Due to a shortage of suitable donor livers, there is a need for expansion of the liver donor pool [1]. One proposed method of addressing this shortage has been to utilize livers from donation after circulatory death (DCD) donors. In these donors, declaration of death is made following cessation of circulation as determined by heartbeat, blood pressure, and/or electrocardiography [2]. This is followed by a super-rapid recovery procurement technique, during which the blood is flushed and the organ is cooled in situ prior to placement on ice. This is contrasted with donation after brain death (DBD) donors where, although the donor’s heart is still beating, brain death has been declared based on neurological criteria. DCD donors are commonly further classified as controlled (cDCD) or uncontrolled (uDCD) [3]. cDCD livers are generally considered less desirable than those recovered from DBD donors, as they are associated with higher rates of graft loss, ischemic cholangiopathy (IC), and inferior recipient survival [4, 5]. Therefore, there is significant interest in the development of novel organ procurement and preservation techniques to help improve outcomes associated with cDCD liver transplantation.

The current mainstay of organ preservation in liver transplantation is static cold storage (SCS) [6]. SCS in carefully selected DBD liver grafts have relatively low rates of known transplant complications such as early allograft dysfunction (EAD), primary non-function (PNF), and IC [6–9]. However, SCS alone in the cDCD context is associated with a higher incidence of graft complications and poorer recipient outcomes when compared with SCS in DBD livers [6]. IC is of particular concern with DCD livers (incidence of approximately 16% DCD vs. 3% DBD) [4, 10]. It has been postulated that warm ischemia (around the time of procurement) and vascular congestion contributes to microthrombus formation and subsequent biliary ischemia, leading to IC [5, 11, 12]. Compared with DBD livers, the PNF rate is greater in DCD livers (odds ratio of 3.6), as is the rate of total biliary complications (26% DCD vs. 16% DBD), and graft failure (odds ratio of 1.9) [4, 10]. These poorer outcomes contribute to higher rates of non-utilization of cDCD grafts for liver transplantation [13].

In normothermic regional perfusion (NRP) protocols, warm oxygenated perfusion with blood is restored in situ after declaration of circulatory death using an extracorporeal membrane oxygenation circuit. Although many technical variants exist, the circuit can be used to perfuse abdominal-only or all abdominal and thoracic organs simultaneously [13, 14]. Although the cellular mechanisms by which NRP works are not yet clear, it certainly allows for in situ assessment of organ function via macroscopic inspection, biopsy, and biochemical evaluation [13–16]. However, NRP does utilize more resources than super rapid recovery (SRR); including increased operating theatre time, disposables, and specifically trained perfusion staff [14].

The adoption of NRP varies significantly worldwide. It is policy to routinely use NRP in cDCD liver transplantation in Italy, France, and Norway, while also permitted for use in various other jurisdictions [13, 14, 17]. Some international centres combine NRP with additional ex-vivo machine perfusion technologies. The goal of NRP utilization is primarily to increase utilization of deceased donor organs and reduce mortality on the liver transplant waiting list. This systematic review and meta-analysis aims to compare outcomes from transplanted livers using NRP cDCD donors with non-NRP cDCD donors, as well comparing cDCD NRP outcomes with outcomes from DBD donation. We hypothesise that NRP improves the outcomes of cDCD livers and yields outcomes comparable to DBD livers.

Table 1 summarises the bias appraisal of each study as per the Newcastle Ottawa Scale. Four of the studies received an overall appraisal of “good,” and seven studies received an overall appraisal of “poor.” Of these seven studies, five studies received “poor” appraisal because they did not control for confounders between the two groups and hence failed to score points in the comparability domain. Two of the included studies received a “poor” appraisal in any of the other scoring domains.

Table 2 summarises the characteristics of each study included in the NRP vs. non-NRP for cDCD donation analysis. Three of the studies utilized NRP alone, and one study utilized NRP in combination with dual hypothermic oxygenated machine perfusion (D-HOPE) for some of the transplanted livers. The comparator groups are a mix of SCS alone and in combination with machine perfusion. The number of livers transplanted in the NRP and non-NRP groups totalled 702 and 505 respectively.

Table 3 summarises the characteristics of studies included in the comparison of cDCD with NRP vs. DBD donation. Two studies utilized NRP in combination with ex-vivo machine perfusion, whilst six studies utilize NRP alone. The comparator groups all utilized standard DBD techniques, except for one which utilized D-HOPE for some DBD transplants. The number of transplants in the cDCD with NRP and DBD groups totalled 402 and 1,037 respectively.

The outcomes examined in this systematic review were chosen because of their clinical importance and their past association of these outcomes with DCD liver transplantation. In the comparison of the NRP and non-NRP groups for cDCD livers, NRP is unanimously associated with lower rates of IC, PNF, HAT, and other biliary complications in conjunction with lower rates of recipient death and graft loss. The discard rate in each group was comparable, suggesting that the improved outcomes seen with NRP were not due to selection bias in the discard of organs in the NRP group. The analysis of utilization is potentially confounded by the fact that the comparison is performed after a decision to proceed to donation is already made. NRP utilization is often associated with more liberal organ acceptance criteria in terms of donor age, agonal time, and graft steatosis. Grafts of poorer quality that would not commonly be utilized as part of the non-NRP denominator are compared with some livers that were considered appropriate for procurement only because NRP was available. Hence, NRP is associated with a greater overall utilization, but a similar non-utilization rate from the point of intended recovery. This parameter is not captured in the reported data, but the advantage may be inferred. Ideally, further meta-analysis on donor and recipient factors such as degree of steatosis, MELD scores, BMI, and specific NRP protocols would have been included; however, the included articles did not consistently provide this data, and the articles that did were notably heterogenous with varied graft management options in addition to NRP.

