Many steatotic donor livers are discarded because of the well-documented association with primary nonfunction, further accentuating the critical organ shortage. However, most such livers exhibit near-normal biochemical function before retrieval, suggesting that cold ischemia is the critical factor rendering steatotic livers nonviable. Various pathogenic mechanisms have been proposed to explain this susceptibility to cold preservation. The crucial injury may be to the sinusoidal lining cell, with altered plasma membrane fluidity (1). Alternatively, fat-laden hepatocytes are more susceptible to cold injury possibly owing to decreased antioxidant defense mechanisms (2). Mitochondrial oxidative phosphorylation is impaired in fatty livers during preservation and may explain the poor energy metabolism in fatty livers that leads to graft failure (3). Impaired microcirculation (4) and decreased ATP content are other possible factors in the poor outcome of these organs (5).
Normothermic organ preservation has been proposed as a means to minimize preservation injury in marginal organs (6, 7). Extended preservation of livers is possible using normothermic perfusion (8) including livers from nonheart-beating donors (9). We recently reported successful transplantation of porcine livers subjected to 40 min of warm ischemia and 20 hr of normothermic preservation (10). Given the sensitivity of steatotic livers to cold preservation, normothermic preservation is a logical approach in this group of donor organs, and reliable preservation would be a major advance given the rising incidence of hepatic steatosis in the donor pool (11). To test normothermic preservation, it was necessary first to develop a model of hepatic steatosis in the clinically relevant porcine model.
All the donors with steatotic livers displayed normal liver biochemistry before retrieval. Perfusion hemodynamics (pump speed, flow, and pressures) were similar for both groups. Both groups showed similar correction of perfusate base excess (Fig. 1a) and ongoing bile production (Fig. 1b). Factor V levels decreased but were consistent with sustained production (Fig. 1e). Liver enzyme levels increased similarly for both groups, reaching a plateau approximately at 24 hr (Fig. 1f and g). Triglyceride levels in the perfusate increased sharply in the steatotic group before returning to baseline, whereas in the normal group, there was a progressive increase in the perfusate triglyceride (Fig. 1d). The perfusate blood glucose level during perfusion of the steatotic livers did not decreased less than 8 mmol/L (Fig. 2a); thus, the steatotic livers did not require supplementary glucose in marked contrast to normal livers in which a increasing glucose infusion was required, typically from 12 hr onward (P=0.03; Fig. 2b). This suggests a difference in metabolism between the two groups confirmed by nearly double the urea production in steatotic livers (P=0.002; Fig. 2c). Baseline hepatic steatosis was significantly different (0.8% group 1 and 28% group 2, P=0.005). Steatosis in the group 2 livers seemed to reduce, whereas conversely an increase in steatosis in the group 1 livers was recorded (P=0.09; Fig. 1h). The reduction in steatosis was accompanied by a decrease in the average size of the lipid deposits. Both groups had approximately 15% steatosis at 48 hr. Hematoxylin-eosin staining confirmed reduced steatosis with an altered distribution of lipid—with time, lipid is absent from the peripheries but remains near the central vein (Fig. 3). Similar mild histologic changes (hepatocellular vacuolisation, sinusoidal dilation, and congestion) were noted in both groups and became more evident with time.
We report successful extended normothermic preservation of steatotic porcine livers. The livers maintained perfusate homeostasis, bile, and Factor V production and displayed similar perfusion hemodynamics to normal livers. One steatotic liver had a high level of bile production initially, and several livers had minimal bile production toward the end of the preservation period. The bile duct was cannulated with an intravenous giving set and bile allowed to drain into a graduated collecting tube. The bile duct was essentially denuded, and ischemic necrosis of the bile duct may well account for the decline in bile output. Ischemic necrosis of the bile duct remains the Achilles heel of liver transplantation (12). Steatotic and control groups showed similar increases in perfusate liver enzymes and similar mild histologic changes. The diabetic donor is reflected in reduced glucose use and increased production of urea, but this metabolic difference does not prevent sustained function ex vivo. Similar observations have been made previously in isolated perfused rat livers taken from diabetic animals (13). Hepatic steatosis itself has a clear association with insulin resistance but a direct causal relationship has not been established, and the mechanisms of lipid and glucose control in the liver are complex (14).
