects model metachrane’s Library (inception analyses were used to calto October 2014). The culate the pooled mean terms and filters’ syntax for each database are provided difference in net change in levels of the solutes of interest in the Supplementary Material. We also searched and the risk ratio (RR) for improvement in the grade of heClinicalTrials.gov using similar terms and the bibliograpatic encephalopathy and all-cause mortality. All pooled phies of retrieved articles. The search strategy was limited estimates are displayed with a 95% CI. Existence of heteroto human studies with no restrictions on language, sample geneity among study effect sizes was examined using the I2 size, duration of study, or year of publication. index and the c2 test P value. An I2 index of 50% or more was used to indicate medium-to-high heterogeneity.14 SubData Extraction and Assessment of Bias group analyses were performed according to certain study We included parallel-arm randomized controlled trials characteristics, including the albumin dialysis modality that enrolled patients with acute liver failure, acute-on(MARS and Prometheus vs SPAD) and the number of dialchronic liver failure, or chronic decompensated liver disysis sessions (,5 vs $5). All analyses were performed usease, comparing the safety and efficacy of albumin dialysis ing the Cochrane’s Collaboration Review Manager systems (MARS, Prometheus, or SPAD) to standard med(version 5.3, Copenhagen, Denmark: The Nordic Cochrane ical therapy or to each other. Centre, The Cochrane Collaboration). The following data were extracted in duplicate: country of origin, year of publication, population setting (acute, RESULTS acute-on-chronic, and chronic liver failure), study design, Characteristics of the Trials and Risk of Bias study period, intervention groups, primary outcome, sumAssessment mary characteristics of study participants (sex, mean age, A total of 3688 potentially relevant citations were identified serum albumin, total bilirubin, creatinine, prothrombin and screened (Fig 2). Seventy articles were retrieved for time/international normalized ratio, and Model for Endevaluation, of which 10 randomized controlled Stage Liver Disease score), summary characteristics of trials fulfilled eligibility criteria comprising 7 trials of the albumin dialysis system used (including percentage
384
Tsipotis et al
A
B
C
Figure 1. Extracorporeal nonbiologic liver support albumin dialysis systems. (A) Molecular adsorbent reticulating system (MARS); (B) fractionated plasma separation and adsorption (Prometheus); and (C) single-pass albumin dialysis. Reproduced with permission from Struecker et al.3
MARS7,8,15-19 and 3 trials of Prometheus.12,20,21 There were no trials of SPAD. Two trials compared MARS to Prometheus, head-to-head, and to standard medical therapy.20,21 All 10 trials were conducted in Europe except for one that included a participating site in the United States.19 Characteristics of the individual trials are presented in Table 1. The trials spanned 13 years and varied in sample size (13-179 patients). Seven trials included patients with acute-on-chronic liver failure,8,12,15,16,18-20 with 2 trials of patients with acute liver failure7,17 and 1 trial of patients with decompensated chronic liver disease.21 The cause of liver disease ranged among studies with alcohol ingestion8,12,15,16,18-21 and viral hepatitis7,8,12,15,16,18,19,21 being the most frequent causes. Other reported causes included cardiogenic shock,17 autoimmune hepatitis, drug-induced7,8,16,19 and toxin-induced7 liver disease, primary8,15,19,21 and secondary15 biliary cirrhosis, hemochromatosis,21 Wilson disease,8 porphyria,21 graft vs host disease,8 nonalcoholic steatohepatitis,8 alpha-1 antitrypsin deficiency,8 and Budd-Chiari syndrome.15 The mean age of the study participants ranged from 40 to 62 years, and overall, 60% of participants were men. In terms of measures of liver dysfunction, the baseline serum total bilirubin level ranged from 11.1 to 25.7 mg/dL and the mean international normalized ratio from 0.8 to 8.3. The mean Model for End-Stage Liver Disease score ranged from 17 to 38, predicting a 3-month mortality risk of 6% to 53%.22 Regarding the characteristics of the albumin dialysis system (Table 2), the percentage of albumin dialysate used ranged from 10% to 20%, with a range of 1 to 10 treatment sessions, lasting 6 to 8 hours each. Anticoagulation was used in 6 trials; otherwise, it was either not reported or used variably.7,12,16,19 In almost all the trials, as part of standard medical therapy, patients in both groups received hemodiafiltration,7,15 hemofiltration,16 continuous venovenous hemofiltration,7,17,18 hemodialysis,7,21 or an unspecified form of kidney replacement therapy8,12
when indicated. Although the duration of the individual trials ranged from 1 to 360 days, when reported, mortality was ascertained at 30 to 360 days. Figure 3 displays the risk-of-bias assessment summary. Although all 10 trials exhibited a lower risk of selection bias (ie, random sequence generation and allocation concealment), there was a higher risk of performance bias (ie, lack of blinding of participants and personnel), detection bias (ie, blinding of outcome assessment, mainly the improvement in hepatic encephalopathy), and attrition bias (ie, incomplete outcome data). Effect of Albumin Dialysis on Removal of LiverRelated Solutes There were 6 trials with 7 experimental arms that reported on changes in total bilirubin levels with 329 analyzable patients. As shown in Figure 4A, by meta-analysis, compared with standard medical therapy, albumin dialysis resulted in a significant net decrease in the mean total bilirubin level by 8.0 mg/dL (95% CI 210.6 to 25.4; P ,.001). There was a nonsignificant net decrease in the mean level of ammonia (mean difference 22.5 mM; 95% CI, 222.4 to 17.3; P ¼.80; 3 trials) and bile acids (mean difference 24.9 mM; 95% CI, 237.5 to 27.6; P ¼ .77; 3 trials). The delivery of less than 5 or 5 or more albumin dialysis sessions did not further enhance the net reduction in mean total bilirubin levels (mean difference 27.9; 95% CI, 214.7 to 21.2; P ¼ .02, and mean difference 27.8; 95% CI, 210.6 to 25.1; P , .001, respectively). In a sensitivity analysis, however, a minimum of 3 sessions of albumin dialysis were required to significantly improve total bilirubin levels compared with fewer than 3 sessions (mean difference 29.3 mg/dL; 95% CI, 213.0 to 25.6; P , .001 vs 23.8 mg/dL; 95% CI, 29.8 to 2.1; P ¼ .21). In subgroup analyses, compared with standard medical therapy, MARS and Prometheus resulted in a significant net decrease in the mean total bilirubin level by 7.6
Albumin Dialysis
385
Figure 2. Study selection flow diagram.
mg/dL (95% CI, 213.0 to 22.3; P ¼ .005; 6 trials) and 8.1 mg/dL (95% CI, 211.2 to 24.9, P ,.001; 3 trials), respectively. In the 2 trials comparing Prometheus to MARS,20,21 compared with MARS, the Prometheus system did not further reduce the mean total bilirubin level (mean difference 4.6; 95% CI, 24.2 to 13.4; P ¼ .30). Effect of Albumin Dialysis on Hepatic Encephalopathy Only 3 trials of MARS reported on the rate of improvement in the West-Haven grade of hepatic encephalopathy.8,18,19 By meta-analysis, compared with standard medical therapy, MARS resulted in a significant 55% increased likelihood of improvement in hepatic encephalopathy (RR, 1.55; 95% CI, 1.16-2.08; P ¼ .003; Fig 4B). There were too few trials to conduct meaningful subgroup analyses according to the number of treatment sessions. Effect of Albumin Dialysis on Mortality All 10 trials reported on mortality rates with 634 analyzable patients. By meta-analysis, compared with standard
medical therapy, albumin dialysis had no significant effect on mortality (RR, 0.95; 95% CI, 0.84-1.07; P ¼ .42; Fig 4C). However, in a sensitivity analysis that excluded 1 trial with the longest follow-up period of 1 year,7 albumin dialysis resulted in a nonsignificant trend toward a 13% risk reduction in mortality (RR, 0.87; 95% CI, 0.75-1.02; P ¼ .08). In a subgroup analysis based on the number of sessions delivered, in studies that delivered 5 or more albumin dialysis sessions, there was a nonsignificant trend toward a 13% risk reduction in mortality (RR, 0.87; 95% CI, 0.741.02; P ¼ .08) compared with trials that provided less than 5 sessions where these was no observed effect on mortality (RR, 1.08; 95% CI, 0.89-1.31; P ¼ .42). On subgroup analyses, there was no demonstrable mortality benefit of either MARS or Prometheus compared with standard medical therapy (RR, 0.97; 95% CI, 0.85-1.11; P ¼ .69; and RR, 0.87; 95% CI, 0.66-1.14; P ¼ .31, respectively). Furthermore, in the 2 head-to-head trials of the 2 albumin dialysis systems, Prometheus did not result in a significant mortality risk reduction compared with MARS (RR, 0.67; 95% CI,
386
Table 1. Characteristics of the Randomized Controlled Trials Included in the Meta-analysis
Study
Year
Country
No. Population Setting Patients
2000 Germany
Heemann et al16
2002 Germany
El Banayosy et al17 Sen et al18
2004 Germany 2004 United Kingdom
Acute-on-chronic liver failure
18
Laleman et al20
2006 Belgium
Acute-on-chronic liver failure
18
Acute-on-chronic liver failure
70
Decompensated liver disease 2012 Germany, Spain, Acute-on-chronic United Kingdom, liver failure Italy, Belgium, Croatia, and Austria 2013 19 European Acute-on-chronic sites liver failure 2013 France Acute liver failure
24
Hassanein et al19 2007 United States, Germany, and Denmark Dethloff et al21 2008 Denmark Kribben et al12
~ ares et al8 Ban Saliba et al7
Acute-on-chronic liver failure Acute-on-chronic liver failure Acute liver failure
13 24 27
145
179 102
MARS 1 SMT 6 HDF vs SMT 6 HDF MARS 1 SMT vs SMT 6 HF MARS 1 SMT vs SMT 6 CVVH MARS 1 SMT 6 CVVH vs SMT 6 CVVH MARS 1 SMT vs Prometheus 1 SMT vs SMT MARS 1 SMT vs SMT
47
38
—
—
25.7
4.0
30
30
52
54
—
—
—
—
30
30
62
70
—
1.8
11.1
2.0
14
NR
45
72
18
2.0
20.3
1.4
7
90
51
67
26
—
27.0
1.3
3
NR
52
56
31
2.0
14.2
1.7
5
180
MARS vs Prometheus vs HD Prometheus 1 SMT 6 RRT vs SMT 6 RRT
55
63
—
2.1
12.5
1.1
1
180
50
63
28
2.0
25.5
2.3
90
90
MARS 1 SMT vs SMT 6 RRT MARS 1 SMT 6 RRT vs SMT 6 RRT
51
69
25
0.8
25.1
—
28
90
40
43
38
8.3
11.2
—
360
360
Abbreviations: CVVH, continuous venovenous hemofiltration; HF, hemofiltration; HD, hemodialysis; INR, international normalized ratio; MARS, molecular adsorbent recirculating system; MELD, Model for End-Stage Liver Disease; RRT, renal replacement therapy; SMT, standard medical therapy.
