Molecular adsorbent recirculating system: Albumin dialysis-based extracorporeal liver assist device



    Corresponding author
    1. Department of Gastroenterology, GB Pant Hospital, New Delhi, India
      Dr D Kapoor, Department of Gastroenterology, GB Pant Hospital, New Delhi 110002, India. Email:
    Search for more papers by this author

Dr D Kapoor, Department of Gastroenterology, GB Pant Hospital, New Delhi 110002, India. Email:


Abstract  Extracorporeal liver assist devices have been used for more than five decades to support patients with liver failure. Numerous modifications have been made to both biological as well as mechanical liver assist devices. Possibly, an ideal liver assist device would be one that would perform optimal detoxification and synthetic functions of the liver, be simple to set up and yet be cost-effective. An albumin dialysis-based device that uses a hybrid albumin-impregnated membrane to get rid of albumin-bound toxins that circulate in abundance in liver failure, called the molecular adsorbent recirculating system (MARS) has been in clinical use for nearly four years now. Results with the use of this device in both acute and acute-on-chronic liver failure have shown consistent improvement in biochemical profile, resolution of encephalopathy, correction in hemodynamics, reduction in intracranial pressure and some improvement in the synthetic function of the liver. In a number of studies, albeit of small sample size, survival advantage has also been observed. The timing of initiation of therapy with MARS, duration of treatment, frequency of sessions and ‘maintenance therapy’ are still some of the unresolved issues with the use of this device. Large multicentric trials on the use of this technique are expected to throw light on these issues and help optimize the potential of this liver assist device.

© 2002 Blackwell Publishing Asia Pty Ltd


The liver plays a pivotal role in regulating a wide range of key metabolic and homeostatic activities. The failure of hepatic function seriously affects these regulatory pathways as well as having an adverse bearing on the functioning of other organ systems culminating in multiorgan failure MOF.1

Traditionally, liver failure has been classified as acute (ALF; developing in the absence of any pre-existing liver disease)2 or acute on chronic (AoCLF; in which a precipitant decompensates previously well compensated liver disease).3 In addition, there is a chronic decompensated state of liver disease that indolently progresses to end-stage liver disease. Although liver transplantation is possibly the only therapeutic modality available for end-stage liver disease, the hepatic func-tion in both acute and acute-on-chronic liver disease may be salvaged, provided that the failing liver is given a chance to regenerate and, in the interim, the metabolic perturbations leading to MOF and death are prevented.

An ever-increasing shortage of organs for liver transplantation has forced clinicians, intensivists and hepatologists to look for alternative methods of management, short of liver transplantation such as the liver assist devices (LAD). A suitable artificial liver should have the capacity to perform the detoxification as well as the synthetic functions of a normal liver, while at the same time provide a milieu for regeneration of the native liver and prevent the cascade of events that culminates in MOF and death.4

Most of the liver support therapies are used extracorporeally, akin to devices providing renal replacement therapy. They may be mainly divided into biological and non-biological LAD (Table 1). While biological liver support therapies utilize hepatocytes or whole organs derived from animal or human sources,5 non-biological approaches are based on dialysis, filtration, and adsorption techniques.

Table 1.  Extracorporeal liver assist devices
 Bio-artificial liver
 Extracorporeal liver assist device
 Plasma exchange with or without hemodiafiltration
 Charcoal hemoperfusion
 Exchange resins
 Sorbent dialysis
 Fractionated plasma separation and administration
 Molecular Absorbent Recirculating System

Of the various types of biological devices available, the maximum experience has been with capillary hollow-fiber systems bioreactor modules. The bioartificial liver or BAL (HepatAssist 2000; Circe Biomedical, Lexington, MA, USA) developed by Demetriou4 and the extracorporeal liver-assist device or ELAD (Vitaglen, La Jolla, CA, USA) developed by Susmann et al.6 are the devices that have been used most extensively, but mainly in ALF. The patients treated with these devices have shown improvement in mental status, decrease in intracranial pressure and an accompanying improvement in biochemical parameters.7,8 The Sussman device was used in a randomized controlled trial that enrolled patients with acute liver failure.9 However, there was no survival advantage when ELAD was compared with standard medical therapy. Moreover, the experience with these devices in AoCLF is limited.

