Background
Acute liver failure (ALF) is defined as rapid and progressive development of severe acute liver injury with impaired liver synthetic function without a previous history of liver disease. Worsening encephalopathy and cerebral edema escalating in brain stem herniation [
1] as well as infection and inflammation with development of systemic inflammatory response syndrome (SIRS) frequently contributing to multiple-organ failure in this setting [
2] are the main courses of death. The complexity of metabolic abnormalities resulting from ALF is still incompletely understood hence morbidity and mortality among patients with ALF without liver transplantation remains as high as 85% [
3].
Animal models of ALF remain the mainstay of research in development of more efficient artificial [
4,
5] or bioartificial [
6,
7] liver assist devices as well as preclinical trials in newly developed therapeutic approaches. Animal models of ALF are mainly based on surgical techniques such as devascularisation [
8], ischemia [
9], extensive/total liver resection [
10,
11] or medical interventions such as hepatic intoxication with acetaminophen [
12], amanitin [
13] or galactosamine [
14]. Unfortunately all aforementioned models involve various limitations, thus affecting morbidity in the assessment of a given intervention. Although anhepatic models have been criticized for their irreversibility and lack of circulating products of cell necrosis or inflammatory mediators through the native liver, it is commonly considered to serve as a suitable in vivo model for testing the effectiveness of liver assist devices due to high reproducibility and the total absence of remaining functional liver parenchyma [
15‐
17]. But surprisingly the hereby demonstrated survival benefit of the already established devices could not be reproduced in clinical practise. Considering this fact we decided to revaluate a simple dummy device to verify if experimentally demonstrated survival benefits are properly controlled for side-effects.
Preliminary pigs studies have been carried out to establish a highly reproducible model of total hepatectomy [
11] delivering long-term survival under standardized intensive care conditions [
18]. It has been demonstrated experimentally and clinically that moderate hypothermia is organ protective in specific life threatening conditions such as cardiac arrest [
19], traumatic brain injury [
20] or ALF [
21]. It is also well known that the use of extracorporeal circuits like hemofiltration, plasmapheresis or liver assist devices result in moderate peripheral cooling of the connected person. Although recent ALF animal models have certainly been controlled for hypothermia [
22], various animal studies evaluating liver assist devices or other therapeutical interventions did not report the exact courses of body temperature while connection to the devices [
15,
23] or did not even test their extracorporeal devices against a dummy device group [
23,
24]. Assuming that these studies might not be properly controlled for extracorporeal device associated transient hypothermia, we performed the present study to analyze a dummy device effects on hemodynamics, body temperature and survival.
Methods
Animal model of ALF
After approval by the institutional review board for animal experiments, ten female German Landrace pigs weighing 36 ± 4 kg underwent total hepatectomy following frontoparietal trepanation for intracranial pressure (ICP) monitoring. All experiments were performed according to the international principles governing research on animals and under the supervision of a veterinarian, who set the guidelines for minimizing pigs suffering.
Anesthesia
Intramuscular premedication was administered using atropine 0.1% (0.05 mg/kg), ketamine (7 mg/kg), azaperone (10 mg/kg) and diazepam (1 mg/kg). A core body temperature of 38 ± 0.5°C measured by a rectal probe was aimed by using warming blankets throughout the experiment. A stomach tube (Argyle™, Tyco Healthcare, Tullamore, Ireland) was placed for intestinal drainage. After oral intubation with a cuffed endotracheal tube (Lo-Contour™ Magill, Mallinckrodt Medical, Athlone, Ireland) the pigs were ventilated with pressure-controlled ventilation modus (Galileo Gold, Hamilton Medical, Rhaezuens, Switzerland). Arterial blood gas analysis (ABL 625, Radiometer Copenhagen, Denmark) including blood lactate measurement was performed hourly and ventilation was adjusted accordingly. Continuous infusion of ketamine (15 mg/kg/h), fentanyl (0.02 mg/kg/h) and midazolam (0.9 mg/kg/h) was administered to maintain deep anesthesia throughout the experiment. Character of respiration, heart rate, eye movement and pain stimulus was used to confirm depth of anesthesia; if any of these parameters indicated a lessening of anesthesia, infusion rates of anaesthetic agents were increased.