Tsipotis et al
Mitzner et al15
Study Groups
Mean Serum Mean Mean Mean Total Serum Study Follow-up Age Men MELD Mean Bilirubin Creatinine Duration Period (y) (%) Score INR (mg/dL) (mg/dL) (d) (d)
387
Albumin Dialysis
Table 2. Characteristics of Albumin Dialysis in the Trials Included in the Meta-analysis Study Mitzner et al15 Heemann et al16 Banayosy et al17 Sen et al18 Laleman et al20 Hassanein et al19 Dethloff et al21 Kribben et al12 ~ ares et al8 Ban Saliba et al7
Albumin Dialysis System MARS MARS MARS MARS MARS and Prometheus MARS MARS and Prometheus Prometheus MARS MARS
No. of Sessions
Duration Per Session (h)
5 10 7 4 3 3 1 8 7 1
6-8 6 8 8 6 6 6 6 7 8
Anticoagulation Use Yes Minimal Yes Yes Yes — Yes Yes (variable) Yes Yes (variable)
Dialysate albumin Strength (%) 10 15 20 20 20 17 20 — — 20
Abbreviation: MARS, molecular adsorbent recirculating system.
0.15-2.98; P ¼ .30). Although there was no evidence of significant heterogeneity among study effect sizes in all the analyses, the results of the subgroup and sensitivity analyses need to be interpreted with caution. Safety Analysis There was a significant inconsistency among studies in the reporting of adverse events. Among the trials that reported on the following safety events, albumin dialysis was not associated with an increased risk of respiratory adverse events (RR, 1.44; 95% CI, 0.83-2.50; P ¼ .19),7,8,12,16,19-21 gastrointestinal bleeding (RR, 0.90; 95% CI, 0.60-1.36; P ¼ .62),7,8,12,19-21 and severe infections or sepsis (RR, 1.14; 95% CI, 0.85-1.55; P ¼ .38).7,8,16,19-21 In 3 trials,16,19,21 there were 8 catheter-related episodes of bleeding or hematomas among the 59 patients treated with MARS, one of which was fatal.19 Hematologic-related adverse events associated using albumin dialysis included thrombocytopenia, coagulopathy, and thrombotic episodes.7,8,12,16,17,19,21 Although 1 trial reported 2 (5%) of 39 patients experiencing treatment-related thrombocytopenia compared with none (0%) among the 31 patients receiving standard medial therapy,19 in the 3 trials where absolute changes in platelet count could be analyzed,12,17,21 compared with standard medical therapy, albumin dialysis did not result in a significant net decrease in the mean platelet count (mean difference 15,840 per microliter; 95% CI, 224,000 to 55,690; P ¼ .44). Among the 3 trials reporting on serious adverse events,7,19,21 only in 1 study were these events more frequently observed in patients assigned to albumin dialysis compared with standard medical therapy (51% vs 26%).19 Almost half of these events were ascribed to the therapy. In the other 2 studies, there were either no reported serious adverse events21 or they were not significantly different between the albumin dialysis and standard medical therapy group (32% vs 29%).7 DISCUSSION Albumin dialysis is the best-studied extracorporeal nonbiologic liver support system for the management of patients with acute, acute-on-chronic, and chronic decompensated liver failure as a potential bridge or destination therapy. It is based on the removal of unwanted albumin-bound and water-soluble substances. In the pre-
sent systematic review, we summarize qualitatively the scarce literature on the 3 available albumin dialysis systems, mainly MARS, SPAD, and Prometheus, and conduct a quantitative analysis of all 10 randomized controlled trials published to date that compare the efficacy and safety of any of these albumin dialysis systems. Most trials included patients with acute-on-chronic liver failure, and the majority of study participants were men. In 8 of the 10 trials, patients in the standard medical treatment arm received some form of kidney replacement therapy. The albumin dialysis experimental treatment arms were restricted to MARS and Prometheus, and no randomized clinical trials of SPAD were identified. To examine the detoxifying capacity of the albumin dialysis systems, we chose surrogate efficacy end points, mainly the improvement in circulating levels of albumin-bound molecules (bilirubin) and water-soluble molecules (ammonia), grades of hepatic encephalopathy, and a hard clinical endpoint, all-cause mortality. In terms of removal of liver-related solutes, compared with standard medical therapy, albumin dialysis resulted in a significant net reduction in serum total bilirubin levels, but in the few trials with analyzable data, albumin dialysis did not significantly reduce levels of ammonia and bile acids. Albumin dialysis also resulted in a significant improvement in grades of hepatic encephalopathy (only observed in trials of MARS with analyzable data). Of note, however, all the patients included in the analysis of hepatic encephalopathy had acute-on-chronic liver failure, and these results should be interpreted in this clinical context. Although there was no significant impact of albumin dialysis on all-cause mortality, among the 6 trials that delivered 5 or more albumin dialysis sessions, there was a nonsignificant trend toward a 13% risk reduction in mortality, suggestive of a dose-dependent effect. The inconsistency in the reporting of adverse events did not allow for a robust safety analysis. Except for 1 trial,19 serious adverse events among patients receiving albumin dialysis were not more frequently reported. Previous studies have found that albumin-bound bilirubin is an important fraction of total bilirubin (up to 90%) in patients with advanced liver disease and is not detected in healthy subjects.23 When hepatic excretion of conjugated bilirubin is impaired, albumin-bound bilirubin appears in serum and becomes a larger component of total
388
Tsipotis et al
Random sequence generation (selection bias) Allocation concealment (selection bias) Blinding of participants and personnel (performance bias) Blinding of outcome assessment (detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias 0% Low risk of bias
Unclear risk of bias
25%
50%
75%
100%
High risk of bias
Figure 3. Risk of bias assessment graph: each 100% stacked column represents a low, unclear, and high risk of bias domain across all the trials included in the meta-analysis.
