Artificial Circulation of the Liver: Machine Perfusion as a Preservation Method in Liver Transplantation



Due to the sharp increase in liver transplant candidates and the subsequent shortage of suitable donor livers, an extension of the current donor criteria is necessary. Simple cold storage, the current standard in organ preservation has proven to be insufficient to preserve extended criteria donor livers. Therefore a renewed interest grew toward alternative methods for liver preservation, such as hypothermic machine perfusion and normothermic machine perfusion. These “new” preservation methods were primarily assessed in rat models, and only a few clinically relevant large animal models have been described so far. This review will elaborate on these alternative preservation methods. Anat Rec, 291:735–740, 2008. © 2008 Wiley-Liss, Inc.


In solid organ transplantation, organs are exposed to an obligatory ischemic period between the retrieval from an organ donor until the transplantation into the recipient. Therefore, organs need to be preserved and protected against the deleterious effects of the ischemic period. The ideal method of preservation of the organ would prolong preservation time, reduce the rate of primary graft nonfunction after transplantation, would allow assessing the viability of the organ before transplantation and would facilitate a more extended use of marginal organs. Considering these characteristics, simple cold storage (SCS) fails to be the ideal preservation method for marginal organs. During SCS, organs are first flushed with a cold (4°C) preservation solution and then immersed and stored in this solution until their transplantation. In contrast to SCS, machine perfusion (MP) preserves the organ by a constant perfusion through its blood vessels with a machine perfusion solution (MPS). MP has been shown to have potential benefits compared with SCS. This review will elaborate on these benefits for the liver.

In clinical kidney transplantation, MP has become a superior method over SCS to preserve kidneys from marginal or extended criteria donors (Wight et al.,2003). Accordingly, MP promises to be a better method to preserve livers from such donors. However, MP of livers is only used experimentally and not yet clinically.

At the start of experimental and clinical transplantation of kidneys and livers, MP was explored as a method to preserve these organs (Belzer,1973). This was based on the hypothesis that organs are best preserved in a condition as physiological as possible by providing oxygen together with sufficient nutrients to support a constant metabolism. Initially this preservation was conducted at body temperature. Only later, the development of hypothermia and SCS was introduced, where the organ was cooled in vivo through a flush out with a cold preservation solution and subsequently stored on ice and further cooled topically (0–4°C). By this means, the metabolism was slowed down, which resulted in a lower need for nutrients and oxygen. Moreover, transportation of the organ to other transplant centers became easier. The successful introduction of a hypothermic preservation solution such as the Collins solution was responsible for the replacement of MP by SCS to preserve kidneys in most transplant centers worldwide.

Later, University of Wisconsin (UW) preservation solution was shown to be superior to Collins for both kidney and liver preservation (Ploeg et al.,1992). UW is now widely used as preservation solution in solid organ transplantation.

Liver transplantation has become widely applicable for patients with irreversible liver failure. Nowadays results of liver transplantation are excellent and as a consequence more patients have been placed on the waiting list. The current organ donor pool is insufficient to fulfill the demand of organ donors. Extended criteria donors, such as steatotic donors, older donors and non heart beating donors (NHBD), represent a potential additional pool of liver donors. NHBD livers have been exposed to a certain period of warm ischemia (WI; no or impaired circulation) before the procurement. Because SCS fails to optimally preserve extended criteria organs, alternative preservation methods are again searched for. As a result more research on MP has been performed. In contrast to SCS, MP organs can be monitored over time; viability markers can be identified. The microcirculation is better preserved due to a continuous circulation that is maintained during the MP. With SCS, the sinusoids become constricted due to the hypothermia, which prevents penetration of the preservation solution in the tissues and may cause an impaired microcirculation upon reperfusion. In contrast, with MP, nutrients (and oxygen) are continuously supplied. It is possible to reduce waste products by dialysis in the circuit or by changing the perfusion fluid. In addition, the whole organ is accessible for different types of pharmacological intervention over the entire period of MP. In the following sections, we will describe two different types of liver machine perfusion currently applied in experimental settings: hypothermic (4–10°C) and normothermic (at normal body temperature).


