Patients with severe HH will be treated with standard medical therapy (i.e. vasopressor support with norepinephrine in refractory hypotension = RR mean < 65 mmHg, positive inotropic support with dobutamine if the central venous oxygen saturation < 70%, renal replacement therapy in case of severe metabolic acidosis and/or renal failure, antibiotic treatment in case of suspected or proven infection, mechanical ventilation in case of severe hypoxemia or hypercapnia and/or GCS <= 8.

20 patients will be allocated by randomization to the MARS arm. Additionally to standard medical therapy they will receive 4 MARS sessions on three consecutive days, MARS® therapy will be applied for at least 12 hours per session. Thereafter, MARS® treatment will be continued if the patient still has increasing aminotransferase levels, requires vasopressor support or suffers from cholestasis (defined as serum bilirubin levels > 5 mg/dL) for 3 sessions again. There will be a maximum of 7 MARS ® sessions per patient.

Molecular adsorbent recirculating system (MARS®) can be used in patients with acute liver failure for bridging to liver transplantation. Studies reported that MARS® therapy improved the hemodynamic situation in patients with acute and acute on chronic liver failure. Several groups observed an increase in arterial pressure, systemic vascular resistance index, a decrease in portal pressure and improvement of renal blood flow. Furthermore, studies demonstrated that MARS therapy reduces ammonia levels and improves hepatic encephalopathy.

The primary endpoint will be the difference of the indocyanine plasma disappearance rate (ICG-PDR) at day 7. Only in case of major differences in baseline values of ICG-PDR we will use the change from baseline to day 7 of ICG-PDR (delta ICG-PDR) as outcome. Assuming normal distribution of ICG-PDR, we will formally test the null-hypothesis of no difference between intervention group and control using an independent sample t-test. The assumption of a normal distribution will be founded on visually inspecting a histogram and using the Shapiro Wilks test for normal distribution. If the assumption of normality does not hold data will be compared using the Mann-Whitney U-test. A p-level of < 0.05 will be considered statistically significant. Furthermore a linear random coefficient model will be used to incorporate daily measurements of ICG-PDR during the course of the study in terms of a repeated measures design.