Apart from the high cost, concerns regarding the use of bioartificial liver assist devices are immunologic complications; transmission of the porcine endogenous retrovirus (PERV) especially with BAL; infectious complications; and tumor transmigration, especially with ELAD (this device utilizes the C3A human hepatoma cell lines).

The use of the purely mechanical liver assist devices is also fraught with logistical problems (large replacement of fresh frozen plasma in plasma exchange), frequent side-effects,10 modest toxin removal11 and improvement in individual clinical parameters (such as encephalopathy) rather than actual patient survival.12,13 Use of cation exchange resins and sorbent dialysis technique to address these problems has not been successful.14 As a recent controlled trail showed, granulocyte activation, platelet consumption, and disseminated intravascular coagulation was no different between the Biologic-DT system and conventional charcoal hemoperfusion.15

The molecular adsorbent recirculating system (MARS; University of Rostock, Rostock, Germany), which is the main focus of this review, is possibly the extracorporeal device that has found most widespread use in intensive care units and hepatology centers worldwide. The list of conditions for which this device has been used has been steadily increasing (Table 2).

Table 2.  Indications of MARS therapy
  1. MARS, molecular adsorbent recirculating system; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis.

Acute on chronic liver failure
Acute liver failure following drug overdosage, toxin exposure
Primary non-function or delayed graft function following
Liver failure following hepatic resection, cardiac surgery
Liver failure following burns, sepsis, trauma
Chronic cholestatic liver disease (PBC, PSC)
Graft vs host disease
Wilson's fulminant hepatic failure

Principle of molecular adsorbent recirculating system therapy

The removal of water-soluble substances is well established by dialysis procedures. However, an unsolved problem is the low dialyzability of lipophilic albumin-bound toxins (ABT; Table 3). Such toxins have been shown to be responsible for the induction of hepatic encephalopathy in chronic liver diseases as well as acute liver failure. The MARS membrane (surface area approx. 1.3 m2, thickness approx. 100 nm and pore size approx. 50 kDa) has an asymmetric structure with large caverns on the dialysate side and small pores towards the skin layer of the membrane, which is in contact with the patient's blood. When the membrane (MARS Flux; Teraklin, Rostock, Germany) is preperfused with albumin at the time of priming the MARS device, albumin molecules in the dialysate solution permeate from the dialysate into the membrane to be retained by the internal part of the membrane. It has been shown that albumin attached to polymers enhances its affinity to albumin-bound toxins.16 The passage of albumin-bound toxins from the patient's blood is carried out via active de-ligandization of plasma albumin, transport of the ligands across the membrane and ligandization of dialysate albumin (Fig. 1). The removal of ABT depends on physicochemical interactions between the ABT in the plasma, the MARS membrane and the dialysate, especially in the skin layer region of the membrane (passage by ‘flip-flop’ across the ‘shuttle molecule’). Continuous desorption of ABT on the dialysate side maintains a constant gradient that permits passage of toxins from the patient's blood side to the dialysate side. The toxins bound to dialysate albumin are disposed of by passage over anion exchanger (AE) and activated charcoal (AC) columns, and sent back to the MARS membrane (recirculated) for the next cycle of toxin passage.

Table 3.  Lipophilic substances bound to albumin that are increased in liver failure (potential ABT)
  1. ABT, albumin-bound toxins.

Aromatic amino-acids
Bile acids
Digoxin-like substances
Endogenous benzodiazepine-like substances
Short and medium chain fatty acids
Nitric oxide
Figure 1.

Schematic representation of the molecular adsorbent recirculating system (MARS) membrane and the kinetics of transfer of albumin-bound toxins (ABT). The openings (caverns) on the dialysate side are much larger, permitting easy passage and binding of albumin close to the blood side of the membrane (skin layer). The ABT are deligandized from the plasma albumin and bind to free albumin on the skin layer, which is at a higher concentration, and are subsequently removed to the dialysate side of the circuit. Albumin in essence acts as a ‘shuttle molecule’. Water-soluble toxins are transferred to the dialysate by mechanisms of diffusion and convection as in conventional dialysis. (Modified from 17 with the kind permission of the authors.)

