In situ normothermic perfusion of livers in controlled circulatory death donation may prevent ischemic cholangiopathy and improve graft survival

Livers from controlled donation after circulatory death (DCD) donors suffer a higher incidence of nonfunction, poor function, and ischemic cholangiopathy. In situ normothermic regional perfusion (NRP) restores a blood supply to the abdominal organs after death using an extracorporeal circulation for a limited period before organ recovery. We undertook a retrospective analysis to evaluate whether NRP was associated with improved outcomes of livers from DCD donors. NRP was performed on 70 DCD donors from whom 43 livers were transplanted. These were compared with 187 non‐NRP DCD donor livers transplanted at the same two UK centers in the same period. The use of NRP was associated with a reduction in early allograft dysfunction (12% for NRP vs. 32% for non‐NRP livers, P = .0076), 30‐day graft loss (2% NRP livers vs. 12% non‐NRP livers, P = .0559), freedom from ischemic cholangiopathy (0% vs. 27% for non‐NRP livers, P < .0001), and fewer anastomotic strictures (7% vs. 27% non‐NRP, P = .0041). After adjusting for other factors in a multivariable analysis, NRP remained significantly associated with freedom from ischemic cholangiopathy (P < .0001). These data suggest that NRP during organ recovery from DCD donors leads to superior liver outcomes compared to conventional organ recovery.


| INTRODUC TI ON
The success of liver transplantation as a treatment for patients with liver disease is limited by a shortage of suitable donor organs, such that of those patients listed for a liver transplant in 2014 and 2015 in the UK, 13% died and 8% were removed from the list within 2 years without undergoing liver transplantation. 1 The situation is similar in the United States where 12% of patients listed in 2013 died and 17% were removed from the list in the subsequent 2 years. 2 This high waiting list attrition has led surgeons to use organs that were previously considered to be at higher risk of failure, 3 balancing the risk of death without a transplant against the risk of complications from a suboptimal graft. 4 Livers donated after circulatory death (DCD) are one such higher risk category.
Following withdrawal of life-supporting treatment in a potential DCD donor, the autonomic response to falling cerebral perfusion is characterized by release of catecholamines which cause an alpha-adrenergic mediated peripheral and mesenteric vasoconstriction, resulting in reduced blood flow to the liver. 5,6 Hence during the withdrawal period before circulatory arrest the liver suffers significant ischemia. This is then compounded by a further period of warm ischemia after circulatory arrest (the asystolic period) during which the donor is verified dead and is transferred to the operating room where, following a rapid laparotomy, the organs are flushed and cooled in situ with preservation fluid.
Cold perfusion reduces, but does not stop the metabolic processes that were active in the warm, and leads to further depletion of high energy phosphates that occurs rapidly during warm ischemia. The successive insults of warm ischemia followed by cold storage account for the suboptimal nature of livers donated after circulatory death (DCD), with a higher incidence of primary nonfunction and initial poor function compared to those from brain dead donors. In addition, DCD donor livers are associated with a higher incidence of biliary complications and in particular nonanastomotic biliary strictures ("ischemic cholangiopathy") which can be seen in 20% to 30% of DCD livers. 7,8 DCD donors comprise 41% of deceased donors in the UK, and 17% in the United States. 1,2 The high proportion of DCD donors has forced increasing utilization of DCD donor livers to address the waiting list mortality, such that 24% of transplanted livers in the UK in 2016 and 2017 were from DCD donors; this compares to just 6% in the United States in 2016. 1,2,9 In situ normothermic regional perfusion is a technique that restores a circulation of oxygenated blood to the abdominal organs via cannulas in the aorta and vena cava using an extracorporeal circuit containing a membrane oxygenator, heater, and pump. Restoration of a blood supply arrests the ischemic damage to the liver and allows it to recover before being cooled down for transport to the recipient center. In this way, it converts DCD donation into a situation more akin to that seen in DBD donation. In an effort to improve the outcomes of DCD liver transplantation, we started a program of normothermic regional perfusion (NRP) in two UK centers in 2011 and 2012. 10,11 Several groups have described encouraging outcomes of DCD organ transplantation following NRP, [12][13][14] initially with kidneys but, latterly, cases of successful liver transplantation. [15][16][17][18][19] Originally pioneered in uncontrolled DCD donation in Spain and Taiwan, 20,21 NRP is increasingly used for controlled DCD donation in Spain, France and the UK. 11,19,22 In contrast to Spain and France, where premortem cannulation and heparinization are permitted, current guidance does not permit either in the UK. This paper describes the experience of the two pioneering UK centers with the use of NRP for DCD liver transplantation and focuses on the recognized complications of DCD liver transplantation, namely early allograft dysfunction and ischemic cholangiopathy.
NRP was used in two settings, one solely by the abdominal transplant team and the other in collaboration with cardiac surgeons to facilitate heart transplantation from DCD donors. 23

