Machine perfusion in liver transplantation

Abstract

Background

Liver transplantation is the only chance of cure for people with end‐stage liver disease and some people with advanced liver cancers or acute liver failure. The increasing prevalence of these conditions drives demand and necessitates the increasing use of donated livers which have traditionally been considered suboptimal. Several novel machine perfusion preservation technologies have been developed, which attempt to ameliorate some of the deleterious effects of ischaemia reperfusion injury. Machine perfusion technology aims to improve organ quality, thereby improving outcomes in recipients of suboptimal livers when compared to traditional static cold storage (SCS; ice box).

Objectives

To evaluate the effects of different methods of machine perfusion (including hypothermic oxygenated machine perfusion (HOPE), normothermic machine perfusion (NMP), controlled oxygenated rewarming, and normothermic regional perfusion) versus each other or versus static cold storage (SCS) in people undergoing liver transplantation.

Search methods

We used standard, extensive Cochrane search methods. The latest search date was 10 January 2023.

Selection criteria

We included randomised clinical trials which compared different methods of machine perfusion, either with each other or with SCS. Studies comparing HOPE via both hepatic artery and portal vein, or via portal vein only, were grouped. The protocol detailed that we also planned to include quasi‐randomised studies to assess treatment harms.

Data collection and analysis

We used standard Cochrane methods. Our primary outcomes were 1. overall participant survival, 2. quality of life, and 3. serious adverse events. Secondary outcomes were 4. graft survival, 5. ischaemic biliary complications, 6. primary non‐function of the graft, 7. early allograft function, 8. non‐serious adverse events, 9. transplant utilisation, and 10. transaminase release during the first week post‐transplant. We assessed bias using Cochrane's RoB 2 tool and used GRADE to assess certainty of evidence.

Main results

We included seven randomised trials (1024 transplant recipients from 1301 randomised/included livers). All trials were parallel two‐group trials; four compared HOPE versus SCS, and three compared NMP versus SCS. No trials used normothermic regional perfusion.

When compared with SCS, it was uncertain whether overall participant survival was improved with either HOPE (hazard ratio (HR) 0.91, 95% confidence interval (CI) 0.42 to 1.98; P = 0.81, I² = 0%; 4 trials, 482 recipients; low‐certainty evidence due to imprecision because of low number of events) or NMP (HR 1.08, 95% CI 0.31 to 3.80; P = 0.90; 1 trial, 222 recipients; very low‐certainty evidence due to imprecision and risk of bias).

No trials reported quality of life.

When compared with SCS alone, HOPE was associated with improvement in the following clinically relevant outcomes: graft survival (HR 0.45, 95% CI 0.23 to 0.87; P = 0.02, I² = 0%; 4 trials, 482 recipients; high‐certainty evidence), serious adverse events in extended criteria DBD liver transplants (OR 0.45, 95% CI 0.22 to 0.91; P = 0.03, I² = 0%; 2 trials, 156 participants; moderate‐certainty evidence) and clinically significant ischaemic cholangiopathy in recipients of DCD livers (OR 0.31, 95% CI 0.11 to 0.92; P = 0.03; 1 trial, 156 recipients; high‐certainty evidence). In contrast, NMP was not associated with improvement in any of these clinically relevant outcomes. NMP was associated with improved utilisation compared with SCS (one trial found a 50% lower rate of organ discard; P = 0.008), but the reasons underlying this effect are unknown.

We identified 11 ongoing studies investigating machine perfusion technologies.

Authors' conclusions

In situations where the decision has been made to transplant a liver donated after circulatory death or donated following brain death, end‐ischaemic HOPE will provide superior clinically relevant outcomes compared with SCS alone. Specifically, graft survival is improved (high‐certainty evidence), serious adverse events are reduced (moderate‐certainty evidence), and in donors after circulatory death, clinically relevant ischaemic biliary complications are reduced (high‐certainty evidence). There is no good evidence that NMP has the same benefits over SCS in terms of these clinically relevant outcomes. NMP does appear to improve utilisation of grafts that would otherwise be discarded with SCS; however, the reasons for this, and whether this effect is specific to NMP, is not clear. Further studies into NMP viability criteria and utilisation, as well as head‐to‐head trials with other perfusion technologies are needed.

In the setting of donation following circulatory death transplantation, further trials are needed to assess the effect of these ex situ machine perfusion methods against, or in combination with, normothermic regional perfusion.

Plain language summary

Can 'perfusion' machines improve the quality of livers donated for transplantation?

Key messages

– Cold machine perfusion improves liver transplantation when compared with the standard ice‐box technique.

– Warm machine perfusion does not seem to have these benefits, but might allow the transplant of donated livers that would otherwise not be used.

What is the issue?

Liver transplantation is the only chance of cure for thousands of people with liver failure or advanced liver cancers. Both of these conditions are becoming more common worldwide. This has caused an imbalance between the number of people needing a new liver, and the number of high‐quality livers which are donated. Increasingly, surgeons are driven to transplant livers which may be considered 'suboptimal'. Whilst receiving one of these livers is better than staying on the transplant waiting list, the outcomes are worse than with optimal organs. Many people are researching perfusion machines which pump liquid containing oxygen and nutrients through the liver in the time period between the death of the donor and the implant of the liver. These perfusion machines vary in lots of ways including the temperature that they keep the organ at.

What did we want to find out?

We wanted to know which of these techniques is the best for improving the quality of donated livers.

What did we do?

We performed a rigorous search for clinical trials which compared perfusion machines. We planned to include trials which compared perfusions machines with each other, or compared with standard ice‐box preservation. Our primary outcomes were death, quality of life, and serious side effects (serious adverse events). We also investigated secondary outcomes of how long the transplanted liver survived, bile duct (thin tubes that go from the liver to the small intestine) damage, and what proportion of the donated livers could be transplanted.

What did we find?

We found six trials with 854 transplant recipients from 1124 donated livers.

Main results

No machine was shown to reduce death, and no trials looked at quality of life. Compared with the standard ice‐box technique, cold machine perfusion improved the survival of the liver, reduced the number of serious adverse events, and reduced damage to the bile ducts. Warm machine perfusion with oxygen did not have these benefits. Warm machine perfusion appeared to increase the proportion of donated livers which could be transplanted, but more research is needed to understand why.

What are the limitations of the evidence?

We found a limited number of trials and some were of mixed quality. The reported data were also insufficient for all planned analyses. None of the trials looked at a machine perfusion technique which is applied in the donor before the organs are removed (termed normothermic regional perfusion).

How up to date is this evidence?
The review includes studies published to 10 January 2023.

Summary of findings

Summary of findings 1. Hypothermic oxygenated perfusion (HOPE) compared with static cold storage (SCS) for preservation of livers prior to transplantation.

Hypothermic oxygenated perfusion (HOPE) compared with static cold storage (SCS) for preservation of livers prior to transplantation
Patient or population: liver transplant recipients Setting: liver transplantation Intervention: hypothermic oxygenated perfusion (HOPE) Comparison: static cold storage (SCS)
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with static cold storage (SCS)	Risk with hypothermic oxygenated perfusion (HOPE)
Overall participant survival follow‐up: median 12 months	Moderate		HR 0.91 (0.42 to 1.98)	482 (4 RCTs)	⊕⊕⊝⊝ Lowa	It is uncertain whether HOPE improves overall survival due to imprecision because of low numbers of deaths.
	940 per 1000	945 per 1000 (885 to 974)
Quality of life ‐ not measured	No studies assessed this outcome.		—	—	—	—
Serious adverse events follow‐up 90 days	449 per 1000	426 per 1000 (255 to 617)	OR 0.91 (0.42 to 1.98)	156 (2 RCTs)	⊕⊕⊕⊝ Moderateb	HOPE reduces serious adverse events compared to SCS; however, the exact magnitude of this effect is unclear due to wide CIs.
Graft survival follow‐up: 12 months	Moderate		HR 0.45 (0.23 to 0.87)	482 (4 RCTs)	⊕⊕⊕⊕ High	HOPE improves graft survival compares to SCS alone.
	840 per 1000	925 per 1000 (859 to 961)
Ischaemic biliary complications total follow‐up: median 6 months	104 per 1000	36 per 1000 (14 to 88)	OR 0.32 (0.12 to 0.83)	326 (2 RCTs)	⊕⊕⊕⊕ High	In livers donated following circulatory death, HOPE reduces clinically significant ischaemic cholangiopathy compared to SCS alone.
Ischaemic biliary complications (donated following brain death (DBD))	35 per 1000	12 per 1000 (1 to 105)	OR 0.33 (0.03 to 3.19)	170 (1 study)	—	—
Ischaemic biliary complications (donated following circulatory death (DCD))	179 per 1000	64 per 1000 (23 to 168)	OR 0.31 (0.11 to 0.92)	156 (1 study)	—	—
Early allograft dysfunction total follow‐up: 7 days	398 per 1000	188 per 1000 (132 to 260)	OR 0.35 (0.23 to 0.53)	482 (4 RCTs)	⊕⊕⊕⊕ High	HOPE reduces early allograft dysfunction compared to SCS alone. As peak post‐transplant transaminases contribute to this outcome, and there is a washout effect during perfusion, this outcome is of questionable clinical relevance.
Adverse events considered non‐serious follow‐up: 12 months	3 studies reported a lower number of total adverse events in the HOPE group compared with the SCS group.		—	312 (3 RCTs)	—	Quantitative synthesis could not be performed.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; HR: hazard ratio; OR: odds ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_436878198874408136.

^a Downgraded two levels due to imprecision. 
^b Downgraded one level due to imprecision.

Summary of findings 2. Normothermic machine perfusion (NMP) compared with static cold storage (SCS) for preservation of livers prior to transplantation.

