The 24-Hour Normothermic Machine Perfusion of Discarded Human Liver Grafts

Transcription

1 ORIGINAL ARTICLE VOGEL ET AL. The 24-Hour Normothermic Machine Perfusion of Discarded Human Liver Grafts Thomas Vogel, 1 Jens G. Brockmann, 2 Alberto Quaglia, 3 Alireza Morovat, 4 Wayel Jassem, 3 Nigel D. Heaton, 3 Constantin C. Coussios, 5 and Peter J. Friend 6 1 Department of General and Visceral Surgery, University Hospital M unster, M unster, Germany; 2 King Faisal Specialist Hospital and Research Centre, Riyadh, Saudi Arabia; 3 Institute of Liver Studies, King s College London, London, UK; 4 Department of Clinical Biochemistry, Oxford University Hospitals National Health Service Trust, Oxford, UK; and 5 Institute of Biomedical Engineering and 6 Nuffield Department of Surgical Sciences, University of Oxford, Oxford, UK Donor organ shortage necessitates use of less than optimal donor allografts for transplantation. The current cold storage preservation technique fails to preserve marginal donor grafts sufficiently. Evidence from large animal experiments suggests superiority of normothermic machine preservation (NMP) of liver allografts. In this study, we analyze discarded human liver grafts that underwent NMP for the extended period of 24 hours. Thirteen human liver grafts which had been discarded for transplantation were entered into this study. Perfusion was performed with an automated device using an oxygenated, sanguineous perfusion solution at normothermia. Automated control was incorporated for temperature-, flow-, and pressure-regulation as well as oxygenation. All livers were perfused for 24 hours; parameters of biochemical and synthetic liver function as well as histological parameters of liver damage were analyzed. Livers were stratified for expected viability according to the donor s medical history, procurement data, and their macroscopic appearance. Normothermic perfusion preservation of human livers for 24 hours was shown to be technically feasible. Human liver grafts, all of which had been discarded for transplantation, showed levels suggesting organ viability with respect to metabolic and synthetic liver function (to varying degrees). There was positive correlation between instantly available perfusion parameters and generally accepted predictors of posttransplant graft survival. In conclusion, NMP is feasible reliably for periods of at least 24 hours, even in highly suboptimal donor organs. Potential benefits include not only viability testing (as suggested in recent clinical implementations), but also removal of the time constraints associated with the utilization of high-risk livers, and recovery of ischemic and other preretrieval injuries (possibly by enabling therapeutic strategies during NMP). Liver Transplantation AASLD. Received July 11, 2016; accepted October 3, The number of patients referred for liver transplantation is increasing for various reasons including epidemiological factors and an expansion of indications for liver transplantation over the last few decades. At the Abbreviations: ALT, alanine aminotransferase; BE, base excess; BM, blood glucose measurement; BMI, body mass index; CIT, cold ischemia time; DCD, donation after circulatory death; ECD, extended criteria donor; FS, flow sensor; HA, hepatic artery; IVC, inferior vena cava; LDH, lactate dehydrogenase; NA, not available; NHBD, non heart-beating donor; NMP, normothermic machine perfusion; PPV, proportional pinch valve; PS, pressure sensor; PV, portal vein; S, sampling port; SD, standard deviation; Shunt S, Terumo shunt sensor; SOPV, shut-off pinch valve; WIT, warm ischemia time. Address reprint requests to Peter J. Friend, M.D., Nuffield Department of Surgical Sciences, University of Oxford, Churchill Hospital, Oxford OX3 7LE, United Kingdom. Telephone: 1 44 (0) ; peter.friend@nds.ox.ac.uk same time, the number of good quality grafts from deceased organ donors has remained static at best. As a result, acceptance criteria of liver grafts have broadened and organs from extended criteria donors (ECDs) form a large portion of the donor organ pool. These organs are known to be more prone to primary nonfunction or delayed graft function, with potentially fatal consequences. (1,2) The use of increasingly suboptimal donor organs necessitates new approaches to organ viability assessment, preservation, and repair, in order that the use of such organs is not associated with an unacceptable compromise with respect to survival. (3) Organ transplantation in its current practice is hampered by time constraints in several ways. Indeed logistic (rather than medical) considerations often determine recipient selection if the status of an organ is deemed to be marginal. (4) In addition, innovative approaches to graft modification and improvement ORIGINAL ARTICLE 207