Several studies have shown that preservation of the normal liver by HMP results in reduced hepatocellular damage and improved liver function after 24 h of preservation (Kim et al.,1997; Southard et al.,2000; Xu et al.,2004,2005). The advantages of HMP have been attributed to the continuous supply of nutrients and oxygen to the liver, even resulting in the “resuscitation” of NHBD organs (Lee et al.,2003). Bessems et al. (2005a) also mentioned an improvement of preservation by HMP of normal and NHBD livers.

Different Machine Perfusion Solutions (MPS) and Their Additives

The preservation solution is used to protect the graft by maintaining a normal osmotic pressure, buffering of the pH and by addition of substrates known to decrease the ischemia reperfusion injury such as oxygen radical scavengers. The preservation solution mainly used for HMP of the liver is the one currently used for clinical kidney HMP: the modified University of Wisconsin solution, that is, UW-gluconate (UW-G; Belzer et al.,1982). The main components of this solution are: reduced gluthatione (to reduce the production of reactive oxygen species); potassium/sodium content of 25 mM and 100 mM, respectively; glucose; adenosine (to provide the graft with sufficient energy substrates); hydroxyethyl starch (HES) as colloid; and gluconate as impermeant.

In 1990, Pienaar et al. successfully transplanted canine livers that were preserved for 72 h and had a 100% survival. They, however, changed UW-G by inverting its potassium/sodium ratio and mentioned that a high potassium concentration (125 mM)/low sodium (25 mM) concentration was necessary for the success of the transplants while an opposite low potassium/high sodium ratio resulted in a poor outcome. Pienaar et al. (1990) used oxygen and the pH was kept stable during HMP.

Experiments in our laboratory were conducted using pig livers in our preclinical transplant model. The settings of the HMP were based on clinical HMP for kidney. In our laboratory, UW-G with the original low potassium/high sodium content was used as MPS (Monbaliu et al.,2006). The donor liver was preserved with HMP without oxygen for 4 hr and perfused through the portal vein and hepatic artery. We obtained a survival of 2/8 by following the animals until postoperative day 3.

Dutkowski et al. (2006) transplanted rat NHBD livers with UW-G and obtained a graft survival of 100%. The set-up of their experiments was different in that they applied SCS before HMP.

Guarrera et al. (2005) added five compounds to UW-G. Alpha-ketoglutarate was included to protect the mitochondria by providing energy substrate during the reoxygenation. L-arginine was added as a nitric oxide precursor and N-acetylcysteine as a glutathione precursor which serves as an antioxidant. Nitroglycerin was added as a nitric oxide donor that regulates the vasodilation and prostaglandin E1 was included because it inhibits the aggregation of platelets and acts as a membrane stabilizer and a vasodilator. The solution was referred to as Vasosol. This resulted in successful transplantation of a limited number of pig livers (n = 3; Guarrera et al.,2005) with adequate liver function and survival until postoperative day 5.

Recently another new MPS, Polysol, was developed. Polysol is based on components of culture media (Bessems et al.,2005a,b; Bessems et al.,2006). Polysol contains nutrients such as glucose, amino acids, vitamins and nutrients for the lipid metabolism (i-inositol, choline chloride, and linoeic acid methyl ester). In contrast to UW-G, Polysol contains all nutrients necessary to fully support the ongoing metabolism. In search for the best composition of Polysol, different colloids were tested and a comparison was made between Polysol-HES and Polysol-PEG (polyethylene glycol). Polysol-HES resulted in inferior hepatocyte function and inferior oxygen consumption compared with Polysol-PEG. HES is claimed to cause microcirculatory disturbances by the formation of rouleaux in SCS (Morariu et al.,2003).