| MATERIAL S AND ME THODS
This retrospective study used prospectively collected data collated from the UK Transplant Registry and hospital records on patients undergoing transplantation using livers recovered from DCD donors after a period of NRP. NRP was considered in donors when trained staff were available, and was initially conducted as part of an approved clinical research study in which donor families consented to NRP treatment (12% for NRP vs. 32% for non-NRP livers, P = .0076), 30-day graft loss (2% NRP livers vs. 12% non-NRP livers, P = .0559), freedom from ischemic cholangiopathy (0% vs. 27% for non-NRP livers, P < .0001), and fewer anastomotic strictures (7% vs. 27% non-NRP, P = .0041). After adjusting for other factors in a multivariable analysis, NRP remained significantly associated with freedom from ischemic cholangiopathy (P < .0001). These data suggest that NRP during organ recovery from DCD donors leads to superior liver outcomes compared to conventional organ recovery.

K E Y W O R D S
clinical research/practice, donors and donation: donation after circulatory death (DCD), extracorporeal membrane oxygenation (ECMO), liver transplantation/hepatology, surgical technique of the donor and the recipients consented to receive an organ treated by NRP. Latterly, NRP has been performed as part of a service evaluation by the National Health Service Blood and Transplant (NHSBT) organization in the UK. In this latter period, when the safety of the technique had been confirmed, recipients gave consent for an organ from a DCD donor, regardless of whether the donor had undergone NRP or not. Interest in utilization of hearts from DCD donors led to a bias in favor of younger donors at one center where the cardiothoracic teams were able to call on different staff to perform the perfusions.

| Normothermic regional perfusion (NRP)
NRP was undertaken in two contexts, one to assess and retrieve the abdominal organs alone and the other to also assess and recover the heart for transplantation; the abdominal organs received the same treatment in each case. All donors were controlled DCD donors (Maastricht III). 24 In all cases the mode of withdrawal of treatment was left to the intensivist caring for the patient, but usually comprised extubation and discontinuation of inotropes. Death was confirmed no less than 5 minutes after cessation of the circulation following which the patient was transferred to the operating room. Abdominal NRP The intention was to restore a circulation to the abdominal organs for 2 hours before in situ perfusion with cold University of Wisconsin preservation solution. Where TA-NRP was performed and cardiac output was restored, the extracorporeal perfusion was stopped at 30 to 60 minutes and the heart allowed to support the limited thoracoabdominal circulation while its function was evaluated. If the cardiac function was considered inadequate and failing to sustain an adequate mean arterial pressure, extracorporeal perfusion was recommenced.
The decision to use the liver was based on subjective and biochemical factors. The appearance of the liver during NRP was assessed, with cirrhotic and severely steatotic livers being declined.
Alanine transaminase (ALT) was measured in the perfusate every 30 to 60 minutes during NRP. In our early experience an ALT over 200 iu/L would result in the liver being declined, but latterly perfusion ALTs up to 500 iu/L have been accepted, providing there was no continued rise in ALT between the first and second hour. Perfusate lactate concentrations were measured every 30 minutes, as part of the blood gas profile used to manage gas delivery to the circuit. A fall in lactate was also considered encouraging, but deoxygenated lactate-rich blood draining back into the circuit from nonperfused areas together with the lactate content in supplementary Hartmann's solution made this less reliable as an indicator; in later perfusions Hartmann's was not administered once NRP had commenced. Typically, the lactate fell between 3 and 14 mmol/L over the 2 hours of perfusion.

| NRP and comparator cohorts
A contemporaneous comparator cohort comprised all DCD liver transplants performed at both centers since the start of the NRP program. Livers subject to ex situ machine perfusion were excluded.