The quantification of steatosis is problematic with no clear consensus as to the optimal method (15). Several staining techniques exist to identify lipid deposition (16), but Oil Red O, which is specific to triglyceride, is widely adopted as the optimal stain for hepatic steatosis. Analysis typically requires an experienced histopathologist but large discrepancies between pathologists have been reported (17). Several semiautomated, computer-assisted image analysis techniques have been published (17–20) noting overestimation of steatosis by pathologists in comparison with computer image analysis. Here, by using Oil Red O staining, we have used a simple image analysis technique using free software that avoids subjective labeling of steatotic deposits to quantify the reduction in steatosis observed in the hematoxylin-eosin histology.
Notwithstanding the small numbers in this study, we may cautiously state that prolonged normothermic preservation seems to reduce the degree of steatosis. Normothermic preservation may, therefore, improve the outcome of transplanting steatotic livers by the avoidance of cold ischemia or by a reduction in steatosis (or a combination). The average size of lipid deposits decreases with time and may reflect a conversion of macrosteatosis to microsteatosis or a preferential reduction in the larger lipid deposits.
Steatosis in donor livers is increasingly common with estimated prevalence ranging between 9% and 26% in 1998 (16) to an alarming 51% in 2006 (11). With increasing steatosis, there is a corresponding increased risk of primary nonfunction (15). Two large studies have determined that mild macrosteatosis (25%  or even as little as 15% ) is independently associated with graft nonfunction, whereas microsteatosis of any degree can be considered safe (21). Thus, reduction in steatosis should have the potential to expand the available donor pool. Although the degree of steatosis generated in our porcine model is low (28%) in comparison with that of small animal models, this level of steatosis has clinical relevance.
Our results are supported by recent studies in a rat model, suggesting that normothermic perfusion may allow “defatting” of livers and metabolic conditioning of livers for transplantation (23). Sandwiched hepatocyte cultures rendered fatty by incubation with fatty acid-rich medium could be successfully “defatted” by a range of agents including ligands for peroxisome proliferator-activated receptors to stimulate lipid export, an insulin-mimetic visfatin known to decrease triglyceride levels and forskolin to stimulate cyclic AMP-driven β oxidation of lipids and ketogenesis. Individually, each agent decreased hepatocyte triglyceride levels, and a cocktail of all the agents achieved a 31% decrease in triglyceride levels in 48 hr. By using an isolated oxygenated normothermic perfusion circuit, Lean and Zucker rat livers were perfused for 3 hr with a control cell culture medium or a medium containing the defatting cocktail. With the control cell culture medium, a 30% reduction in triglyceride levels was detected, and the addition of the defatting cocktail yielded a 65% reduction. The decrease with the control culture medium alone may be explained by the hourly exchange of perfusate (of the 50 mL perfusate, 30 mL was replaced with fresh medium each hour). Hepatocytes in the periportal region were “defatted” in preference to those in the perivenous region, a result mirrored in our experiments. With reduction in hepatocyte triglyceride levels, there was a corresponding 40% increase in perfusate triglyceride and an increase in bile production.
A potentially important limitation of our circuit is that it recirculates mobilized triglyceride. By adopting a perfusate exchange protocol, a greater reduction in steatosis may be possible although changes in biochemistry would be harder to interpret. The recirculation of perfusate may explain the gradual increase in lipid deposition in the control arm of our experiments. The normal livers perfused with a hyperglycemic perfusate in the presence of insulin should lay down glycogen stores, and once these stores is full (approximately 5% of liver mass), additional glucose is channeled into triglyceride secretion as very low-density lipoproteins. These very low-density lipoproteins will simply recirculate (an increasing rising triglyceride level was observed in the perfusate) and act as a fatty load further increasing hepatic lipid deposition. Future work could therefore incorporate several modifications—a perfusate exchange protocol, a filter to extract mobilized circulating lipid, a perfusate with a lower glucose (dextrose) load, or a cocktail of defatting agents. Further work is now required to demonstrate that normothermic preservation can facilitate the successful transplantation of donor livers with steatosis, but these initial results are encouraging.