Normothermic machine perfusion (NMP) compared with static cold storage (SCS) for preservation of livers prior to transplantation
Patient or population: liver transplant recipients Setting: liver transplantation Intervention: normothermic machine perfusion (NMP) Comparison: static cold storage (SCS)
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with static cold storage (SCS)	Risk with normothermic machine perfusion (NMP)
Overall participant survival follow‐up: 12 months	Moderate		HR 1.08 (0.31 to 3.80)	222 (1 RCT)	⊕⊝⊝⊝ Very lowa	It is uncertain whether NMP impacts overall participant survival.
	980 per 1000	978 per 1000 (926 to 994)
Quality of life ‐ not measured	No studies assessed this outcome		—	—	—	—
Serious adverse events (SAEs)	Nasralla 2018 reported 21 SAEs across 121 recipients of NMP livers, and 36 SAE across 101 recipients of SCS livers. Markmann 2022 reported "liver specific" serious adverse event; 7/153 in the NMP group and 11/146 in the SCS group		—	522 (2 RCTs)	—	NMP is not associated with an increase in SAEs compared with SCS; there are no safety concerns.
Graft survival follow‐up: 12 months	Moderate		HR 1.20 (0.44 to 3.29)	522 (2 RCTs)	⊕⊕⊝⊝ Lowb	NMP does not appear to improve graft survival compared to SCS.
	975 per 1000	970 per 1000 (920 to 989)
Ischaemic biliary complications (on protocol imaging) follow‐up: 6 months	108 per 1000	86 per 1000 (32 to 216)	OR 0.78 (0.27 to 2.27)	155 (1 RCT)	⊕⊝⊝⊝ Very lowc	NMP does not appear to reduce radiological ischaemic biliary complications, and the effect on clinically significant ischaemic biliary complications is not known.
Early allograft dysfunction follow‐up: 7 days	300 per 1000	146 per 1000 (86 to 240)	OR 0.40 (0.22 to 0.74)	540 (3 RCTs)	⊕⊕⊕⊝ Moderated	NMP reduces early allograft dysfunction compared with SCS alone. As peak post‐transplant transaminases contribute to this outcome, and there is a washout effect during perfusion, this outcome is of questionable clinical relevance.
Adverse events considered non‐serious	574 per 1000	554 per 1000 (421 to 679)	OR 0.92 (0.54 to 1.57)	222 (1 RCT)	⊕⊕⊕⊝ Moderatee	NMP does not appear to reduce the proportion of recipients experiencing adverse events compared with SCS.
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; HR: hazard ratio; OR: odds ratio.
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
See interactive version of this table: https://gdt.gradepro.org/presentations/#/isof/isof_question_revman_web_436879195689026772.

^a Downgraded one level due to risk of bias and two levels due to imprecision (because of low numbers of deaths). 
^b Downgraded two levels dues to risk of bias and imprecision. 
^c Downgraded one level due to risk of bias and two levels due to imprecision. 
^d Downgraded one level due to risk of bias; however, this bias may skew results in favour of SCS. 
^e Downgraded one level due to risk of bias; however, this may bias results in favour of SCS.

Background

Description of the condition

Liver transplantation is the only chance of cure for people with end‐stage liver disease and some people with advanced liver cancers or acute liver failure. Continuous advances in transplant medicine have resulted in excellent outcomes during recent years; the mean five‐year post‐liver transplantation survival is now around 80% in adults in the UK and USA (Kwong 2020; NHSBT 2019). However, the lack of available organs results in significant mortality for those on the waiting list (Kwong 2020; NHSBT 2019). In addition, many people who could receive a life‐saving transplant do not meet current inclusion criteria, as the precious and sparse resource of human organs is carefully allocated.

Currently, approximately one‐third of retrieved organs in the UK do not proceed to transplant (NHSBT 2019). Optimal liver grafts are donated following brainstem death (where warm ischaemic time is low) from standard criteria donors, who are younger than 50 years old, without hepatic steatosis or viral hepatitis (Saidi 2013). Recent attempts to expand the donor pool have resulted in the increasing use of suboptimal organs from expanded criteria donors, who are elderly; have hepatic steatosis, malignancies, or viral hepatitis; and those who donated following circulatory death (DCD) (Saidi 2013). However, it is well documented that these grafts have worse outcomes (both short‐term graft function and long‐term graft survival) if additional measures are not taken to improve preservation and optimise function of the liver transplant (Briceño 2009; Foley 2011; Laing 2016; Nemes 2016).

DCD grafts pose a specific set of challenges, due to the damage brought on by warm ischaemia. This leads to a large increase in the prevalence of ischaemic biliary complications, which generally present in the first six months after transplant with obstructive jaundice and recurrent infection, often resulting in either retransplantation or death (Goussous 2021; Guichelaar 2003). These ischaemic biliary complications have been termed the 'Achilles heel' of DCD liver transplantation (Croome 2020).

Machine perfusion was first described nearly a century ago (Dutkowski 2008). There has been a resurgence of interest since the early 2000s, in the hope that machine perfusion could better preserve and optimise suboptimal grafts, thereby, improving graft and recipient survival following transplantation. Machine perfusion could also potentially allow some of the poorer quality retrieved livers, which currently are not used for transplantation, to be used, further expanding the donor pool (NHSBT 2019).

Description of the intervention

It has long been known that reducing the temperature of organs can prolong their preservation time (Collins 1969). Clinically, this is achieved using a preservation solution to perfuse and surround the liver, which is then packaged in an ice box (termed 'static cold storage', SCS). This is a satisfactory method for preserving high‐quality liver grafts, which are reasonably resistant to ischaemia reperfusion injury. However, as the number of donors with expanded criteria and the number of organ donations after circulatory death have risen to try and match the demand for donor organs, there has been increased interest in alternative preservation methods, which may prove superior to standard SCS (Kwong 2020; NHSBT 2019).

Several machine perfusion technologies and strategies have been developed, which aim to improve organ preservation between donor and recipient, and optimise high‐risk grafts. These include normothermic machine perfusion (NMP), hypothermic oxygenated machine perfusion (HOPE), controlled oxygenated rewarming, and normothermic regional perfusion (Dutkowski 2014; Dutkowski 2015; Guarrera 2015; Hessheimer 2019; Hoyer 2016; Nasralla 2018; Oniscu 2014). The basic components of all of these perfusion circuits are a membrane oxygenator to deliver oxygen and remove carbon dioxide, a pump (which can provide continuous or pulsatile flow), a heater‐cooler unit, a reservoir, and infusion pumps.

A detailed description of every proposed machine perfusion protocol is outside the scope of this current Cochrane Review, but has been recently reviewed (Czigany 2019). Perfusion can be performed in the donor during transportation of the organ from donor to recipient, or prior to implantation at the recipient hospital (end‐ischaemic). Some perfusions are normothermic, and some are hypothermic, but the exact target temperature used in different protocols varies. Hypothermic techniques/strategies can be completed without an oxygen carrier, whereas normothermic techniques/strategies require an oxygen carrier, which is often, but not always, human red blood cells (Matton 2018).

Evidence from the kidney transplantation literature provides hope that machine perfusion may be able to improve clinical outcomes (Tingle 2019). In deceased donor renal transplantation, non‐oxygenated hypothermic machine perfusion (during transport) improved early function after kidney transplantation, and increased long‐term graft survival in organs procured after brain death (donated following brain death; DBD), and after circulatory death (DCD), when compared to SCS (Tingle 2019).

How the intervention might work

Traditional SCS aims to preserve transplant organs by lowering their metabolic rate. In general, for every 10 °C drop, the metabolism rate is halved (Wilson 2006). Therefore, at 4 °C, the enzymatic rate is approximately 10% of that at 37 °C (Wilson 2006). SCS works by removing blood and clots from the liver graft, and replacing them with an acellular preservation solution in a hypothermic environment. However, in this oxygen depletion system, metabolism does not completely halve. This results in adenosine triphosphate (ATP) depletion and ischaemic injury ensues, priming the graft for further reperfusion injury (Martin 2019).

Hypothermic oxygenated machine perfusion uses commercially available machines to deliver oxygenated perfusate while maintaining the principle of lowering metabolic activity through hypothermia. Protocols in current use describe end‐ischaemic HOPE in the recipient hospital (Dutkowski 2014; Dutkowski 2015; Guarrera 2015; van Rijn 2018).

There are multiple proposed beneficial effects of performing HOPE. First, there is a physical washout benefit that helps to clear the microcirculation in the liver, which includes diluting waste products and blood remnants. The Porte Group, from the Netherlands, demonstrated that HOPE increased ATP content more than 15‐fold, which remained elevated after reperfusion (van Rijn 2017; Westerkamp 2016). There also appears to be reduced expression of proinflammatory cytokines (Schlegel 2014a), downregulation of Kupffer cell activity (Guarrera 2011; Schlegel 2013a), and reduced vascular resistance (Lee 2002; Op den Dries 2014). The delivery of oxygen, and recovery of mitochondria before restoration of normothermia (which is not achieved by NMP), is thought to be critical in preventing reperfusion injury (Schlegel 2013b; Schlegel 2014b).

NMP aims to maintain physiological conditions and a metabolically active liver by providing oxygen and nutrients. NMP replenishes liver ATP, and allows reperfusion to take place in a controlled environment, isolated from the recipient's immune and coagulation systems (Brockmann 2009; Xu 2012). NMP can be performed in transit, therefore, avoiding ATP depletion and loss of metabolic homeostasis, which occurs with prolonged SCS (Martin 2019; Nasralla 2018). Finally, as NMP maintains a metabolically active organ, theoretically, it has benefits over hypothermic perfusion as a tool for viability assessment, and therefore, it could improve organ utilisation (Mergental 2016; Watson 2018).

Controlled oxygenated rewarming aims to capitalise on the aforementioned benefits of both HOPE and NMP, including early mitochondrial recovery and viability testing (von Horn 2017). Another key factor is the concept of rewarming injury. In animal models, the process of rapid rewarming (and resulting rapid increase in metabolic demands) has been shown to be detrimental (Minor 2019). In theory, controlled oxygenated rewarming allows early recovery of mitochondria, avoidance of rewarming injury, and maintenance of a metabolically active organ for prolonged preservation and viability testing.

Normothermic regional perfusion in the donor is an alternative approach. Cannulae are placed to access the subdiaphragmatic aorta and inferior vena cava, and oxygenated blood is circulated through abdominal organs while they remain in situ. Animal models have shown that normothermic regional perfusion replenishes ATP and improves antioxidant levels (Net 2001; Net 2005). Legislation in some countries allows cannulation and heparinisation of potential DCD donors premortem; in this setting, normothermic regional perfusion can greatly reduce warm ischaemic time (Hessheimer 2019). This is the only technique that is performed prior to in situ cold flush. Breaking the chain between perimortem warm ischaemia and cold ischaemia may lead to improved outcomes.

Why it is important to do this review

Several machine perfusion technologies have been developed which aim to improve utilisation and outcomes of donated livers. Improved techniques, which allow prolonged preservation, could also have logistical benefits and allow a shift towards operating in daylight hours. In addition, enabling longer preservation times could unlock the use of machine perfusion as a drug‐delivery platform to further optimise liver quality.

Despite promising clinical data, it is unclear if machine perfusion functions better than SCS. It remains unclear which of these perfusion methods, and which perfusion protocols, result in the best clinical outcomes for the various types of available grafts. Trials often compare a new perfusion technique with standard static cold storage, with a lack of head‐to‐head comparisons amongst perfusion protocols. The use of network meta‐analysis would help to answer these questions. To our knowledge, there are no previous systematic reviews or network meta‐analyses of randomised clinical trials in liver machine perfusion.