The question can be raised whether the metabolism is still active at 4°C and whether the addition of such products/nutrients is necessary. Our data show persistence of oxygen consumption during 24 hr of oxygenated HMP (Fig. 1), an indication that the liver remains metabolically active even at low temperature and thus justifying the addition of metabolism supporting nutrients. The oxygen pressure is measured in the MPS and in the outflow (vena cava inferior). The metabolic activity is then calculated by using the following formula: oxygen consumption = (pO2 MPS − pO2 outflow)/liver weight.

Figure 1.

Oxygen consumption as a surrogate for the metabolic activity of the liver. The oxygen consumption (mmHg/g liver) was calculated according to the following formula: oxygen consumption = (pO2 MPS − pO2 outflow)/liver weight. Porcine livers (n = 5) were perfused with the following setting: Portal Vein: 5 mmHg, 0.5 mL/g liver/min, hepatic artery: 20 mmHg and with oxygen (300 mmHg) first at 4°C during 24 hr and then rewarmed (37°C). The metabolic activity is ±30% at 4°C compared with the isolated warm reperfusion at 37°C.

As seen in Figure 1, there still 30% of oxygen consumption at 4°C compared with the warm reperfusion (mimicking the in vivo conditions and metabolism).

Minor et al. (2006) used HTK as MPS. HTK is a solution based on histidine, tryptophan, ketoglutarate and it has a potassium content of 9 mM and a sodium content of 15 mM. HTK is frequently used for SCS. They demonstrated that HTK, as a MPS, was equivalent to UW-G (Olschewski et al.,2003).

Various Machine Perfusion Settings

As with the MPS, there has been no consensus on the optimal machine perfusion settings for liver preservation: various pressures and flows, double or single vessel perfusion, retrograde or antegrade, pulsatile or not pulsatile, with or without oxygen, position of the liver and changes in the temperature have been described. The pressures reported in the literature for the venous inflow vary from 3–18 mmHg (Table 1). The target flows vary from 0.14 to 0.5 mL/min/g liver. For the arterial inflow, pressures varying from 20–80 mmHg have been reported and they can be either pulsatile or not. The flows vary from 0.1–1.2 mL/min/g liver but are less frequently described in the literature.

Table 1. List of all mentioned pressure, flows, species, and temperature during machine perfusion (P = pulsatile)
ReferenceSpeciesPVHATemp (°C)
Hypothermic machine perfusion
Slapak et al.,1969Canine12 mmHg76/23 mmHg (P)11
Brettschneider et al.,1968Canine4.8 mL/min/g1.2 mL/min/g4
Belzer et al.,1970Porcine5–8 mmHg60/40 mmHg (P)8–10
Gellert et al.,1985Porcine2 mmHg80/40 mmHg (P)10
0.5 mL/min/g0.125 mL/min/g
Pienaar et al.,1990Canine16–18 mmHg (30BPM)5
Boudjema et al.,1991Rabbit15–25 mmHg5
Yamamoto et al.,1991Porcine0.5–0.6 mL/min/g7
Rossaro et al.,1992Rat0.5 mL/min/g6–10
van Gulik et al.,1994Canine16–18 mmHg5
Kim et al.,1997Rat11 mmHg4
0.5 mL/min/g
Iwamoto et al.,2000Porcine7–8 mmHg50–60 mmHg8
20–30 mmHg
Southard et al.,2000Rat0.14 mL/min/g4
Compagnon et al.,2001Rat0.4 mL/min/g0.1 mL/min/g4
Uchiyama et al.,2001Porcine 30–60 mmHg (P or not)8
Dutkowski et al.,2003Rat4.48 ± 0.54 mmHg3–6
0.44 mL/min/g liver
Lauschke et al.,2003Rat0.5 mL/min/g4
Guarrera et al.,2005Human/Porcine3–5 mmHg12–18 mmHg3–5
0.7 mL/g liver/min = Targetflow
Jain et al.,2005Porcine10–12 mmHg30 mmHgHypotherm
0.3 mL/min/g liver0.1 mL/min/g liver
't Hart et al.,2007Rat4 mmHg (25% Normothermic)50 mmHg (P) 20–30 mmHg (P)1–3
350 mL/min80 mL/min
Our proposed settingPorcine5 mmHg20 mmHg4–7
0.5 mL/min/g
Normothermic machine perfusion
Friend et al.,2001Porcine5–8 mmHg70–90 mmHg39
1.3–1.7 L/min1.6–2.2 L/min
Schon et al.,2001Porcine15 cm water = 11.07 mmHg100 mmHg37
250 mL/min150 mL/min