In vitro experiments with the MARS set-up have revealed that the clearance of various water- and lipid-soluble moieties across various components of the circuit is different, with bile acids being cleared predominantly by the activated charcoal, bilirubin by the anion exchanger and creatinine by the standard dialyzer. 17 Radiolabeling of albumin with Texas Red and study of its distribution on the membrane by fluorescent microscopy has shown that no physical transfer of albumin (size: 65 kDa) occurs across the MARS membrane. 18 If the dialysate is albumin-free, hardly any transfer of lipophilic moieties occurs from the blood side to the dialysate side. Also, for a given albumin-bound ligand, there exists equilibrium in its concentration between the plasma, dialysate and the membrane ( Fig. 2 ). This phenomenon also has been shown to occur in vitro (C. Steiner, University of Rostock, pers. comm. 2000).

Figure 2.

Changes in the concentration of albumin and bilirubin in the plasma and the dialysate after a state of equilibrium has been reached in an in vitro molecular adsorbent recirculating system (MARS)-like two-loop circuit. The concentration of the plasma albumin remains fairly constant, while that of the dialysate decreases and then plateaus, representing the loss to the membrane (which can also be thought to be a separate compartment). Similarly, the fall in the plasma bilrubin level is much more than the addition of bilirubin to the dialysate side, again signifying the importance of the membrane compartment. Therefore, in effect, equilibrium exists not just between the dialysate and the plasma but between the dialysate, membrane and the plasma.

Molecular adsorbent recirculating system circuit

The components of the MARS extracorporeal circuit are depicted in Fig. 3. The device is piggybacked onto a conventional veno-venous hemofiltration (CVVH) or hemodialysis (HD) equipment. Two pumps drive the extracorporeal circuit. The first pump pushes the blood from the venous access (generally a double lumen femoral or jugular catheter) through the hollow compartment of the MARS cartridge and is provided by the HD or the CVVH device. The blood flow rates are kept at 150–250 mL/min, depending on the hemodynamic status of the patient. Anticoagulation is provided by heparin, the dosage of which is adjusted to keep the activated clotting time (ACT) between 160 and 190 s. We have used prostacyclin (PGI2) in exceptional cases in which coagulaopathy and/or thrombocytopenia have contraindicated use of heparin. The closed loop albumin circuit, filled with 600 mL of 20% human albumin, is driven by the pump fitted with the MARS monitor (MARS monitor; Teraklin). Conventionally, patients are treated in sessions of 6–8 h. It is possible that the efficacy of toxin removal dwindles with passage of time because the concentration of albumin coating the skin layer on the dialysate side is likely to fall progressively. Patients are treated every day or, more often, on alternate days, depending on the clinical and biochemical profile.

Figure 3.

Schematic representation of the molecular adsorbent recirculating system (MARS) circuit. The patient is connected to the blood circuit containing the MARS membrane. The albumin containing dialysate solution circulates in a closed loop circuit and the albumin is on-line regenerated by passage over the charcoal column and anion exchanger. A dialysis/hemofiltration module is incorporated into the secondary circuit to add all the benefits of transfer of water-soluble toxins (low molecular weight substances), as in conventional hemodialysis/hemofiltration. (Reproduced with permission from Teraklin, Rostock, Germany.)

Results of clinical studies utilizing molecular adsorbent recirculating system

The indications for which MARS has been used in clinical practise have already been mentioned (Table 2). The overall experience spans more than a few thousand treatment sessions in more than 500 patients. The results of MARS therapy can be best assessed by the impact on functioning of individual organ systems.

Effect on hemodynamics

The effect of MARS on systemic hemodynamics is striking. The most impressive changes occur in mean arterial pressure (MAP: an increase of 10–20 mmHg reported in most series.19–21 The increase in MAP is associated with an increase in systemic vascular resistance index (SVRI)20,21 and a decreased cardiac output.20 Interestingly, even patients with ALF tolerate the MARS treatment quite well, with little hemodynamic instability, as is often observed with other extracorporeal devices.22 The basis for this improvement might be effective removal of nitric oxide–albumin adduct (S-nitrosothiol), high circulating levels of which might be in part responsible for the hyperdynamic circulation in liver failure.23 Our initial results also point to a decreased total body nitric oxide production in patients treated with MARS (Fig. 4).

