Hypothermic machine perfusion (HMP) is widely used to preserve kidneys for transplantation with improved results over cold storage (CS). To date, successful transplantation of livers preserved with HMP has been reported only in animal models. In this, the first prospective liver HMP study, 20 adults received HMP-preserved livers and were compared to a matched group transplanted with CS livers. HMP was performed for 3–7 h using centrifugal perfusion with Vasosol® solution at 4–6°C. There were no cases of primary nonfunction in either group. Early allograft dysfunction rates were 5% in the HMP group versus 25% in controls (p = 0.08). At 12 months, there were two deaths in each group, all unrelated to preservation or graft function. There were no vascular complications in HMP livers. Two biliary complications were observed in HMP livers compared with four in the CS group. Serum injury markers were significantly lower in the HMP group. Mean hospital stay was shorter in the HMP group (10.9 ± 4.7 days vs. 15.3 ± 4.9 days in the CS group, p = 0.006). HMP of donor livers provided safe and reliable preservation in this pilot case-controlled series. Further multicenter HMP trials are now warranted.
The shortage of organ donors has led to increased transplantation of organs from extended criteria donors (ECD). In orthotopic liver transplantation (OLT), organs from elderly, steatotic or donation after cardiac death (DCD) donors are at higher risk of moderate to severe preservation injury (PI) and PI-related biliary complications (1). Severe PI may contribute to primary nonfunction (PNF) and a need for retransplantation in 3–7% of all liver transplants (2). Lesser degrees of PI can cause early allograft dysfunction (EAD), often associated with the need for prolonged intensive care, cholestasis, increased morbidity and infectious complications.
Static hypothermic cold storage (CS), currently the only technique in clinical use for liver preservation, has been essentially unchanged for two decades. In addition to University of Wisconsin solution (UW), Custodiol® (HTK) and Celsior® solutions have become available, but without evidence of superiority (3).
In kidney transplantation, graft preservation with hypothermic machine perfusion (HMP) is widely used (4,5). HMP provides continuous circulation of preservation solution and metabolic substrates for adenosine triphosphate (ATP) generation. A ‘washout effect’ removes and dilutes waste products from direct endothelial/parenchymal contact, in turn stabilizing the microcirculation. There is strong evidence that renal HMP improves early graft function (6) and have improved tissue ATP concentrations upon reperfusion (7). Furthermore, recent reports suggest a long-term graft survival benefit for HMP kidneys (8,9). HMP also permits pharmacologic manipulation and pretransplant assessment of grafts (7) and has enhanced the use of ECD kidneys (5).
Studies of HMP in animal models of liver transplantation, including our own (10), have been promising, but except for Brettschneider and Starzl's early attempts at liver HMP with a combination of autologous blood and preservation solution (11), there have been no clinical trials of liver HMP (12–15). Here, we describe a prospective cohort pilot study designed to evaluate the safety and feasibility of liver preservation with HMP in 20 liver transplant patients.
This was a prospective cohort study of liver HMP in 20 patients undergoing OLT at our center. These patients were compared to a control group of 20 patients from the same time period who received UW static cold stored livers. Control subjects were matched for donor and recipient age, Model for End-Stage Liver Disease (MELD) score, and cold and warm ischemia times (Table 1).
Table 1. Donor, recipient and preservation characteristics of machine perfusion and cold storage subjects
Machine perfusion (HMP)
Cold storage (CS)
Number of subjects
39.4 ± 2.5
45.6 ± 2.1
AST peak (lU/mL)
213.9 ± 92.8
70.4 ± 11.2
AST last (lU/mL)
99.4 ± 45.2
39.3 ± 18.0
Na+ peak (mmol/L)
150.6 ± 1.8
155.7 ± 2.6
Na+ last (mmol/L)
141.2 ± 1.6
148.5 ± 2.1
HBV core positive
Length of stay (days)
6.3 ± 1.0
6.1 ± 1.3
Recipient and preservation characteristics
55.4 ± 6.2
52.7 ± 8.9
Indications for Tx
HCC concommitant (HCV or EtOH)
Lab MELD score
17.2 ± 7.4
16.8 + 6.8
Cold ischemic time (h)
9.4 ± 2.1
8.9 ± 2.8
Machine perfusion time (h)
4.3 ± 0.9
Percent of CIT that is perfusion time
47.47 ± 12.7
Warm ischemic time (min)
44.3 ± 6.5
45.1 ± 6.7
Adult patients between 18 and 65 years of age undergoing isolated primary OLT between July 2004 and February 2008 with MELD scores not greater than 35 who were not in the ICU at the time of organ offer were eligible. Patients were excluded if the expected cold ischemia time was less than 4 h or if the estimated total travel time from the donor hospital was greater than 3 h. Recipients of grafts from DCD or donors older than 65 years, or those with greater than 25% macrovesicular steatosis on biopsy, or significant liver dysfunction or ischemic injury were also excluded. Estimated percentages of both macro and microsteatosis were similar in the groups. At times logistics of the operating room ruled out enrollment as an HMP subject.