MATERIALS AND METHODS
The objectives of this study were to establish whether normothermically perfused steatotic livers function normally and to investigate whether normothermic preservation reduces the degree of hepatic steatosis. Livers were divided into two contemporary groups: group 1, normal livers (n=5) and group 2, steatotic livers (n=3). Two-way analysis of variance with repeated measures was used to compare the groups.
Induction of Hepatic Steatosis
Large animal models of hepatic steatosis are few in number. In dogs, hepatic steatosis has been produced by a high fat (51%) choline deficient diet (24) and in pigs by the administration of intragastric alcohol (25). Our group attempted to obtain porcine hepatic steatosis using intragastric alcohol with only limited success (data not shown). Therefore, an alternate method was developed seeking to mimic the conditions in human known to predispose to fatty liver—namely diabetes and a diet high in fat and carbohydrate.
White Landrace pigs weighing approximately 25 kg were fed a specially formulated diet rich in fat (20% by volume) and carbohydrate for 3 weeks before a tunneled central venous line was inserted under general anesthesia. A permanent diabetic state was then induced by the intravenous administration of 125 mg/kg streptozotocin (Sigma-Aldrich, UK). Within 24 hr, the animals were hyperglycemic and ketotic. This state was maintained for a further 2 weeks during which the high fat diet continued. Small doses of insulin were required to avoid diabetic ketoacidosis but where possible blood glucose levels were allowed to remain increased (typically more than 12 mmol/L). Supplementary intravenous hydration with saline was required in the initial period where poor oral fluid intake was observed.
Under general anesthesia, a midline abdominal incision was made, the bile duct was divided, and the hepatic vessels were isolated. Donor blood was collected, and a peripheral liver biopsy was performed with sections snap frozen in liquid nitrogen and fixed in 10% formalin. Heparin (20,000 Units; CP Pharmaceuticals, UK) was given intravenously and allowed to circulate. The infrarenal aorta and portal vein were cannulated with 20-Fr. cannula (Bard, UK and Medtonics, UK respectively) and perfused with cold hyperosmolar citrate solution (Soltran, Baxter, UK) through the portal vein by gravity and aorta at 150 mm Hg pressure commenced. Simultaneously, the suprahepatic inferior vena cava (IVC) was divided into the pericardium and the aorta cross-clamped in the chest. After 2 L infusion through the aorta and 1 L through the portal vein, the liver was removed and placed in a bowl of 0.9% saline ice slush. The suprahepatic IVC was cannulated with a 28-Fr. cannula (Medtronics, UK) with its orifice positioned at the level of the hepatic veins. Pressure monitoring lines (V-Green 1.5-mm internal diameter, Baxter) were placed in the IVC (directly) and portal vein (via a tributary). The hepatic artery was cannulated with a 10-Fr. cannula (Medtonics, UK) with pressures measured from a side port. The bile duct was cannulated with an intravenous giving set (RCM2071B, Baxter), and bile output was monitored. Before connection to the circuit, a second biopsy was performed. During liver donor hepatectomy, retrieval, backtable, and reperfusion on the circuit, the intraparenchymal temperature was continuously recorded using a 38-mm long, 0.8-mm diameter needle thermocouple (HYP-2, Omega Engineering Ltd, UK) connected to a specially calibrated datalogger (HH611A, Omega Engineering Ltd) with ±0.05°C accuracy recording the temperature every minute. Recordings were subsequently downloaded to a computer for graphical analysis. The needle was inserted to the hilt in the same anatomical location for each experiment (close to Hartmann's pouch of the gallbladder entering segment V) and secured to the gallbladder wall using 3/0 Ethilon sutures (Ethicon, UK). The operating theater temperature was recorded continuously and was not significantly different between experiments (21.4°C±0.7°C). The average time to from starting the aortic flush to the liver being retrieved to the backtable was 16±4 min, and the average time to connection to the circuit and reperfusion was 76±11 min. The lowest temperature reached was 7.6°C±1.8°C. The cooling curves obtained are in keeping with those reported in human donor liver retrieval (26).