2 VOGEL ET AL. LIVER TRANSPLANTATION, February 2017 during the organ preservation period are strongly timedependent, and a means to extend preservation times might allow these techniques to transfer from the experimental to the clinical setting. These approaches include pharmacological, metabolic, or genetic interventions. (5-7) The current universal standard for organ preservation is static cold storage. Although limited in terms of the duration of preservation, cold storage has the major advantages of simplicity, portability, and affordability. However, in recent years, with increasing use of marginal organs the limitations of static cold storage are a major factor influencing patient and graft survival rates. The ideal method of organ preservation would do the following: 1. Enable accurate discrimination between transplantable and nontransplantable organs. 2. Allow for recovery from/repair of preexisting damage within the donor organ (eg, caused by warm ischemia or steatosis). 3. Allow for prolonged preservation times. There is increasing evidence from large animal models of transplantation and now preliminary evidence from clinical trials that organ preservation by normothermic machine perfusion (NMP) is superior to static cold storage. (8-10) Preservation of organs by replicating the physiological environment may help to solve many of the challenges of organ preservation. Experimentally, this has been shown to be effective, enabling successful transplantation of livers under circumstances that would not be survivable with conventional preservation methods. (11) The avoidance of cooling and, potentially, the repair of preexisting injury might allow more livers from ECDs to be transplanted The members of the research team that were responsible for the organ perfusions and transplantation have associations with the device manufacturer OrganOx. Peter J. Friend and Constantin C. Coussios, in addition to being full-time academics at the University of Oxford, receive consultancy payments as nonexecutive medical and technical directors of OrganOx. Thomas Vogel, Jens G. Brockmann, Constantin C. Coussios, and Peter J. Friend are shareholders of OrganOx, Ltd. Nigel D. Heaton is an employee of and is on the speaker s bureau of Astellas. Constantin C. Coussios owns stock in, is an employee of, consults for, and holds intellectual property rights for OrganOx. Copyright VC 2016 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI /lt safely. The objective of this study was to evaluate the technical feasibility of NMP of high-risk human liver grafts for an extended preservation period, long enough to achieve the potential benefits mentioned above. Preclinical data are presented and analyzed from a series of human liver grafts, declined for clinical transplantation, and perfused on a NMP device for 24 hours. Patients and Methods Thirteen human liver grafts were included in this study. All organs were retrieved for the purpose of transplantation, but discarded during postretrieval assessment at the donor hospital or at the transplant center. No organ offered for inclusion in the trial was rejected by the investigators. Perfusions were performed at the Oxford Transplant Centre. Hemodynamic and biochemical perfusion parameters were investigated. Perfusate was sampled at regular intervals during the perfusion period and, at the end of each perfusion, livers were sectioned for histological evaluation of the reperfusion injury. The study was evaluated and approved by the national ethics review committee of the United Kingdom (REC reference 09/H0605/ 85). PREPARATION OF LIVER GRAFTS All livers were retrieved during standard multiorgan recovery at various donor hospitals. Packaging and shipment followed routine clinical practice and was maintained until the start of the process of perfusion. Upon arrival at the research laboratory, livers were immediately assessed and prepared for connection to the NMP circuit. Back-table preparation included the following: cholecystectomy was performed the diaphragmatic remnant was excised and phrenic veins oversewn; capsular tears, if present, were repaired; accessory arteries were reconstructed, where necessary to provide a single inflow (by anastomosis to splenic or gastroduodenal arteries; Fig. 1A,B). Cannulation of inflow and outflow vessels was carried out (Fig. 1C,D). Usually, a 20-Fr and 10-Fr cannula was inserted into the portal vein (PV) and hepatic artery (HA), respectively. A 28-Fr cannula was placed in the inferior vena cava (IVC) via the infrahepatic IVC, with its orifice positioned at the level of the hepatic veins. The suprahepatic IVC was closed with a running suture (or ligated if long enough). The bile duct was flushed and cannulated with a 6-Fr or 8-Fr tube in order to collect 208 ORIGINAL ARTICLE

3 LIVER TRANSPLANTATION, Vol. 23, No. 2, 2017 VOGEL ET AL. FIG. 1. (A) Liver with accessory left HA. (B) Magnification of anastomosis to a branch of proper HA. (C) Cannulation of PV, HA, IVC, and bile duct. (D) Cannulae of appropriate size. and monitor bile production. All cannulae were fixed with ligatures at the distal end of each vessel in order to minimize the loss of vessel length. Back-table preparation and priming of the perfusion machine were usually carried out in parallel. The average time needed for graft preparation was minutes, depending on vascular variations. Back-table preparation of all liver grafts was performed with livers immersed in ice cold preservation solution (usually University of Wisconsin). Directly before connection of the liver to the NMP device, this was flushed out with approximately 1000 ml of crystalloid solution (Sterofundin; B. Braun Medical, Ltd., Sheffield, UK) at room temperature. PERFUSION DEVICE The perfusion apparatus used in this trial was the preclinical prototype of a fully automated NMP system based on the principle of autoregulation of hepatic perfusion as described previously. (12,13) The underlying perfusion methodology had been extensively tested in large animal models of machine perfusion and organ preservation. (10,11,14,15) The system was designed to be capable of supporting organs for prolonged preservation periods by perfusion with an oxygenated blood-based perfusate and nutrients at normal body temperature (Fig. 2). Basic tubing components of the perfusion circuit were sourced from cardiopulmonary bypass suppliers. The principle components were the following: a blood reservoir (CAPIOX 1500 ml Flexible Venous Reservoir; Terumo, UK), centrifugal pump (BP 50 Centrifugal Pump; Medtronic, Watford, UK), and membrane oxygenator (Quadrox-i, Maquet Cardiopulmonary, Tyne & Wear, UK and Dideco EOS Extracorporeal Membrane Oxygenation M Phisio oxygenator; Sorin, Italy). Medical grade silicone tubing with internal diameters of 1/4 and 3/8 inches was used throughout the circuit (Raumedic AG, Helmbrechts, Germany). Heat exchange was achieved via in-line peltier heating elements. Perfusion pressures and flows were monitored continuously for the HA, PV, and IVC. Pressures (portal artery, caval artery, and HA) were measured in-line the cannulae using single-use pressure sensors (PSs; Pendo TECH, Princeton, NJ). Hemodynamic control was based on caval and HA pressures, and achieved by adjustment of the centrifugal pump speed and inflowcontrol into the reservoir bag by an adjustable pinch valve. The circuit provided dual inflow into the liver: the cannula within the HA was fed directly from the centrifugal pump, whereas the PV received blood under gravitational pressure from the reservoir. In this way, pressures within both systems were maintained within physiological limits. The system was also regulated via closed loop control for perfusate temperature and partial pressures of oxygen and carbon dioxide (adjusted by means of a variable air/o 2 supply). ORIGINAL ARTICLE 209