bilirubin, thus its potential removal by albumin dialysis in patients with advanced liver disease. We found no significant difference in the net reduction of total bilirubin levels between MARS and the Prometheus system, and increasing the number of treatment sessions to 5 or greater did not further affect bilirubin levels, suggesting a ceiling effect. In a sensitivity analysis, a minimum of 3 sessions of albumin dialysis were required to significantly improve bilirubin levels. This could be attributed to the fact that the maximum clearance ability of albumin dialysis in removing bilirubin is reached after 3 sessions. In the 3 trials examining the removal of serum bile acids, which are 80% protein bound in patients with liver cirrhosis,24 albumin dialysis resulted in a nonsignificant net decrease in circulating levels of this biomarker. We also found a nonsignificant net decrease in levels of ammonia in 3 trials with analyzable data. However, this might have been, in part, because of the fact that both the experimental and control groups received some form of kidney replacement therapy, which would have removed this water-soluble small molecular weight solute. Although ammonia is the best-characterized neurotoxin in the pathophysiology of hepatic encephalopathy,25 an elevated serum ammonia level is not required to make the diagnosis and is not specific for this disorder. Serum nitrotyrosine is a byproduct of superoxide and nitric oxide generation26 and has been shown to be a promising biomarker of hepatic encephalopathy.27 Although the majority of circulating nitrotyrosine is free and not protein bound,28 none of the trials included in our systematic review examined the potential removal of this biomarker by albumin dialysis. The lack of improvement in patient survival is not surprising as all the trials were clearly underpowered to examine this hard clinical end point, especially in such clinical settings where the underlying acute severity of illness and coexisting comorbidities are typically dominant predictors of clinical outcomes. This is further confounded by the inadequate safety analysis because of the inconsistent reporting of adverse events across the trials. Although we found no published trials of SPAD, a query of the US-based clinical trials registration Web site (http:// clinicaltrials.gov/) identified 2 randomized controlled trials
(NCT01079104 and NCT01079091) of Hepa Wash system (Hepa Wash GmbH, Munich, Germany), a form of SPAD. These 2 trials have an estimated planned enrollment of 244 participants, with the primary outcome of 30-day mortality. The current status of these trials is unknown. Our results confirm those of a previous meta-analysis6 comparing MARS to standard medical therapy with regard to bilirubin removal, improvement of hepatic encephalopathy, and lack of survival benefit. Although both studies used similar methodologies, the previous meta-analysis included nonrandomized trials.6 In contrast to our study, another recently published metaanalysis29 found a mortality benefit of artificial liver support systems over standard medical therapy in patients with acute-on-chronic liver failure. The study did not focus on albumin dialysis and included studies that compared albumin dialysis, hemoperfusion, hemofiltration, and the BioLogic-DT sorbent suspension dialyzer to standard medical therapy. Strengths of our synthesis include the large number of trials, the strict methodology used for the selection of the studies and the accurate implementation of the methods of the protocol. Our synthesis focused exclusively on comparing albumin dialysis systems to either standard medical therapy or to each other. However, there are also important limitations that should be emphasized. Because of the nature of the intervention, double blinding was not possible in the trials, which might have been a source of bias in the assessment of hepatic encephalopathy. Although there was no demonstrable heterogeneity in the effect estimates of individual trials, methodologic heterogeneity was salient, including etiology and severity of liver disease, timing of outcome assessment, and variable follow-up period. In addition, the limited number of trials examining Prometheus, the variable use of kidney replacement therapies in the individual trials, and the variability in the albumin dialysis protocols all created further obstacles in drawing safe conclusions. Finally, the variable timing of the reporting of outcomes among studies likely affected the generalization of the results. In conclusion, albumin dialysis, including MARS and the Prometheus system, results in an increase in the removal of
389
Albumin Dialysis
A Study or Subgroup 4.5.1 MARS vs SMT Dethloff 2008 Mitzner 2000 Laleman 2006 Hassanein 2007 Sen 2004 Banayosy 2004 Subtotal (95% CI)
Albumin Dialysis SMT SD Total SD Total Mean Mean -3.6 19.15 -9.5 13.8 -5.2 7.01 0.3 15.86 -13.58 8.28 -2.98 6.29
Mean Difference IV, Random, 95% CI
8 -4.40 [-27.01, 18.21] 5 -7.70 [-29.51, 14.11] -5.60 [-18.89, 7.69] 6 -0.30 [-8.81, 8.21] 31 9 -16.65 [-24.67, -8.63] 13 -7.36 [-12.90, -1.82] 72 -7.64 [-12.97, -2.31]
0.8 26.42 -1.8 22.36 0.4 15.06 0.6 19.6 3.07 9.06 4.38 8.19
8 8 6 39 9 14 84
Mean Difference IV, Random, 95% CI
Heterogeneity: Tau² = 14.92; Chi² = 7.88, df = 5 (P = 0.16); I² = 37% Test for overall effect: Z = 2.81 (P = 0.005) 4.5.2 Prometheus vs SMT Dethloff 2008 Laleman 2006 Kribben 2012 Subtotal (95% CI)
-2.7 16.09 -11.7 10.58 11 -8
8 6 68 82
0.8 26.42 0.4 15.06 9 0
8 6 77 91
-3.50 [-24.94, 17.94] -12.10 [-26.83, 2.63] -8.00 [-11.26, -4.74] -8.09 [-11.24, -4.94]
Heterogeneity: Tau² = 0.00; Chi² = 0.46, df = 2 (P = 0.79); I² = 0% Test for overall effect: Z = 5.04 (P < 0.00001) Total (95% CI)
-7.97 [-10.58, -5.35]
154
175
Heterogeneity: Tau² = 0.83; Chi² = 8.36, df = 8 (P = 0.40); I² = 4% Test for overall effect: Z = 5.97 (P < 0.00001) Test for subgroup differences: Chi² = 0.02, df = 1 (P = 0.89), I² = 0%
B Study or Subgroup 4.8.1 MARS vs SMT
Albumin Dialysis SMT Events Total Events Total 15 24 9
Bañares 2013 Hassanein 2007 Sen 2004 Subtotal (95% CI)
24 39 9 72
13 12 6
-100 -50 0 50 Favors Albumin Dialysis Favors SMT
Risk Ratio IV, Random, 95% CI
34 31 9 74
100
Risk Ratio IV, Random, 95% CI
1.63 [0.96, 2.77] 1.59 [0.96, 2.64] 1.46 [0.91, 2.35] 1.55 [1.16, 2.08]
31 48 Total events Heterogeneity: Tau² = 0.00; Chi² = 0.11, df = 2 (P = 0.95); I² = 0% Test for overall effect: Z = 2.98 (P = 0.003) 4.8.2 Prometheus vs SMT Subtotal (95% CI)
0
72
Total (95% CI)
0
Not estimable
74
1.55 [1.16, 2.08]
0
0 Total events Heterogeneity: Not applicable Test for overall effect: Not applicable
31 48 Total events Heterogeneity: Tau² = 0.00; Chi² = 0.11, df = 2 (P = 0.95); I² = 0% Test for overall effect: Z = 2.98 (P = 0.003) Test for subgroup differences: Not applicable
C Study or Subgroup 4.11.1 MARS vs SMT
SMT Albumin Dialysis Total Events Total Events
Banayosy 2004 Bañares 2013 Dethloff 2008 Hassanein 2007 Heemann 2002 Laleman 2006 Mitzner 2000 Saliba 2013 Sen 2004 Subtotal (95% CI)
7 49 2 19 1 0 6 44 5
14 90 8 39 12 6 8 53 9 239
9 51 3 17 6 0 5 37 5
0.01
Risk Ratio IV, Random, 95% CI
13 89 8 31 12 6 5 49 9 222
0.1 1 10 100 Favors SMT Favors Albumin Dialysis
Risk Ratio IV, Random, 95% CI
0.72 [0.38, 1.37] 0.95 [0.73, 1.23] 0.67 [0.15, 2.98] 0.89 [0.56, 1.40] 0.17 [0.02, 1.18] Not estimable 0.79 [0.49, 1.26] 1.10 [0.90, 1.34] 1.00 [0.44, 2.29] 0.97 [0.85, 1.11]
Total events 133 133 Heterogeneity: Tau² = 0.00; Chi² = 6.58, df = 7 (P = 0.47); I² = 0% Test for overall effect: Z = 0.40 (P = 0.69) 4.11.2 Prometheus vs SMT Dethloff 2008 Kribben 2012 Laleman 2006 Subtotal (95% CI)
3 41 0
8 77 6 91
3 42 0
8 68 6 82
1.00 [0.28, 3.54] 0.86 [0.65, 1.14] Not estimable 0.87 [0.66, 1.14]
Total events 44 45 Heterogeneity: Tau² = 0.00; Chi² = 0.05, df = 1 (P = 0.82); I² = 0% Test for overall effect: Z = 1.01 (P = 0.31) Total (95% CI)
330
304
0.95 [0.84, 1.07]
177 Total events 178 Heterogeneity: Tau² = 0.00; Chi² = 7.16, df = 9 (P = 0.62); I² = 0% Test for overall effect: Z = 0.81 (P = 0.42) Test for subgroup differences: Chi² = 0.53, df = 1 (P = 0.47), I² = 0%
0.1 0.2 0.5 1 2 Favors Albumin Dialysis Favors SMT
5
10
Figure 4. Forest plot displaying of albumin dialysis vs standard medical therapy on (A) net change in serum mean total bilirubin level, (B) risk ratio for improvement in the West-Haven grade of hepatic encephalopathy, and (C) risk ratio for all-cause mortality. Abbreviations: CI, confidence interval; SMT, standard medical therapy; MARS, molecular adsorbent recirculating system.
390
Tsipotis et al
total bilirubin, an albumin-bound liver solute, and in an improvement in hepatic encephalopathy, relative to standard medical therapy. These changes might confer a longterm survival benefit. These findings call for the design of multicenter, large, randomized controlled trials of efficacy and safety of albumin dialysis aimed at exploring their potential role as bridge or destination extracorporeal supportive therapies for patients with liver failure awaiting liver transplantation or recovery of hepatic function. ACKNOWLEDGMENTS The authors acknowledge the technical assistance of Catherine Guarcello, MS (Stohlman Library, St. Elizabeth’s Medical Center, Boston, MA) and Mark Sherer.