Normothermic Liver Perfusion Circuit
The circuit (Fig. 4) was primed with approximately 1500 mL of blood group identical porcine blood and comprised an oxygenator (Jostra Quadrox HMO 2000 Oxygenator, Maquet Cardiopulmonary, UK), a centrifugal pump (BP50 Centrifugal Pump, Medtronics), a soft shelled reservoir (MVR 800, Medtronics), tubing (PVC and silicon, 1/4- and 3/8-inch internal diameter with polycarbonate connectors, Medtronics), a gate clamp (EW-06833-10, Cole Palmer, UK), and two flow probes (DP 38P, Medtronics). A heat exchanger (Biomedicus, Medtronics) maintained the temperature of the blood at 39°C (normal temperature for a pig). Oxygen and air flow to the oxygenator were adjusted with a target paO2 of 14 to 20 kPa and paCO2 3.5 to 6 kPa. The oxygenator was not changed during the perfusion. During priming, the pH was adjusted to within physiologic range with 20 mmol of sodium bicarbonate and 10 mL of 10% calcium gluconate and 10,000 Units of heparin added. The liver, in an intestinal bag (Vi-Drape, MCD, UK) with fine perforations, was placed onto water-filled balloons sitting on a perforated polycarbonate tray within a stainless steel bowl. Ascites collected in the sump of the bowl was returned to the circuit through a tube passed through a pinch valve under the control of a light sensor that opened the valve when the tube was full, closing when emptied.
Immediately before perfusion of the liver, a prostacyclin (epoprostenol sodium; Flolan, GlaxoWellcome, UK) infusion was commenced at 4 mg/hr. Prostacyclins are known to increase portal flow (27), and in an isolated perfused liver, epoprostenol increased total blood flow and oxygen transport at 30-min reperfusion after warm ischemia (28). An amino acid solution (amino acid component of Nutriflex Lipid+, B Braun, UK) was infused at 15 mL/hr to provide nutrition. The nutritional requirement of the isolated porcine liver is unknown, but by infusing a standard solution, changes to the amino acid profile of the perfusate can be monitored. Human Actrapid (Novo Nordisk, UK) insulin was infused at 15 mL/hr during the first 4 hr of the perfusion then at 1 mL/hr thereafter. The initial high dose was to compensate for the high glucose content of the prime blood, which was stored in dextrose supplemented blood transfusion bags (CPDA Single Blood Collection Systems, Baxter). Additional glucose in the form of a 20% glucose infusion (Macopharma, UK) was provided as required to maintain perfusate glucose within a range of 4 to 8 mmol/L as determined by blood glucose analysis (Accutrend machine and BM-Accutest sticks, Boehringer-Mannheim, UK). To compensate for the lack of enterohepatic recirculation of bile salts, a 1 mL/hr continuous infusion of sodium tauroursodeoxycholate bile salt (Sigma-Aldrich) was used. Cefuroxime antibiotic (750 mg; Zinacef, GlaxoSmithKline, UK) was added to the circuit.
Perfusion was initiated by a progressive increase in pump speed over a few minutes until the IVC pressure was between 0 and 5 mm Hg. The gate clamp was then adjusted to obtain an arterial pressure in the range 85 to 95 mm Hg. Portal perfusion pressure is determined by the vascular resistance of the liver. Previous work has shown that significantly altering the height of the reservoir yields only marginal changes in the portal perfusion pressure. During the first hour, fine adjustment of the pump speed and gate clamp was required to compensate for expansion of the microcirculation during liver warming. Samples were obtained from the hepatic arterial inflow and IVC outflow at 1, 2, 4 hr, and then every 4 hr until the end of perfusion and were spun at 6500 rpm with the plasma stored in liquid nitrogen for later analysis. Additional arterial and venous blood samples were collected hourly and analyzed immediately by blood gas analysis (Bayer 348 RapidLab Blood Gas System, Bayer, UK). Arterial, portal, and IVC pressures and flows were measured, pump speed noted, and bile output recorded hourly. A liver biopsy was performed at 24 and 48 hr. When liver perfusion ceased, random core samples were cut from left and right lobes, deep in segment 4 (central) and the hilum of the liver. The tissue was fixed in 10% formalin for hematoxylin-eosin staining or snap frozen in liquid nitrogen for Oil Red O staining.