4 VOGEL ET AL. LIVER TRANSPLANTATION, February 2017 FIG. 2. (A) Diagram of perfusion circuit (illustration not to scale). (B) Picture of perfusion machine in operating mode. PERFUSION SOLUTION While backbench preparation of the liver was performed, the perfusion circuit was mounted on the perfusion device and primed with approximately 1200 ml of perfusate. The perfusate was based on blood group compatible packed red blood cells (3-4 units) and 500 ml of crystalloid solution (Sterofundin; B. Braun Medical, Ltd.). The perfusate was preheated to normal body temperature (37 8C); this temperature was maintained throughout the perfusion period. During priming of the circuit, before connection of the liver, the perfusate was supplemented with bolus additions: cefuroxime (750 mg), heparin (10,000 IU), and calcium gluconate 10% (10 ml in order to counteract the calcium binding of citrate). ph was adjusted by titration with sodium bicarbonate (20-50 mmol) to the physiologic level. No further corrections of ph were done during the perfusion. Once the perfusate within the circuit was optimized, a sample was obtained for full blood count, electrolytes, urea, creatinine, and liver function tests. During the perfusion the following solutions were infused at constant rates: 1. Total parenteral nutrition solution (Nutriflex; B. Braun Medical, Ltd.; 15 ml/hour) was used as an amino acid and glucose source (lipids were avoided). 2. Heparin (830 IU/hour) to prevent clotting, particularly within the oxygenator. 3. Bile salts (140 mg/hour; Sodium Taurocholate, New Zealand Pharmaceuticals, Palmerston, New Zealand) in isotonic saline, to compensate for the loss of bile salts due to the drainage of bile. 4. Prostacyclin (8 mg/hour; Flolan, GSK, Mississauga, Ontario, Canada) to optimize microperfusion. Flow rates of nutrition (on/off times; amino acid and glucose) and insulin were adjusted according to a sliding scale, based on measured perfusate glucose levels. The liver was placed on a silicone sling in a sterile perfusion chamber, and perfusion commenced. Any fluid that leaked from the liver during perfusion was recirculated by means of an automated roller pump. Bile was collected via the silicone bile tube secured in the common bile duct. DATA COLLECTION During the 24-hour perfusion, regular sampling of perfusate was performed as follows: a full blood count (1.2 ml) was taken every 2 hours for 8 hours and every 4 hours subsequently. Additionally, 200 ml of blood was drawn from the afferent side of the oxygenator every hour and analyzed on a standard blood gas analyzer (ABL 725, Radiometer, Brønshøj, Denmark) reflecting venous blood gases. Arterial blood gases from the efferent side of the oxygenator were analyzed continuously using a Terumo CDI Blood Parameter 210 ORIGINAL ARTICLE

5 LIVER TRANSPLANTATION, Vol. 23, No. 2, 2017 VOGEL ET AL. TABLE 1. Donor Data and Organ Procurement Parameters Parameter Mean Range Remarks Age, years BMI, kg/m CIT, minutes NHBD 9/13 (69%) WIT, minutes Biochemistry Gamma-glutamyltransferase, mmol/l Reference range: ALT, U/L Reference range: Sodium, mmol/l Reference range: Bilirubin, mmol/l Reference range: 3-17 Albumin, g/l Reference range: Monitoring System 500 (Ann Arbor, MI). For this, a CDI510 shunt sensor was inserted into a bypass line and calibrated before priming of the circuit using the Terumo CDI540 calibrator. Parameters measured and displayed on the device included ph, partial pressure of carbon dioxide, partial pressure of oxygen, potassium, temperature, oxygen saturation, base excess (BE), and HCO 3. Data of the Terumo blood gas analyzer was transferred to the main control board for storage (all parameters) and used as a feedback signal for the automated temperature and oxygenation control loops. Data for all system parameters were stored once every second. A further 6 ml of perfusate were drawn off the circuit for complete biochemistry analysis every 2 hours for the first 8-hour perfusion period and every 4 hours thereafter. Samples were centrifuged at 12,000 rpm for 8 minutes and 3 aliquots of supernatant of 1000 ml each were deep frozen ( 80 8C). HISTOLOGY Biopsies were obtained during back-table preparation before the start of machine perfusion. At the end of perfusion, livers were sectioned and random samples were obtained from the right and left liver lobes, central region, and hilar region. Samples were placed in formalin and submitted for paraffin embedding and hematoxylin-eosin staining, as per standard protocols. Histological examination was carried out by a single specialist liver histopathologist (A.Q.), using a BX51 Olympus microscope (Waltham, MA). The sections were scored for neutrophil clusters (0 5 none; 1 5 occasional, found after very scrupulous search at high magnification [4003]; 2 5 scattered, easily seen at magnification; 3 5 scattered in most lobules; 4 5 many neutrophils clusters in close proximity on the verge of confluence; 5 5 dense confluent sheets); apoptosis or necrosis (0 5 absent; apoptotic bodies per magnification field [highest number per field recorded]; 2 5 >2 apoptotic bodies per magnification field [highest number per field recorded]; 3 5 focal hepatocyte necrosis, 1-2 foci per magnification field [highest number per field recorded]; or more foci per magnification field [highest number per field recorded]; 5 5 focal confluent hepatocellular necrosis; 6 5 multiple foci of confluent necrosis; 7 5 areas of necrotic parenchyma are most extensive than the viable parenchyma in the sample examined; 8 5 a limited amount of viable parenchyma is identified); mitosis (number of mitoses in a 10 high-power field at magnification); canalicular cholestasis (0 5 none; 1 5 mild; 2 5 moderate; 3 5 severe); ductular cholestasis (0 5 none; 1 5 mild; 2 5 moderate; 3 5 severe); and signs of cholangitis (0 5 absent; 1 5 present). Steatosis was assessed as large droplet steatosis (a single large steatotic vacuole pushing the hepatocyte nucleus to the periphery) and small droplet steatosis (1 or more small cytoplasmic vacuoles mostly associated with a centrally placed nucleus) as 2 separate yet identical scores (0 5 none; 1 5 <26%; %-50%; %-75%; 4 5 >75%). Large droplet steatosis was scored first. The small droplet steatosis was scored afterward, this second score representing the percentage of hepatocytes without large droplet steatosis. STATISTICAL ANALYSIS Basic statistical data analysis was performed using IBM SPSS Statistics, version 21 (IBM, Armonk, NY). Data were analyzed using descriptive methods and presented as percentages or mean values 6 standard deviation (SD). Statistical analysis was performed using Student t test and v 2 analysis. P < 0.05 was considered significant. ORIGINAL ARTICLE 211