The study protocol was approved under the Western IRB. All patients gave informed consent after discussion with an attending surgeon familiar with the technique. Patients who met all criteria but declined to participate as HMP subjects or were ruled out due to logistical reasons were identified as potential control subjects. The HMP technique, perfusate and trial were performed under a Food and Drug Administration Investigational Device Exemption and registered with the U.S. National Institutes of Health at Clinicaltrials.gov (NCT00879268).
Liver machine perfusion technique and effluent sampling
Preclinical experience with development of our technique has been previously described (10). We utilized a modified Medtronic PBS® (Minneapolis, MN) for HMP. After standard 4L UW aortic and portal in situ flush, donor livers were packaged and transported to our center. Upon arrival, standard bench preparation was performed. Cannulation of the portal and arterial systems were accomplished with reusable metallic perfusion cannulas of appropriate size (Waters Medical Systems, Rochester, MN). Atraumatic cannulation was performed as far away from potential anastomotic sites as possible. All benchwork and cannulation of the liver was supervised by a staff surgeon.
HMP was performed for a minimum of 3 and a maximum of 7 hours, depending on the duration of the donor hepatectomy procedure and how long the liver had already been exposed to static cold preservation at the time of cannulation. We attempted to keep the total cold ischemia time under 12 h.
The portal vein (PV) and the hepatic artery (HA) underwent continuous centrifugal HMP with Vasosol® (Preservation Solutions Inc., Elkhorn, WI), a novel solution based on the chemistry of Belzer Machine Perfusion Solution (Belzer-MPS; TransMed, Elkhorn, WI) with added antioxidants, metabolic substrates and vasodilators (Table 2). A total recirculating perfusate volume of 3 L was used. Flow rates were adjusted for graft weight (0.667 mL/g liver/min). Hypothermia (4–8°C) was monitored with intraparenchymal temperature probes (myocardial probes) in the right and left lobe of the graft. Perfusion pressure was continuously monitored using a pressure transducer in the perfusion circuit. PV and HA pressures were measured directly via indwelling angiocatheters attached to the transducer on the perfusion device. The infrahepatic vena cava was clamped with a small vascular clamp to direct effluent to exit the liver in a stream like fashion from the suprahepatic vena cava where an 8 French feeding tube is placed to sample effluent from the hepatic venous return. Flow rates were increased if the parenchymal temperature in either lobe rose above 7°C. The organ cassette does not require cooling, hypothermia is maintained endothermically with partial submersion in the basin of cold effluent. Active oxygenation was not utilized in this system. Figure 1 shows a schematic (A) and photograph (B) of the HMP setup.
Table 2. Composition of vasosol solution
Vasosol® Base Solution3 (1 L)
aVasosol® base solution is identical to BelzerMPS™.
Effluent was collected from the suprahepatic vena cava cannula every 30 min during HMP and was analyzed for aspartate aminotransferase (AST), alanine aminotrasferase (ALT) and lactate dehydrogenase (LDH) using an Olympus AU 2700® analyzer (Olympus America Inc, Center Valley, PA). After 2 h of HMP, effluent levels of AST, ALT and LDH had generally reached steady state; therefore, we compared effluent at 2 h with peak posttransplant serum AST and ALT.
Liver transplantation and postoperative care
OLT was performed using standard techniques without venovenous bypass. Immediately prior to implantation, grafts in both groups were flushed with 1 L of Hextend (BioTime, Inc., Berkeley, CA) solution via the PV to wash out residual preservation solution. Liver core needle biopsies were obtained; moments before crossclamp, after preservation and 60 to 90 min postreperfusion.
Routine daily laboratory studies were performed during the hospital stay and results recorded. Complications, radiographic studies, liver biopsies and their results were also recorded. Postoperative care was delivered in accordance with the protocols of the Center for Liver Disease and Transplantation at Columbia University Medical Center.