Electrolytes, liver function tests, urea, and full blood counts including hemoglobin, hematocrit, white blood cell count, and platelet counts were measured using standard clinical methods. The Factor V assay (Oxford Hemophilia and Thrombosis Centre, Oxford, UK) involves dilutions of standard and test plasma mixed with substrate plasma, which is deficient in factor V. In the presence of activated factor V, factor Xa rapidly cleaves the two peptide bonds in prothrombin to form thrombin that in turn cleaves fibrinogen to form a fibrin clot. A modified prothrombin clotting time assay is performed with the ability of the test and standard plasmas to correct the prolonged prothrombin clotting time of the substrate plasma compared.
Paraffin-mounted, formalin-fixed sections were sectioned at 3 μm before hematoxylin and eosin staining using a standard technique. Briefly, slides were dewaxed by passage through xylene before rehydrating by means of a decreasing alcohol gradient to water, stained with Gill's hematoxylin, differentiated in 1% acid alcohol, counterstained with 1% eosin, and dehydrated, cleared, and mounted in di-N-butylphthalate in xylene. Snap frozen tissue was cut at 6 to 7 μm using a Bright Instruments 5040 Microtome (Bright Instrument Company Ltd., UK) in a Bright Instruments OTF cryocabinet (Bright Instrument Company Ltd., UK) and mounted on multiwell slides.Oil Red O staining was performed according to standard technique. Briefly, the slides were fixed in 10% formal saline, rinsed in 70% ethanol, stained in a fresh saturated stock solution of Oil Red O in isopropanol together with water in a 6:4 ratio, washed in 70% ethanol, counterstained with hematoxylin, and mounted in Aqumount (Lerner Laboratories, Pittsburgh, PA). Photomicrographs were taken using a Leitz Laborlux S microscope (Leica Microsystems, UK) fitted with a Zeiss Axiocam digital camera (Carl Zeiss Ltd, UK) with the images captured using KS400 imaging software (Imaging Associates, UK).
Image Analysis of Steatosis
Images were loaded into Image J (http://rsb.info.nih.gov/ij/index.html ) and analyzed using a purpose written macro. Briefly, the photograph was converted from a Tagged Image File Format to a Red, Green, Blue image stack, the Blue stack isolated, and subtracted from the original image leaving only the Red and Green stacks. This image was then converted to grayscale and analyzed using the Binary Threshold algorithm on Image J. This isodata algorithim sets an arbitrary test intensity threshold then computes the average intensity of pixels below and the average intensity of pixels above the threshold. The average of the two averages (above and below) is taken and compared with the test threshold. The whole process is repeated until the test threshold is greater than the composite average. This identifies all the Red regions, eliminating the need to subjectively set a threshold for “Red.” Image J counts Red regions, provides a mean size of Red region, and reports the area occupied by Red as a percentage of the whole image. To confirm that the technique provides good results, the map of the Red regions can be superimposed on the original image (Fig. 5a). In some photographs, there are large areas of portal tracts or fibrotic tissue, which cannot take up the Oil Red O stain or where tissue fails to occupy the whole image. Steatosis is underestimated if simply calculating the percentage of the total slide identified as Red lipid deposits, so a standard mask was applied centered on the central vein and only the region within the mask was analyzed (Fig. 5b).
The authors thank the assistance of D. Guerriero, Nuffield Department of Surgery, University of Oxford with the normothermic perfusions.
1. Fukumori T, Ohkohchi N, Tsukamoto S, et al. The mechanism of injury in a steatotic liver
graft during cold preservation. Transplantation
1999; 67: 195.
2. Taneja C, Prescott L, Koneru B. Critical preservation injury in rat fatty liver
is to hepatocytes, not sinusoidal lining cells. Transplantation
1998; 65: 167.
3. Caraceni P, Bianchi C, Domenicali M, et al. Impairment of mitochondrial oxidative phosphorylation in rat fatty liver
exposed to preservation-reperfusion injury. J Hepatol
2004; 41: 82.
4. Teramoto K, Bowers JL, Kruskal JB, et al. Hepatic microcirculatory changes after reperfusion in fatty and normal liver transplantation
in the rat. Transplantation
1993; 56: 1076.
5. Fukumori T, Ohkohchi N, Tsukamoto S, et al. Why is fatty liver
unsuitable for transplantation
? Deterioration of mitochondrial ATP synthesis and sinusoidal structure during cold preservation of a liver
. Transplant Proc
1997; 29: 412.
6. Jamieson RW, Friend PJ. Normothermic organ preservation
. Transplant Rev
2006; 20: 172.