6 VOGEL ET AL. LIVER TRANSPLANTATION, February 2017 Age, years TABLE 2. Organ Procurement Parameters and Reasons for Discard for Individual Livers Donor Sex BMI, kg/m 2 NHBD WIT, minutes CIT, minutes Remarks Reason For Discard* Steatosis Macroscopic Steatosis Microscopic Liver 1 51 Male 23 Yes Steatosis, time to Mild Mild arrest 1 hour Liver 2 42 Male NA Yes Donor medical history, None None transaminase levels Liver 3 76 Female 35 No NA 964 Severe steatosis Severe Moderate Liver 4 48 Male NA No NA 445 Steatosis Moderate Severe Liver 5 70 Male 25 Yes arteries Incidental tumor None None in donor Liver 6 73 Female NA Yes Steatosis, donor Mild-moderate Mild age, DCD Liver 7 53 Female NA Yes Donor medical None None history, DCD Liver 8 72 Female 28 Yes arteries Steatosis Moderate Mild Liver 9 58 Female 35 No NA 730 Severe steatosis Severe Very severe Liver Male 35 Yes arteries Steatosis Moderate None Liver Female NA Yes 13 Severe steatosis, Moderate Mild donor age, DCD Liver Male 27 Yes Logistical reasons Mild None Liver Female 26 No NA arteries Steatosis Mild Mild *As given by the transplant center. Steatosis judged on macroscopic appearance in the research laboratories at the Oxford Transplant Centre. Steatosis judged on microscopic appearance. Perfusion postponed for logistical reasons. Results DONOR DATA Thirteen donor livers were included in this trial. Demographic data and predonation liver function parameters are shown in Table 1. Organ procurement data and reasons for discard for individual livers are shown in Table 2. The mean age of the organ donors was 62 years, ranging from 42 to 76 years. Organ procurement was performed after cardiac arrest (donation after circulatory death [DCD]) in 9 of 13 (69%) liver donors with a mean warm ischemia time (WIT) of 11.3 minutes (range, 3-15 minutes). WIT comprised the time from circulatory arrest to the perfusion of the organs with cold preservation solution in situ (ie, the asystolic or acirculatory warm period). Mean cold ischemia time (CIT) was 570 minutes, ranging from 280 to 964 minutes. All liver grafts had been declined for transplantation by every UK liver transplant unit. Steatosis was the reason for discard in 9 of 13 (69%) organs. One organ was rejected for transplantation because of raised transaminase levels in the donor. One DCD organ was discarded because the agonal period (from withdrawal of ventilation to cardiac arrest) was judged excessive. Two livers were discarded for nonliver-related reasons. In 1 case, the incidental finding of a solid lung tumor led to discard of all organs for transplantation. Liver 12 was discarded for logistical reasons at the recipient center. Upon arrival at the research laboratories, livers were classified for macroscopic appearance of steatosis by 2 independent liver transplant surgeons. This classification correlated well with reasons for discard of the liver grafts and with microscopic assessment of steatosis in 10 of 13 liver grafts. In 3 livers, steatosis was overjudged as being moderate although microscopy revealed mild (2 grafts) or no steatosis (1 graft). PERFUSION General A total of 11 of 13 livers were perfused for 24 hours. Two liver perfusions were performed using standard cardiac surgery oxygenators (Quadrox-i, Maquet). In these perfusions, performance of oxygenators decreased after 19 hours, and perfusions were therefore stopped at this time point. For all subsequent perfusions, oxygenators were used that are specified for longer-term perfusion; these are made for use in extracorporeal membrane oxygenation therapy (Sorin Dideco 212 ORIGINAL ARTICLE