| Definitions
Ischemic cholangiopathy was defined as the presence of any nonanastomotic biliary stricture on endoscopic or magnetic resonance cholangiopancreatography (ERCP or MRCP) in the absence of arterial thrombosis or stenosis. Cholangiography was undertaken when clinically indicated by pruritus, cholangitis, raised bilirubin or ALP posttransplant. Protocol cholangiograms were also performed as part of the initial evaluation of NRP. Patients with evidence of strictures but without symptoms sufficiently severe to warrant retransplantation were treated symptomatically, often with attempts at endoscopic or percutaneous duct clearance of any debris.
An anastomotic stricture was defined as any stricture at the biliary anastomosis requiring treatment, and anastomotic leaks were defined as a bile leak confirmed at laparotomy; no suspected leak was managed without surgery in either group. Glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-Epi) equation. 25 Organ damage was graded separately by the donor surgeon and recipient center as minor (trivial) or moderate (needs repair); the worst degree of damage reported is recorded here.

| Statistical analysis
Descriptive data were presented as median, interquartile range or range (continuous variables) or number, percentage (categorical variables) and were compared using the Kruskal-Wallis test (continuous variables) and Fisher's exact test (categorical variables). Analysis of variance (ANOVA) was used to compare CKD-Epi GFR over time between NRP and comparator cohorts. A random effect was included in the ANOVA model to allow for correlation between GFR results at different time points for the same patient. Kaplan Meier estimates of graft survival were compared using the log-rank test.
Graft survival was defined as the time from transplant to graft failure with patient deaths with a functioning graft censored. Logistic regression was used to determine the factors associated with developing ischemic cholangiopathy. Analyses were undertaken using SAS/ STAT, version 9.4 (SAS Institute Inc., Cary, NC) and Prism version 7.0c (GraphPad, La Jolla, CA). longer withdrawal periods and with higher biochemistry thresholds were used. There was no difference in UK donor liver index between NRP and non-NRP livers, nor in the donors where the livers were not used. 3 The US donor risk index for livers was higher in the non-NRP livers reflecting the longer cold ischemic time and fewer local donors (Table 1). 4,26 Of the 43 cases where NRP was performed and the liver transplanted, TA-NRP was performed in 10 cases from which 9 hearts were also transplanted.

| Donor demographics and timings
The median donor age in the NRP liver transplants was 41, compared to 50 in the comparator group (Table 1). There was a greater proportion of head injury liver donors undergoing NRP (23% NRP vs. 12% non-NRP) and more donors dying from hypoxic brain injury in the non-NRP comparator cohort (37% non-NRP, 28% NRP), but these differences were not significant (P = .5358). A greater proportion of the liver donors were in the transplant center in the NRP group compared to the non-NRP comparators (77% vs. 16%, P < .0001), and similarly most NRP livers were retrieved by the transplanting center, unlike comparator livers (93% NRP vs. 55% non-NRP, P < .0001).
While the median withdrawal period was slightly shorter in the NRP livers (NRP 13 minutes, non-NRP 14 minutes, P = .0707) the median asystolic period was slightly longer (NRP 16 minutes, non-NRP 13 minutes, P < .001) reflecting the extra time taken to establish the donor on the extracorporeal circuit. NRP was performed for a median of 123 minutes ( Figure 1). As a consequence of distant procurement being more common in the comparator livers, the median cold ischemic times experienced by the comparator livers were 62 minutes longer than the NRP livers (444 vs. 382 minutes, P = .004).

| Recipient demographics and outcomes
The liver recipients in each study group were of similar age, severity of liver disease and had similar indications for transplantation ( Table 2). Although there was less deterioration in renal function in recipients of livers recovered using NRP, with a median fall in estimated glomerular filtration rate of 13 mls/min compared to 26 mls/min in the non-NRP liver recipients, the difference in CKD-GFR did not reach significance (P = .6229; Table 2).

| Factors determining ischemic cholangiopathy
Given the differences in donor and recipient populations in each group, a multivariate analysis was undertaken to determine whether NRP was a significant factor in preventing ischemic cholangiopathy (IC). Tables 3-5 (Table 6).
Although NRP was highly significant (P < .0001) in not developing IC, no meaningful odds ratio could be determined since there were no cases of IC in livers from donors where organ recovery used NRP.