It is vital to analyse the data from randomised clinical trials to identify the impact of machine perfusion, and to compare different perfusion strategies, to establish the role of these techniques in clinical practice.

Objectives

To evaluate the effects of different methods of machine perfusion (including hypothermic oxygenated machine perfusion (HOPE), normothermic machine perfusion (NMP), controlled oxygenated rewarming, and normothermic regional perfusion) versus each other or versus static cold storage (SCS) in people undergoing liver transplantation.

Methods

Criteria for considering studies for this review

Types of studies

We included randomised clinical trials that compared preservation methods in deceased donor liver transplantation if at least one intervention group used machine perfusion. We anticipated that some trials would randomise at the recipient level, whereas some trials would randomise the liver graft itself into either intervention or control (especially those studies where perfusion was initiated at the donor hospital). We included both types of trials. We planned to include trials that fulfilled our inclusion criteria, even if they did not measure any of our outcomes of interest. We planned to use quasi‐randomised studies and observational studies for the report on harms (in a narrative way).

Types of participants

Inclusion criteria

Recipients of whole, deceased donor liver transplants from donors following either circulatory (DCD) or brain death (DBD). We planned to include trials of recipients of donors who met both standard criteria and expanded criteria (Saidi 2013). All adult recipients of liver transplants were eligible, including those receiving a retransplant. We included trials investigating liver utilisation proportions, which randomised liver grafts to a perfusion group; in this setting, the participant was the randomised donor liver. We also planned to include trials where only a subset of the participants were eligible; but no such trials were identified. We only included trials in humans.

Exclusion criteria

We planned to exclude any studies with unethical retrieval of livers (ISTTOT 2008). Therefore, we assessed whether publications identified in our searches have since been withdrawn due to fraud or unethical organ retrievals. We excluded case series comparing perfusion techniques to retrospective cohorts. We excluded trials where grafts had been split or reduced, or used as part of a multivisceral transplant. We excluded preclinical studies investigating machine perfusion technologies where none of the grafts proceeded to transplant.

Types of interventions

Several machine perfusion technologies have been described, and any of these were eligible for inclusion. Specifically, we planned to compare the following methods: SCS (control), HOPE, NMP, controlled oxygenated rewarming, or normothermic regional perfusion. We expect that in the future, some trials may combine these techniques (e.g. normothermic regional perfusion followed by NMP). We grouped studies comparing HOPE via both hepatic artery and portal vein, or via portal vein only.

Types of outcome measures

We assessed outcomes at the maximum follow‐up, unless stated otherwise, for each of the following outcomes.

Primary outcomes

• Overall participant survival

• Quality of life, assessed using any validated scale

• Serious adverse events. We accepted individual complications and serious adverse events defined by:

  • Clavien‐Dindo classification, grade III or higher (Clavien 2009; Dindo 2004);

  • International Conference on Harmonization–Good Clinical Practice (ICH‐GCP) guideline: any untoward medical occurrences that resulted in death, were life‐threatening, required inpatient hospitalisation or prolongation of existing hospitalisation, and resulted in persistent or significant disability or incapacity (ICH‐GCP 2016).

Secondary outcomes

• Graft survival

• Ischaemic biliary complications, within six months and at maximum follow‐up (e.g. ischaemic‐type biliary lesions (Foley 2011))

• Primary non‐function of the graft

• Early allograft function, measured with a validated model (seven days) (e.g. Early Allograft Dysfunction or Model for Early Allograft Function criteria (Jochmans 2017; Olthoff 2010; Pareja 2015))

• Adverse events not considered serious by the aforementioned definitions

• Transplant utilisation (proportion of grafts allocated to an intervention that proceeded to transplant)

• Transaminase release during the first week post‐transplant (participant serum) (until seven days)

We prespecified that we would not draw conclusions on the superiority of a perfusion technique based solely on outcomes such as early allograft dysfunction and transaminase release post‐transplant; these laboratory values are of questionable clinical relevance in machine perfusion studies, as proteins are washed out of the graft during machine perfusion. We have not included transaminase release in our summary of findings tables, and only discussed this outcome narratively. These outcomes are included as they have been used as the primary outcome in landmark machine perfusion trials (Markmann 2022; Nasralla 2018). Some clinicians consider these markers to be important in the field of liver machine perfusion, when trials powered for overall participant survival or graft survival are difficult to conduct.

We also prespecified that we would not draw conclusions on the superiority of a technique based on graft utilisation, taken in isolation. We considered graft utilisation in the context of the outcomes in grafts that are transplanted. If a certain preservation technique increases graft utilisation while improving, or maintaining recipient outcomes, this is clearly of benefit.

Search methods for identification of studies

Electronic searches

The Cochrane Hepato‐Biliary Group (CHBG) Information Specialist searched the CHBG Controlled Trials Register via the Cochrane Register of Studies Web on 16 May 2022 and 10 January 2023. We searched the Cochrane Central Register of Controlled Trials (2022, Issue 5), MEDLINE Ovid (1946 to 10 January 2023), Embase Ovid (1974 to 10 January 2023), LILACS (Bireme; 1982 to 10 January 2023); Science Citation Index Expanded (Web of Science; 1900 to 10 January 2023); and Conference Proceedings Citation Index (Web of Science; 1990 to 10 January 2023). The latter two were searched simultaneously through the Web of Science.

Appendix 1 gives the search strategies with the date range of the searches.

Searching other resources

We searched the following online trial registries to identify ongoing and unpublished trials: ClinicalTrial.gov (clinicaltrials.gov/), European Medicines Agency (EMA; www.ema.europa.eu/ema/), World Health Organization International Clinical Trial Registry Platform (trialsearch.who.int/Default.aspx), and the US Food and Drug Administration (FDA; www.fda.gov).

We also searched the reference lists of relevant studies and clinical practice guidelines and searched the reference lists of recent reviews on liver machine perfusion (e.g. Czigany 2019 and Schlegel 2019).

Data collection and analysis

We performed this review in line with recommendations from the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021a).

Selection of studies

Two review authors (ST and JD) independently screened titles identified by our search strategy. We discarded reports that did not fulfil the inclusion criteria of this review; however, initially, we retained reports of studies and reviews that may have included relevant data or references to relevant trials. Two review authors (ST and JD) then independently assessed retrieved abstracts, and if necessary, the full text of these studies to determine which studies satisfied the inclusion criteria. One review author (CW) was available to resolve any differences in opinion around trial selection, but there were none.

Where multiple publications existed for a single randomised clinical trial, we listed all of them under a single, main study reference. We also planned to include trials that were only available as conference abstracts. We planned to include trials in the review, regardless of whether measured outcome data were reported in a usable way, only if we could ensure that these studies were not withdrawn because of ethical or other reasons. A PRISMA flowchart was generated to detail the output of our search and study selection (Page 2021a; Page 2021b).

Data extraction and management

Two review authors (ST and JD) independently extracted necessary data from the included trials, using a prepiloted data extraction form. This included information on trial characteristics, included participants, type(s) of machine perfusion, additional potential effect modifiers, outcome data (including measurement time point), and source of funding. We planned to resolve any discrepancies with the senior review author (CW).

Assessment of risk of bias in included studies

Two review authors (ST and JD) independently assessed risk of bias, and where differences of opinion arose, CW arbitrated. We used the Cochrane RoB 2 tool, as outlined in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021b).

We assessed the effect of assignment to the intervention using RoB 2 (Higgins 2021b). Therefore, we performed analyses based on the intention‐to‐treat (ITT) principle, which includes all randomised participants, regardless of the interventions that participants actually received.

In brief, we assessed the following sources of bias in the individually randomised trials, at the outcome level, using RoB 2 (Higgins 2021b; Sterne 2019).

• Bias arising from the randomisation process

• Bias due to deviations from intended interventions

• Bias due to missing outcome data

• Bias in measurement of the outcome

• Bias in selection of the reported result

As per the Cochrane Handbook for Systematic Reviews of Interventions guidance, we used RoB 2 to assess the risk of bias for only a subset of outcomes (Higgins 2021b). There is currently no guidance on how to select outcomes for which to perform risk of bias assessments; we have selected outcomes based on the most contemporaneous advice from Cochrane. The risk of bias assessments fed into one domain of the GRADE approach for assessing certainty of a body of evidence (Schünemann 2021). For the risk of bias assessment, we focussed on results of the trials that contributed information that users of the review will find most useful; overall participant survival at maximum follow‐up, quality of life, serious adverse events, graft survival, ischaemic biliary complications (ischaemic‐type biliary lesions) within one year, and adverse events considered non‐serious.

We carefully studied the latest (i.e. at the time of review production) guidance on preliminary consideration for assessing risk of bias, the signalling questions to be used, and the response options for the signalling questions, such as yes, probably yes, no, probably no, and no information (Higgins 2021b; Sterne 2019).

Overall risk of bias

The overall rating assigns one of three levels of judgement:

• low risk of bias: the trial was at low risk of bias for all domains for this result;

• some concerns: the trial was judged to raise some concerns in at least one domain for this result, but was not at high risk of bias for any of the remaining domains;

• high risk of bias: the trial was at high risk of bias in at least one domain for this result, or the study was judged to have some concerns for multiple domains in a way that substantially lowered confidence in the result.

We used the same levels of overall risk of bias judgements across different trials for each of the domains listed as we used for the individual domains, that is low risk of bias, some concerns, or high risk of bias. Judging a result to be at a particular level of risk of bias for an individual domain implied that the result has an overall risk of bias at least this severe. We followed the guidance on preliminary consideration for assessing risk of bias on how to record risk of bias in trial data obtained through different sources (e.g. unpublished data, correspondence with a trialist, etc.).

Displaying risk of bias judgements

We used the most recently developed RoB 2 Excel tool. An algorithm in Excel maps the responses to the signalling questions per outcome, and proposes a risk of bias judgement for each domain. We stored the Excel spreadsheets for each trial outcome on a secure server, which are available upon request. In the most up‐to‐date version of Review Manager Web at the time of writing (RevMan Web 2020), RoB 2 assessments (and their justifications) are only visible if a trial is included in the 'analysis' section for that outcome. Therefore, where only one trial gives usable data for an outcome or interest, a forest plot and analysis has still been generated in order to include RoB 2 assessments and their justifications.