Figure 4.

Total body production of nitric oxide (NO) before and after molecular adsorbent recirculating system (MARS) therapy (single session). For comparison, the NO production in a patient with preascitic cirrhosis is shown. The decrease in NO production is sustained even after cessation of the MARS treatment and is possibly one of the mechanisms of improved systemic hemodynamics after MARS therapy.

Effect on metabolic function

Consistent improvement in biochemical profile has been shown following single and multiple treatment sessions with MARS, with significant decrease in blood urea, creatinine, ammonia, lactate, bilrubin, bile acids, aroamatic amino-acids, short and medium chain fatty acids and copper.17,19,24,25 Hyponatremia, which is often observed in the critically ill patients with liver failure, is also known to be corrected following MARS therapy.19,26 The biochemical improvement is sustained even after completion of MARS therapy. Clinical improvement often lags behind biochemical improvement.

Effect on hepatic function

The data regarding improvement in synthetic function of the liver following MARS have shown that changes may not occur after single treatment sessions (unlike improvement in biochemical profile). In patients responding to MARS therapy, improvements have been observed in the levels of Factor VII, cholinesterase, antithrombin III, albumin and prothrobmin time index.17,27,28 The change in the level of transaminases has been insignificant in patients with AoCLF,20 although in a mixed patient population studied by Awad et al. a significant decrease in ALT was observed.28 Significant improvement in pruritus following MARS therapy may be due to effective removal of bile acids and other pruritogenic compounds by the use of this technique. Incubation of porcine hepatocytes in vitro with plasma samples from patients with AoCLF before and after MARS therapy (single session) has shown that there is improved viability, proliferation and metabolic capability of the cells after treatment.27

Effect on cerebral function

Significant improvement in mental status has been reported after MARS treatment in patients with hepatic encephalopathy in the setting of ALF22,28 and AoCLF.20,27 As mentioned, treatment with MARS shifts the amino acid profile favorably towards branched chain amino acid, resulting in an increase in Fischer's index.24,28 Impressive reductions in grade of encephalopathy,27 intracranial pressure28 and oxygen saturation of jugular venous bulb blood (SjVO2)20 has been reported after MARS. Part of the benefit from MARS may be due to impressive reduction in plasma ammonia levels achieved by treatment. Recently, Schmidt et al. showed an increase in middle cerebral artery mean flow velocity using transcranial Doppler following MARS treatment.21

Effect on renal function

Improvement in renal function has been reported in a number of studies due to the use of this technique.17,19,26 The first randomized controlled trial comparing MARS with hemodiafiltration in patients with type I hepatorenal syndrome (HRS) showed impressive survival advantage with the use of the former.26 Therefore, MARS is arguably the most promising treatment for type I HRS short of liver transplantation.29 Favorable alterations in renal hemodynamics may be one of the reasons for improved outcome in this situation. Indeed, plasma renin activity (PRA) has been shown to decrease following MARS treatment.21

Effect on patient outcome

Although the number of randomized controlled trials comparing MARS with standard medical therapy or other interventions are very few, most studies have shown unequivocal benefit with respect to survival and indicators of severity of liver disease following treatment with MARS (Table 4). Some of these studies have been highlighted in Table 5. The MARS has been used successfully as a bridge while patients wait for transplantation in a number of clinical situations,31 as well as providing liver support in anhepatic patients for up to 70 h.28

Table 4.  Improvement in parameters of severity of liver disease following MARS therapy: data from the University College London Study
Severity criteriaPre-MARSPost-MARSP
  1. Data from seven patients with AoCLF and multiorgan failure treated with a mean of 5.1 sessions of MARS. All values are median (range). Five of the seven patients (70%) were discharged from the hospital.

  2. MARS, molecular adsorbent recirculating system; MELD, Mayo end-stage liver disease; Df, discriminant function; SOFA, sequential organ failure assessment; APACHE II, acute physiology and chronic health evaluation; AoCLF, acute-on-chronic liver failure.

