Abstract
Existing liver perfusion models are largely limited by high degrees of ischemic and reperfusion injury and the lack of standardization. To establish a highly standardized perfusion model and minimize reperfusion injury, a porcine liver perfusion model was developed using an artificial heart pump (Buecherl Artificial Heart). This model is characterized by pneumatically driven and pressure controlled blood pumps with pulsating flow characteristics. The perfusion parameters and the integrity of the perfused organ were assessed using hemodynamic and hepatic function tests. In eight porcine liver perfusion experiments the system allowed maintaining stable and physiologic organ function over 3 hours by bile production (5.5 ±3.1 ml/30 minutes, resp. 22.9 ±8.4 ml cumulative at 180 minutes), oxygen consumption (2.2 ±0.2 ml/min/100 g overall mean) and significantly better liver enzyme levels (AST 19.5 ± 10.1 U/l/100 g, ALT 2.1 ± 0.8 U/l/min, LDH 57.8 ± 24.2 U/l/100 g) compared to previous studies. It was also possible to reduce the circulating blood volume to 1,000 ml and to create a compact perfusion system that is adoptable to other organ systems such as the kidneys. The compact size and the absence of magnetic components also allow a use for advanced imaging techniques. In conclusion this optimized perfusion system provides a sound basis for future studies in the area of hepatotoxicity and pharmacology.
Introduction
Drug discovery and development processes consists of a series of steps that start with the demonstration of pharmacological effects in experimental cell and animal models and end with drug safety and efficacy studies. A main limitation of many newly discovered compounds is often an unacceptably high level of toxicity with the liver as the primary target organ. Therefore, approaches to study hepatic toxicity in the early phase of drug discovery represent an important step towards rational drug development. Variety of in vitro liver models have been developed in the past years. Next to their role in toxicology, they may also be used in transplantation medicine. In this respect, the establishment of liver transplantation as terminal treatment of severe liver diseases (Boillot et al., 2002; Grewal, 2002) led to a variety of new aspects in basic hepatological research. As the donor livers undergo warm and cold ischemia due to the surgical process of explantation, the investigation of the ischemia- and reperfusion-related injuries and consecutive development of reperfusion injury is most important for improving graft survival (St. Peter et al., 2002).
Animal experiments but also hemoperfusion of isolated organs are suitable techniques to assess the preservation process or kinetics of pharmacological compounds. The technical development of this method in recent years made it possible to keep an explanted liver alive in a defined environment over several hours. It has been shown that this technology offers many advantages towards other experimental in vitro models such as precision cut liver slices or hepatocyte cell cultures (Groneberg et al., 2002; Sadri, 2002). As we could report earlier while following a multiorgan harvesting protocol and using slaughtered pigs this experimental setup contributes to the 3-R concept in animal experimentation by reducing animal numbers for research purposes (Grosse-Siestrup et al., 2002a). However, presently established setups of isolated perfused livers are mostly performed using rat organs (Brouwer and Thurman, 1996). Other organ perfusions are done as direct bedside-treatment of liver failure patients (Schon et al., 1997).
The studies performed in human and rodents demonstrated many similarities in physiological components and functions, but also considerable differences. When attempting to assess reperfusion-related injuries, an animal species must be chosen that serves as an appropriate model for the possible human situation. In this respect, the pig liver proved to be anatomically and physiologically closer to the human liver when compared to rodent livers (Vilei et al., 2001). Not only organ dimensions are similar in pigs and humans but also the bile composition of the pig resembles human bile, as described in a comparative study between these 2 species (Kobayashi et al., 1998).
In previous studies performed at our laboratory, we reported the use of commonly used roller pumps, which produce a linear flow (Dittrich et al., 2004; Fehrenberg et al., 2004; Grosse-Siestrup et al., 2003; Hochel et al., 2003) or of tube valve pumps to produce pulsating flow characteristics in the organs (Grosse-Siestrup et al., 2001, 2002b). The tube valve pumps showed, however, considerable limitations like the lack of pressure control, volume control only by frequency. Therefore, the goal of the present study was to optimize the isolated liver perfusion system by using pneumatically driven artificial heart pumps which allow pressure control. These pumps proved in the past to be safe and powerful in extracorporeal perfusion of cardiothoracic surgery patients. We believe that these pumps substantially improve the biological features of the normothermic hemoperfusion process of the porcine liver and other organs.