SUPPLEMENTARY DATA Supplementary data related to this article can be found at http://dx.doi.org/10.1053/j.ackd.2015.05.004. REFERENCES 1. Hassanein TI, Schade RR, Hepburn IS. Acute-on-chronic liver failure: extracorporeal liver assist devices. Curr Opin Crit Care. 2011;17(2):195-203. 2. Stange J. Extracorporeal liver support. Organogenesis. 2011;7(1):64-73. 3. Struecker B, Raschzok N, Sauer IM. Liver support strategies: cutting-edge technologies. Nat Rev Gastroenterol Hepatol. 2014;11(3): 166-176. 4. Tolwani A, While K. Extracorporeal liver support and the kidney. In: Critical Care Nephrology. 2nd ed. Philadelphia, PA: Saunders Elsevier; 2009:1061-1067. 5. Gambro. Molecular adsorbent recirculating system., http://www.gambro. com/en/global/Therapies/Molecular-Adsorbent-Recirculating-Systems1/ ?tab¼documentTab#tabList. Accessed November 28, 2014. 6. Vaid A, Chweich H, Balk EM, Jaber BL. Molecular adsorbent recirculating system as artificial support therapy for liver failure: a meta-analysis. ASAIO J. 2012;58(1):51-59. 7. Saliba F, Camus C, Durand F, et al. Albumin dialysis with a noncell artificial liver support device in patients with acute liver failure: a randomized, controlled trial. Ann Intern Med. 2013;159(8):522-531. 8. Banares R, Nevens F, Larsen FS, et al. Extracorporeal albumin dialysis with the molecular adsorbent recirculating system in acute-on-chronic liver failure: the RELIEF trial. Hepatology. 2013;57(3):1153-1162. 9. Sauer IM, Goetz M, Steffen I, et al. In vitro comparison of the molecular adsorbent recirculation system (MARS) and single-pass albumin dialysis (SPAD). Hepatology. 2004;39(5):1408-1414. 10. Ringe H, Varnholt V, Zimmering M, et al. Continuous veno-venous single-pass albumin hemodiafiltration in children with acute liver failure. Pediatr Crit Care Med. 2011;12(3):257-264. 11. Rifai K, Ernst T, Kretschmer U, et al. Prometheus—a new extracorporeal system for the treatment of liver failure. J Hepatol. 2003;39(6): 984-990. 12. Kribben A, Gerken G, Haag S, et al. Effects of fractionated plasma separation and adsorption on survival in patients with acute-onchronic liver failure. Gastroenterology. 2012;142(4):782-789.e3.
13. Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol. 2005;5:13. 14. Huedo-Medina TB, Sanchez-Meca J, Marin-Martinez F, et al. Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychol Methods. 2006;11(2):193-206. 15. Mitzner SR, Stange J, Klammt S, et al. Improvement of hepatorenal syndrome with extracorporeal albumin dialysis MARS: results of a prospective, randomized, controlled clinical trial. Liver Transplant. 2000;6(3):277-286. 16. Heemann U, Treichel U, Loock J, et al. Albumin dialysis in cirrhosis with superimposed acute liver injury: a prospective, controlled study. Hepatology. 2002;36(4 pt 1):949-958. 17. El Banayosy A, Kizner L, Schueler V, Bergmeier S, Cobaugh D, Koerfer R. First use of the Molecular Adsorbent Recirculating System technique on patients with hypoxic liver failure after cardiogenic shock. ASAIO J. 2004;50(4):332-337. 18. Sen S, Davies NA, Mookerjee RP, et al. Pathophysiological effects of albumin dialysis in acute-on-chronic liver failure: a randomized controlled study. Liver Transplant. 2004;10(9):1109-1119. 19. Hassanein TI, Tofteng F, Brown RS Jr, et al. Randomized controlled study of extracorporeal albumin dialysis for hepatic encephalopathy in advanced cirrhosis. Hepatology. 2007;46(6):1853-1862. 20. Laleman W, Wilmer A, Evenepoel P, et al. Effect of the molecular adsorbent recirculating system and Prometheus devices on systemic haemodynamics and vasoactive agents in patients with acute-on-chronic alcoholic liver failure. Crit Care. 2006;10(4): R108. 21. Dethloff T, Tofteng F, Frederiksen HJ, Hojskov M, Hansen BA, Larsen FS. Effect of Prometheus liver assist system on systemic hemodynamics in patients with cirrhosis: a randomized controlled study. World J Gastroenterol. 2008;14(13):2065-2071. 22. Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology. 2003;124(1):91-96. 23. Weiss JS, Gautam A, Lauff JJ, et al. The clinical importance of a protein-bound fraction of serum bilirubin in patients with hyperbilirubinemia. N Engl J Med. 1983;309(3):147-150. 24. Morita M, Amuro Y, Hada T, et al. Protein-bound and unbound bile acids in serum of patients with liver cirrhosis. Hepatol Res. 1998;10(3):217-224. 25. Butterworth RF. Pathogenesis of hepatic encephalopathy and brain edema in acute liver failure. J Clin Exp Hepatol. 2015;5(Suppl 1):S96-S103. 26. Beckman JS. Oxidative damage and tyrosine nitration from peroxynitrite. Chem Res Toxicol. 1996;9(5):836-844. 27. Montoliu C, Cauli O, Urios A, et al. 3-nitro-tyrosine as a peripheral biomarker of minimal hepatic encephalopathy in patients with liver cirrhosis. Am J Gastroenterol. 2011;106(9):1629-1637. 28. Frost MT, Halliwell B, Moore KP. Analysis of free and protein-bound nitrotyrosine in human plasma by a gas chromatography/mass spectrometry method that avoids nitration artifacts. Biochem J. 2000; 345(pt 3):453-458. 29. Zheng Z, Li X, Li Z, Ma X. Artificial and bioartificial liver support systems for acute and acute-on-chronic hepatic failure: a metaanalysis and meta-regression. Exp Ther Med. 2013;6(4):929-936.
INTRODUCTION Over the past several decades, a number of extracorporeal liver support therapies have been conceived to either bridge patients with liver failure to transplantation or serve as potential destination therapies for patients with acute or fulminant hepatic failure while awaiting tissue regeneration and recovery of liver function. These extracorporeal therapies are divided in 2 major categories, biologic and nonbiologic support systems. Biologic systems incorporate liver cells or tissues that simulate the excretory, synthetic, and metabolic functions of the liver,1,2 whereas nonbiologic systems use artificial membranes and adsorbents to detoxify the blood in patients with liver failure. Albumin dialysis is the beststudied and most promising nonbiologic system and is based on the removal of unwanted albumin-bound and water-soluble substances, such as bilirubin, bile acids, ammonia, nitrotyrosine, and fatty acids. There are 3 available albumin dialysis systems (Fig 1)3: the molecular adsorbent reticulating system or MARS (Teraklin, Rostock, Germany), single-pass albumin dialysis (SPAD), and the Prometheus system (Fresenius, Bad Homburg, Germany). MARS is a commercially available albumin dialysis system that removes protein-bound and water-soluble From Division of Nephrology, Department of Medicine, St. Elizabeth’s Medical Center, Boston, MA; and Department of Medicine, Tufts University School of Medicine, Boston, MA. Financial Disclosure: The authors declare that they have no relevant financial interests. Address correspondence to Bertrand L. Jaber, MD, MS, Division of Nephrology, Department of Medicine, St. Elizabeth’s Medical Center, 736 Cambridge Street, Boston, MA 02135. E-mail: [email protected] Ó 2015 by the National Kidney Foundation, Inc. All rights reserved. 1548-5595/$36.00 http://dx.doi.org/10.1053/j.ackd.2015.05.004
382
toxins. It comprises 2 separate dialysis circuits (Fig 1A); the first circuit contains exogenous human serum albumin, which is in contact with the patient’s blood through a semipermeable membrane (molecular weight cutoff of 50-60 kDa). Water-soluble and protein-bound toxins in the blood pass through this membrane. The toxin-enriched albumin solution is then passed through another dialyzer to remove water-soluble toxins using a counter-current bicarbonate-based dialysate. Albuminbound toxins are removed by 2 adsorbent cartridges that contain activated charcoal and an anion exchanger. The regenerated albumin solution is then ready for new uptake of toxins from the blood. Although MARS has been used in Europe for the treatment of acute-onchronic liver failure, severe alcoholic hepatitis, severe pruritus because of cholestasis, and intoxication with protein-bound substances, in the United States, it has been cleared for use in the treatment of drug overdose, poisoning, and hepatic encephalopathy but is not indicated for the treatment of chronic liver disease or as a bridge to liver transplantation.4,5 Relative contraindications to the use of MARS include severe sepsis, coagulopathy, and bleeding. A previous metaanalysis of small, randomized, and quasi-randomized trials published in 2012 demonstrated a clinical benefit of MARS compared with standard medical therapy in terms of lowering serum total bilirubin levels (net change 27.0 mg/dL; 95% confidence interval [CI], 210.4 to 23.7; P , .001) and improving hepatic encephalopathy (odds ratio, 3.0; 95% CI, 1.9-5.0; P , .001); however, there was no observed mortality benefit (odds ratio, 0.91; 95% CI, 0.64-1.31; P ¼ .62).6 Two randomized controlled trials of MARS have since been completed.7,8 Single-pass albumin dialysis is an extracorporeal liver support system that uses a conventional dialysis circuit in which an exogenous albumin solution is passed once
Advances in Chronic Kidney Disease, Vol 22, No 5 (September), 2015: pp 382-390
Albumin Dialysis
383
of albumin dialysate used, blood flow rate, anticoagulation use, duration of treatment, and number of treatment sessions), and duration of follow-up. The efficacy end points of interest were changes in circulating levels of total bilirubin, ammonia, and bile acids; improvement in hepatic encephalopathy; and all-cause mortality. To assess these end points, we extracted the mean values of the solutes at baseline and at the end of the study period and the net change from baseline, the number of patients experiencing an improvement in the West-Haven grade of hepatic encephalopathy (as defined in individual trials), and the number of patients who died. Safety end points of interest included gastrointestinal bleeding, catheter-related bleeding, thrombocytopenia, infections and sepsis, pulmonary adverse events, and serious adverse events. Disagreements were resolved through consensus. The corresponding authors of 3 trials were contacted for data clarification. When indicated, values reported as medians with ranges were converted to estimates of means with standard deviations.13 The Cochrane’s Collaboration tool was used to assess risk of bias, which covers 6 domains of bias (selection, performance, detection, attrition, reporting, and other bias). Two authors (A.S. CLINICAL SUMMARY and E.T.) independently rated each domain’s risk Use of albumin dialysis, including MARS and the of bias as low, high, or unPrometheus system, as supportive treatment for liver clear, with the implication failure is successful at removing albumin-bound of a third author (B.J.) in molecules, such as bilirubin and at improving hepatic case of disagreement. encephalopathy.
