Extracorporeal liver assist device to exchange albumin and remove endotoxin in acute liver failure: Results of a pivotal pre-clinical study

Background & Aims In acute liver failure, severity of liver injury and clinical progression of disease are in part consequent upon activation of the innate immune system. Endotoxaemia contributes to innate immune system activation and the detoxifying function of albumin, critical to recovery from liver injury, is irreversibly destroyed in acute liver failure. University College London-Liver Dialysis Device is a novel artificial extracorporeal liver assist device, which is used with albumin infusion, to achieve removal and replacement of dysfunctional albumin and reduction in endotoxaemia. We aimed to test the effect of this device on survival in a pig model of acetaminophen-induced acute liver failure. Methods Pigs were randomised to three groups: Acetaminophen plus University College London-Liver Dialysis Device (n = 9); Acetaminophen plus Control Device (n = 7); and Control plus Control Device (n = 4). Device treatment was initiated two h after onset of irreversible acute liver failure. Results The Liver Dialysis Device resulted in 67% reduced risk of death in acetaminophen-induced acute liver failure compared to Control Device (hazard ratio = 0.33, p = 0.0439). This was associated with 27% decrease in circulating irreversibly oxidised human non-mercaptalbumin-2 throughout treatment (p = 0.046); 54% reduction in overall severity of endotoxaemia (p = 0.024); delay in development of vasoplegia and acute lung injury; and delay in systemic activation of the TLR4 signalling pathway. Liver Dialysis Device-associated adverse clinical effects were not seen. Conclusions The survival benefit and lack of adverse effects would support clinical trials of University College London-Liver Dialysis Device in acute liver failure patients.

Critical care management of the porcine acetaminophen-induced model of acute liver failure has been previously published [1]. The following is a summary of the critical care protocols for ventilation; intravenous fluid therapy; maintenance of acidbase status, electrolyte balance, and normoglycaemia; and cardiovascular support.

Ventilation
Pressure-controlled intermittent positive pressure ventilation with a circle breathing system and a mixture of oxygen and medical air was used throughout. Gas flow rate was maintained throughout at 20-22 ml/kg/min. At beginning of anaesthesia, ratio of oxygen to medical air was set to maintain percentage of inspired oxygen at 30-36%, positive end expiratory pressure was 5 cmH2O, inspired pressure was 12-15 cmH2O and respiratory rate was 10-13 breaths per minute. Three consecutive sighs were introduced manually every one to two hours. Ventilator controls were adjusted to maintain partial pressure of carbon dioxide in arterial blood between 35-45 mmHg, partial pressure of oxygen in arterial blood between 100-120 mmHg and tidal volume between 6-10 ml/kg.

Intravenous fluid therapy and correction of electrolyte and acid-base abnormalities
Intravenous fluid therapy with crystalloid solutions was administered at a constant rate of 5ml/kg/hr throughout. Crystalloid solutions used were compound sodium 4 lactate from induction of anaesthesia; 0.9% sodium chloride from 12 hours after onset of acetaminophen dosing; and 0.18% sodium chloride with 4% glucose if arterial blood sodium concentrations exceeded 145 mmol/l and continued to increase towards 150 mmol/l. 6% hydroxyethyl starch ('Voluven 6%', Fresenius Kabi Ltd, Cheshire, UK) was administered intravenously at 1 ml/kg/hr from onset of acetaminophen dosing. Rate was then adjusted to maintain mean arterial blood pressure as detailed below.
1 ml/kg/hr of a parenteral nutrition solution containing 10% glucose and 5.9% amino acids ('Vamin 9 Glucose', Fresenius Kabi Ltd, Cheshire, UK) was administered intravenously from onset of acetaminophen dosing until study end. Further intervention to maintain normoglycaemia was not required.
Potassium chloride was added to intravenous crystalloid solutions, as required, if arterial blood potassium concentrations fell below 3.5 mmol/l. Potassium infusion was stopped if arterial blood potassium exceeded 3.5 mmol/l and continued to increase towards 5 mmol/l. 8.4 % Sodium bicarbonate was given by intravenous infusion to correct metabolic acidosis.
10 % calcium gluconate was given slowly to effect by intravenous infusion if arterial blood calcium fell below 1.1 mmol/l.