Materials and Methods
Animals
White German landrace female pigs (age 6 months; weight 50 ± 5 kg were used with approval of the official veterinarian institutions. Animals used in this study received humane care and study protocols complied with the guidelines for the care and use of laboratory animals of German authorities. The animals were sacrificed by final exsanguinations during the blood harvesting procedure.
Organ Harvesting
A total of 8 liver perfusion experiments were done. All organs were harvested under general anesthesia following a standardized procedure. After preparation of the hepatic vessels the animals were heparinised and after 5 minutes, blood harvesting started. Portal vein and hepatic artery were cannulated and gravity flushed with 4°C Euro Collins preservation solution (2500 ml and 500 ml). Warm ischemia was 6.3 ± 2.8 minutes (mean ± SEM). Common bile duct and caudal vena cava were cannulated. Suprahepatic vena cava was ligated. During a 1-hour period of cold ischemia organs were stored at 4°C under sterile conditions until warm rinsing.
Blood Harvesting
Prior to bleeding systemic anticoagulation was done with heparin (15.000 IU, Liquemin N, Roche, Grenzach-Wyhlen). Collection of autologous blood was performed by cannulation of the cervical vessels on 1 side. Blood harvesting was conducted until 1.5 liter of blood were collected or until the regular heart function was lost. Then, the hepatic vessels were opened and the animal died. Again the sterile-collected blood was anticoagulated with heparin (15.000 U/l).
Perfusion System
The system consisted of a blood and a dialysis circuit (Figure 1) both connected through 3 parallel fiber dialysis modules (F7, Fresenius Medical Care). Blood was pumped from the blood reservoir through the dialysis modules by a 30 ml “artificial heart” pump (BERLIN HEART AG, Berlin, Germany), driven by a pneumatic HEIMES driving unit (989 M booster, Dr. Heimes GmbH, Aachen, Germany). After dialysis and oxygenation the main portion of the blood stream was directed into the portal vein. The pressure was boostered by a second smaller pump (10 ml) for pressure optimized arterial blood supply via A. hepatica. Venous blood (Vena Cava) was drained back into the reservoir and a complete blood circuit was established. Before perfusion the circuit is filled with 1 liter of autologous blood. The portal vein flow was measured 3 of 8 experiments.
The dialysis circuit was filled with 7500 ml of dialysate solution (Table 1). This solution was pumped with 1500 ml/min through the dialysis modules and back to the dialysate reservoir by a mono-head roller pump (Type 10-00-00, Stöckert, München, Germany). The dialysate solution was warmed up by a water bath connected to a heat exchanger and permanently enriched with 97.5% O2 and 2.5% CO2 by an oxygenating module (Lilliput 1, Dideco, Mirandola, Italy) resulting in a 100% oxygen saturation in the blood.
A heating water bath kept the temperature of the blood heat exchanger module and the organ chamber at 38.5°C. The organ chamber consisted of a clear PMMA shell mimicking the diaphragmatic curvature with the water bath underneath. The livers were wrapped in sterile plastic bags inside the chamber with the hilus upside. Residual extravasation of the liver hilus with its blood and lymphatic vessels was drained back into the blood circuit.
Perfusion Procedure
Before organ connection the dialysate and the shortcut organ perfusion circuit were filled with dialysate and saline, respectively. The liver was placed in the warmed (38.5°C) organ chamber connected to the perfusion circuit and initially rinsed with a 37°C oxygenated saline solution (3 L). The warm rinsing solution was replaced by autologous blood and reperfusion started. For pressure controlled perfusion the mean pressure value was adjusted to 8 mmHg in the portal vein and 80 mmHg in the hepatic artery. During perfusion blood, bile and dialysate samples were taken every 30 minutes over a 3-hour period and blood gas analysis was performed. In parallel blood samples were taken for clinical chemical analysis.
Liver Radiography
In order to evaluate the liver perfusion in this model, X-rays were taken initially and at the end of the perfusion process. For this purpose a bolus injection of 50 ml and 10 ml Ultravist (Schering, Berlin, Germany) was given in the portal vein and the hepatic artery respectively and a radiograph was taken immediately (Siremobil Iso C, Siemens, Germany, resolution 470,000 pixels).
Liver Histology
Histology was carried out on formalin-fixed liver specimen. Then 5-μm sections were cut from paraffin-embedded tissues and mounted on glass slides. Routine histology was then carried out using hematoxylin & eosin (H&E) staining.
