The surgical resection model of ALF was undertaken in 23 pigs weighing 35–40 kg. This modified model  combined 70% resection of the hepatic parenchyma and 20 min total venous hepatic occlusion. Two hours after surgery, 20 animals were randomized into a control group (ALF group) and a group treated by FPSA (FPSA group). The animals in the FPSA group were connected to the elimination device (Prometheus system, Fresenius Medical Care AG, Germany), as described below, for 6 hours of treatment 2 hours after surgery had been completed . In the SHAM group (three animals) a laparotomy alone was performed. Twelve hours after surgery, all animals were euthanized by a lethal dose of thiopental (sodium pentothal, Abbott Laboratories, UK) and 7.45% potassium chloride (Braun). In the FPSA group, two animals died postoperatively as a consequence uncontrollable hemorrhage. In total, 21 animals (10 in the control group, 8 in the FPSA treated group and 3 in the SHAM group) completed the study.
Animals were premedicated with intramuscular ketamine 10 mg/kg (Narkamon, Léčiva, Prague, Czech Republic), atropine 0.01 mg/kg (Atropin Biotika, Hoechst-Biotika, Martin, Slovakia), azaperone 4 mg/kg (Stresnil, Janssen Pharmaceutical, Beerse, Belgium) and medetomidine 25 μg/kg (Domitor, Pfizer, NY, USA) 20 minutes before induction. General anesthesia was induced with intravenous fentanyl 4 μg/kg (Fentanyl Torrex, Torrex Chiesi, Vienna, Austria) and etomidate 0.3 mg/kg (Hypnomidate, Janssen Pharmaceutica); then the trachea was intubated. Positive pressure ventilation was maintained (Siemens Servo 900 C ventilator, Siemens, Elema, Sweden) in volume controlled mode, with a fraction of inspired oxygen of 0.4, positive end expiratory pressure of 5 cm H2O, frequency 16 breaths/min, tidal volume 6–8 mL/kg so as to achieve normocapnia (paCO2 4.6–5.3 kPa). Maintenance anesthesia was provided using isoflurane (Forane, Abbott Laboratories, UK) with an intravenous infusion of 6–10 μg/kg/h fentanyl. Neuromuscular blockade was achieved by means of boluses of 0.02 mg/kg pipecuronium bromide (Arduan, Gedeon Richter, Budapest, Hungary) given throughout surgery. Crystalloid and colloid solutions were used to support the circulation and address hemorrhage; an infusion of norepinephrine (Léčiva) was administered when needed. The protocol for anesthesia and intravascular volume therapy was the same in each experimental group. Augmentin 1.2 g (amoxicillin and clavulanic acid, SmithKline Beecham Pharmaceuticals, London, UK) and famotidine 20 mg (Quamatel, Gedeon Richter) were administered as antibiotic prophylaxis and to suppress gastric acid secretion, respectively. Catheters were inserted into the femoral artery and vein under direct surgical vision to allow connection of the FPSA extracorporeal device, and to allow invasive blood pressure monitoring and collection of blood samples. The internal jugular vein was identified surgically, and a thermodilution pulmonary artery catheter (7F Arrow, PA, USA) was positioned to monitor hemodynamic variables.
After liver resection, mechanical ventilation was changed to pressure controlled mode, to achieve normocapnia with a tidal volume of 6 to 8 mL/kg. The body temperature was maintained between 36.5°C and 38.0°C using a forced-air warming system (WarmAir, CSZ Medical). Subsequent sedation and analgesia were provided by infusions of propofol 6 mg/kg/h (Fresenius Kabi, Bad Homburg, Germany) and fentanyl 2 mg/kg/h. Crystalloid and colloid solutions were given as needed. Blood glucose concentration was maintained above 4.5 mmol/l using a continuous infusion of 40% glucose solution (Braun). The circulation was supported when necessary with a norepinephrine infusion to maintain a mean arterial pressure >65 mmHg.