7 LIVER TRANSPLANTATION, Vol. 23, No. 2, 2017 VOGEL ET AL. FIG. 3. Blood flows for individual liver perfusions. The bold line indicates mean blood flow. Blood flow did not change during 24 hours of normothermic perfusion. (A) Portal flow by gravity. (B) Arterial blood flow (calculated). (C) Total IVC blood flow. EOS950) and provided stable performance for 24 hours. In terms of hepatic arterial anatomical anomalies, 4 (31%) livers had accessory right or left arteries. In all cases, FIG. 4. Pressures during 24 hours of normothermic perfusion for individual livers. The bold line indicates mean pressure. Control was achieved by regulation of pump speed and inflow to portal venous blood reservoir to keep values within physiologic range. Portal pressure was based on the principle of autoregulation. (A) Portal pressure, (B) HA pressure, and (C) IVC pressure. All pressures were measured in-line showing little variation between livers. arterial reconstruction was performed before connection of the liver. The standard procedure was by anastomosis to 1 of the celiac trunk branches, which allowed single ORIGINAL ARTICLE 213

8 VOGEL ET AL. LIVER TRANSPLANTATION, February 2017 FIG. 5. (A) Bile production started on average after 1 hour and maintained stable in most livers throughout the perfusion. (B) In 3 livers, bile production was almost absent whereas bile production varied between 3 ml/hour and 11.2 ml/hour for the remaining. arterial cannulation (Fig. 1B). Vessels were patent in all cases after 24 hours of perfusion and bleeding was minimal during normothermic perfusion with fully heparinized perfusate at normal blood pressures. Hemodynamic Parameters PRESSURES/FLOWS The mean portal blood flow was L/minute throughout 24 hours of normothermic perfusion with little variation between livers (Fig. 3A). Mean arterial and total perfusion blood flows were L/ minute and L/minute, respectively (Fig. 3B,C). The (nonpulsatile) arterial pressure was under automated control and remained within the physiologic range during the entire perfusion period for all perfusions. Mean portal, HA, and IVC pressures were FIG. 6. Oxygen consumption calculated according to Fick s principle. All livers showed evidence of oxygen consumption throughout the perfusion. Mean O 2 consumption varied from 1.01 to 2.26 ml O 2 /minute adjusted to 100 g liver weight (B). NB O 2 consumption could not be calculated for liver 3 due to a failure of the external blood gas analyzer during perfusion mm Hg, mm Hg, and mm Hg, respectively (Fig. 4A-C). Metabolic Function BILE PRODUCTION Total bile production in 24 hours and mean hourly bile production showed marked differences between livers, as shown in Fig. 5. In general, bile production started within 1-2 hours after beginning NMP and lasted throughout the 24-hour period (Fig. 5A). Mean hourly bile production ranged from 0.5 to 11.2 ml/hour (Fig. 5B). In 3 livers, bile production was almost absent. OXYGEN CONSUMPTION Oxygen consumption was calculated according to Fick s principle based on arterial and venous blood gas 214 ORIGINAL ARTICLE

9 LIVER TRANSPLANTATION, Vol. 23, No. 2, 2017 VOGEL ET AL. FIG. 7. Glucose utilization was calculated based on on/off times of the nutrition infusion pump under control of perfusate glucose levels. analysis and total blood flow. All livers showed evidence of oxygen consumption throughout the perfusion (Fig. 6). Mean O 2 consumption varied between 1.01 and 2.26 ml O 2 /minute per 100 g liver weight. GLUCOSE UTILIZATION Infusion of parenteral nutrition (Nutriflex plus, B. Braun Medical, Ltd.) was started at a fixed rate once perfusate glucose levels dropped below 12 mmol/l and paused again when levels exceeded 15 mmol/l. Nutrition was deferred during the first 4 hours in all perfusions due to high starting glucose levels. Thereafter, glucose levels could be kept within the desired range by algorithmcontrolled automatic adjustments of on/off times of nutrition administration and insulin perfusion flow rates based on manual blood glucose measurement (BM) reader input. On the basis of on/off times of the infusion pump, total glucose administration and consumption was calculated (Fig. 7). The mean glucose utilization varied between livers ranging from 3.3 mg of glucose per kg liver tissue per minute up to 21.1 mg/kg/minute. Biochemistry ALANINE AMINOTRANSFERASE Alanine aminotransferase (ALT) levels were measured at regular intervals in plasma samples of perfusate (Fig. 8). Differences were seen in both the start level of ALT and the increase/decrease during subsequent 24-hour normothermic perfusion. Absolute levels ranged from 816 to 8121 U/L at the beginning of normothermic perfusion FIG. 8. ALT levels in perfusate were measured regularly in plasma samples of perfusate. Differences exist in both the start value of ALT and increments, during subsequent 24-hour normothermic perfusion. Shown are (A) absolute values in perfusate and (B) changes from postreperfusion levels to end values. (hour 0); end levels after 24 hour of machine perfusion ranged from 817 to 8982 U/L. The amount of ALT increase did not correlate with end-perfusion levels. LACTATE DEHYDROGENASE Lactate dehydrogenase (LDH) plasma levels are shown in Fig. 9. Perfused livers showed a high variation in endperfusionldhlevels(fig.9a).meanlevelattheendof perfusion was 13,593 U/L with livers 3, 6, and 11 exhibiting very high values (32,520, 20,031, and 27,850 U/L, respectively). However, LDH increase (Fig. 9B) did not correlate with end-perfusion levels, and LDH increase was low or absent for most livers during the perfusion period. ph ph was adjusted before connection of the liver to the device as described above. After beginning of perfusion, ORIGINAL ARTICLE 215