Measures of treatment effect

We used odds ratios (OR) with 95% confidence intervals (CI) for dichotomous outcomes. We analysed participant survival and graft survival as time‐to‐event data, and performed meta‐analysis using the general inverse‐variance method, as described in Chapter 10 of the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2021). Where hazard ratios (HR) were not available, we used techniques described by Tierney 2007 to estimate HRs, in line with guidance from the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021c).

For quality of life, we planned to pool the results using standardised mean difference (SMD) with 95% CI. However, no trials reported on this outcome.

Unit of analysis issues

We anticipated that included trials would randomise either at the level of the transplant recipient or at the level of the liver graft. For trials randomising transplant recipients, we did not foresee any unit of analysis issues. For trials randomising liver grafts and reporting on graft utilisation, the unit of analysis was the liver grafts for this outcome. Such trials also reported on recipient outcomes; in this case, the unit of analysis was the transplant recipient. We did not expect cluster or cross‐over trials, and none were found. We analysed the results using ITT analysis wherever possible.

Dealing with missing data

Our preferred method of dealing with missing data was to contact trial authors and attempt to obtain missing data from them directly. When this failed, we planned to impute missing data using methods outlined in the Cochrane Handbook for Systematic Reviews of Interventions (e.g. last observation carried forward; imputing an assumed outcome, such as assuming all were poor outcomes; imputing the mean; imputing based on predicted values from a regression analysis (Higgins 2021d)). We planned to select an imputation method depending on the data available; if multiple methods were appropriate, we planned to perform sensitivity analyses with each.

For missing standard deviations, we planned to impute these from P values, 95% CIs, or from other studies with similar designs, participant groups, and numbers (Furukawa 2006). As described above, we analysed survival as time‐to‐event data. In all cases, this could be performed directly using HRs, or estimating these using the available time‐to‐event analyses (such as number of events and P values from logrank tests) using methods described by Tierney 2007.

Assessment of heterogeneity

We assessed clinical and methodological heterogeneity as described in Subgroup analysis and investigation of heterogeneity. We visually assessed heterogeneity using forest plots generated from pairwise comparison meta‐analyses. We paid particular attention to the following potential effect modifiers when assessing heterogeneity: DCD compared to DBD, expanded criteria donors compared to standard criteria donors, steatotic grafts, age of donors, duration of cold ischaemic time or perfusion time, type of perfusion fluid used, and recipient factors (age, indication, retransplant). For most of these, there was an insufficient number of trials to perform meaningful subgroup analyses.

Then, we analysed the presence and extent of heterogeneity using the standard I² statistic (the percentage of the variability in effect estimates that was due to heterogeneity rather than chance). Strict cut‐offs for the I² statistic are often not useful, especially in this review where the number of trials for each comparison was low (making the Chi² test, which underlies the I² statistic, underpowered). However, we used the following thresholds as a guide (Deeks 2021):

• 0% to 40%: might not be important;

• 30% to 60%: may represent moderate heterogeneity;

• 50% to 90%: may represent substantial heterogeneity;

• 75% to 100%: considerable heterogeneity.

Assessment of reporting biases

As there were fewer than 10 trials in the analysis, funnel plots could not be used to assess for small‐study bias (Higgins 2021a).

Data synthesis

We performed all data analyses following instructions from the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2021a). For pairwise analyses, we used Review Manager Web (RevMan Web 2020). If meta‐analysis was not possible, we planned to use alternative methodologies to report results, but these were not required.

We performed individual pairwise meta‐analyses when there was direct evidence from at least two trials comparing the same preservation techniques. We performed this using random‐effects models for our main analyses. We used the methods of Bucher 1997, which are endorsed by the Cochrane Handbook for Systematic Reviews of Interventions (Chaimani 2021), to perform an indirect comparison between two machine perfusion techniques (HOPE and NMP), where the only trials available compared them to a common control preservation technique (SCS).

Subgroup analysis and investigation of heterogeneity

We planned to perform the following subgroup analyses; however, sufficient data were rarely available to allow this.

• Trials at overall low risk of bias compared to trials with some concerns, and trials at high risk of bias, as trials with some concern and at high risk of bias may overestimate or underestimate the intervention effects (Higgins 2021b).

• Trials without for‐profit funding compared to trials with some concern or at risk of for‐profit funding, as conflicts of interest can introduce bias, including publication bias (Lundh 2017).

• Donation following circulatory death (stratified by Maastricht criteria) compared to donation following brainstem death, as these grafts have different patterns of ischaemic injury, and therefore, treatment effects may vary (Thuong 2016).

• Standard criteria donors compared to extended criteria donors, as standard criteria organs may be more resilient to ischaemia reperfusion injury, meaning preservation effects may vary; this was not possible as they either included only extended criteria, or did not give separate results for standard versus extended criteria donors.

• Continuous perfusion during transport compared to perfusion at the recipient centre (end‐ischaemic), as the prior cold ischaemic time in the latter group may alter treatment effect size; this was not possible as all HOPE trials were end‐ischaemic, and all non‐pilot NMP trials were continuous.

We assessed statistical heterogeneity as described above. We planned to perform this assessment for the overall cohort, and within each subgroup, providing insights into the source of any heterogeneity present. However, due to the small number of trials such analyses were either not possible, or would be meaningless due to very low statistical power.

Sensitivity analysis

We assessed the sensitivity of our results to changes in methodology, by performing the following sensitivity analyses of our outcome measures.

• Fixed‐effect model (for pairwise meta‐analyses)

• Including only trials at low risk of bias; this was not possible as meta‐analyses either included only trials at low risk of bias, or no trials at low risk of bias

• Analyses excluding trials for which we imputed data

Summary of findings and assessment of the certainty of the evidence

We created two summary of findings tables on two comparisons (i.e. HOPE versus SCS for liver transplant recipients and NMP versus SCS for liver transplant recipients). We presented the following outcomes: overall participant survival at maximum follow‐up, quality of life, serious adverse events, graft survival, ischaemic biliary complications (within six months), early allograft dysfunction (seven days), and adverse events considered non‐serious. We included these outcomes in the summary of findings tables even if there were no data available. We used the five GRADE factors (i.e. risk of bias, inconsistency, imprecision, indirectness, and publication bias) to assess the certainty of evidence as it related to the trials that contributed data for the prespecified outcomes (Schünemann 2013).

Regarding 'risk of bias', we used the overall judgement for an outcome result. 'Low' risk of bias indicates 'no limitation (the certainty is not rated down)'; 'Some concerns' indicates either 'no limitation' or 'serious limitation (the certainty is rated down one level)'; and 'High' risk of bias indicates either 'serious limitation' or 'very serious limitation (the certainty is rated down two levels)'.

We used the methods and recommendations described in Chapter 14 of the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2021), using GRADEpro GDT software (GRADEpro GDT). We justified all decisions to downgrade the certainty of evidence using footnotes, and made comments to aid the reader's understanding of the review where necessary.

We assessed the certainty of evidence as falling into one of these levels of evidence.

• High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.

• Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.

• Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect.

• Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Two review authors (ST and JD) independently assessed the certainty of evidence, and disagreements were resolved by a third review author (CW). Next to each outcome in the summary of findings tables, we have provided the duration of the analysed follow‐up data (including its mean and range where possible).

Results

Description of studies

Results of the search

The database searches identified 2502 records and clinical trial registries identified 179 records on 16 May 2022 and 10 January 2023 (Figure 1). After exclusion of duplicates, we screened 2493 records and excluded 2406 records based on title and abstract. We retrieved the full text of the remaining 87 records and assessed them for eligibility. We excluded 22 trials (32 records) with reason for exclusion (Characteristics of excluded studies table) and identified 11 ongoing studies (19 reports) investigating machine perfusion technologies (Characteristics of ongoing studies table). Overall, seven trials were included in the review, comprising 34 individual reports (Characteristics of included studies table). One study (two records) is awaiting classification (Characteristics of studies awaiting classification table).

1.

PRISMA flow diagram for the identification of included studies in the meta‐analysis of machine perfusion in liver transplantation (Page 2021a; Page 2021b). Date of last search 10 January 2023.

Our search identified no quasi‐randomised studies or observational studies which would add additional information on intervention harms to the review.

Included studies

Full details for the seven included trials are available in the Characteristics of included studies table. These seven trials included 1024 transplant recipients from 1301 randomised/included livers. Trials were completed in Europe (Czigany 2021; Ghinolfi 2019; Nasralla 2018; Ravaioli 2022; Schlegel 2023; van Rijn 2021) or the US (Markmann 2022).

All were two‐arm randomised clinical trials comparing a machine perfusion technique with SCS. Two large trials (Markmann 2022; Nasralla 2018) and one pilot trial (Ghinolfi 2019) compared NMP with SCS. The other four trials compared HOPE with SCS (Czigany 2021; Ravaioli 2022; Schlegel 2023; van Rijn 2021). No trials used normothermic regional perfusion, all DCD retrievals used the standard super rapid retrieval technique. Machines used for NMP were the OrganOx Metra (Nasralla 2018), Liver Assist from Organ Assist (Ghinolfi 2019), and the Transmedics portable Organ Care System Liver device (Markmann 2022); all of these performed oxygenated perfusions at 37 °C with a blood‐based perfusate. Machines used for HOPE were the Liver Assist from Organ Assist (Czigany 2021; Schlegel 2023; van Rijn 2021) and the Bridge to Life Vitasmart (Ravaioli 2022).

All trials of HOPE performed this as an end‐ischaemic intervention, in the recipient hospital (Czigany 2021; Ravaioli 2022; Schlegel 2023; van Rijn 2021). van Rijn 2021 perfused both the hepatic artery and portal vein (sometimes termed dual HOPE), whereas Czigany 2021, Ravaioli 2022, and Schlegel 2023 perfused only the portal vein. The two largest NMP trials initiated NMP in the donor hospital and continued throughout transport (Markmann 2022; Nasralla 2018), whereas the pilot trial investigated end‐ischaemic NMP (Ghinolfi 2019).

SCS was generally performed according to pre‐established local protocols, as this treatment is the current standard of care. Regarding NMP versus SCS trials: Nasralla 2018 used either University of Wisconsin solution (UW) or histidine‐tryptophan‐ketoglutarate (HTK) as the SCS solution, Ghinolfi 2019 used Celsior solution, and Markmann 2022 did not specify the cold storage solution used. The SCS solution used in HOPE versus SCS trials was either HTK (Czigany 2021; Schlegel 2023), UW (Schlegel 2023; van Rijn 2021), Celsior solution (Ravaioli 2022), or Institut Georges Lopez solution (IGL‐1) (Schlegel 2023).

Median total preservation times ranged from 5 hours and 38 minutes to 11 hours and 54 minutes (further details in Characteristics of included studies table). The longest total preservation times were achieved in the NMP arm of Nasralla 2018.