Formulas Used
Oxygen consumption was calculated using the equation:
where Ah_tHb = total Hemoglobin A. hep. [g/dl], Ah_sO2 = Oxygen saturation A. hep. [%], Vc_sO2 = Oxygen saturation V. cava [%], Ah_pO2 = oxygen pressure A. hep. [mmHg], Vc_pO2 = oxygen pressure V. cava [mmHg], QB_Ah = blood flow A. hep. [ml/min], QB_Vp = blood flow V. port. [ml/min]. Hepatic artery resistance was calculated dividing the mean arterial pressure by the mean arterial flow.
Statistical Analysis
The data was assembled in Microsoft EXCEL and then transferred into an adequate statistical program for further analyses (GraphPad Prism, San Diego, USA). Mean, standard deviation, standard error of the mean, median, maximum, minimum, 1st and 3rd Quartile (Q1 is the first 25% of the data and Q3 is the first 75% of the data) were used to describe the results.
Results
Liver Weight
The organ weight was measured before and after perfusion and the data are presented in Table 2. After organ preservation, liver weight ranged from 1072 g to 1665 g (mean value at 1277 g). After perfusion a weight gain of 14 ± 8.9% was observed. No correlation between weight gain and any other parameter measured in this study was detected.
Hemodynamics
Adjusting the HEIMES driving unit to a mean pressure of about 80 mmHg in the hepatic artery and approximately 8 mmHg in portal vein resulted in a stable arterial blood flow of about 18 ml/min 100 g−1 for the hepatic artery and of approximately 42 ml/min 100 g−1 for the portal vein (Table 2). The blood flow showed pulsating flow characteristics throughout the experiments as illustrated in Figure 2. A slight loss of pressure caused in the dialysis modules occurred as expected.
After 60 minutes of reperfusion a steady state of the hepatic artery resistance was observed for all perfused livers (Figure 3). After an initial decrease this parameter remained constant throughout perfusion with a median value at approximately 0.4 mmHg/ml/min 100 g−1. Except for 1 liver, this parameter did not exceed 0.6 mmHg/ml/min 100 g−1.
Oxygen Consumption, Blood Electrolytes and Blood Parameters
A comparison of oxygen consumption in 3 livers showed a slight decrease between 30 minutes and 60 minutes of perfusion but remained constant until the end of perfusion (Table 2).
The values for sodium and potassium are described in Table 3. Assessment of blood electrolyte levels refers to Plonait and Bichhardt (1988). The levels of sodium in general and in between arterial and venous blood did not notably vary during the course of the experiments. A slight increase in the potassium concentration of about 1 mmol/L was noticeable, but remained within the normal range for the pig (2.3–6.3. mmol/L).
Hemoglobin ranged from about 5 to 9 g/dl between different organs depending on initial hemoglobin values in blood and were constant throughout the perfusion period (Table 3). The PCV remained constant during perfusion at 20 ± 5%.
As an indicator for hemolysis the level of free hemoglobin (fHb) was measured in arterial blood. A slowly but constant increase of fHb 4-fold of initial values was observed in all perfused livers depending on the initial fHb value.
Bile Production
The results for the bile production are summarized in Table 3. After an initial adaptation period of about 30 minutes, the livers started bile production with a constant increase for all livers during the entire experimental period. The amount of bile produced during perfusion considerably varied between perfused livers when comparing livers at the same observation time. Except for 1 liver that produced only about 5 ml bile over the entire experimental period, the bile flow was 5.5 ± 2.03 ml/min/100 g liver weight and the total bile amount after 3 hours of perfusion was 22.9 ± 8 ml (Table 3).
pH, Enzymes, and Glucose
pH levels in arterial and venous blood decreased slightly during perfusion always remaining into the normal range (7.35 to 7.45, Table 3). AST, ALT, and LDH levels in plasma obtained from arterial blood increased constantly during perfusion. Whereas the increase in AST was approximately 20-fold after 3 hours of perfusion (Figure 3A), ALT (Figure 3B), and LDH (Table 3) showed a 2-fold increase only.
Since glucose was supplied by the dialysate only minor deviations were observed for the glucose levels in the blood circuit.
Radiography
Radiographic evaluation revealed a detailed perfusion of portal venous and arterial vessels with no alterations during the 3-hour perfusion experiment (Figure 4). These results are representative for all perfusion experiments since cannulation of vessels was performed with a standardized procedure and organ resistance did not vary notably between livers.