10 VOGEL ET AL. LIVER TRANSPLANTATION, February 2017 FIG. 10. ph was adjusted before connection of the liver and maintained stable for all liver perfusions without further corrections. Shown are the mean ph 6 SD for 13 normothermic liver perfusions. FIG. 9. LDH levels during 24 hours of normothermic perfusion. Perfused livers showed a high variation in end-perfusion LDH levels (A). LDH increase (B) did not correlate with endperfusion levels, and LDH increase was low or absent for most livers during the perfusion period. no further corrections were done. ph remained stable in all liver perfusions, the greatest decrease in ph was seen in liver 13 (ph drop from 7.49 to 7.13 after 24 hours; Fig. 10). Mean ph change for all livers was (hour 1/hour 24). Similar findings were observed for changes in HCO 3 and BE (data not shown). HISTOLOGY Representative figures of 2 sets of specimens from 2 livers are shown in Fig. 11. A neutrophil infiltrate was seen in 8 of 13 preperfusion liver samples (62%), but no neutrophils were found in postperfusion samples. The degree of steatosis was variable in this set of biopsies (Fig. 12A,B). In 2 cases, there was no steatosis, either small or large droplet type, in either preperfusion or postperfusion specimens. In 3 cases, there was no steatosis (small or large droplet-type) in the preperfusion specimen and a mild degree of steatosis (of both types in 2 specimens and large droplet only in 1 specimen) was noted in the postreperfusion specimen. In 1 case, the opposite was observed, with mild large and small droplet steatosis in the preperfusion specimen and no steatosis at all in the postperfusion samples. In most (6/7) of the remaining 7 cases, small and large droplet steatosis coexisted in both prereperfusion and postreperfusion specimens, without appreciable preperfusion to postperfusion change, and in combinations which indicated that at least 50% of hepatocytes were affected by steatosis in most cases. Regardless of the steatosis status, apoptosis or necrosis was identified in all livers except 1 (Fig. 12C). Confluent necrosis was present in 6 livers. In 1 of these, it was present in both preperfusion and postperfusion samples. In 4 livers, it was present in the postperfusion specimens only. In 1 case, confluent necrosis was present in the preperfusion sample, although the postperfusion sample had multiple foci of hepatocellular necrosis. In the other 3 livers, nonconfluent foci of hepatocellular necrosis were present in the postperfusion specimens. In the 2 remaining livers, the changes were minimal and in the form of occasional apoptotic bodies in postperfusion specimens only. Whether the presence of necrosis in the postperfusion samples alone represents true occurrence or sampling variation remains uncertain. No mitoses were found in most (20/26) specimens and where present did not seem to correlate with other signs of liver injury. Canalicular or cholangiolar cholestasis or cholangitis was present in 216 ORIGINAL ARTICLE

11 LIVER TRANSPLANTATION, Vol. 23, No. 2, 2017 VOGEL ET AL. FIG. 11. Light microscopy of 2 liver grafts preperfusion and postperfusion (hematoxylineosin). (A,B) Liver 12. Both preperfusion and postperfusion show mild changes only. (C,D) Liver 3. Both perfusion and postperfusion samples show confluent hepatocellular necrosis with a tendency to spare periportal hepatocytes. Large droplet steatosis is also present. very few specimens (1 out 26 specimens, for each feature) and never coexisted. Discussion Static cold storage is the current universal preservation method for all transplant organs. The principles underlying cold preservation are the slowing of metabolism (by cooling) and reduction of cell swelling due to the composition of preservation solutions. Despite this, organs sustain injury at several consecutive stages: 1. Warm ischemia prior to preservation (in case of DCD or hemodynamic instability). 2. Cold preservation injury. 3. Ischemic rewarming during surgical implantation. 4. Ischemia/reperfusion injury. (16) Even with the most effective preservation solutions, cold storage aggravates graft injury at the time of transplantation. This is due to the separate factors, first of the duration of ischemia and, second, the specific effects of cooling. (17,18) For this reason, preservation time must be minimized within current organ preservation practice. Preventing (rather than slowing) the depletion of cellular energy charge has the potential to improve graft quality substantially and is a major objective of normothermic oxygenated preservation methods. In contrast to cold preservation, the purpose of normothermic preservation is to maintain cellular metabolism by replicating the physiological environment. On the other hand, NMP is associated with increased costs and complexity. Any system failure poses an immediate risk of serious harm to the organ; system reliability will be of utmost importance to the success of any clinical device. On the basis of the experience of numerous liver perfusions in large animal models of liver transplantation, our group has developed a mobile, automated perfusion device for NMP. The system relies on the principle of autoregulation of hepatic blood flows, as described before. The automated system was tested extensively with porcine liver perfusions in which we found it to be as effective as the previous manually controlled variants of the same circuit. Using this device, we have shown in animal studies of porcine liver transplantation that longterm liver preservation, in excess of 40 hours, is feasible with successful transplantation ORIGINAL ARTICLE 217