Patients in both groups were followed clinically, with all complications and outcomes recorded for 12 months posttransplantation. The major endpoints of this study were the mean incidences of PNF, EAD and patient and graft survival at 1 month and 1 year. In accordance with the Clinical Trials in Organ Transplantation study group, EAD was defined as the presence of at least one of the following at 7 days after liver transplantation: serum bilirubin ≥10 mg/dL (16) and international normalized ratio (INR) ≥1.6 or alanine aminotranferease (ALT) >2000 in the first 7 postoperative days (POD). Surrogate endpoints included biliary and vascular complications, serum markers of liver and renal function (AST, ALT, total bilirubin (TBili), serum creatinine (SCr), INR and hospital length of stay. These values within the effluent were drawn from the 2 h effluent sampling. The peak values of AST, ALT and INR were determined within the first 48 h postreperfusion, whereas the peak values for TBili and SCr were determined within the first 5 days postreperfusion.
All statistics were calculated with Graph Pad Prism 5. All data were tested for normality using the D’Agostino & Pearson omnibus normality test. Data passing normality were compared using unpaired or paired t-tests and Mann–Whitney or Wilcoxon matched pairs test when it did not. Results are given as the mean ± standard deviation (SD) unless otherwise stated. And p-values less than 0.05 were considered significant.
HMP was performed for a minimum of 3 and a maximum of 7 h. While our target HMP time was at least 4 h, four patients had expedient hepatectomy that resulted in early termination of HMP. All 20 HMP grafts functioned immediately, with intraoperative bile production. There were no episodes of PNF in either group. EAD occurred in one HMP patient (5%) and five control subjects (25%, p = 0.08) (Table 3). There were no vascular complications in the HMP group and one (HA stenosis) in the control group. Two HMP patients had biliary complications (one early leak from a size-mismatched graft that was corrected with external drainage and resolved, and one mild anastomotic stricture in a patient with a Roux-en-Y hepaticojejunostomy that was serially dilated and resolved). Four patients in the control group had biliary complications (one early leak due to ductal necrosis, and three strictures above the confluence requiring very proximal hepaticojejunostomy; one of these patients died of septic complications after surgery).
Table 3. Patient outcomes
Machine perfusion (HMP)
Cold storage (CS)
1p = 0.08, 2p = 0.006.
*Technically met criteria but occurred in the setting of early acute cellular rejection.
Early allograft dysfunction
Vascular complications (total)
Hepatic artery stenosis
Biliary complications (total)
Early bile leak
Hospital length of stay (days)
10.9 ± 4.72
15.3 ± 4.9
Actual graft and patient survival
Deaths with fuctional grafts
Cardiovascular death at 1 month
Recurrent cancer at 5 months
Pneumonia and sepsis at 3 months
Recurrent HCV and sepsis at 7 months
Mean hospital length of stay was significantly lower in the HMP group (10.9 ± 4.7 days vs. 15.3 ± 4.9 days in the CS group, p = 0.006). At 12 months, there were two deaths in each group; all deaths were unrelated to preservation or graft function (Table 3).
Hypothermia (4–8°C) was monitored with intraparenchymal temperature probes (myocardial probes) in the right lobe (RL) and left lobe (LL) of the graft. Temperatures were maintained at a mean of 5.3 ± 0.10 SEM (RL) 4.1 ± 0.15 SEM (LL) degrees Celsius (Figure 2A). Perfusion pressure was continuously monitored using a pressure transducer in the perfusion circuit. PV and HA pressures were stably maintained throughout the HMP procedure time across all patients (PV; 2.9 ± 0.08 SEM mmHg, HA; 5.5 ± 0.15 SEM mmHg) (Figure 2B). Effluent solutes (Na+, K+ and Ca2+) were measured across the duration of the procedure (Figure 2C). All solutes varied little across the HMP time (Na+ 84.23 ± 0.8 SEM mmol/L, K+ 37.17 ± 0.61 SEM mmol/L, Ca2+ 1.83 ± 0.07 SEM mmol/L), indicating a homeostasis of metabolism for the duration of perfusion. Though active oxygenation was not utilized in the HMP circuit, pO2 and pCO2 levels remained stable at means of 137.2 ± 4.8 SEM mmHg and 12.9 ± 0.20 SEM mmHg, respectively (Figure 2D) due to ambient air interchange at the organ chamber. Effluent AST and ALT (IU/mL) were measured serially for the duration of perfusion, neither fluctuated with large variation (AST range 307.0–609.6 IU/mL; mean 409.4 ± 27.09 SEM IU/mL, ALT range 145.5–384.0 IU/mL; mean 104.8 ± 4.87 SEM IU/mL) (Figure 2E). Cellular injury markers lactate and LDH were also measured in the perfusate for the duration of HMP (Figure 2F). Lactate and LDH levels increased (Lactate range 3.15–4.16 mmol/L; mean 3.76 ± 0.12 SEM mmol/L, LDH range 83.22–125.8 IU/L; mean 104.8 ± 4.87 SEM IU/L) for the duration of perfusion.