Child–Pugh score 12 (11–14)11 (10–12)< 0.02
MELD 25 (21–32)15 (9–21)<  0.002
Maddrey's Df102 (39–192)57 (18–137)< 0.02
SOFA 13 (10–16)11 (9–14)< 0.002
APACHE II 10 (7–13) 6 (2–9)< 0.002
Table 5.  Outcome data on patients treated with MARS
Ref. no.Patient profileNo. MARS sessionsOutcome
  • ↑increased; ↓decreased; ALF, acute liver failure; AoCLF, acute-on-chronic liver failure; DF, delayed graft function; FHF, fulminant hepatic failure; PNF, primary non-function; HRS, hepatorenal syndrome; HE, hepatic encephalopathy; HDF, hemodiafiltration; MAP, mean arterial pressure; MARS, molecular adsorbent recirculating system; SVRI, systemic vascular resistance index; CO, cardiac output; CBF, cerebral blood flow; Tx, transplantation; PTI, prothrombin time index; AT III, antithrombin III; UCL, University College London; UNOS, United Network for Organ Sharing.

  •  Mixed patient population (ALF + AoCLF).

 Novelli et al.2210 patients with diverse etiologies:
6.4 (1–24)Survival 70%, 2 of the patients required
 Tx. Improved biochemistry and GCS;
 INR improved in patients with FHF
 Awad et al.289 patients: 5 UNOS I,
 73 (12–132) hours
Improved initial outcome in 6 patients;
 3 UNOS I bridged to Tx; in ICP from
 37 to 13 mmHg; ↑in Fischer's ratio and
 Factor VII levels
 Stange et al.1713 patients: 8 UNOS IIA,
 5 UNOS IIB; 8 with HRS
5 (2–14)Survival 70%; improved biochemistry
 and HE. ↑in PTI, AT III,
 Mitzner et al.26RCT: 13 patients with type I HRS;
 8-MARS, 5-HDF
5.2 (2–14)7-day survival: 68% MARS, 0% HDF;
 Improved biochemistry, oliguria, PTI
 and serum sodium
 Stange et al.2726 patients: 16 UNOS IIA,
8 (2–14) in UNOS 2A,
 4 in UNOS 2B
Survival: 100% in UNOS IIA, 44% in
 UNOS IIB. Improved biochemistry
 and PTI, AT III.
 Mitzner et al.198 patients with HRS:
 all UNOS IIA with ascites
5.9 (4–14)Hospital survival: 63%, decrease in CP
 score from 13.2 to 9.4; nil ascites, ↓ in
 HE Gd., 2.8–0.8
 Sorkine et al.208 patients3 (1–8)Survival:63%; ↓ in HE Gd: II/III to
 normal; improved biochemistry, ↑in
 MAP, SVRI, ↓ in CO
 Schmidt et al.218 patients: all with HE1Survival: 50%, HE reversal in 3 patients;
 ↑ in MAP and CBF
 Lamesch et al.3017 patients, 9 with AoCLF4.3 (2–9)Survival: 53%; improved biochemistry;
 all patients weaned off mech.
 UCL data7 patients: 4 with type I HRS,
 3 with type II HRS
5.1 (3–9)3 month survival: 57%; improved
 biochemistry; ↓ in HE Gd: 2.4–0.8

Safety profile

Apart from mild platelet loss, there are no known reported adverse effects of MARS therapy.32 Even the thrombocytopenia is asymptomatic and seldom requires a patient to be weaned off therapy. Unlike some of the other extracorporeal devices, use of MARS is not associated with loss of clotting factors, growth factors, hormone binding proteins or acute phase reactants.18 However, one has to be aware that the concentration of albumin-bound drugs can decrease following MARS treatment and relevant drug levels must be requisitioned for such patients.30


The MARS continues to be one of the most popular extracorporeal devices used for providing liver support. It has been successfully used in a wide range of conditions from ALF to AoCLF. From the cost and logistics point of view, it has an edge over contemporary bio-artificial devices. Moreover, it has an excellent safety profile. However, the paucity of randomized, controlled trials comparing MARS with standard medical therapy or other interventions prevents more liberal use of this technique. Two large multicentric trials are currently in progress in the USA and Europe and results of these are eagerly awaited.


The author gratefully acknowledges the help, support and guidance of Drs Rajiv Jalan and Christian Steiner, Prof. Roger Williams and staff at the intensive care unit of the University College London Hospitals for undertaking patient studies using MARS.