12 VOGEL ET AL. LIVER TRANSPLANTATION, February 2017 FIG. 12. Scores of necrosis and steatosis for individual liver grafts before and after perfusion. (A,B) The majority of livers were affected by steatosis to a varying degree, with only little differences in preperfusion and postperfusion samples of liver grafts. (C) Some livers revealed substantial levels of necrosis in postperfusion biopsies. (manuscript submitted). Also, the first clinical data have now been published recently. (19) In this phase 1 study, 20 human livers were transplanted after a median time of NMP of 9.3 hours. The study presented here has 2 objectives: first, to investigate the hemodynamic and biological behavior of human liver grafts during NMP; second, to test the stability and reliability of hardware components and control algorithms of the device with human liver grafts for an extended preservation period of 24 hours. In (pressure-controlled) hypothermic machine perfusion experiments, a progressive decrease of perfusion flow has been described. (20) This was not the case in this series: pressures and flows in this study were stable throughout the perfusion period. Arterial and portal flows increased slightly during the first 2 hours and remained stable for the rest of the perfusion. In the experimental setting (porcine), we have described an indicative value of portal pressures; increased portal pressures during NMP correlated with unsuccessful transplantationinaporcinetransplantation model. (11) In this series, we did not find large differences in portal pressures between perfused livers. This might reflect the quality of organs included in this study, but some organs were discarded for liver-unrelated reasons (eg, malignancy), and we might expect a difference between these and livers discarded following prolonged WITs. This was not observed, and there is no obvious explanation. Besides portal pressure profiles, we also found oxygen consumption to be a marker of limited utility within this series. There was a trend toward higher oxygen consumption in good-quality livers based on overall judgment (including histology, bile production, etc.). However, these differences did not reach statistical significance. Published data on oxygen consumption in isolated perfused organs are conflicting: in a study by Dutkoswski et al. higher oxygen consumption was correlated to poor outcome, (21) whereas other groups have found the opposite. (22,23) Bile production is thought to be a valid marker of liver viability in vivo as well as in experimental (re-) perfusion models. In a similar study using discarded human livers, the Groningen group suggested bile production to be the most indicative factor of hepatic function. (24) In this study, we demonstrated good correlation between bile production during normothermic perfusion and postperfusion histologic examination. Bile production was significantly correlated with better histological grading (P ). This difference became significant after 4 hours of normothermic perfusion and lasted for the rest of the perfusion period. Overall the level of necrosis seen in postperfusion biopsies was high, but we also found livers with almost no signs of necrosis, except minor findings of sparse apoptotic liver cells. A possible explanation is the 218 ORIGINAL ARTICLE

13 LIVER TRANSPLANTATION, Vol. 23, No. 2, 2017 VOGEL ET AL. prolonged period of cold ischemia to which all livers were exposed to before normothermic perfusion was started (nearly all discards took place once the donor organs had been transported to the transplant unit). The mean CIT in this study was 9.5 hours. In porcine liver perfusion studies, even short periods of cold ischemia when applied before NMP showed negative impact on perfusion parameters. Evidence of greater graft injury was seen in ischemically damaged porcine livers after only 4 hours of cold preservation. (15,25) This finding corroborates the previously suggested need for the perfusion device to be mobile and capable of being carried to the donor site in the clinical setting. In this study, we have used blood-based perfusate solution: human packed red blood cells diluted in crystalloid solution. For experimental purposes, the supplyofbloodwasrestrictedtoexpiredredblood cells, which possibly accounts for perfusion details observed during this study: the level of potassium measured in most perfusions was high (data not shown). Potassium levels were high already after priming of the system before connection of the liver to the device; during perfusion a mean increase of no more than 15% was observed. This finding is supported by the course of lactate levels; mean lactate level before connection of the liver was mmol/l. Under NMP, lactate was cleared (mean end value mmol/l). This is in clear contrast to what is described in hypothermic machine perfusion where lactate usually increases constantly over the perfusion period and is a sensitive marker of tissue ischemia. (26,27) A number of conclusions may be drawn from this study: first, it is possible to perfuse high-risk livers consistently for 24 hours. Second, there is evidence that, overall, there is little evidence of deterioration of the functional and histological characteristics of damaged livers during prolonged perfusion. Prolonged normothermic preservation might allow time for recovery (eg, cellular adenosine triphosphate repletion although we present very limited evidence in this regard), and it might also offer a useful platform for therapeutic delivery (eg, steatosis reduction). Prolonged NMP might remove the critical time constraints associated with the management of high-risk livers (particularly valuable in parts of the world where transport times effectively exclude the use of such livers from outlying donor hospitals). Finally (and possibly of greatest value), NMP allows functional parameters to be measured as a marker of the viability of the organ: prolonged NMP might allow clinicians to accept a high-risk organ on the basis that key prognostic information will become available before the need to commit a patient to the risks of surgery. Normothermic preservation technology has now progressed to the phase of clinical implementation, but much research is still needed, including studies of discarded livers as well as experimental models, in order to maximize its benefits and understand its limitations. REFERENCES 1) Abt PL, Desai NM, Crawford MD, Forman LM, Markmann JW, Olthoff KM, Markmann JF. Survival following liver transplantation from non-heart-beating donors. Ann Surg 2004;239: ) D alessandro AM, Hoffmann RM, Knechtle SJ, Odorico JS, Becker YT, Musat A, et al. Liver transplantation from controlled non-heart-beating donors. Surgery 2000;128: ) Bahra M, Neuhaus P. Liver transplantation in the high MELD era: a fair chance for everyone? Langenbecks Arch Surg 2011; 396: ) Fischer-Fr ohlich CL, Lauchart W. Expanded criteria liver donors (ECD): effect of cumulative risks. Ann Transplant 2006; 11: ) Brooks AD, Ng B, Liu D, Brownlee M, Burt M, Federoff HJ, Fong Y. Specific organ gene transfer in vivo by regional organ perfusion with herpes viral amplicon vectors: implications for local gene therapy. Surgery 2001;129: ) Cheung ST, Tsui TY, Wang WL, Yang ZF, Wong SY, Ip YC, et al. Liver as an ideal target for gene therapy: expression of CTLA4Ig by retroviral gene transfer. J Gastroenterol Hepatol 2002;17: ) Kozaki K, Uchiyama M, Nemoto T, Degawa H, Takeuchi H, Matsuno N, et al. Usefulness of a combination of machine perfusion and pentoxifylline for porcine liver transplantation from non-heart-beating donors with prolonged hypotension. Transplant Proc 1997;29: ) Perera T, Mergental H, Stephenson B, Roll GR, Cilliers H, Liang R, et al. First human liver transplantation using a marginal allograft resuscitated by normothermic machine perfusion. Liver Transpl 2016;22: ) Watson CJ, Kosmoliaptsis V, Randle LV, Russell NK, Griffiths WJ, Davies S, et al. Preimplant normothermic liver perfusion of a suboptimal liver donated after circulatory death. Am J Transplant 2016;16: ) Butler AJ, Rees MA, Wight DG, Casey ND, Alexander G, White DJ, Friend PJ. Successful extracorporeal porcine liver perfusion for 72 hr. Transplantation 2002;73: ) Brockmann J, Reddy S, Coussios C, Pigott D, Guirriero D, Hughes D, et al. Normothermic perfusion: a new paradigm for organ preservation. Ann Surg 2009;250: ) ImberCJ,StPeterSD,LopezdeCenarruzabeitiaI,PigottD,James T, Taylor R, et al. Advantages of normothermic perfusion over cold storage in liver preservation. Transplantation 2002;73: ) St Peter SD, Imber CJ, Lopez I, Hughes D, Friend PJ. Extended preservation of non-heart-beating donor livers with normothermic machine perfusion. Br J Surg 2002;89: ) Friend PJ, Imber C, St Peter S, Lopez I, Butler AJ, Rees MA. Normothermic perfusion of the isolated liver. Transplant Proc 2001;33: ) Reddy SP, Bhattacharjya S, Maniakin N, Greenwood J, Guerreiro D, Hughes D, et al. Preservation of porcine nonheart-beating donor livers by sequential cold storage and warm perfusion. Transplantation 2004;77: ORIGINAL ARTICLE 219