Histology
To correlate functional parameters to morphological findings, histology was carried out and HE-staining demonstrated an intact structure of the organ (Figure 5).
Discussion
Hemodynamics
With the perfusion system presented here it was possible to maintain a constant mean hepatic pressure as well as a stable perfusion flow for a period of 3 hours. The adaptation of the organ to the reperfusion situation occurred about 1 hour after begin of reperfusion. At this time artery hepatic resistance achieved a steady state. This parameter is generally applied as an indicator for good organ preservation prior to reperfusion and good supply during reperfusion. Several authors used hepatic artery resistance to compare the preservation ability of different solutions or to characterize their perfusion models (Gibelin et al., 2000; Eisele et al., 2001). A comparison of the data published by Gibelin et al. and Eisele et al. for the hepatic artery resistance with the data obtained in our model, however, is difficult since the authors worked with different species and/or considerably longer preservation periods. Nevertheless, the data obtained in this model for the hepatic artery resistance are in accordance with the values described for pigs (Travis et al., 1996). In comparison with the previous model developed in our laboratory (Grosse-Siestrup et al., 2001) no relevant differences between the values for hepatic artery resistance in both models were detected, but a steady state of this parameter was only achieved with the present model. This difference can be explained by the fact that perfusion pressure was a constant in the present model, whereas in our previous model this parameter was a variable. Thus a controlled perfusion pressure allowed a stable perfusion of the pig livers.
A weight gain of 14% (mean value) was observed at the end of perfusion. The weight gain was possibly due to edema, caused by postischemic injury. Noteworthy, is that although this parameter can be easily performed it is seldom described in publications. In one article, where the beneficial effect of machinery preservation of livers in a pig model was tested, a weight gain of 18% was observed in one of the experimental groups after 2 hours’ machinery cold preservation (Iwamoto et al., 2000). In comparison to our previous perfusion system no differences related to this parameter were observed.
Oxygen Consumption, Bile Production, Electrolytes, and Hemoglobin
Oxygen consumption is essential for living cells and, therefore, one useful tool in accessing hepatocytes injury in the perfusion model. A slight decrease on the level of oxygen consumption during the first hour of perfusion was observed. A similar behavior concerning oxygen consumption was described before (Butler et al., 2002). It is possible that during the first hour of reperfusion a higher degree of oxygen consumption was necessary for regenerating adenosine triphosphate (ATP) storage in the liver cells. Although cold preservation was performed in order to keep ATP depletion to a minimum, it was described that there is still a considerable energetic cell activity even at 1°C (St. Peter et al., 2002).
To validate this assumption, however, measurements of the energy status of the livers in our model before, during and after perfusion have to be performed. The values for this parameter were than stable through the entire perfusion period. Moreover, comparing the values obtained in this study with the values published for pigs (Travis et al., 1996) no differences were observed. In contrast to our data, in another study performed with an isolated pig liver (using roller pumps), where organs were preserved in different manners, but also perfused with blood, oxygen consumption decreased during the first 3 hours of perfusion (Adham et al., 1997). The authors explained this situation by relating to lowered metabolic functions in their model. Thus oxygen consumption in our method demonstrated a metabolic activity of the livers and this was constant throughout the perfusion period.
Bile production is related to the metabolic functions of the liver (Adham et al., 1997; Butler et al., 2002; St. Peter et al., 2002). Since only working cells are able to generate bile, this parameter might contribute to evaluate the metabolic performance of the isolated perfused liver. In our system we observed a constant increase on bile production during the 3 hours perfusion process with a cumulative total of 23 ml, which is consistent with physiological bile flow data of the pig of about 8.0 ± 0.75 ml/h. Compared to our previous study (Grosse-Siestrup et al., 2002b) bile flow was comparably higher (0.55 ± 0.2 vs. 0.31 ± 0.11 ml/min/1000 g liver weight). Our findings are in accordance to data described in the literature for bile production in similar perfusion models (Adham et al., 1997; Butler et al., 2002; St. Peter et al., 2002). Studies with a considerable longer perfusion period showed a decrease on bile production and composition that was referred to a consumption of substrate, a lack of hormonal stimuli and a deficient biliary drainage. Butler et al. and Adham et al. (1997) discussed the fact that bile production might be proportionally related to oxygen consumption and a decrease on the amount of bile might be related to the degree of liver damage. A correlation of bile production and oxygen consumption in our system was not possible to be performed, since the oxygen consumption could only be calculated for 3 livers.