Surgical procedure and randomization
Animals were kept under standard laboratory conditions and fasted overnight before surgery. They received 2 g ceftriaxon (Rocephin
® , Hoffmann-La Roche, Basel, Switzerland) prior to surgery. The superior vena cava through the jugular veins (Multi-Lumen Central Venous Catheter, Arrow International, Reading, PA, USA) and the internal carotid artery (Leadercath, Vygon, Écouen, France) were instrumented to measure mean arterial pressure (MAP) and central venous pressure. Following frontoparietal cranial trepanation, a probe was inserted in the frontal brain parenchyma to measure ICP (Camino
® MPM-1 monitor, Integra Neurosciences, Plainsboro, NJ, USA). The abdominal cavity was entered through a midline incision and a urinary catheter (Gentle-Flo™, Tyco Healthcare, Tullamore, Ireland) was placed by cystostomy. Total hepatectomy was performed according to a recently published technique [
11] with a Y-graft vascular prosthesis (Uni-Graft
® K DV, ITV, Denkendorf, Germany) interposition. Intraoperative blood loss was substituted with porcine erythrocyte and fresh-frozen plasma units. After stabilisation of the hemodynamic situation pigs were randomized in two groups receiving standardized intensive care therapy alone (n = 5) or additional periodic connection to the extracorporeal dummy device system (n = 5).
Cyclic connection of the pigs to the extracorporeal dummy device system started approximately twelve hours after hepatectomy for a time period of twelve hours. This connection cycles were continued every twelve hours for twelve hours. To simulate the experimental setting of an artificial or bioartificial liver assist device, pigs underwent plasma separation (TPE, Prismaflex, Gambro, Hechingen, Germany) with a mean blood flow rate of 120 mL/min and a plasma separation rate of 20 mL/min using a plasma filter (TPE 2000, Gambro, Hechingen, Germany) with a maximal pore size of 0.5 μm. The total extracorporeal volume of the device system was 125 mL ± 10%. The device system including the plasma filter was washed and primed according to the manufacturer's instruction. Heparin (250 U/h) was administered as necessary to avoid clotting in the extracorporeal circuit of the dummy device. The separated plasma fraction was completely returned to the animal via the implanted central vein catheter without any further plasma filtration/detoxification or plasma exchange/replacement.
Preparation of donor fresh-frozen plasma and erythrocyte units
Blood was collected in blood bag systems (500 mL, Compoflex® Fresenius HemoCare, Bad Homburg, Germany) and centrifuged at 2,500 g for 20 minutes (Heraeus Cryofuge 5500i, Thermo Electron Corporation, Langenselbold, Germany). Plasma fraction was pressed into separate bags and shock-frozen at minus -80°C. Erythrocytes were conserved with 100 mL SAG-Mannitol and stored at 4°C.
Standardized intensive care therapy
Animals remained in general anesthesia receiving pressure-controlled ventilation until conclusion of the study protocol (15-30 breaths/minute, tidal volume 6-12 mL/kg and oxygen concentration 0.3-1.0, depending on oxygenation). Monitoring throughout the experiment included electrocardiogram, ICP, MAP, central venous pressure, oxygen saturation and core body temperature. Urinary output, haemoglobin, lactate, serum electrolytes, acid-base balance, blood gase analysis and blood glucose levels were monitored hourly and immediately corrected as required. The selected laboratory parameters like prothrombin time, albumin, plasma protein, creatinine, bilirubin and ammonia were measured before, after and every eight hours after hepatectomy until death. All blood samples were obtained from the arterial carotid catheter. Norepinephrine, in combination with fresh-frozen plasma, hydroxyethylstarch 6% (Voluven
® HES 130/0.4, Fresenius, Bad Homburg, Germany) and sodium chloride solution 0.9% were used to ensure hemodynamic stability. Algorithms of fluid management, ventilation and intensive care medication have already been reported in detail [
18]. Blood glucose levels were maintained > 100 mg/dL with glucose 20% solution. Packed erythrocyte units were given if haemoglobin levels trend to decline below 6 g/dL. To prevent spontaneous bleeding eight fresh-frozen plasma units were given within 24 hours. Pigs received furosemide (maximum 1,000 mg/d) to maintain diuresis as long as possible. Antibiotic prophylaxis with 2 g ceftriaxon was continued daily. Death was defined as a decline of MAP below 30 mmHg under maximal vasopressor support.
Laboratory analysis
All biochemical parameters as prothrombin time, creatinine, albumin, bilirubin, ammonia, total plasma protein and blood count were measured by the certified laboratories of the University Hospital Tuebingen (Zentrallabor, Innere Medizin IV, University Hospital Tuebingen, Germany). Sample analysis was conducted within 1 hour of collection at each time point.
Endotoxin was measured by a Limulus-Amoebozyte-Lysate assay (Charles River Endosafe, Charleston, SC, USA) and performed according to the manufacturer's instruction. In brief, all endotoxin samples were obtained from the arterial catheter and collected in sterile pyrogen-free vacuum tubes (Endo Tube ET, Chromogenics AB, Moelndal, Sweden). Subsequently probes were centrifuged at 3,000 g for 10 minutes. The supernatant was immediately transferred into biopur-grade reaction tubes (Eppendorf AG, Hamburg, Germany) and stored at -80°C.