through the dialysate compartment and then discarded.9 This liver support system has shown some promise in a case series of children with acute liver failure.10 The Prometheus system is a variant of albumin dialysis and combines fractionated plasma separation, adsorption, and hemodialysis.11 It uses a 250-kDa semipermeable membrane generating an albumin-containing plasma-like solution. The patient’s albumin-containing plasma solution is then adsorbed on 2 albumin-detoxifying columns before reuniting with blood cells. Diffusive hemodialysis is then performed to remove water-soluble solutes. This system is advantageous as it relies on endogenous rather than exogenous albumin. The HELIOS trial represents the largest multicenter clinical experience using the Prometheus system.12 To provide an update on MARS and review the scarce literature on the other 2 albumin dialysis modalities, SPAD and the Prometheus system, we conducted a meta-analysis of all randomized controlled trials published to date that compared the efficacy and safety of any of the 3 albumin dialysis systems, including head-tohead comparisons. METHODS
Literature Search and Study Selection The literature search and study selection were performed independently by 2 authors (A.S. and E.T.). With the assistance of a medical librarian, the following electronic data However, survival benefit has not been established yet. Data Synthesis and bases were searched for Additional experience is required to guide its optimal use relevant citations: PubMed, Analysis and address safety concerns. Ovid, EMBASE, and CoRandom-effects model metachrane’s Library (inception analyses were used to calto October 2014). The culate the pooled mean terms and filters’ syntax for each database are provided difference in net change in levels of the solutes of interest in the Supplementary Material. We also searched and the risk ratio (RR) for improvement in the grade of heClinicalTrials.gov using similar terms and the bibliograpatic encephalopathy and all-cause mortality. All pooled phies of retrieved articles. The search strategy was limited estimates are displayed with a 95% CI. Existence of heteroto human studies with no restrictions on language, sample geneity among study effect sizes was examined using the I2 size, duration of study, or year of publication. index and the c2 test P value. An I2 index of 50% or more was used to indicate medium-to-high heterogeneity.14 SubData Extraction and Assessment of Bias group analyses were performed according to certain study We included parallel-arm randomized controlled trials characteristics, including the albumin dialysis modality that enrolled patients with acute liver failure, acute-on(MARS and Prometheus vs SPAD) and the number of dialchronic liver failure, or chronic decompensated liver disysis sessions (,5 vs $5). All analyses were performed usease, comparing the safety and efficacy of albumin dialysis ing the Cochrane’s Collaboration Review Manager systems (MARS, Prometheus, or SPAD) to standard med(version 5.3, Copenhagen, Denmark: The Nordic Cochrane ical therapy or to each other. Centre, The Cochrane Collaboration). The following data were extracted in duplicate: country of origin, year of publication, population setting (acute, RESULTS acute-on-chronic, and chronic liver failure), study design, Characteristics of the Trials and Risk of Bias study period, intervention groups, primary outcome, sumAssessment mary characteristics of study participants (sex, mean age, A total of 3688 potentially relevant citations were identified serum albumin, total bilirubin, creatinine, prothrombin and screened (Fig 2). Seventy articles were retrieved for time/international normalized ratio, and Model for Endevaluation, of which 10 randomized controlled Stage Liver Disease score), summary characteristics of trials fulfilled eligibility criteria comprising 7 trials of the albumin dialysis system used (including percentage
384
Tsipotis et al
A
B
C
Figure 1. Extracorporeal nonbiologic liver support albumin dialysis systems. (A) Molecular adsorbent reticulating system (MARS); (B) fractionated plasma separation and adsorption (Prometheus); and (C) single-pass albumin dialysis. Reproduced with permission from Struecker et al.3
MARS7,8,15-19 and 3 trials of Prometheus.12,20,21 There were no trials of SPAD. Two trials compared MARS to Prometheus, head-to-head, and to standard medical therapy.20,21 All 10 trials were conducted in Europe except for one that included a participating site in the United States.19 Characteristics of the individual trials are presented in Table 1. The trials spanned 13 years and varied in sample size (13-179 patients). Seven trials included patients with acute-on-chronic liver failure,8,12,15,16,18-20 with 2 trials of patients with acute liver failure7,17 and 1 trial of patients with decompensated chronic liver disease.21 The cause of liver disease ranged among studies with alcohol ingestion8,12,15,16,18-21 and viral hepatitis7,8,12,15,16,18,19,21 being the most frequent causes. Other reported causes included cardiogenic shock,17 autoimmune hepatitis, drug-induced7,8,16,19 and toxin-induced7 liver disease, primary8,15,19,21 and secondary15 biliary cirrhosis, hemochromatosis,21 Wilson disease,8 porphyria,21 graft vs host disease,8 nonalcoholic steatohepatitis,8 alpha-1 antitrypsin deficiency,8 and Budd-Chiari syndrome.15 The mean age of the study participants ranged from 40 to 62 years, and overall, 60% of participants were men. In terms of measures of liver dysfunction, the baseline serum total bilirubin level ranged from 11.1 to 25.7 mg/dL and the mean international normalized ratio from 0.8 to 8.3. The mean Model for End-Stage Liver Disease score ranged from 17 to 38, predicting a 3-month mortality risk of 6% to 53%.22 Regarding the characteristics of the albumin dialysis system (Table 2), the percentage of albumin dialysate used ranged from 10% to 20%, with a range of 1 to 10 treatment sessions, lasting 6 to 8 hours each. Anticoagulation was used in 6 trials; otherwise, it was either not reported or used variably.7,12,16,19 In almost all the trials, as part of standard medical therapy, patients in both groups received hemodiafiltration,7,15 hemofiltration,16 continuous venovenous hemofiltration,7,17,18 hemodialysis,7,21 or an unspecified form of kidney replacement therapy8,12
when indicated. Although the duration of the individual trials ranged from 1 to 360 days, when reported, mortality was ascertained at 30 to 360 days. Figure 3 displays the risk-of-bias assessment summary. Although all 10 trials exhibited a lower risk of selection bias (ie, random sequence generation and allocation concealment), there was a higher risk of performance bias (ie, lack of blinding of participants and personnel), detection bias (ie, blinding of outcome assessment, mainly the improvement in hepatic encephalopathy), and attrition bias (ie, incomplete outcome data). Effect of Albumin Dialysis on Removal of LiverRelated Solutes There were 6 trials with 7 experimental arms that reported on changes in total bilirubin levels with 329 analyzable patients. As shown in Figure 4A, by meta-analysis, compared with standard medical therapy, albumin dialysis resulted in a significant net decrease in the mean total bilirubin level by 8.0 mg/dL (95% CI 210.6 to 25.4; P ,.001). There was a nonsignificant net decrease in the mean level of ammonia (mean difference 22.5 mM; 95% CI, 222.4 to 17.3; P ¼.80; 3 trials) and bile acids (mean difference 24.9 mM; 95% CI, 237.5 to 27.6; P ¼ .77; 3 trials). The delivery of less than 5 or 5 or more albumin dialysis sessions did not further enhance the net reduction in mean total bilirubin levels (mean difference 27.9; 95% CI, 214.7 to 21.2; P ¼ .02, and mean difference 27.8; 95% CI, 210.6 to 25.1; P , .001, respectively). In a sensitivity analysis, however, a minimum of 3 sessions of albumin dialysis were required to significantly improve total bilirubin levels compared with fewer than 3 sessions (mean difference 29.3 mg/dL; 95% CI, 213.0 to 25.6; P , .001 vs 23.8 mg/dL; 95% CI, 29.8 to 2.1; P ¼ .21). In subgroup analyses, compared with standard medical therapy, MARS and Prometheus resulted in a significant net decrease in the mean total bilirubin level by 7.6
Albumin Dialysis
385
Figure 2. Study selection flow diagram.