| D ISCUSS I ON
This paper describes our experience with liver transplantation from controlled DCD donors who have undergone in situ normothermic regional perfusion (NRP) before organ recovery, and compares those cases to a contemporaneous cohort of liver transplants from DCD donors who did not undergo NRP but were also transplanted in the two study centers. The study has shown that NRP is an independent factor in reducing the incidence of ischemic cholangiopathy in DCD livers following transplantation, so much so that no liver from a donor undergoing NRP prior to recovery developed cholangiopathy.
Our study has also shown that lower recipient sodium at the time of surgery, older donor age, and a donor outside the local hospital was independently associated with the development of ischemic cholangiopathy.
The study is limited by being a retrospective analysis of prospectively collected data, rather than being a randomized controlled trial.
We have compensated for this by including a contemporaneous com- Analysis of variance (ANOVA) was used to compare CKD-Epi GFR over time and between the two groups. A random effect was included in the model to allow for correlation between GFR at different time points for the same patient. Data are median (IQR) or number (percentage). a UKELD = United Kingdom end-stage liver disease score. 39  week in comparator cohort due to primary nonfunction, hepatic artery thrombosis or death, and one in NRP cohort with hepatic artery thrombosis. d Denominators exclude livers with early failure from consideration for anastomotic stricture or ischemic cholangiopathy. e Includes 2 patients dying from PNF without a retransplant. f Model for early allograft function. 28 Data not available for two patients in NRP group, and for eight patients in non-NRP group including four with PNF.
cholangiopathy has been reported in controlled DCD livers receiving ante-mortem heparinization and undergoing "super-rapid recovery," where thrombi should not be forming. 30 It is also possible that bile ducts are more sensitive to ischemia than hepatocytes, with less regenerative capacity, and benefit most from early reperfusion in the donor and avoidance of consecutive periods of warm followed by cold ischemia.
Another possible explanation for the absence in cholangiopathy relates to the composition of the bile. Treatment withdrawal in the donor is followed by a period of reduced organ perfusion secondary to catecholamine release in response to falling systemic pressure and cerebral perfusion. 5 With decreasing perfusion, and decreasing oxygenation, the donor becomes more acidotic, and the ability of the cholangiocytes to produce bicarbonate may be impaired. By restoring a circulation promptly following arrest it is possible that the biliary tree has time to recover and produce the bicarbonate "umbrella" that has been proposed to stop direct bile salt injury of the biliary epithelium. 34,35 Several additional observations are noteworthy. The non-NRP livers suffered a numerically higher (7%) incidence of primary nonfunction, compared to no primary nonfunction in the NRP livers (P = .1347). This is in spite of the liver utilization rate in NRP-treated livers being 61%, in contrast to the UK national rate without NRP which varied between 27% and 36% per annum in the period of the study, 1,36 suggesting that selection bias was probably not the reason for the better initial outcomes. Instead the difference, albeit not significant in this small study, was probably accounted for by the ability to test liver viability and function in the period after the warm ischemic insults that characterize the withdrawal and asystolic periods. 5,6 The ability to test viability of the liver in situ is the main benefit of NRP. Our small series does not provide sufficient information to define viability criteria, but we followed similar parameters to those published to help assess uncontrolled DCD donors in Spain, 20 namely lactate fall as a marker of function and ALT release as a marker of damage, although we adopted a more liberal interpretation of ALT levels to inform liver utilization. It is likely that with more experience clear parameters will be defined. In the original Cambridge series, a shunt was used to divert blood away from the oxygenator for a few minutes until adequate mixing of the blood with the heparinized prime solution had occurred, 10 while Edinburgh also used heparin-coated components in its circuit.
In the current circuit only the leucocyte filter is bypassed for the first 2 minutes of perfusion. Ensuring adequate heparinization at the start of NRP is paramount if clots and emboli are to be avoided, and this may be best achieved by premortem heparinization in countries where this is permitted.
NRP has other benefits, and in this series it was used to facilitate DCD heart transplantation. 37 We have also been able to use livers with longer withdrawal periods than currently accepted by many centers, 38 reassured by the knowledge that the liver's function can be checked in situ before removal. This further increases the available pool of DCD donor livers that may be used to address the high mortality of those on the waiting list. The functional assessment also allows for the discard of livers that may develop primary nonfunction but which previously would have been used based solely on predonation data.
In summary, we have described 43 cases of liver transplantation from DCD donors subject to NRP, with improved early allograft function, absence of ischemic cholangiopathy, and improved graft survival. The multivariate analysis emphasizes the potential of NRP as an independent factor preventing ischemic cholangiopathy, and the study supports continued implementation and evaluation of the technique.

ACK N OWLED G M ENTS
The authors wish to acknowledge the invaluable support of Andy  CI, confidence interval; NRP, Normothermic regional perfusion. NRP use was statistically significant after adjusting for other factors (P = .0001). However, we were not able to produce meaningful odds ratios as there were no cases of ischemic cholangiopathy in the NRP group.