Trials included: only DCD livers (van Rijn 2021), only DBD donors (Czigany 2021; Ghinolfi 2019; Ravaioli 2022; Schlegel 2023), or a mix of the two (Markmann 2022; Nasralla 2018). Trials including DBD livers typically restricted to only expanded criteria donors (Czigany 2021; Ghinolfi 2019; Ravaioli 2022), whereas Nasralla 2018 and Schlegel 2023 also included standard criteria DBD and Markmann 2022 selected livers with a moderate risk profile.

As detailed in our protocol, we emailed trial authors for clarification where needed. This was only required for Ravaioli 2022, who kindly provided raw data on graft survival and mortality to allow them to be analysed as time‐to‐event variables.

Excluded studies

We excluded 22 screened studies; 18 were not randomised, two were preclinical studies where the liver was not transplanted, and two presented no primary data (review/editorial); further details on these studies are available in the Characteristics of excluded studies table. None of these studies provided any useful data regarding reports of harm.

Studies awaiting classification

One study (two records) is awaiting classification (Characteristics of studies awaiting classification table).

Ongoing studies

Eleven studies are ongoing (Characteristics of ongoing studies table).

Risk of bias in included studies

We assessed bias using the RoB 2 tool at the outcome level (Higgins 2021b). These judgements, along with justifications, are given alongside relevant analyses and in the Risk of bias (tables). The RoB 2 Excel spreadsheet which was used to generate these judgements is available upon request.

Bias arising from the randomisation process

All trials reported suitable methods for randomisation. Most trials used centralised electronic randomisation (Czigany 2021; Markmann 2022; Nasralla 2018; Ravaioli 2022; van Rijn 2021). Ghinolfi 2019 used sealed envelopes and Schlegel 2023 used either centralised electronic randomisation or Microsoft Excel, depending on the centre. Six trials were at low risk of bias in this domain, across all assessed outcomes (Czigany 2021; Ghinolfi 2019; Nasralla 2018; Ravaioli 2022; Schlegel 2023; van Rijn 2021).

Markmann 2022 allocated recipients to interventions and concealed the allocation until recipients were enroled into the trial. However, allocation was revealed to the retrieving/implanting team prior to the visual assessment of the liver in the donor. The implanting surgeon could decline the liver for transplantation at that stage, with knowledge of the assigned group, before the intervention (NMP or SCS) had begun. When being assessed in the donor (before retrieval of the liver), the implanting surgeon was significantly more likely to accept a DCD liver for a recipient who was randomised to machine perfusion. As this decision happened prior to retrieval and initiation of preservation, this highlights a lack of equipoise amongst implanting surgeons. Therefore, the baseline risk profile of donors was different, prior to initiation of NMP or SCS, with significantly more DCD liver transplants in the NMP group.

Bias due to deviations from intended interventions

In machine perfusion trials, the intervention is initiated, performed, and completed before the implant of the liver takes place. Therefore, issues of 'non‐adherence', and participants in the control group later seeking out the intervention, cannot occur. In six trials all/nearly all participants received the intervention that they were assigned to and an appropriate ITT analysis was performed, making them low risk of bias for this domain across all assessed outcomes (Czigany 2021; Ghinolfi 2019; Nasralla 2018; Ravaioli 2022; Schlegel 2023; van Rijn 2021).

As discussed above, in Markmann 2022, many more DCD livers which were allocated to NMP ended up receiving this intervention, whereas many DCD livers allocated to SCS never received this intervention. In addition, Markmann 2022 used per‐protocol analyses for their primary analyses. For some outcomes, an ITT analysis is also given (judgement of some concerns in this domain), but for other outcomes, only per‐protocol was available (judgement of high risk of bias).

Bias due to missing outcome data

In five trials there were few missing data, and this was clearly reported; these trials were considered low risk in this domain (Czigany 2021; Ghinolfi 2019; Ravaioli 2022; Schlegel 2023; van Rijn 2021).

In Nasralla 2018, the unit of randomisation was the liver, but recipient outcomes were clearly only available for transplanted grafts. A significantly higher proportion of participants were transplanted in the NMP group — as fewer grafts were discarded in the NMP group, it may be that the baseline risk profile of transplanted grafts was higher than for SCS livers. However, it was unclear whether the extra grafts which were transplanted in the NMP group were genuinely lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore caused bias in favour of SCS for recipient outcomes. For this reason, Nasralla 2018 was at some concerns in this domain.

Markmann 2022 was also biased in this domain for similar reasons. In addition, data on loss to follow‐up was not clear for some outcomes. For one‐year graft survival (98% for NMP versus 99% for SCS) the numbers at risk at one year (eFigure 2 in the supplemental content of the primary reference for the study) were 104 in the NMP group and 105 in the SCS group; it is not clear how participants were lost to follow‐up, so this outcome was at high risk in this domain.

Bias in measurement of the outcome

Four trials were at low risk across all outcomes, except adverse events (Czigany 2021; Ghinolfi 2019; Ravaioli 2022; van Rijn 2021). Adverse events in these trials were assessed by teams which were not blinded (due to the nature of the intervention) — knowledge of assignment to treatment group may have altered the decision to report an event as an adverse event. Nasralla 2018 was low risk of bias across all assessed outcomes, including adverse events as two independent blinded clinicians assessed adverse events.

Schlegel 2023 was at low risk across all outcomes included in the review. Some outcomes were added post‐hoc, and not predefined in the protocol. Due to the risk of bias that this introduces, these post‐hoc outcomes were discussed narratively and not included in any meta‐analyses.

Markmann 2022 was at low risk of bias in this domain for the outcome of participant survival. For graft survival, Markmann 2022 was graded 'some concerns' as the definition was not prespecified, it is not clear whether this was death‐censored. Markmann 2022 was at high risk of bias for the outcomes serious adverse events and ischaemic biliary complications; clinical teams clearly believed that SCS was inferior to NMP (higher chance of rejecting livers prior to retrieval if the recipient was randomised to SCS) — these same teams may have been more likely to order investigations/biopsies in participants receiving SCS organs. No information was given on the number of investigations performed in each group, and no protocol biopsies/imaging were mentioned. This bias persisted even though the radiologists were blinded.

Bias in selection of the reported result

Six trials were at low risk of bias in this domain across every assessed outcome (Czigany 2021; Ghinolfi 2019; Nasralla 2018; Ravaioli 2022; Schlegel 2023; van Rijn 2021).

Markmann 2022 was at high risk of bias in this domain for the outcome ischaemic biliary complications. Trial authors combined both non‐anastomotic strictures and bile leaks, and labelled them as 'ischaemic biliary complications'. Anastomotic strictures are not included in this definition — including bile leaks in the definition of ischaemic biliary complications is not standard. This definition was not prespecified in the protocol, which stated that they would investigate biliary strictures as an endpoint. Combining two distinct outcomes (non‐anastomotic strictures and bile leaks), post‐hoc, was judged to put them at high risk of bias.

Other risk of bias

Markmann 2022 received for‐profit funding from Transmedics who led the trial design and were responsible for data collection and generating the trial report; this raised concern about potential bias, especially for certain outcomes where the report used non‐standard definitions. van Rijn 2021 received not‐for‐profit funding to run the trial, in addition to Bridge to Life providing UW preservation solution; this element is for‐profit, but Bridge to Life had no other role in the trial. All other trials received not‐for‐profit funding (Czigany 2021; Ghinolfi 2019; Nasralla 2018; Ravaioli 2022; Schlegel 2023).

Effects of interventions

See: Table 1; Table 2

Overall participant survival

van Rijn 2021 provided an HR and CIs which could be used directly in meta‐analysis. Czigany 2021, Nasralla 2018, Ravaioli 2022, and Schlegel 2023 did not provide HRs but did perform time‐to‐event analyses which allowed estimation of HRs (Tierney 2007).

It is uncertain whether HOPE reduces overall participant survival compared with SCS (HR 0.91, 95% CI 0.42 to 1.98; P = 0.81, I² = 0%; 4 trials, 482 recipients; Analysis 1.1; low‐certainty evidence due to imprecision because of low number of events). There was no evidence that this effect differed in DCD versus DBD grafts (P = 0.36). Sensitivity analyses using a fixed‐effect model and removing trials with for‐profit funding yielded similar results. It remains uncertain whether NMP reduces participant survival compared with SCS (HR 1.08, 95% CI 0.31 to 3.80; P = 0.90; 1 trial, 222 recipients; Analysis 2.1; very low‐certainty evidence due to imprecision and risk of bias).

1.1. Analysis.

Comparison 1: Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS), Outcome 1: Overall participant survival

2.1. Analysis.

Comparison 2: Normothermic machine perfusion (NMP) versus static cold storage (SCS), Outcome 1: Overall participant survival

An indirect comparison of HOPE versus NMP using the results of Analysis 1.1 and Analysis 2.1 was performed with the techniques of Bucher 1997; it is uncertain whether HOPE impacts on overall participant survival when compared with NMP (indirect HR 0.84, 95% CI 0.19 to 3.68; P = 0.83; very low‐certainty evidence due to the NMP versus SCS comparison).

Ghinolfi 2019 investigated participant survival at six months and found that 10/10 participants in the NMP group were alive and 9/10 in the SCS group were alive (overall risk of bias for this outcome was low); as there were no events in the NMP group an HR could not be calculated. Markmann 2022 did not provide an ITT analysis for overall participant survival. Their per‐protocol analysis revealed overall participant survival of 94% in the NMP group versus 93.7% in the SCS group (no P value reported); overall risk of bias for this outcome was high risk.

Quality of life

No trials reported on any quality of life measures.

Serious adverse events

Czigany 2021 and Ravaioli 2022 both reported the number of recipients experiencing serious adverse events based on Clavien‐Dindo classification within three months. Meta‐analysis of these studies confirmed a reduction in serious adverse events with HOPE versus SCS in extended criteria DBD livers (OR 0.45, 95% CI 0.22 to 0.91; P = 0.03, I² = 0%; 2 trials, 156 recipients; Analysis 1.2; moderate‐certainty evidence). Sensitivity analysis with a fixed‐effect model did not alter these results.

1.2. Analysis.

Comparison 1: Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS), Outcome 2: Serious adverse events (90‐day follow‐up)

Schlegel 2023 examined serious adverse events (Clavien‐Dindo 3a or greater) within the first year post‐transplant and found 44/85 recipients in the HOPE group versus 46/85 in the SCS group experienced a serious adverse event. Schlegel 2023 performed a post‐hoc analysis of serious adverse events (Clavien‐Dindo 3b or greater) which blinded assessors defined as liver‐graft‐related; they reported a reduction in such events (7/85 in the HOPE group versus 17/85 in the SCS group; P = 0.027).