The introduction of hemodialysis into isolated organ perfusion attenuated variations in several parameters resulting in a stable environment to the organ. The blood pH was kept close to the physiological value of 7.4 by hemodialysis throughout the experiment, despite the fact that the liver is a net acid producer and the pH tends to fall in a recirculating perfusion system. No relevant changes in the sodium levels were detected throughout perfusion and other electrolyte levels were within the normal range for pigs. Potassium levels that are influenced by the organ preservation solution were initially low and only slightly increased during perfusion. An increase on potassium in the perfusate is usually referred to cell damage (Travis et al., 1996). At least 2 ways of cell damage were possible in our system: hepatocellular injury and hemolysis. Hemolysis is a problem in all artificial hemoperfusion systems and the increase of free hemoglobin (fHB) in our system supports this assumption. Further histological examination of cell damage a might be a way to approach this question. The potassium levels were just like sodium within the normal range for pigs.
Total hemoglobin, PCV, and other red cell variables were determined in order to ascertain that no undesired blood dilution or other pathological processes occurred on the blood during perfusion as well as to provide a constant oxygen supply. Perfusion conditions were interpreted as stable because there was no significant change in these parameters.
Enzymes and Glucose
Several authors used the enzymes AST, ALT and LDH as indicators of liver injury (Schon et al., 1993, 1997; Iwamoto et al., 2000; Ricciardi et al., 2001; Butler et al., 2002). Starting from enzyme levels in the zero sample drawn after the blood circuit was initially closed for perfusion and a constant perfusate volume over the perfusion perfusion process we concluded that every change in liver enzyme levels may be interpreted as hepatocellular reaction. The concentrations of AST and ALT increased during perfusion in our model, but in the last hour of perfusion both enzymes were stable. Noteworthy is that although increased values of both transaminases studied were observed, compared to similar perfusion systems (Schon et al., 1993,1997; Iwamoto et al., 2000) the release of AST and to a lesser extent ALT was noticeably lower in our system. With respect to LDH, the values obtained were above the normal range, but in the expected range for this kind of experimental model (Schon et al., 1997; Butler et al., 2002). Although a constant increase of LDH is often associated with hepatocellular injury, one study on the isolated perfused kidney concluded that high levels of LDH indicated not only cell damage but were also the result of a larger volume of renal tissue perfused (Herrera et al., 2000). There is no published evidence that LDH released in the liver is dependent on the quantity of tissue perfused until now.
An increase on the glucose concentration in plasma was observed in the first hour of perfusion. This was also described by Travis et al. (1996), as possibly, due to high levels of glucose in the dialysate or a changed glycogen metabolism (Travis et al., 1996). Histological investigation of tissue samples in the future might contribute in understanding this phenomenon.
Radiography
Radiographs clearly showed adequate perfusion conditions of the organ and proved to be a simple way to verify cannulation of vessels and standardized conditions in isolated organ perfusion. Altogether with this system we were able to perform several necessary modifications, which allowed us to reduce the amount of circulating blood, perform a pressure controlled, hemodynamically optimized perfusion and provide a pulsating blood flow.
An enhanced metabolic activity of the livers was demonstrated (bile production and oxygen consumption) and a lower degree on cell damage as measured by the release of AST, ALT, LDH, and potassium was observed when compared to our previous perfusion model. Compared to porcine organs the perfusion of the isolated rat liver is still more easily performed such that there is an extensive database for a broad spectrum of scientific issues. Nevertheless porcine organ perfusion may find its place in between basic toxicological research and human clinical trials when further evidence apart from the rat is needed. Other issues concern metabolic pathways of drugs or substances where the gap between these substances and known metabolites has to be closed.
Most of these questions can be addressed in an peracute setting as described and the close proximity of porcine organ perfusion to human conditions e.g., organ mass, metabolic speed, bile composition, and enzymatic equipment, as characterized thus far, may help to reduce further extensive toxicological studies and justify the more complex porcine organ perfusion process. Therefore this study was focused on the optimization and facilitation of the perfusion process and by the use of pneumatic and pulsating flow driving units and a compact blood circuit as it is necessary for easy assembly and further standardization and accreditation of the perfusion process.
Footnotes
ACKNOWLEDGMENTS
We thank V. Essig for excellent technical assistance.