Although the heterogeneity of interventions of the included studies is recognized, we considered that this was outweighed by the benefit of such analysis in a practical sense; cDCD NRP vs. non-NRP liver transplantation studies with strictly no additional perfusion technologies are limited and are unlikely to become available in the future as it would be unethical to withhold treatment from these organs when global standards permit their use and the emerging evidence supports their effectiveness. In the clinical setting, it is also more practical to compare cDCD NRP vs. non-NRP livers that may have been treated with other perfusion technologies as this is more reflective of current practice.

In the comparison of NRP cDCD vs. DBD, NRP cDCD livers perform equally well as DBD livers, exhibiting comparable complication and survival rates. A previous systematic review by De Beule et al. reported an overview of NRP for liver and kidney transplantation, including cDCD transplantation [34]. That review compared outcomes of NRP against SCS for cDCD livers. Although the authors reported lower rates of IC, EAD, biliary strictures (of any type), and anastomotic biliary strictures, there was no difference in PNF, 1 year patient survival or HAT. Additionally, no comparison could be made for cDCD NRP vs. DBD livers at that time. The conclusion made by the authors was that NRP could possibly provide benefits for reducing biliary complications for cDCD donation. In our review, the rate of discard with NRP DCD was comparable to DCD liver programs around the world. Haque et al. describes a 30% discard rate for all DCD liver donation within the US, and Oniscu et al. describes a 29.6% discard rate for non-NRP DCD donation in the UK. [17, 35]. Oniscu et al. did however describe a lower discard rate of 18.3% for NRP DCD donation in the UK. The same study also reported a higher overall utilization rate when using NRP for liver grafts. This was attributed to two main factors; the ability for functional evaluation of organs in situ, and a higher acceptance rate of the initial graft offer when NRP is known to be utilized. This review has focussed on liver transplantation, however, previous analyses have shown improved post-transplant outcomes and organ utilization for other abdominal organs, such as the kidneys, when employing NRP compared to standard DCD techniques [36, 37]. All studies included in our cDCD NRP vs. non-NRP analysis reported liver transplant outcomes only, and no studies were found meeting the inclusion criteria which reported multiple organ donation outcomes. NRP circuits may be configured in a manner that allows simultaneous perfusion to multiple other abdominal and thoracic organs, allowing the potential benefits of NRP to be extended to other transplanted grafts. Further studies looking at the outcomes of multiple grafts from the same NRP donor may be beneficial.

A notable limitation of our study is that all included studies were observational, as no randomised controlled trials satisfied the inclusion criteria. The need for randomised trials to provide high quality evidence of the benefit of NRP has been previously outlined, although conducting such studies is now arguably unethical in the context of the results demonstrated above [38]. Additionally, more than half of the included studies are classified as “poor” according to the Newcastle Ottawa Scale due to the nature of the scoring system of the scale. Any paper that does not score in the comparability domain receives an automatic “poor” designation, although they may score well in all other respects. Importantly, the majority did specify that there was no statistically significant difference between the donor and recipient groups in a variety of metrics, however this demonstration is not considered sufficient to score points for comparability on the Newcastle Ottawa Scale. Another limitation is that although this review examines the use of NRP compared to non-NRP, we were unable to make any direct comparison of NRP vs. SCS, HOPE, or NMP. Hence, the outcome may be slightly confounded by livers receiving a combination of NRP, HOPE, and NMP in addition to NRP. The number of studies currently published is insufficient to facilitate direct comparisons between each technology combination. Ideally, the effect of NRP on recipient outcomes would be isolated from the effects of other ex-vivo perfusion technologies, however this is not currently possible with the available data. The control groups in each comparison (non-NRP cDCD and DBD groups) also contain liver grafts treated with HOPE or NMP in addition to standard SCS. The inclusion of these technologies in the control groups may lead to an underestimate of NRP effect. One included study contained 2 uDCD livers which could not be separated from our cDCD with NRP vs. DBD analysis. The decision was made to include this study even with the increased risk of bias, as the effect of only 2 livers in the sample size was highly unlikely to alter the results in any meaningful way and their inclusion allowed for the inclusion of 49 additional cDCD livers to increase the power of our analysis. As uDCD livers are of poorer quality, the inclusion of these livers would more likely disadvantage the NRP analysis than advantage it, making the positive effect of NRP results even more persuasive.

The most important future analysis should focus on the effect of NRP to increase utilization from the point of organ offer due to the more liberal acceptance criteria (principally on account of acceptance of more advanced donor age, longer agonal times, and higher rates of steatosis). Direct comparisons of NRP with ex-vivo machine perfusion may also be useful. It is certainly possible that some combination of NRP, HOPE, and NMP will provide the optimal combination of maximal utilization and acceptable recipient outcomes, but this will be challenging to investigate robustly on account of the possible number of combinations [13]. It should be noted that a randomised controlled trial examining NMP vs. SCS for liver transplantation demonstrated no change in biliary complication rate, graft survival, or patient survival rates whilst increasing the number of transplantable grafts by 20% [39].

In summary, this review demonstrates that the use of NRP in cDCD liver transplantation is associated with lower rates of many significant post operative complications as well as improved graft and patient survival. NRP cDCD outcomes were comparable to DBD outcomes. The use of NRP appears to also increase the utilization of cDCD livers for transplantation, although non-utilization rates of recovered DCD livers are similar between NRP and standard techniques following donation. NRP has the potential to allow for the expansion of the donor pool and improvement of outcomes so reducing the mortality for those patients needing liver transplantation.