Cardiovascular support
Direct mean arterial blood pressure (MAP) was monitored continuously with the aim to maintain MAP at >70 mmHg for the first 12 hours following onset of acetaminophen dosing, then >60 mmHg until onset of irreversible acute liver failure (ALF), then >50 mmHg until study end. If MAP fell below these targets three possible actions were taken. First, a 5ml/kg intravenous bolus of 6% hydroxyethyl starch was given over 15 minutes. This was repeated up to 3 times as long as central venous pressure (CVP) remained <12 mmHg. Following achievement of MAP targets, continuous rate infusion of 6% hydroxyethyl starch was increased by 1 ml/kg/hr. Second, intravenous noradrenaline infusion was initiated at 0.1 μg/kg/min and increased by 0.1 μg/kg/min every 15 minutes until MAP targets were achieved.
Third, intravenous terlipressin infusion was initiated at 2 mg/24 hours when noradrenaline exceeded 0.5 μg/kg/min and further increased to 4 mg/24 hours when noradrenaline exceeded 1.0 μg/kg/min. If following irreversible ALF, MAP >50 mmHg was maintained for 1 h, reduction in 6% hydroxyethyl starch infusion was attempted if CVP was >9 mmHg. However if CVP was <9 mmHg, decrease in noradrenaline infusion was attempted. Both reductions were attempted whilst maintaining MAP >50 mmHg.

University College London-Liver Dialysis Device (UCL-LDD): technical details
UCL-LDD is a novel artificial extracorporeal liver support device including two haemofilters, Filter 1 and Filter 2. All components of the device were obtained from Gambro Dialysatoren GmbH, Rostock, Germany and used following the 6 recommendations of this company. A schematic of the device is shown in Fig. 1.
Filter 1 was 'SepteX TM ', a haemofilter with a polyarylethersulfone 'high cut-off' membrane (effective surface area 1.1 m 2 , pore diameter ~10 nm, membrane thickness 50 µm, internal diameter 215 µm). SepteX TM was included to achieve extraction of albumin, along with bound toxins, by haemofiltration [2]. Treatment with both UCL-LDD and CD was via a femoral haemodialysis catheter.
'PrimaSol BGK 2/0' was used for the replacement fluid and dialysate. Blood flow rate was set at 150ml/min, replacement (pre blood pump) volume at 1 l/h and dialysate rate at 35ml/kg/h, with zero net fluid removal. The circuit was anticoagulated continuously with 1000 iu heparin/h.

Aims
To determine the efficacy of albumin removal and endotoxin adsorption from pigs with acetaminophen (APAP)-induced acute liver failure (ALF) by the high cut-off filter, SepteX TM (Gambro Dialysatoren GmbH, Rostock, Germany) and the endotoxin adsorption filter, OXiris TM (Gambro Dialysatoren GmbH, Rostock, Germany) in University College London-Liver Dialysis Device (UCL-LDD).

Methods
The efficacy of albumin removal and endotoxin adsorption from pigs with APAPinduced ALF by the SepteX TM and OXiris TM haemofilters within UCL-LDD, was assessed by exchanging and thus comparing these filters with a standard continuous renal replacement filter, HF1000 TM (Gambro Dialysatoren GmbH, Rostock, Germany). Twelve pigs were induced to ALF with APAP and divided into four treatment groups of n=3. Each group underwent extracorporeal device treatment with a combination of two haemofilters beginning 2 h after onset of ALF.
-Group 1 was treated with SepteX TM plus OXiris TM that is UCL-LDD -Group 2 was treated with SepteX TM plus HF1000 TM -Group 3 was treated with HF1000 TM plus OXiris TM -Group 4 was treated with HF1000 TM plus HF1000 TM that is the Control Device Endotoxin and albumin concentrations were compared between treatment groups and over time using a linear mixed effects model. Analysis was carried out using IBM SPSS version 20 (New York, USA) and significance was set at the 5% level.