mg/dL (95% CI, 213.0 to 22.3; P ¼ .005; 6 trials) and 8.1 mg/dL (95% CI, 211.2 to 24.9, P ,.001; 3 trials), respectively. In the 2 trials comparing Prometheus to MARS,20,21 compared with MARS, the Prometheus system did not further reduce the mean total bilirubin level (mean difference 4.6; 95% CI, 24.2 to 13.4; P ¼ .30). Effect of Albumin Dialysis on Hepatic Encephalopathy Only 3 trials of MARS reported on the rate of improvement in the West-Haven grade of hepatic encephalopathy.8,18,19 By meta-analysis, compared with standard medical therapy, MARS resulted in a significant 55% increased likelihood of improvement in hepatic encephalopathy (RR, 1.55; 95% CI, 1.16-2.08; P ¼ .003; Fig 4B). There were too few trials to conduct meaningful subgroup analyses according to the number of treatment sessions. Effect of Albumin Dialysis on Mortality All 10 trials reported on mortality rates with 634 analyzable patients. By meta-analysis, compared with standard
medical therapy, albumin dialysis had no significant effect on mortality (RR, 0.95; 95% CI, 0.84-1.07; P ¼ .42; Fig 4C). However, in a sensitivity analysis that excluded 1 trial with the longest follow-up period of 1 year,7 albumin dialysis resulted in a nonsignificant trend toward a 13% risk reduction in mortality (RR, 0.87; 95% CI, 0.75-1.02; P ¼ .08). In a subgroup analysis based on the number of sessions delivered, in studies that delivered 5 or more albumin dialysis sessions, there was a nonsignificant trend toward a 13% risk reduction in mortality (RR, 0.87; 95% CI, 0.741.02; P ¼ .08) compared with trials that provided less than 5 sessions where these was no observed effect on mortality (RR, 1.08; 95% CI, 0.89-1.31; P ¼ .42). On subgroup analyses, there was no demonstrable mortality benefit of either MARS or Prometheus compared with standard medical therapy (RR, 0.97; 95% CI, 0.85-1.11; P ¼ .69; and RR, 0.87; 95% CI, 0.66-1.14; P ¼ .31, respectively). Furthermore, in the 2 head-to-head trials of the 2 albumin dialysis systems, Prometheus did not result in a significant mortality risk reduction compared with MARS (RR, 0.67; 95% CI,
386
Table 1. Characteristics of the Randomized Controlled Trials Included in the Meta-analysis
Study
Year
Country
No. Population Setting Patients
2000 Germany
Heemann et al16
2002 Germany
El Banayosy et al17 Sen et al18
2004 Germany 2004 United Kingdom
Acute-on-chronic liver failure
18
Laleman et al20
2006 Belgium
Acute-on-chronic liver failure
18
Acute-on-chronic liver failure
70
Decompensated liver disease 2012 Germany, Spain, Acute-on-chronic United Kingdom, liver failure Italy, Belgium, Croatia, and Austria 2013 19 European Acute-on-chronic sites liver failure 2013 France Acute liver failure
24
Hassanein et al19 2007 United States, Germany, and Denmark Dethloff et al21 2008 Denmark Kribben et al12
~ ares et al8 Ban Saliba et al7
Acute-on-chronic liver failure Acute-on-chronic liver failure Acute liver failure
13 24 27
145
179 102
MARS 1 SMT 6 HDF vs SMT 6 HDF MARS 1 SMT vs SMT 6 HF MARS 1 SMT vs SMT 6 CVVH MARS 1 SMT 6 CVVH vs SMT 6 CVVH MARS 1 SMT vs Prometheus 1 SMT vs SMT MARS 1 SMT vs SMT
47
38
—
—
25.7
4.0
30
30
52
54
—
—
—
—
30
30
62
70
—
1.8
11.1
2.0
14
NR
45
72
18
2.0
20.3
1.4
7
90
51
67
26
—
27.0
1.3
3
NR
52
56
31
2.0
14.2
1.7
5
180
MARS vs Prometheus vs HD Prometheus 1 SMT 6 RRT vs SMT 6 RRT
55
63
—
2.1
12.5
1.1
1
180
50
63
28
2.0
25.5
2.3
90
90
MARS 1 SMT vs SMT 6 RRT MARS 1 SMT 6 RRT vs SMT 6 RRT
51
69
25
0.8
25.1
—
28
90
40
43
38
8.3
11.2
—
360
360
Abbreviations: CVVH, continuous venovenous hemofiltration; HF, hemofiltration; HD, hemodialysis; INR, international normalized ratio; MARS, molecular adsorbent recirculating system; MELD, Model for End-Stage Liver Disease; RRT, renal replacement therapy; SMT, standard medical therapy.
Tsipotis et al
Mitzner et al15
Study Groups
Mean Serum Mean Mean Mean Total Serum Study Follow-up Age Men MELD Mean Bilirubin Creatinine Duration Period (y) (%) Score INR (mg/dL) (mg/dL) (d) (d)
387
Albumin Dialysis
Table 2. Characteristics of Albumin Dialysis in the Trials Included in the Meta-analysis Study Mitzner et al15 Heemann et al16 Banayosy et al17 Sen et al18 Laleman et al20 Hassanein et al19 Dethloff et al21 Kribben et al12 ~ ares et al8 Ban Saliba et al7
Albumin Dialysis System MARS MARS MARS MARS MARS and Prometheus MARS MARS and Prometheus Prometheus MARS MARS
No. of Sessions
Duration Per Session (h)
5 10 7 4 3 3 1 8 7 1
6-8 6 8 8 6 6 6 6 7 8
Anticoagulation Use Yes Minimal Yes Yes Yes — Yes Yes (variable) Yes Yes (variable)
Dialysate albumin Strength (%) 10 15 20 20 20 17 20 — — 20
Abbreviation: MARS, molecular adsorbent recirculating system.
0.15-2.98; P ¼ .30). Although there was no evidence of significant heterogeneity among study effect sizes in all the analyses, the results of the subgroup and sensitivity analyses need to be interpreted with caution. Safety Analysis There was a significant inconsistency among studies in the reporting of adverse events. Among the trials that reported on the following safety events, albumin dialysis was not associated with an increased risk of respiratory adverse events (RR, 1.44; 95% CI, 0.83-2.50; P ¼ .19),7,8,12,16,19-21 gastrointestinal bleeding (RR, 0.90; 95% CI, 0.60-1.36; P ¼ .62),7,8,12,19-21 and severe infections or sepsis (RR, 1.14; 95% CI, 0.85-1.55; P ¼ .38).7,8,16,19-21 In 3 trials,16,19,21 there were 8 catheter-related episodes of bleeding or hematomas among the 59 patients treated with MARS, one of which was fatal.19 Hematologic-related adverse events associated using albumin dialysis included thrombocytopenia, coagulopathy, and thrombotic episodes.7,8,12,16,17,19,21 Although 1 trial reported 2 (5%) of 39 patients experiencing treatment-related thrombocytopenia compared with none (0%) among the 31 patients receiving standard medial therapy,19 in the 3 trials where absolute changes in platelet count could be analyzed,12,17,21 compared with standard medical therapy, albumin dialysis did not result in a significant net decrease in the mean platelet count (mean difference 15,840 per microliter; 95% CI, 224,000 to 55,690; P ¼ .44). Among the 3 trials reporting on serious adverse events,7,19,21 only in 1 study were these events more frequently observed in patients assigned to albumin dialysis compared with standard medical therapy (51% vs 26%).19 Almost half of these events were ascribed to the therapy. In the other 2 studies, there were either no reported serious adverse events21 or they were not significantly different between the albumin dialysis and standard medical therapy group (32% vs 29%).7 DISCUSSION Albumin dialysis is the best-studied extracorporeal nonbiologic liver support system for the management of patients with acute, acute-on-chronic, and chronic decompensated liver failure as a potential bridge or destination therapy. It is based on the removal of unwanted albumin-bound and water-soluble substances. In the pre-
sent systematic review, we summarize qualitatively the scarce literature on the 3 available albumin dialysis systems, mainly MARS, SPAD, and Prometheus, and conduct a quantitative analysis of all 10 randomized controlled trials published to date that compare the efficacy and safety of any of these albumin dialysis systems. Most trials included patients with acute-on-chronic liver failure, and the majority of study participants were men. In 8 of the 10 trials, patients in the standard medical treatment arm received some form of kidney replacement therapy. The albumin dialysis experimental treatment arms were restricted to MARS and Prometheus, and no randomized clinical trials of SPAD were identified. To examine the detoxifying capacity of the albumin dialysis systems, we chose surrogate efficacy end points, mainly the improvement in circulating levels of albumin-bound molecules (bilirubin) and water-soluble molecules (ammonia), grades of hepatic encephalopathy, and a hard clinical endpoint, all-cause mortality. In terms of removal of liver-related solutes, compared with standard medical therapy, albumin dialysis resulted in a significant net reduction in serum total bilirubin levels, but in the few trials with analyzable data, albumin dialysis did not significantly reduce levels of ammonia and bile acids. Albumin dialysis also resulted in a significant improvement in grades of hepatic encephalopathy (only observed in trials of MARS with analyzable data). Of note, however, all the patients included in the analysis of hepatic encephalopathy had acute-on-chronic liver failure, and these results should be interpreted in this clinical context. Although there was no significant impact of albumin dialysis on all-cause mortality, among the 6 trials that delivered 5 or more albumin dialysis sessions, there was a nonsignificant trend toward a 13% risk reduction in mortality, suggestive of a dose-dependent effect. The inconsistency in the reporting of adverse events did not allow for a robust safety analysis. Except for 1 trial,19 serious adverse events among patients receiving albumin dialysis were not more frequently reported. Previous studies have found that albumin-bound bilirubin is an important fraction of total bilirubin (up to 90%) in patients with advanced liver disease and is not detected in healthy subjects.23 When hepatic excretion of conjugated bilirubin is impaired, albumin-bound bilirubin appears in serum and becomes a larger component of total
388
Tsipotis et al
Random sequence generation (selection bias) Allocation concealment (selection bias) Blinding of participants and personnel (performance bias) Blinding of outcome assessment (detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias 0% Low risk of bias
Unclear risk of bias
25%
50%
75%
100%
High risk of bias
Figure 3. Risk of bias assessment graph: each 100% stacked column represents a low, unclear, and high risk of bias domain across all the trials included in the meta-analysis.