Microdialysis and ICP monitoring
Microdialysis and ICP monitoring was initiated within the first postoperative hour. Two burr holes were created over the right frontal bone under sterile conditions. The dura mater was punctured, and the microdialysis membrane and ICP sensor were placed separately into the frontal lobe. To prevent leakage of cerebrospinal fluid (CSF), the burr holes were sealed with bone wax. The CMA70 microdialysis catheter (CMA Solna Sweden) consists of a thin, double lumen, 60 mm plastic tube with a 10 mm semipermeable 20 kDa membrane at the tip. After insertion into the brain tissue, the catheter was connected to a microinfusion pump (CMA/106 Microinjection pump; CMA Microdialysis AB, Stockholm, Sweden) and continuously perfused with artificial CSF at a rate of 0.3 μL/min. The dialysate was collected in microvials, each containing approximately 18 μL aliquots for further analysis (CMA 600 Microdialysis Analyzer, CMA Microdialysis). The analyzer measures glucose, lactate, pyruvate and glutamate concentrations in each sample, with >99% recovery when using a buffer of known concentration at the selected perfusion rate [8, 9]. We collected dialysate samples every 60 min for analysis; the first sample was collected during the second postoperative hour. Specimens were frozen at −80°C for later analysis of glutamine concentration using a Perkin-Elmer reverse-phase high pressure liquid chromatography system with fluorescence detection and precolumn o-phthalaldehyde derivatization . ICP monitoring was performed using an intraparenchymal sensor (Codan Microsensor, Johnson and Johnson, USA).
The Prometheus extracorporeal system was primed with 0.9% NaCl, prior to its connection to the venous dialysis catheter in the right femoral vein. The device eliminates water-soluble and protein-bound toxins and breakdown products of metabolism. In the FPSA circuit, venous blood passes through a separator with a pore size of 250 kDa (AlbuFlow, Fresenius). The separated plasma than passes through a neutral resin absorbent column (Prometh01, Fresenius) and an anion exchange resin absorber (Prometh02, Fresenius) to remove albumin-bound toxins. The plasma phase is then returned to the filter and dialyzed as whole blood in a high flow dialyzer (F60S, Fresenius) to remove water-soluble toxins  before being returned to the femoral vein. The treatment was performed with a blood flow of 200 mL/min and a plasma flow of 300 mL/min. Sodium citrate was used as an anticoagulant. The dose of sodium citrate was determined according to serum ionized calcium concentration.
Regression models were constructed separately for each parameter in each group of observations. The models consisted of a polynomial of at most fifths degree with fitted intercepts; terms were calculated using the least squares method. The model was developed by initially including all terms, then the term with the lowest significance was sequentially removed until all terms demonstrated significance on an α = 5% level. The intercept was not removed. The first-degree term was retained in models where none of the terms demonstrated significance. The fits were calculated for the chosen time points, including the standard error, and 95% confidence intervals (95%CI). In addition, analysis of variance (ANOVA) was undertaken using the same model. For each given time point, a two-sample t-test was undertaken to examine the following hypotheses: H0, the difference between fits was not significant; H1, the difference between fits was significant. The t-test was calculated using the following parameters: mean difference (difference between the compared fits); degrees of freedom (degrees of freedom corresponding to the error term in ANOVA, i.e. the number of observations minus the number of terms included in the regression model); and standard deviation (standard error of the fit multiplied by the square root of [degrees of freedom +1]). A p-value (the smallest level to conclude significance) was calculated from the test statistic T. The acceptability of the alternative hypothesis was evaluated on an α = 5% level (p <0.05). The regression model was developed in Q-DAS Destra v.10 (Q-DAS Asset GmbH & Co. KG, Weinheim, Germany). The fits and associated parameters were calculated in Minitab 16.2.2. The t-test was performed and data plotted using Microsoft MS Excel 2010.
Preoperative treatment, surgery and postoperative care were undertaken according to the protection against cruelty to animals, law no. 312/2008 Coll., and the protection, breeding and the use of experimental animals, decree no. 207/2004 Coll. The study was approved by the experts and ethics committee of the Institute of Clinical and Experimental Medicine in Prague, Czech Republic.