Results
Measured plasma endotoxin concentrations are illustrated in Fig. 2 for the four treatment groups. In all four groups there was a significant 362 ± 39 % increase in 9 plasma endotoxin from 2.6 ± 0.7 EU/ml at baseline to 11.6 ± 3.8 EU/ml at ALF (p < 0.001). In the CD and SepteX TM plus HF1000 TM groups, endotoxin concentrations continued to rise significantly through the 6 h treatment period by 98 ± 50 % (p < 0.001) and 81 ± 7 % (p = 0.004) respectively. However in the UCL-LDD and the HF1000 TM plus OXiris groups there was no significant increase in endotoxin concentrations after initiation of treatment.
Effluent albumin concentrations were significantly higher with SepteX TM than without SepteX TM (p = 0.032). Without SepteX TM effluent albumin concentrations at 2 and 6 h of treatment were 0.42 ± 0.21 g/l and 0.68 ± 0.14 g/l respectively. With SepteX TM effluent albumin concentrations at 2 and 6 h of treatment were 1.22 ± 0.10 g/l and 1.20 ± 0.15 g/l respectively. Effluent albumin concentrations were not significantly affected by time of treatment. With an effluent volume of 2 l/h, over a 6 h treatment period, total albumin extraction with SepteX TM was 2.42 ± 0.14 g/h.

Conclusion
In pigs with APAP-induced ALF, treatment with UCL-LDD achieves albumin extraction at a rate of 2.42 ± 0.14 g/h due to the presence of the SepteX TM filter and prevents further increase in plasma endotoxin concentrations due to the presence of the OXiris TM filter.
Binding of IL-18 and IL-1β to IL-18 receptor/ IL-18 receptor accessory protein and type 1 IL-1 receptor/Il-1 receptor accessory protein respectively activates NF-κB and AP-1 pathways resulting in SEAP expression. Assay details were as for TLR4 reporter assays with the following exceptions. Cells were seeded at an initial density of ≈50 000 cells/well and incubated for 24 h prior to addition of 20 µl plasma samples. Plasma from normal pigs after the initial surgical set up required for the porcine acute liver failure model were used as positive controls, as IL-18 has previously been shown to be elevated in acutely stressed pigs [3,4]. SEAP activity was detected by addition of 40 µl supernatant to 160 µl alkaline phosphatase detection medium.

Liver histopathology results
Post mortem examination of liver specimens from pigs included in the main study to assess the efficacy of treatment with University College London-Liver Dialysis Device (UCL-LDD) compared to a Control Device (CD) in pigs with acetaminopheninduced acute liver failure (ALF) was performed. Acute centrilobular to midzonal hepatocyte degeneration and necrosis was confirmed in all ALF pigs irrespective of whether they were treated with UCL-LDD (APAP-UCL-LDD group) or CD (APAP-CD group), but not in any of the Control pigs (sham induction to ALF with water) treated with CD (Control-CD group) (Fig. 3). Necrosis was graded as mild, moderate or severe according to the degree and percentage of parenchyma affected. Median 12 grade for both APAP-UCL-LDD and APAP-CD groups was moderate with 4 out of 9 APAP-UCL-LDD pigs and 3 out of 7 APAP-CD having severe necrosis. There was no significant difference in severity of necrosis between APAP-UCL-LDD and APAP-CD.

Results for Il-18/Il-1β reporter cell assays
No significant IL-18/IL-1β bioactivity was detected in any of the plasma samples obtained after onset of acute liver failure from pigs treated with either University College London-Liver Dialysis Device (APAP-UCL-LDD group) or Control Device (APAP-CD group) when compared to normal control pigs treated with Control Device (Control-CD group) (Fig. 4).