Nasralla 2018 and van Rijn 2021 provided data on the total number of serious adverse events, but not the number of people experiencing serious adverse events. van Rijn 2021 reported 101 serious adverse events across 78 recipients of HOPE livers, and 132 serious adverse events across 78 recipients of SCS livers (overall low risk of bias). Nasralla 2018 reported 21 serious adverse events across 121 recipients of NMP livers and 36 serious adverse events across 101 recipients of SCS livers (overall some concerns of bias). Markmann 2022 did not provide details on overall serious adverse events but provided information on liver‐specific serious adverse events (predefined as primary non‐function, ischaemic biliary complications, hepatic vascular complications, or liver graft infection within 30 days); the trial authors reported 7/153 in the NMP group versus 11/146 in the SCS group experienced liver‐specific serious adverse events (overall some concerns of bias). Ghinolfi 2019 provided no information on the number of serious adverse events in either group. Overall, there is good evidence that NMP is safe and does not increase the rate of serious adverse events.

Graft survival

van Rijn 2021 provided an HR and CIs which could be used directly in meta‐analysis. Czigany 2021, Nasralla 2018, Markmann 2022, Ravaioli 2022, and Schlegel 2023 performed time‐to‐event analyses (Kaplan‐Meier with logrank tests) which allowed estimation of HRs (Tierney 2007).

Four trials compared HOPE with SCS (Czigany 2021; Ravaioli 2022; Schlegel 2023; van Rijn 2021). There was an improvement in graft survival with HOPE compared to SCS alone (HR 0.45, 95% CI 0.23 to 0.87; P = 0.02, I² = 0%; 4 trials, 482 recipients; Analysis 1.3; high‐certainty evidence). On subgroup analysis, HOPE improved graft survival versus SCS in the "extended criteria" DBD group (HR 0.30, 95% CI 0.11 to 0.82; P = 0.02, I² = 0%; 2 trials, 156 recipients; Czigany 2021; Ravaioli 2022; Analysis 1.3; high‐certainty evidence). Sensitivity analysis using a fixed‐effect model, and removing trials which received for‐profit funding, did not alter these results. There was no evidence of a differing treatment effect in livers from DCD or DBD donors (P = 0.51, I² = 0%; Analysis 1.3).

1.3. Analysis.

Comparison 1: Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS), Outcome 3: Graft survival

In contrast, NMP was not associated with a reduction in graft survival (HR 1.20, 95% CI 0.44 to 3.29; P = 0.72, I² = 0%; 2 trials, 522 recipients; Analysis 2.2; low‐certainty evidence due to imprecision and risk of bias). Sensitivity analyses using a fixed‐effect model, and removing Markmann 2022 (due to risk of bias and for‐profit funding), did not change these conclusions.

2.2. Analysis.

Comparison 2: Normothermic machine perfusion (NMP) versus static cold storage (SCS), Outcome 2: Graft survival

An indirect comparison of HOPE versus NMP using the results of Analysis 1.3 and Analysis 2.2 was performed with the techniques of Bucher 1997; HOPE was associated with improved graft survival compared to NMP (indirect HR 0.38, 95% CI 0.11 to 1.23; P = 0.106; low‐certainty evidence due to the NMP versus SCS comparison).

Markmann 2022 described graft survival at 12 months; however, they only provided a per‐protocol analysis and gave no specific details on loss to follow‐up (although number at risk in their eFigure 2 in the supplemental content of the primary reference for the study suggested significant loss to follow‐up); sensitivity analysis removing this trial from Analysis 2.2 did not change the conclusions. Ghinolfi 2019 investigated graft survival at six months and found one graft loss out of the 10 recipients in the NMP group and no graft losses in the 10 recipients in the SCS group (overall risk of bias for this outcome was low risk); as there were no events in the SCS group an HR could not be calculated.

Ischaemic biliary complications

van Rijn 2021 assessed ischaemic biliary complications in livers from DCDs; crucially they specified clinically significant ischaemic cholangiopathy (with jaundice, recurrent cholangitis, or cholestatic liver "function" test derangement) and two blinded radiologists reviewed imaging with central blinded review of all clinical cases. This confirmed a reduction in ischaemic cholangiopathy with HOPE versus SCS in DCD livers at six months (OR 0.31, 95% CI 0.11 to 0.92; P = 0.03). Schlegel 2023 reported non‐anastomotic strictures at 12 months in DBD donors. Overall, HOPE decreased ischaemic biliary complications compared with SCS (OR 0.32, 95% CI 0.12 to 0.83; P = 0.02; 2 trials, 326 recipients; Analysis 1.4; high‐certainty evidence)

1.4. Analysis.

Comparison 1: Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS), Outcome 4: Ischaemic biliary complications

Nasralla 2018 assessed non‐anastomotic strictures on protocol magnetic resonance cholangiopancreatography at six months (defined as stricture greater than 70% luminal diameter); it is unclear how many of these were clinically significant. They found minimal differences between the two groups (OR 0.78, 95% CI 0.27 to 2.27; P = 0.65; 1 trial, 155 recipients; Analysis 2.3; very low‐certainty evidence due to imprecision and risk of bias). This trial included DBD and DCD grafts and there was no difference when these subgroups were examined separately. At one year, they described two 'clinically relevant' recipients with ischaemic cholangiopathy (no specific definition given); one in the NMP group and one in the SCS group.

2.3. Analysis.

Comparison 2: Normothermic machine perfusion (NMP) versus static cold storage (SCS), Outcome 3: Ischaemic biliary complications (on protocol imaging)

Markmann 2022 assessed ischaemic biliary complications with NMP versus SCS. Their definition of ischaemic biliary complications in the final manuscript was non‐anastomotic strictures or bile leaks (but not anastomotic strictures). Including anastomotic bile leaks in the definition of ischaemic biliary complications is not standard, and was not prespecified in their protocol (which only included non‐anastomotic strictures). There was no breakdown of the number of recipients experiencing non‐anastomotic strictures, and therefore true differences in the rate of ischaemic biliary complications could not be assessed. They also provided no definition for non‐anastomotic strictures or bile leaks, or whether these were clinically significant. Results discussed below regarding transplant utilisation also highlighted that surgeons performing transplantation in the trial believed that NMP was superior to SCS; teams (which were unblinded) may therefore have been more likely to request subsequent imaging for recipients in the SCS group — no information was given on how many recipients in each group received imaging. Markmann 2022 reported a reduction in their composite outcome of non‐anastomotic strictures plus bile leaks of 4/153 in the NMP group and 14/146 in the SCS group (overall high risk of bias).

Ghinolfi 2019 reported no ischaemic biliary complications in either group over six months of follow‐up (10 per group). Ravaioli 2022 and Czigany 2021, which both investigated HOPE versus SCS in grafts from DBD donors, did not include ischaemic biliary complications as an outcome.

Primary non‐function

All seven trials reported on primary non‐function. It is uncertain whether HOPE reduces primary non‐function compared with SCS (OR 0.32, 95% CI 0.07 to 1.43; I² = 0%, P = 0.13; 4 trials, 482 recipients; Analysis 1.5; low‐certainty evidence due to imprecision). Sensitivity analyses using a fixed‐effect model, and removing trials with for‐profit funding yielded similar results.

1.5. Analysis.

Comparison 1: Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS), Outcome 5: Primary non‐function

Ghinolfi 2019 and Markmann 2022 reported no recipients with primary non‐function in either group. Nasralla 2018 reported 1 episode/121 recipients with primary non‐function in the NMP group versus 0 episodes/101 recipients with primary non‐function in the SCS group.

Early allograft dysfunction

All trials assessed early allograft dysfunction at one week using the Olthoff 2010 definition. Part of this definition relates to peak transaminases (at any time within the first week), which are potentially flawed in machine perfusion studies given the preservation washout effects; trials did not provide sufficient data on which recipients were diagnosed with early allograft dysfunction based solely on peak transaminase levels to allow us to adjust for this. As a result, it is unclear whether reducing early allograft dysfunction in this setting represents a real difference in graft injury or confers no benefit to the transplant recipient.

HOPE reduced early allograft dysfunction compared with SCS (OR 0.35, 95% CI 0.23 to 0.53; P < 0.001; 4 trials, 482 recipients; Analysis 1.6; high‐certainty evidence). This effect was similar in DCD and extended criteria DBD subgroups (Analysis 1.6). Sensitivity analyses using a fixed‐effect model, and removing studies with for‐profit funding yielded similar results.

1.6. Analysis.

Comparison 1: Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS), Outcome 6: Early allograft dysfunction

NMP probably reduces early allograft dysfunction compared with SCS (OR 0.40, 95% CI 0.22 to 0.74; P = 0.23, I² = 33%; 3 trials, 540 recipients; Analysis 2.4; moderate‐certainty evidence due to risk of bias). Sensitivity analyses using a fixed‐effect model, and removing Markmann 2022 (for‐profit funding and high risk of bias), did not change these conclusions.

2.4. Analysis.

Comparison 2: Normothermic machine perfusion (NMP) versus static cold storage (SCS), Outcome 4: Early allograft dysfunction

An indirect comparison of HOPE versus NMP using the results of Analysis 1.6 and Analysis 2.4 was performed with the techniques of Bucher 1997; there was no evidence of a difference between HOPE and NMP on early allograft dysfunction (indirect OR 0.88, 95% CI 0.42 to 1.83; P = 0.74; low‐certainty evidence due to the NMP versus SCS comparison).

Adverse events not considered serious

Two trials reported the number of participants experiencing any adverse event (Czigany 2021; Nasralla 2018). Czigany 2021 suggested no evidence of a difference in proportion of recipients with any adverse event between HOPE and SCS groups (OR 0.18, 95% CI 0.01 to 4.03; P = 0.28; 1 trial, 6 recipients; low‐certainty evidence due to risk of bias and imprecision). This result is imprecise (wide CIs) as almost all participants in each group experienced at least one adverse event. Czigany 2021 also reported on adverse events using the Comprehensive Complication Index, a validated metric to summarise adverse events in surgical trials (Slankamenac 2013). They found a reduction in the Comprehensive Complication Index with HOPE compared with SCS at both 90 days (median: 32 in HOPE group versus 52 in SCS group; P = 0.02) and six months (median: 35 in HOPE group versus 56 in SCS group; P = 0.03) (moderate‐certainty evidence).