Peak patient serum AST (Figure 3A), ALT (Figure 3B), TBili (Figure 3C) and SCr (Figure 3D) levels were all significantly lower in the HMP group when compared to the CS group (AST: HMP, 1154 ± 79.5 SEM IU/mL, CS, 3339 ± 755.1 SEM IU/mL, p = 0.011, ALT: HMP, 560.0 ± 79.5 SEM IU/mL, CS, 1358 ± 270.2 SEM IU/mL, p = 0.044, TBili: HMP, 6.59 ± 1.3 SEM mg/dL, CS, 9.46 ± 1.6 SEM mg/dL, p = 0.042, SCr: HMP, 1.34 ± 0.12 SEM mg/dL, CS, 2.00 ± 0.22 SEM mg/dL, p = 0.013). Peak INR values (HMP, 1.79 ± 0.12 SEM, CS, 1.92 ± 0.11 SEM) held normal distributions between the two groups and were not significantly different (Figure 3E).
Serum markers of liver damage and renal function (AST, ALT, TBili and SCr) were lower in HMP patients compared to CS patients over the course of 14 POD (Figure 4A–D). INR trends showed very similar results in both the CS and HMP groups (Figure 4E). Recovery to normality after transplantation can be used to determine of the speed of recovery of the patient as well as the functionality of the transplanted graft. We calculated the time (days) to normal function by the Columbia University Medical Center standard clinical laboratory definitions for normality: AST (12–38 IU/mL), ALT (7–41 IU/mL), TBili (0.3–1.3 mg/dL), SCr (0.6–1.12 mg/dL) and INR (0.87–1.16). All times were lower in patients in the HMP group when compared to patients in the CS group (Figure 4E), with both SCr and TBili times being calculated significantly lower in the HMP group (p = 0.041, p = 0.023, respectively).
Though the 60–90 min postreperfusion allograft biopsy from the HMP group showed an increase in endothelial swelling, common with machine perfusion (17), there were no significant differences between groups in histological appearance (data not shown). Graft weight increased in all grafts (mean weight increase 198.1 ± 25.6 SEM g), though was not significantly correlated with perfusion time (data not shown).
Peak serum AST and ALT postreperfusion was correlated with levels of effluent AST, ALT and LDH after 2 h of HMP to determine whether these values could be used as predictors of postoperative graft damage. Figure 5 shows the correlation plots of the various effluent values as compared to either peak AST (Figure 5A–C) or peak ALT (Figure 5D–F). Peak recipient AST was significantly correlated to the 2 h effluent AST (r = 0.76, p = 0.0002), ALT (r = 0.79, p = 0.0002) and LDH (r = 0.78, p = 0.0002). Peak recipient ALT levels were significantly correlated to the 2 h effluent AST (r = 0.74, p = 0.0004), though held a greater correlation to both effluent ALT (r = 0.84, p < 0.0001) and effluent LDH (r = 0.82, p < 0.0001).
Having previously described our experience with liver HMP in a preclinical setting (10), we sought to bring this approach into the clinical arena with a phase 1, prospective cohort trial. Indeed, preclinical animal studies of HMP for liver preservation have been promising, showing increased ATP production and across-the-board decreases in liver damage markers (14,18,19). The delayed evolution to clinical use has been due partly to the challenge of developing a user-friendly portable HMP system and partly to the severe ramifications of PNF and EAD in liver recipients. In direct response to these barriers, we attempted to prove that the hurdles preventing this technology from moving into clinical practice can be overcome.
Our study, the first to compare HMP to standard cold preservation in human liver transplantation, shows that HMP is safe, may improve graft function and attenuates classical biochemical markers of liver preservation injury. Early molecular results suggest there is a reduction in PI that may be responsible for fewer complications posttransplantation.
Machine perfusion pump characteristics remained stable for the duration of the pump procedure. No large spikes in solutes, pressures or temperatures were recorded indicating that the pump procedure was stable and required little intervention or input from the perfusionist. The continuous centrifugal flow allows for an influx of nutrients as well as oxygen, while flushing cytokines, proteins and toxins from the liver, in effect preventing the damage cascade buildup that occurs within a statically preserved liver.
LDH released in the perfusate has been used as a marker for cellular damage (17); in the HMP group, both lactate and LDH increased during the time course of perfusion. This is not unexpected as this measurement is a cumulative measurement with a finite volume of recirculating perfusate. Though this increase is often related to increasing liver damage, we saw no indication that the levels produced due to machine perfusion affected the functionality of the liver posttransplantation.