14 VOGEL ET AL. LIVER TRANSPLANTATION, February ) Clavien PA, Harvey PR, Strasberg SM. Preservation and reperfusion injuries in liver allografts. An overview and synthesis of current studies. Transplantation 1992;53: ) Carini R, Autelli R, Bellomo G, Albano E. Alterations of cell volume regulation in the development of hepatocyte necrosis. Exp Cell Res 1999;248: ) Petrosillo G, Ruggiero FM, Paradies G. Role of reactive oxygen species and cardiolipin in the release of cytochrome c from mitochondria. FASEB J 2003;17: ) Ravikumar R, Jassem W, Mergental H, Heaton N, Mirza D, Perera MT, et al. Liver transplantation after ex vivo normothermic machine preservation: a phase 1 (first-in-man) clinical trial. Am J Transplant 2016;16: ) van der Plaats A, Maathuis MH, 0 T Hart NA, Bellekom AA, Hofker HS, van der Houwen EB, et al. The Groningen hypothermic liver perfusion pump: functional evaluation of a new machine perfusion system. Ann Biomed Eng 2006;34: ) Dutkowski P, Furrer K, Tian Y, Graf R, Clavien PA. Novel short-term hypothermic oxygenated perfusion (HOPE) system prevents injury in rat liver graft from non-heart beating donor. Ann Surg 2006;244: ) Bessems M, Doorschodt BM, Kolkert JL, Vetelainen RL, van Vliet AK, Vreeling H, et al. Preservation of steatotic livers: a comparison between cold storage and machine perfusion preservation. Liver Transpl 2007;13: ) Jain S, Lee CY, Baicu S, Duncan H, Xu H, Jones JW Jr, et al. Hepatic function in hypothermically stored porcine livers: comparison of hypothermic machine perfusion vs cold storage. Transplant Proc 2005;37: ) Sutton ME, op den Dries S, Karimian N, Weeder PD, de Boer MT, Wiersema-Buist J, et al. Criteria for viability assessment of discarded human donor livers during ex vivo normothermic machine perfusion. PLoS One 2014;9:e ) Reddy S, Greenwood J, Maniakin N, Bhattacharjya S, Zilvetti M, Brockmann J, et al. Non-heart-beating donor porcine livers: the adverse effect of cooling. Liver Transpl 2005;11: ) Bessems M, Doorschodt BM, Dinant S, de Graaf W, van Gulik TM. Machine perfusion preservation of the pig liver using a new preservation solution, polysol. Transplant Proc 2006;38: ) Monbaliu D, Vekemans K, De Vos R, Brassil J, Heedfeld V, Qiang L, et al. Hemodynamic, biochemical, and morphological characteristics during preservation of normal porcine livers by hypothermic machine perfusion. Transplant Proc 2007;39: ORIGINAL ARTICLE