Nasralla 2018 demonstrated that NMP did not reduce the proportion of participants experiencing any adverse event compared with SCS (OR 0.92, 95% CI 0.54 to 1.57; P = 0.76; 1 trial, 222 recipients; moderate‐certainty evidence due to risk of bias).

Three trials reported the total number of adverse events between the two groups (Ravaioli 2022; Schlegel 2023; van Rijn 2021). van Rijn 2021 reported 644 adverse events across 78 recipients in the HOPE group versus 694 adverse events across 78 recipients in the SCS group (no statistical analysis; overall some concerns of risk of bias). Ravaioli 2022 reported 143 adverse events across 55 recipients in the HOPE group versus 190 adverse events across 55 recipients in the SCS group (no statistical analysis; overall high risk of bias). Schlegel 2023 reported 616 adverse events across 85 recipients in the HOPE group versus 574 adverse events across 85 recipients in the SCS group. Two trials did not clearly state either the total number of adverse events or the number of participants experiencing any adverse event (Ghinolfi 2019; Markmann 2022).

Overall, there are no significant safety concerns for either HOPE or NMP based on these results.

Transplant utilisation

There were no cluster randomised trials that could be used to assess transplant utilisation. Czigany 2021, Ghinolfi 2019, Ravaioli 2022, Schlegel 2023, and van Rijn 2021 all performed the randomisation process after the final decision to accept a liver, and therefore all, or nearly all, randomised livers were transplanted.

In contrast, for Markmann 2022 and Nasralla 2018, the randomisation and allocation process occurred before the final decision to transplant the liver. These trials assessed utilisation by dividing the number of livers transplanted in each group by the number of recipients (Markmann 2022) or livers allocated to that group (Nasralla 2018).

In Nasralla 2018 where livers were the unit of randomisation, 121/137 livers undergoing NMP were transplanted and 101/133 livers undergoing SCS were transplanted; a 50% lower rate of organ discard (P = 0.008). It is not clear whether this result represents the ability to prolong preservation time, the use of viability criteria, or a lack of equipoise amongst transplanting surgeons with a preheld belief that NMP was superior to SCS.

Markmann 2022 used recipients as the unit of randomisation. Allocation was revealed to the retrieval team and implanting surgeon before the liver had been visualised. In DBD grafts, Markmann 2022 reported 124 transplants from 154 livers in the NMP group versus 133 transplants from 168 livers in the SCS group (P = 0.762). In DCD grafts, they reported 28 transplants from 55 livers in the NMP group versus 13 transplants from 51 livers in the SCS group (P = 0.007). However, the only difference was the number of livers declined on visual inspection, whilst still in the donor (i.e. prior to being placed on the NMP device or undergoing SCS). Therefore, this effect can only represent a lack of equipoise rather than an effect specific to the machine — the implanting team clearly believed in the superiority of NMP over SCS, or at least its ability to assess viability. Transplant utilisation was not a prespecified outcome, and the trial was not designed to assess utilisation.

It is worth noting that the improved utilisation seen in Markmann 2022 and Nasralla 2018 was not associated with inferior outcomes in the NMP group. However, it is unknown whether similar outcomes would have been achieved if the discarded livers in the SCS group were transplanted.

Transaminase release during the first week post‐transplant

Again, it must be highlighted that whilst machine perfusion interventions appear to reduce peak serum transaminases post‐transplant when compared with SCS this may simply reflect a flush‐out effect during machine perfusion. Even if flush‐out does not represent the entire effect, there are no definitive data to show that reducing post‐transplant transaminase release would lead to improved clinically relevant outcomes. Therefore, we discuss these results only narratively below.

Three trials which compared HOPE versus SCS reported on this outcome (Czigany 2021; Schlegel 2023; van Rijn 2021). Czigany 2021 found that HOPE was associated with a significant reduction in peak alanine transaminase in the first week versus SCS (median 418 in the HOPE group versus 796 in the SCS group; P = 0.030). van Rijn 2021 reported mean peak one‐week alanine transaminase of 1026 in the HOPE group versus 1259 in the SCS group (they performed no statistical analysis on this comparison). Schlegel 2023 analysed area under curves for transaminase release in the first week and found no important differences between HOPE and SCS groups.

Two NMP versus SCS trials reported on peak serum transaminases post‐transplant (Ghinolfi 2019; Nasralla 2018). Nasralla 2018 found a reduction in peak aspartate transaminase in the first week post‐transplant in the NMP group compared with the SCS group (geometric mean: 485 in the NMP group versus 974 in the SCS group; P < 0.001). The pilot trial of Ghinolfi 2019 found no important differences in peak transaminases in the first week, but was not powered for this. Markmann 2022 and Ravaioli 2022 did not report on transaminase release post‐transplant, except indirectly via early allograft dysfunction as reported above.

Discussion

Summary of main results

We identified four randomised clinical trials which compared HOPE with SCS (Czigany 2021; Ravaioli 2022; Schlegel 2023; van Rijn 2021), and three randomised trials comparing NMP with SCS (Markmann 2022; Nasralla 2018; van Rijn 2021).

Trials comparing HOPE with SCS were at low risk of bias across almost all outcomes and showed convincing improvements in outcomes which are relevant to clinicians, patients, and healthcare providers. When compared with conventional SCS, HOPE improved graft survival (HR 0.45, 95% CI 0.23 to 0.87; P = 0.02, I² = 0%; 4 trials, 482 recipients; Analysis 1.3; high‐certainty evidence), reduced serious adverse events in extended criteria DBD liver transplant (OR 0.45, 95% CI 0.22 to 0.91; P = 0.03, I² = 0%; 2 trials, 156 recipients; Analysis 1.2; moderate‐certainty evidence), and reduced clinically significant ischaemic cholangiopathy in recipients of DCD livers (OR 0.31, 95% CI 0.11 to 0.92; P = 0.03; 1 trial, 156 recipients; Analysis 1.4; high‐certainty evidence).

In contrast, trials comparing NMP with SCS found no improvement in any of these three important outcomes. There is evidence that NMP is able to reduce the discard rate of livers (Nasralla 2018 showed a 50% lower rate of organ discard). However, no study to date has been primarily designed to assess this. Whether this reduced discard relates to the capacity for increased preservation time, the ability to assess viability, or a preheld belief by surgeons that NMP is superior is unclear. It is also uncertain whether this effect is specific to NMP, or whether it could be achieved with other machine perfusion technologies.

It is uncertain whether overall participant survival is impacted by either HOPE (HR 0.91, 95% CI 0.42 to 1.98; P = 0.81, I² = 0%; 4 trials, 482 recipients; Analysis 1.1; low‐certainty evidence due to imprecision because of low number of events) or NMP (HR 1.08, 95% CI 0.31 to 3.80; P = 0.90; 1 trial, 222 recipients; Analysis 2.1; very low‐certainty evidence due to imprecision and risk of bias) when compared with SCS. It is worth noting that power to detect differences in overall participant survival are low, due to the low number of deaths. We consider that the improvements in graft survival discussed above would likely lead to improvements in overall participant survival in trials with a larger sample size. No trials reported on quality of life.

We identified no trials randomising recipients to normothermic regional perfusion. In the setting of DCD transplantation, several previous cohort studies have suggested improved liver graft survival and a reduction in ischaemic cholangiopathy when compared to standard organ procurement and SCS (De Beule 2021; Oniscu 2023). This has seen increasing use of normothermic regional perfusion worldwide, with some arguing that the positive results from cohort studies would make a randomised trial unethical (De Beule 2021; Hessheimer 2019; Oniscu 2023; Savier 2020). Due to a lack of randomised clinical trials, how HOPE and NMP compare with normothermic regional perfusion is uncertain. It is also unclear whether the benefits seen with HOPE following standard DCD retrieval (shown in this review) would extend to HOPE following normothermic regional perfusion, or whether one of these techniques is sufficient in isolation.

Another interesting development has been the combination of ex situ techniques; HOPE has been used to resuscitate/optimise livers, prior to a period of NMP for viability assessment (van Leeuwen 2021). In this setting, the liver is often slowly rewarmed from hypothermia to normothermia (termed 'controlled oxygenated rewarming'), a technique which has been shown to salvage livers otherwise deemed untransplantable, with excellent transplant outcomes in a single armed prospective clinical trial (van Leeuwen 2019). These techniques are yet to be studied in randomised clinical trials, although selecting a control arm in such a trial would be challenging as the technique is primarily used for livers which have otherwise been declined for transplant.

We have attempted to focus on outcomes that are the most relevant and important for patients, clinicians, and healthcare organisations. We have not included outcomes such as post‐reperfusion syndrome (a transient phenomenon) or length of hospital stay, following advice from the Cochrane Hepato‐Biliary editorial team. In addition, we have avoided making conclusions based solely on recipient peak transaminases. There is a lack of high‐quality evidence that lower transaminases are a good surrogate for clinically relevant outcomes, and they may simply represent a flush‐out effect during machine perfusion. Although this flush‐out effect is likely to be greater in NMP, where the organ is metabolically active, some transaminases are likely to be flushed out during hypothermic perfusion.

We have not considered cost‐effectiveness, although hypothermic techniques are generally less expensive due to their inherent simplicity, both of the circuit and perfusate constituents/infusions (Zimmermann 2022). In addition, hypothermic techniques do not generally require blood products, which removes pressure from blood donation services, and avoids logistical issues.

The searches were updated by the Cochrane Hepato‐Biliary Editorial team after the peer review process. This identified a pilot study of SCS versus controlled oxygenated rewarming (Minor 2022). As this was a pilot study and did not include a comparison which would contribute to any of the meta‐analyses, this study has been listed as a 'study awaiting classification' following advice from the Cochrane Hepato‐Biliary Editorial team. This will be assessed/included in the update of the review.

Overall completeness and applicability of evidence

We employed robust search strategies, which included screening of reference lists and searching of clinical trial registries, to ensure we identified all relevant trials.

Trials of HOPE versus SCS included both DBD and DCD grafts, which are currently deemed transplantable using SCS. NMP versus SCS trials contained standard and extended criteria DBD livers, as well as DCD grafts. However, the number of transplanted DCD grafts in the NMP trials was low, limiting the applicability of these trials to DCD liver transplant. Whether HOPE or NMP can salvage DCD livers which are considered too high‐risk to be transplanted following SCS remains unknown.

Due to the relatively limited number of present trials it is difficult to perform meaningful subgroup analyses. For example, the optimal method of performing HOPE is still not known, although it seems that both dual (hepatic artery and portal vein) and portal vein only HOPE are effective. It is also unclear what the best method of preservation is for other marginal graft types such as steatotic grafts, but trials are underway to assess this (NCT03930459).