Within this study, all livers underwent a certain amount of CS cold ischemic time, regardless of whether they were used for the HMP or CS groups due to transport from donor centers. Within the HMP group, cold time ranged an average 9.4 h, with the first several those hours being ice time. Peak liver injury markers posttransplantation were significantly reduced in the HMP group, indicating that insult from the initial cold store period may be reversed or that the PI cascade before reperfusion can be ameliorated by HMP, thereby reducing severity of injury. Improving preservation is likely multifaceted but is surely related to the washout effect of HMP as well as the continuous delivery of metabolic substrates and antioxidants to the parenchyma and endothelium.
Following the progression to normality posttransplantation we established a timeline to normality in which the HMP group, in all of the markers examined, experienced shorter normalization times than did the CS group. These data, taken as a generalized picture of liver damage and early function, culminated in the significantly shorter patient length of stays and a reduction in complications posttransplant in the HMP group. We speculate that the improvement in early renal function seen with HMP might allow easier titration of calcineurin inhibitors to therapeutic levels, which might be a factor in the improvement in length of stay.
Although no significant histological differences between groups were apparent, our 60–90 min postreperfusion time point was likely too early to detect any but the most severe parenchymal disruption and necrosis. Further mechanistic investigations, including immunohistochemistry and reverse transcriptase-polymerase chain reaction (RT-PCR) of pro-inflammatory cytokines, are underway and may elucidate additional differences between HMP and CS preservation of the human liver.
The disparity with solutions used between groups results in an inability to conclusively prove that HMP alone was responsible for all of the benefits observed here. The perfusion solution Vasosol has significantly more active components than University of Wisconsin Solution and as such certainly has a beneficial contribution. We choose Vasosol for this trial due to its biochemical basis of the standard kidney pump solution, Belzer MPS. With the added antioxidants, vasodilatory and metabolic support we expected it to further ameliorate damage in an organ that has more dire consequences to ischemic/reperfusion injury. These additional additives as well as their contribution to the benefits of HMP will require further study in the near future. We maintain that the combination of HMP with Vasosol should be the system utilized for liver preservation specifically due to the persistent deterioration of the quality of donor livers available.
It is important to emphasize that HMP occurred in our center during patient preparation and recipient hepatectomy thus only a portion of the cold ischemic period was perfusion time. This simplifies the technique and allows utilization of less portable perfusion devices. The preclinical work of Dutkowski et al. support this and suggest that even a brief period (2 h) of HMP prior to implantation ameliorates a great deal of the mitochondrial and energetic disturbances and injuries associated with cold ischemia (18,20). Further, because peak serum AST and ALT were correlated with measured AST, ALT and LDH levels in perfusion effluent, on-the-go viability testing of effluent chemistry biomarkers could be used in conjunction with ‘hands on’ evaluation and biopsy data. In this manner, centers might be able to use the HMP system and determine, from effluent characteristics, how the liver will respond once transplanted, much in the way that kidneys are evaluated ‘on pump’ with perfusion parameters.
HMP also has the potential to become a donor team/organ procurement organization (OPO) phenomenon. This scenario would require a portable system and cannulation of the graft in the donor operating room, might allow safe extension of what was previously considered the ‘limits of safe cold ischemia’ for liver allografts.
The reduction of PI is likely to be most important in older, steatotic and DCD grafts. As such, our center has begun a human trial with HMP in ECD livers, which are at higher risk of EAD. As DCD and steatotic livers in particular are most susceptible to PI (18–20), we anticipate that the benefits of HMP will be most pronounced in these cases. Large-scale multicenter studies of HMP in liver transplantation are warranted to evaluate the full potential of this dynamic preservation technique.
The authors thank Procent Technologies and their perfusion team, Ben O’Mar Arrington, Tod Brown and Jason Boykin, for their outstanding support and assistance throughout this trial. We thank Nancy Erlich for her editorial assistance. We also acknowledge the contributions of Drs. Sarah Bellemare, Rodrigo Sandoval, Piotr Witkowski, Seth Narins, Abrar Khan and William Stubenbord. Thanks to the Irving Institute for Clinical and Translational Research Department of Biostatistics for aid in statistical analysis. Finally, we thank the staff of the Milstein Hospital operating rooms and Core Laboratory of Columbia University Medical Center.
This study was supported by Grant no. R38OT01301 from the Department of Health and Human Services (HHS), Health Resources and Services Administration (HRSA), Division of Transplantation (DoT). Its contents are solely the responsibility of the authors and do not necessarily represent the official view of the HRSA.