Another reason why it is difficult to compare the different HMP settings are the variable temperatures (ranges from 1 to 11°C). It is known that the fluidity of the phospholipids layers of the cell membrane changes with the temperature (Ehringer et al.,2002), resulting in less flexibility of the cell membrane. Endothelial cell damage is easily provoked by extended cold ischemia time, even in SCS (Natori et al.,1999). In HMP, similar to SCS, there is damage resulting from hypothermia. In HMP, special attention has to be paid to avoid shear stress induced by the installed flow during the HMP and the viscosity of the solution. The flow itself can induce damage to the endothelial cell lining because the sinusoids are more constricted and rigid due to the cold temperature. Moreover, with low temperature, the viscosity of the MPS is higher. All this may result in damage to the sinusoidal endothelial cells. An additional factor could be that endothelial cells are swelling due to blockage of the Na/K-pump and therefore the lumen of the sinusoids is decreased and shear stress is increased.

Another important factor is that livers from different species (rat, dog, pig, and human) have been used for HMP research (for an overview see Table 1). It is difficult to translate the pressures used in rats into porcine liver preservation and human liver preservation. However, it has been documented that similar pressures and flows can be applied both in pigs and humans (Guarrera et al.,2005) with a target flow of 0.7 mL/min/g liver resulting in a pressure of 3–5 mmHg and 12–18 mmHg for the venous and the arterial side, respectively. Of note, Pienaar et al. (1990) and Yamamoto et al. (1991) who successfully preserved livers for 72 hr used a single vessel perfusion of the portal vein with a pulsatile perfusion pressure of 16–18 mmHg/30 beats per minute (Pienaar et al.,1990) and a flow of 0.5–0.6mL/min/g liver (Yamamoto et al.,1991).

Recently, 't Hart et al. (2007) linked the perfusion pressure to the homogeneity of the liver perfusion. They describe that 25% of the physiological flow is best for the endothelial cells integrity and results in homogeneous perfusion, while a lower flow results in heterogeneous perfusion and a higher flow results in massive endothelial cell damage. It has to be mentioned, however, that these results are obtained in a rat model and that homogeneous perfusion in a porcine or human setting might be more difficult to achieve. The large volume of the organ might play a role in compressing the sinusoids. In our lab, we compared two different settings and found that for porcine livers a pressure of 5 mmHg (venous inflow) and 20 mmHg (arterial inflow) was better for the morphological integrity and the ATP-preservation compared with a higher pressure (7 mmHg and 20 mmHg for venous and arterial inflow, respectively). Moreover the addition of oxygen was necessary to obtain nearly intact morphology after 24 hr of HMP. No vacuoles could be found in the parenchyma preserved by MP in the presence of oxygen (Vekemans et al.,2007). It has been described that vacuoles reflect a lower ATP content in the hepatocytes, and it is likely that oxygen plays a pivotal role in the ATP maintenance of the cell and in its morphological integrity. There is an ongoing discussion on whether oxygen could be injurious or not. It could be possible that oxygen increases the generation of reactive oxygen species (ROS). Subsequently ROS will induce an inflammation cascade in the sinusoid.

From this review, it is clear that many aspects in the development toward an optimal preservation with HMP are still to be elucidated.