7. Jamieson RW, Friend PJ. Organ reperfusion and preservation. Front Biosci
2008; 13: 221.
8. St Peter SD, Imber CJ, Lopez I, et al. Extended preservation of non-heart-beating donor livers with normothermic
machine perfusion. Br J Surg
2002; 89: 609.
9. Reddy SP, Bhattacharjya S, Maniakin N, et al. Preservation of porcine non-heart-beating donor livers by sequential cold storage and warm perfusion. Transplantation
2004; 77: 1328.
10. Brockmann J, Reddy S, Coussios C, et al. Normothermic
perfusion: A new paradigm for organ preservation
. Ann Surg
2009; 250: 1.
11. Halon A, Patrzalek D, Rabczynski J. Hepatic steatosis
transplant donors: Rare phenomenon or common feature of donor population? Transplant Proc
2006; 38: 193.
12. Koneru B, Sterling MJ, Bahramipour PF. Bile duct strictures after liver transplantation
: A changing landscape of the Achilles' heel. Liver Transpl
2006; 12: 702.
13. Exton JH, Corbin JG, Harper SC. Control of gluconeogenesis in liver
. V. Effects of fasting, diabetes, and glucagon on lactate and endogenous metabolism in the perfused rat liver
. J Biol Chem
1972; 247: 4996.
14. Postic C, Girard J. Contribution of de novo fatty acid synthesis to hepatic steatosis
and insulin resistance: Lessons from genetically engineered mice. J Clin Invest
2008; 118: 829.
15. Imber CJ, St Peter SD, Handa A, et al. Hepatic steatosis
and its relationship to transplantation
. Liver Transpl
2002; 8: 415.
16. Garcia Urena MA, Colina Ruiz-Delgado F, Moreno Gonzalez E, et al. Hepatic steatosis
transplant donors: Common feature of donor population? World J Surg
1998; 22: 837.
17. Fiorini RN, Kirtz J, Periyasamy B, et al. Development of an unbiased method for the estimation of liver steatosis
. Clin Transplant
2004; 18: 700.
18. Dragunow M, Cameron R, Narayan P, et al. Image-based high-throughput quantification of cellular fat accumulation. J Biomol Screen
2007; 12: 999.
19. Franzen LE, Ekstedt M, Kechagias S, et al. Semiquantitative evaluation overestimates the degree of steatosis
biopsies: A comparison to stereological point counting. Mod Pathol
2005; 18: 912.
20. Marsman H, Matsushita T, Dierkhising R, et al. Assessment of donor liver steatosis
: Pathologist or automated software? Hum Pathol
2004; 35: 430.
21. Zamboni F, Franchello A, David E, et al. Effect of macrovescicular steatosis
and other donor and recipient characteristics on the outcome of liver transplantation
. Clin Transplant
2001; 15: 53.
22. Salizzoni M, Franchello A, Zamboni F, et al. Marginal grafts: Finding the correct treatment for fatty livers. Transpl Int
2003; 16: 486.
23. Nagrath D, Xu H, Tanimura Y, et al. Metabolic preconditioning of donor organs: Defatting fatty livers by normothermic
perfusion ex vivo. Metab Eng
2009; 11: 274.
24. Takahashi K, Hakamada K, Totsuka E, et al. Warm ischemia and reperfusion injury in diet-induced canine fatty livers. Transplantation
2000; 69: 2028.
25. Mets B, Kamban JR, Janicki PK, et al. Alcohol pretreatment does not affect bupivacaine pharmacokinetics in the pig. Pharmacol Res
1996; 33: 291.
26. Villa R, Fondevila C, Erill I, et al. Real-time direct measurement of human liver
allograft temperature from recovery to transplantation
2006; 81: 483.
27. Zardi EM, Picardi A, Ambrosino G, et al. Enhanced portal flow velocity and volume following Iloprost treatment. Acta Biomed
2003; 74(suppl 2): 108.
28. Wolf S, Schon MR, Krautlein K, et al. Effects of epoprostenol and norepinephrine on hemodynamics of the isolated perfused liver
. Transplant Proc
1998; 30: 2334.
29. Abramoff MD, Magelhaes PJ, Ram SJ. Image processing with ImageJ. Biophotonics Int
2004; 11: 36.