After thawing, samples were heat inactivated at 80°C for 10 minutes. Samples underwent cooling for 1 minute at 0°C and ultrasonic bathing for 3 minutes followed by centrifugation at 10,000 g for 60 minutes at 4°C. Supernatants were transferred in micro plates (Microtest 96, Becton Dickinson, Franklin Lakes, NJ, USA) and measured kinetically with the (Endosafe® Endochrome-K™, Charles River Endosafe, Charleston, SC, USA). Spike and recovery assays were performed.
Statistical Analysis
Mean values of the selected variables determined before, during and after hepatectomy were compared by t-Test, (JMP® 4.0, SAS Institute, Cary, NC, USA). A p value < 0.05 was considered significant. Results are reported as mean ± standard deviation (SD). Figures are given as mean ± standard error of mean (SEM) of a minimum of two animals per study group.
Discussion
Based on preliminary pig studies with a model of total hepatectomy [
11] and standardized intensive care therapy [
18], we performed the presented study to evaluate a dummy device effects on hemodynamics, body temperature and survival.
Although the used dummy device system was not created as a therapeutic approach, a significant survival benefit of more than 20 hours was noticed. The courses and values of the selected laboratory parameters ammonia, lactate, creatinine or endotoxin which are appropriate to verify different aspects of detoxification were analysed before and after connection to the dummy device. As it was expected, no significant changes of these parameters were observed during or after dummy device connections. Therefore it was demonstrated that these parameters were neither improved nor changed by cyclic application of the dummy device. Our observations were confirmed by the recently published animal study of Ho et al. [
25] analysing selective plasma filtration in a porcine model of ALF. A nearly identical experimental device setting (plasma separation with return of the undetoxified plasma) served as a control group in which no detoxification aspect was noticed and survival remained comparable to the sham (diseased) control group.
Body temperature of all pigs was aimed to maintain within a range of 38 ± 0.5°C by warming blankets, but this parameter was significantly affected by the cyclic connection to the extracorporeal device. Moderate hypothermia has already been established as a method for organ and brain protection by reducing oxygen metabolism in severe neurosurgical traumas or systemic disorders such as ALF [
26]. Surprisingly the intracranial pressure could not be decreased in our experimental setting. This phenomenon might be associated to the fact that the cyclic hypothermia of 1-1.5°C was not effective for decreasing ICP.
The study might be criticized for the absence of an additionally warmed dummy device group with body temperatures corresponding to those of the presented control group or an additionally cooled control group. But the presented study was not designed to confirm the already experimentally [
27] and clinically [
28] proven merits of therapeutic hypothermia by cooling down the animals to a predefined temperature. Our objective was to analyse the clinical effects of a dummy device to verify their potential impact on survival. Therefore we could demonstrate that even slight episodes of hypothermia which remained considerably above the required interval for a therapeutical hypothermia do significantly enhance hemodynamics and consecutively anhepatic survival.
Although antibiotic prophylaxis was administered daily, a significant rise of endotoxin levels, paralleled by a leukocyte drop (data not shown) occurred at time of death in 60% of the dummy device group animals surviving beyond 60 hours. This observation confirmed the fact that relevant inflammation mediators were not eliminated by binding to the plasma separation filter. The onset of sepsis destabilized the only just compensated circulation of the organism resulting in a sudden uncontrollable circulation failure, contrary to the more progressive course in multiple-organ failure. It therefore marks definitively an endpoint of anhepatic survival which can hardly be prolonged further. Theoretically this phenomenon could also be associated with hypothermia [
29], but it is hardly conceivable that cyclic hypothermia of 1-1.5°C causes a relevant destabilisation of the immune system within this experimental setting.
Standardized intensive care management prolonged anhepatic survival. In combination with slight hypothermia it could be extended up to 88 hours. Therefore the anhepatic porcine model provides a valuable tool for screening functionality and safety of liver support technologies but the efficacy of liver assist device should not be assessed by an anhepatic survival benefit. It should be mentioned that as much as you prolong anhepatic survival the more side-effects like potential transient hypothermia or septical complications will disturb the reproducibility of the model. Significant elimination of hepatotoxic substances or relevant detoxification resulting in increased hemodynamic stability could be better surrogate parameters to verify the effectiveness of any therapeutical approach.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KT, MS, AE, TS, MM, CT carried out the studies. KT, MS, CT designed the study and coordinated the study group. KT, MS, MM, CT drafted the manuscript. AK helped to draft the manuscript and participated in its design. AE, TS carried out the biochemical analysis and helped to draft the manuscript. MS performed the statistical analysis. All authors read and approved the final manuscript.