NMP is based on the assumption that replicating the physiological normal environment is the best method of maintaining a liver ex vivo viable after retrieval (Imber et al.,2002b). NMP preservation is logistically more complex than HMP and includes the risk for bacterial contamination during the preservation. The advantages of hypothermia (decrease of metabolism and the lower need for essential nutrients, oxygen supply and thus oxygen carrying capacity) are lost in NMP.

Normothermic perfusion has been studied frequently but only for a short period of time in an isolated organ perfusion system to assess, for example, hepato-toxic effects of particular compounds (Grosse-Siestrup et al.,2004; Nui et al.,2006). The application in preservation was only attempted by a few research groups (Friend et al.,2001; Schon et al.,2001). In the experiments of Friend, the perfusion is done with heparinized whole blood at a temperature of 39°C; the settings are mentioned in Table 1. To mimic the enterohepatic circulation of bile salts, a 1% solution of sodium, taurocholate (7 mL/hr) was instilled into the perfusate.

Schon et al. (2001) performed NMP for 4 hr with 2l whole pig blood and with 1l of an electrolyte solution. The perfusion setting is mentioned in Table 1. The liver is positioned in an organ chamber subjected to intermittent oscillations (to mimic the negative pressure that the diaphragm movements normally exert on the liver) and the blood is continuously dialyzed. NMP was reported to improve preservation of normal and NHBD pig livers (Imber et al.,2002a,b; Friend et al.,2001; Reddy et al.,2005; Schon et al.,2001).


Extensive research should be done in the identification of parameters to predict viability during MP of liver and before transplantation. Our group showed that vascular resistance during HMP was not a good surrogate for viability and could not discriminate normal livers from ischemic livers (Derveaux et al.,2005). Aspartate amino transferase (AST) measured in the MPS appeared to be a good viability parameter. We observed an increased AST-release during HMP of WI livers compared with normal livers. Possible other parameters that could be explored is the ATP content of the liver, which represents the energy status of the liver during HMP. Liver fatty acid-binding protein (L-FABP) is another possible viability parameter. This small protein (15 kDa) is involved in the intracellular transport of long-chain fatty acids in the liver. L-FABP is regarded as a sensitive marker for liver cell damage after transplantation of NHBD livers (Monbaliu et al.,2005). In kidney, HMP redox active iron (de Vries et al.,2006) is regarded as a viability parameter to evaluate the difference between heart beating donors (HBD) and NHBD. This could also be explored in liver HMP.

During NMP, the identification of viability parameters is perhaps easier as the liver is in a normal metabolic state. The bile production has been suggested as a good viability parameter. Bile analyzed by proton magnetic resonance spectroscopy showed that HBD compared with NHBD had an increased concentration of bile acids, lactate, glucose and phosphatidylcholine and had a decreased concentration of acetate (Habib et al.,2004).


Whether MP is better than or equal to SCS for normal and extended criteria livers remains to be further explored. In many models described in the literature, a rewarming at 37°C is applied to mimic the ischemia reperfusion injury without having to perform a liver transplantation. However, a liver transplant model is superior over an ex vivo assessment to study primary graft nonfunction, in vivo ischemia reperfusion injury, neutrophil sequestration and graft/recipient survival. In addition a lot of research on MP has been done in rat models and not in clinically relevant large animal models. To date, it remains unclear whether MP could protect against major complications of a liver transplant such as occurrence of ischemic type of biliary lesions. It is, however, known that the biliary tree can be preserved when the arterial microcirculation is well perfused. It has even been reported that hepatic artery thrombosis results in biliary complications (Nishida et al.,2006). Only with MP the hepatic artery and its downstream vessels can be perfused and accessed by metabolites that result in optimal preservation of the peribiliary plexus. Moreover, MP could predict disturbances in the hepatic artery by, for example, the vascular resistance. This review concludes that many aspects of MP are still to be elucidated before MP can be applied in a clinical setting. Further evaluation in large animal transplant models is warranted.


K.V. is a postdoctoral researcher of the “Fund for Scientific research-Flanders (FWO)”. J.P. holds a grant from the CAF, Vilvoorde, Belgium.