bilirubin, thus its potential removal by albumin dialysis in patients with advanced liver disease. We found no significant difference in the net reduction of total bilirubin levels between MARS and the Prometheus system, and increasing the number of treatment sessions to 5 or greater did not further affect bilirubin levels, suggesting a ceiling effect. In a sensitivity analysis, a minimum of 3 sessions of albumin dialysis were required to significantly improve bilirubin levels. This could be attributed to the fact that the maximum clearance ability of albumin dialysis in removing bilirubin is reached after 3 sessions. In the 3 trials examining the removal of serum bile acids, which are 80% protein bound in patients with liver cirrhosis,24 albumin dialysis resulted in a nonsignificant net decrease in circulating levels of this biomarker. We also found a nonsignificant net decrease in levels of ammonia in 3 trials with analyzable data. However, this might have been, in part, because of the fact that both the experimental and control groups received some form of kidney replacement therapy, which would have removed this water-soluble small molecular weight solute. Although ammonia is the best-characterized neurotoxin in the pathophysiology of hepatic encephalopathy,25 an elevated serum ammonia level is not required to make the diagnosis and is not specific for this disorder. Serum nitrotyrosine is a byproduct of superoxide and nitric oxide generation26 and has been shown to be a promising biomarker of hepatic encephalopathy.27 Although the majority of circulating nitrotyrosine is free and not protein bound,28 none of the trials included in our systematic review examined the potential removal of this biomarker by albumin dialysis. The lack of improvement in patient survival is not surprising as all the trials were clearly underpowered to examine this hard clinical end point, especially in such clinical settings where the underlying acute severity of illness and coexisting comorbidities are typically dominant predictors of clinical outcomes. This is further confounded by the inadequate safety analysis because of the inconsistent reporting of adverse events across the trials. Although we found no published trials of SPAD, a query of the US-based clinical trials registration Web site (http:// clinicaltrials.gov/) identified 2 randomized controlled trials
(NCT01079104 and NCT01079091) of Hepa Wash system (Hepa Wash GmbH, Munich, Germany), a form of SPAD. These 2 trials have an estimated planned enrollment of 244 participants, with the primary outcome of 30-day mortality. The current status of these trials is unknown. Our results confirm those of a previous meta-analysis6 comparing MARS to standard medical therapy with regard to bilirubin removal, improvement of hepatic encephalopathy, and lack of survival benefit. Although both studies used similar methodologies, the previous meta-analysis included nonrandomized trials.6 In contrast to our study, another recently published metaanalysis29 found a mortality benefit of artificial liver support systems over standard medical therapy in patients with acute-on-chronic liver failure. The study did not focus on albumin dialysis and included studies that compared albumin dialysis, hemoperfusion, hemofiltration, and the BioLogic-DT sorbent suspension dialyzer to standard medical therapy. Strengths of our synthesis include the large number of trials, the strict methodology used for the selection of the studies and the accurate implementation of the methods of the protocol. Our synthesis focused exclusively on comparing albumin dialysis systems to either standard medical therapy or to each other. However, there are also important limitations that should be emphasized. Because of the nature of the intervention, double blinding was not possible in the trials, which might have been a source of bias in the assessment of hepatic encephalopathy. Although there was no demonstrable heterogeneity in the effect estimates of individual trials, methodologic heterogeneity was salient, including etiology and severity of liver disease, timing of outcome assessment, and variable follow-up period. In addition, the limited number of trials examining Prometheus, the variable use of kidney replacement therapies in the individual trials, and the variability in the albumin dialysis protocols all created further obstacles in drawing safe conclusions. Finally, the variable timing of the reporting of outcomes among studies likely affected the generalization of the results. In conclusion, albumin dialysis, including MARS and the Prometheus system, results in an increase in the removal of
389
Albumin Dialysis
A Study or Subgroup 4.5.1 MARS vs SMT Dethloff 2008 Mitzner 2000 Laleman 2006 Hassanein 2007 Sen 2004 Banayosy 2004 Subtotal (95% CI)
Albumin Dialysis SMT SD Total SD Total Mean Mean -3.6 19.15 -9.5 13.8 -5.2 7.01 0.3 15.86 -13.58 8.28 -2.98 6.29
Mean Difference IV, Random, 95% CI
8 -4.40 [-27.01, 18.21] 5 -7.70 [-29.51, 14.11] -5.60 [-18.89, 7.69] 6 -0.30 [-8.81, 8.21] 31 9 -16.65 [-24.67, -8.63] 13 -7.36 [-12.90, -1.82] 72 -7.64 [-12.97, -2.31]
0.8 26.42 -1.8 22.36 0.4 15.06 0.6 19.6 3.07 9.06 4.38 8.19
8 8 6 39 9 14 84
Mean Difference IV, Random, 95% CI
Heterogeneity: Tau² = 14.92; Chi² = 7.88, df = 5 (P = 0.16); I² = 37% Test for overall effect: Z = 2.81 (P = 0.005) 4.5.2 Prometheus vs SMT Dethloff 2008 Laleman 2006 Kribben 2012 Subtotal (95% CI)
-2.7 16.09 -11.7 10.58 11 -8
8 6 68 82
0.8 26.42 0.4 15.06 9 0
8 6 77 91
-3.50 [-24.94, 17.94] -12.10 [-26.83, 2.63] -8.00 [-11.26, -4.74] -8.09 [-11.24, -4.94]
Heterogeneity: Tau² = 0.00; Chi² = 0.46, df = 2 (P = 0.79); I² = 0% Test for overall effect: Z = 5.04 (P < 0.00001) Total (95% CI)
-7.97 [-10.58, -5.35]
154
175
Heterogeneity: Tau² = 0.83; Chi² = 8.36, df = 8 (P = 0.40); I² = 4% Test for overall effect: Z = 5.97 (P < 0.00001) Test for subgroup differences: Chi² = 0.02, df = 1 (P = 0.89), I² = 0%
B Study or Subgroup 4.8.1 MARS vs SMT
Albumin Dialysis SMT Events Total Events Total 15 24 9
Bañares 2013 Hassanein 2007 Sen 2004 Subtotal (95% CI)
24 39 9 72
13 12 6
-100 -50 0 50 Favors Albumin Dialysis Favors SMT
Risk Ratio IV, Random, 95% CI
34 31 9 74
100
Risk Ratio IV, Random, 95% CI
1.63 [0.96, 2.77] 1.59 [0.96, 2.64] 1.46 [0.91, 2.35] 1.55 [1.16, 2.08]
31 48 Total events Heterogeneity: Tau² = 0.00; Chi² = 0.11, df = 2 (P = 0.95); I² = 0% Test for overall effect: Z = 2.98 (P = 0.003) 4.8.2 Prometheus vs SMT Subtotal (95% CI)
0
72
Total (95% CI)
0
Not estimable
74
1.55 [1.16, 2.08]
0
0 Total events Heterogeneity: Not applicable Test for overall effect: Not applicable
31 48 Total events Heterogeneity: Tau² = 0.00; Chi² = 0.11, df = 2 (P = 0.95); I² = 0% Test for overall effect: Z = 2.98 (P = 0.003) Test for subgroup differences: Not applicable
C Study or Subgroup 4.11.1 MARS vs SMT
SMT Albumin Dialysis Total Events Total Events
Banayosy 2004 Bañares 2013 Dethloff 2008 Hassanein 2007 Heemann 2002 Laleman 2006 Mitzner 2000 Saliba 2013 Sen 2004 Subtotal (95% CI)
7 49 2 19 1 0 6 44 5
14 90 8 39 12 6 8 53 9 239
9 51 3 17 6 0 5 37 5
0.01
Risk Ratio IV, Random, 95% CI
13 89 8 31 12 6 5 49 9 222
0.1 1 10 100 Favors SMT Favors Albumin Dialysis
Risk Ratio IV, Random, 95% CI
0.72 [0.38, 1.37] 0.95 [0.73, 1.23] 0.67 [0.15, 2.98] 0.89 [0.56, 1.40] 0.17 [0.02, 1.18] Not estimable 0.79 [0.49, 1.26] 1.10 [0.90, 1.34] 1.00 [0.44, 2.29] 0.97 [0.85, 1.11]
Total events 133 133 Heterogeneity: Tau² = 0.00; Chi² = 6.58, df = 7 (P = 0.47); I² = 0% Test for overall effect: Z = 0.40 (P = 0.69) 4.11.2 Prometheus vs SMT Dethloff 2008 Kribben 2012 Laleman 2006 Subtotal (95% CI)
3 41 0
8 77 6 91
3 42 0
8 68 6 82
1.00 [0.28, 3.54] 0.86 [0.65, 1.14] Not estimable 0.87 [0.66, 1.14]
Total events 44 45 Heterogeneity: Tau² = 0.00; Chi² = 0.05, df = 1 (P = 0.82); I² = 0% Test for overall effect: Z = 1.01 (P = 0.31) Total (95% CI)
330
304
0.95 [0.84, 1.07]
177 Total events 178 Heterogeneity: Tau² = 0.00; Chi² = 7.16, df = 9 (P = 0.62); I² = 0% Test for overall effect: Z = 0.81 (P = 0.42) Test for subgroup differences: Chi² = 0.53, df = 1 (P = 0.47), I² = 0%
0.1 0.2 0.5 1 2 Favors Albumin Dialysis Favors SMT
5
10
Figure 4. Forest plot displaying of albumin dialysis vs standard medical therapy on (A) net change in serum mean total bilirubin level, (B) risk ratio for improvement in the West-Haven grade of hepatic encephalopathy, and (C) risk ratio for all-cause mortality. Abbreviations: CI, confidence interval; SMT, standard medical therapy; MARS, molecular adsorbent recirculating system.
390
Tsipotis et al
total bilirubin, an albumin-bound liver solute, and in an improvement in hepatic encephalopathy, relative to standard medical therapy. These changes might confer a longterm survival benefit. These findings call for the design of multicenter, large, randomized controlled trials of efficacy and safety of albumin dialysis aimed at exploring their potential role as bridge or destination extracorporeal supportive therapies for patients with liver failure awaiting liver transplantation or recovery of hepatic function. ACKNOWLEDGMENTS The authors acknowledge the technical assistance of Catherine Guarcello, MS (Stohlman Library, St. Elizabeth’s Medical Center, Boston, MA) and Mark Sherer.
SUPPLEMENTARY DATA Supplementary data related to this article can be found at http://dx.doi.org/10.1053/j.ackd.2015.05.004. REFERENCES 1. Hassanein TI, Schade RR, Hepburn IS. Acute-on-chronic liver failure: extracorporeal liver assist devices. Curr Opin Crit Care. 2011;17(2):195-203. 2. Stange J. Extracorporeal liver support. Organogenesis. 2011;7(1):64-73. 3. Struecker B, Raschzok N, Sauer IM. Liver support strategies: cutting-edge technologies. Nat Rev Gastroenterol Hepatol. 2014;11(3): 166-176. 4. Tolwani A, While K. Extracorporeal liver support and the kidney. In: Critical Care Nephrology. 2nd ed. Philadelphia, PA: Saunders Elsevier; 2009:1061-1067. 5. Gambro. Molecular adsorbent recirculating system., http://www.gambro. com/en/global/Therapies/Molecular-Adsorbent-Recirculating-Systems1/ ?tab¼documentTab#tabList. Accessed November 28, 2014. 6. Vaid A, Chweich H, Balk EM, Jaber BL. Molecular adsorbent recirculating system as artificial support therapy for liver failure: a meta-analysis. ASAIO J. 2012;58(1):51-59. 7. Saliba F, Camus C, Durand F, et al. Albumin dialysis with a noncell artificial liver support device in patients with acute liver failure: a randomized, controlled trial. Ann Intern Med. 2013;159(8):522-531. 8. Banares R, Nevens F, Larsen FS, et al. Extracorporeal albumin dialysis with the molecular adsorbent recirculating system in acute-on-chronic liver failure: the RELIEF trial. Hepatology. 2013;57(3):1153-1162. 9. Sauer IM, Goetz M, Steffen I, et al. In vitro comparison of the molecular adsorbent recirculation system (MARS) and single-pass albumin dialysis (SPAD). Hepatology. 2004;39(5):1408-1414. 10. Ringe H, Varnholt V, Zimmering M, et al. Continuous veno-venous single-pass albumin hemodiafiltration in children with acute liver failure. Pediatr Crit Care Med. 2011;12(3):257-264. 11. Rifai K, Ernst T, Kretschmer U, et al. Prometheus—a new extracorporeal system for the treatment of liver failure. J Hepatol. 2003;39(6): 984-990. 12. Kribben A, Gerken G, Haag S, et al. Effects of fractionated plasma separation and adsorption on survival in patients with acute-onchronic liver failure. Gastroenterology. 2012;142(4):782-789.e3.
13. Hozo SP, Djulbegovic B, Hozo I. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol. 2005;5:13. 14. Huedo-Medina TB, Sanchez-Meca J, Marin-Martinez F, et al. Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychol Methods. 2006;11(2):193-206. 15. Mitzner SR, Stange J, Klammt S, et al. Improvement of hepatorenal syndrome with extracorporeal albumin dialysis MARS: results of a prospective, randomized, controlled clinical trial. Liver Transplant. 2000;6(3):277-286. 16. Heemann U, Treichel U, Loock J, et al. Albumin dialysis in cirrhosis with superimposed acute liver injury: a prospective, controlled study. Hepatology. 2002;36(4 pt 1):949-958. 17. El Banayosy A, Kizner L, Schueler V, Bergmeier S, Cobaugh D, Koerfer R. First use of the Molecular Adsorbent Recirculating System technique on patients with hypoxic liver failure after cardiogenic shock. ASAIO J. 2004;50(4):332-337. 18. Sen S, Davies NA, Mookerjee RP, et al. Pathophysiological effects of albumin dialysis in acute-on-chronic liver failure: a randomized controlled study. Liver Transplant. 2004;10(9):1109-1119. 19. Hassanein TI, Tofteng F, Brown RS Jr, et al. Randomized controlled study of extracorporeal albumin dialysis for hepatic encephalopathy in advanced cirrhosis. Hepatology. 2007;46(6):1853-1862. 20. Laleman W, Wilmer A, Evenepoel P, et al. Effect of the molecular adsorbent recirculating system and Prometheus devices on systemic haemodynamics and vasoactive agents in patients with acute-on-chronic alcoholic liver failure. Crit Care. 2006;10(4): R108. 21. Dethloff T, Tofteng F, Frederiksen HJ, Hojskov M, Hansen BA, Larsen FS. Effect of Prometheus liver assist system on systemic hemodynamics in patients with cirrhosis: a randomized controlled study. World J Gastroenterol. 2008;14(13):2065-2071. 22. Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver disease (MELD) and allocation of donor livers. Gastroenterology. 2003;124(1):91-96. 23. Weiss JS, Gautam A, Lauff JJ, et al. The clinical importance of a protein-bound fraction of serum bilirubin in patients with hyperbilirubinemia. N Engl J Med. 1983;309(3):147-150. 24. Morita M, Amuro Y, Hada T, et al. Protein-bound and unbound bile acids in serum of patients with liver cirrhosis. Hepatol Res. 1998;10(3):217-224. 25. Butterworth RF. Pathogenesis of hepatic encephalopathy and brain edema in acute liver failure. J Clin Exp Hepatol. 2015;5(Suppl 1):S96-S103. 26. Beckman JS. Oxidative damage and tyrosine nitration from peroxynitrite. Chem Res Toxicol. 1996;9(5):836-844. 27. Montoliu C, Cauli O, Urios A, et al. 3-nitro-tyrosine as a peripheral biomarker of minimal hepatic encephalopathy in patients with liver cirrhosis. Am J Gastroenterol. 2011;106(9):1629-1637. 28. Frost MT, Halliwell B, Moore KP. Analysis of free and protein-bound nitrotyrosine in human plasma by a gas chromatography/mass spectrometry method that avoids nitration artifacts. Biochem J. 2000; 345(pt 3):453-458. 29. Zheng Z, Li X, Li Z, Ma X. Artificial and bioartificial liver support systems for acute and acute-on-chronic hepatic failure: a metaanalysis and meta-regression. Exp Ther Med. 2013;6(4):929-936.