With regard to DCD grafts, none of these trials used normothermic regional perfusion which is becoming standard, or even mandated, in many countries (Hessheimer 2019; Oniscu 2023; Savier 2020). The effect of these interventions, when delivered following normothermic regional perfusion, is not known. This is clearly not a concern when considering the use of machine perfusion in DBD grafts.

Quality of the evidence

The four trials comparing HOPE with SCS were at low risk of bias using RoB 2 across all domains for the majority of assessed outcomes (Czigany 2021; Ravaioli 2022; Schlegel 2023; van Rijn 2021). These were well‐designed trials where the majority of randomised livers underwent their allocated intervention and were transplanted, and were at low risk of bias from for‐profit funding.

Risk of bias in trials comparing NMP with SCS was mixed. In Markmann 2022 and Nasralla 2018, liver grafts allocated to NMP were more likely to be transplanted that those allocated to SCS; therefore, the baseline quality (prior to intervention) of grafts transplanted following NMP in these trials may be lower than those transplanted following SCS, which may underestimate any benefits of NMP when examining recipient outcomes. This may contribute to the lack of benefits seen with NMP in clinically relevant recipient outcomes. Nasralla 2018 was otherwise at low risk of bias for all other domains and received not‐for‐profit funding. Markmann 2022 received for‐profit funding and was deemed high risk for many RoB 2 domains for issues such as the use of non‐standard outcome definitions which differed from their prespecified outcomes. The pilot trial by Ghinolfi 2019 was generally at low risk of bias for assessed outcomes.

Unfortunately, there were no trials directly comparing NMP and HOPE. For some outcomes it was possible to perform an indirect comparison between NMP and HOPE using techniques described by Bucher 1997. As no trials directly compared these interventions, heterogeneity could not be assessed for the HOPE versus NMP comparison.

Potential biases in the review process

The search for randomised clinical trials was performed using systematic methods following consultation with the Cochrane Hepato‐Biliary information specialist. Two review authors independently screened the identified studies prior to inclusion in the review. Two review authors independently used a standardised data extraction form to collect data from included trials. Subgroup analysis was only performed if prespecified in our protocol, to limit bias from multiple comparisons. Due to the limited number of trials it is difficult to assess publication bias, which may introduce bias into our results.

Agreements and disagreements with other studies or reviews

There have been several previous narrative reviews on the topic of machine perfusion in liver transplantation (Czigany 2019; Schlegel 2019). In this section we focus on comparisons with previous systematic reviews and meta‐analyses only, specifically those focused on NMP, HOPE, or both.

Previous meta‐analyses included both randomised clinical trials and retrospective cohort studies (Bellini 2019; Boteon 2018; Jakubauskas 2022; Jia 2020; Liew 2021; Zhang 2019), whereas we have focussed only on randomised clinical trials due to their superior level of evidence. All of these meta‐analyses found that HOPE reduces ischaemic biliary complications, and none of these meta‐analyses found a significant impact of NMP on ischaemic biliary complications (Bellini 2019; Boteon 2018; Jakubauskas 2022; Jia 2020; Liew 2021; Zhang 2019). Our review is concurrent with these findings, but as it only includes randomised clinical trials the evidence is at lower risk of bias.

Previous analyses have been mixed regarding the impact of HOPE on graft survival, with some showing no impact and others showing significant improvements compared with SCS (Bellini 2019; Jia 2020; Liew 2021; Zhang 2019). However, no previous meta‐analysis analysed graft survival as a time‐to‐event variable, instead dichotomising graft survival at one year. No meta‐analysis showed any improvement in graft survival with NMP (Bellini 2019; Boteon 2018; Jakubauskas 2022; Jia 2020; Liew 2021; Zhang 2019). We appropriately analysed graft survival as time‐to‐event data, and use data from randomised trials, to show an improvement in graft survival with HOPE versus SCS; a benefit which is not replicated with NMP.

Authors' conclusions

Implications for practice.

In situations where the decision has been made to transplant a liver donated following circulatory death (DCD) or an extended criteria liver donated following brainstem death (DBD) liver, end‐ischaemic hypothermic oxygenated machine perfusion (HOPE) will provide superior clinically relevant outcomes compared to static cold storage (SCS). Specifically, graft survival is improved (high‐certainty evidence), serious adverse events are reduced (moderate‐certainty evidence), and in DCD donors clinically relevant ischaemic biliary complications are reduced (high‐certainty evidence).

There is no good evidence that NMP has the same benefits over SCS in terms of these clinically relevant outcomes. NMP appears to improve utilisation of grafts that would not otherwise be discarded with SCS, however, the reasons for this are not clear. There are no significant safety concerns relating to either HOPE or NMP.

Implications for research.

Trials in donation following brainstem death (DBD) organ transplant are needed to directly compare various ex situ machine perfusion techniques, including novel techniques such as controlled oxygenated rewarming. A direct comparison of HOPE versus NMP is warranted, in addition to further randomised trials investigating alternative machine perfusion technologies, in isolation or in combination; such trials are ongoing (NCT04644744; NCT04744389). In future machine perfusion studies of recipients of livers from extended criteria DBD donors end‐ischaemic HOPE should be used as the current gold standard 'control' (rather than SCS).

In the setting of donation following circulatory death (DCD) transplantation, further trials are needed to assess the effect of ex situ machine perfusion devices against, or in combination with, normothermic regional perfusion. Randomised clinical trials are also needed to address remaining questions related to how NMP may improve graft utilisation, whether this effect is unique to NMP, or may be seen with other types of machine perfusion. In addition, further work refining, optimising, and validating NMP viability criteria is needed if they are to become widely adopted into clinical practice.

History

Protocol first published: Issue 7, 2021

Risk of bias

Risk of bias for analysis 1.1 Overall participant survival.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.1.1 Donated following brain death

Ravaioli 2022

Low risk of bias

Electronic randomisation through Medidata Balance. Concealed until liver accepted. Well matched groups.

Low risk of bias

All participants received the intervention they were randomised to. Does not specify intention to treat, but as all livers randomised to each group received that intervention the ITT analysis would be identical to per‐protocol analysis.

Low risk of bias

Data available for all/nearly all participants.

Low risk of bias

Hard outcome of patient survival.

Low risk of bias

Protocol available.

Low risk of bias

Low risk of bias in every domain.

Czigany 2021

Low risk of bias

Centralised electronic randomisation. Well matched groups.

Low risk of bias

Open label trial due to nature of the intervention, but no bias arising from deviations from intended interventions. Intention to treat.

Low risk of bias

Data available for all/nearly all participants.

Low risk of bias

Hard outcome of patient survival. Protocol explains outcome assessors blinded.

Low risk of bias

Pre‐published protocol.

Low risk of bias

Low risk of bias in every domain.

Schlegel 2023

Low risk of bias

Clear description of randomisation process, concealment of block size and allocation concealment. There were baseline differences in the groups, but this was likely due to random chance and the relatively small sample size, rather than issues in the randomisation process.

Low risk of bias

Participants and treating physicians were blinded. Surgeons performing machine perfusion clearly not blinded. No deviations to treatments because of trial context. Intention to treat analysis used.

Low risk of bias

Complete outcome data with no loss to follow‐up.

Low risk of bias

Hard outcome of participant survival.

Low risk of bias

This outcome was pre‐specified in the pre‐published protocol.

Low risk of bias

High quality randomised trial.

Subgroup 1.1.2 Donated following circulatory death

van Rijn 2021

Low risk of bias

Computer generated centralised randomisation. Well matched for all key factors.

Low risk of bias

Patients and procurement teams were blinded. Implanting teams could not be blinded. Intention to treat analysis applied.

Low risk of bias

Data available for all/nearly all participants ‐ full description given in Figure S1 of their manuscript.

Low risk of bias

Hard outcome of patient death.

Low risk of bias

Prepublished protocol specified outcome.

Low risk of bias

Low risk of bias in every domain.

Risk of bias for analysis 1.2 Serious adverse events (90‐day follow‐up).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Czigany 2021

Low risk of bias

Centralised electronic randomisation. Well matched groups.

Low risk of bias

Open label trial due to nature of the intervention, but no bias arising from deviations from intended interventions. Intention to treat.

Low risk of bias

Data available on all/nearly all participants.

Low risk of bias

Pre‐specified Clavien‐Dindo with definitions. Protocol explains outcome assessors blinded.

Low risk of bias

Prespecified. Used the well validated Clavien Dindo greater than or equal to 3 with appropriate analysis.

Low risk of bias

Low risk of bias in every domain.

Ravaioli 2022

Low risk of bias

Electronic randomisation through Medidata Balance. Concealed until liver accepted. Well matched groups.

Low risk of bias

All participants received the intervention they were randomised to. Does not specify intention to treat, but as all livers randomised to each group received that intervention the ITT analysis would be identical to per‐protocol analysis.

Low risk of bias

Outcome available for all/nearly all participants.

Low risk of bias

Confirmed with study authors that values in Table 2 correspond to Clavien Dindo graded complications.

Low risk of bias

Protocol available. Used Clavien Dindo classification.

Low risk of bias

Low risk of bias in every domain.

Risk of bias for analysis 1.3 Graft survival.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.3.1 Donated following brain death

Czigany 2021

Low risk of bias

Centralised electronic randomisation. Well matched groups.

Low risk of bias

Open label trial due to nature of the intervention. No significant deviations from intended interventions. Intention‐to‐treat.

Low risk of bias

Full follow‐up to 1 year for most participants. 4 participants which were followed up to 6 months were appropriately censored.

Low risk of bias

Clear definitions of graft survival. Protocol explains outcome assessors blinded.

Low risk of bias

Prepublished protocol. Hard outcome of graft survival with appropriate definition and analysis.

Low risk of bias

Low risk of bias in every domain.

Ravaioli 2022

Low risk of bias

Electronic randomisation through Medidata Balance. Concealed until liver accepted. Well matched groups.

Low risk of bias

Open label due to nature of trial. All participants received the intervention they were randomised to. Does not specify intention to treat, but as all livers randomised to each group received that intervention the ITT analysis would be identical to per‐protocol analysis.

Low risk of bias

Outcome data available for all/nearly all participants.

Low risk of bias

Hard outcome of graft survival with definition.

Low risk of bias

Protocol available. Pre‐specified analysing time‐to‐event with Kaplan‐Meier estimates.

Low risk of bias

Low risk of bias in every domain.

Schlegel 2023

Low risk of bias

Clear description of randomisation process, concealment of block size and allocation concealment. There were baseline differences in the groups, but this was likely due to random chance and the relatively small sample size, rather than issues in the randomisation process.

Low risk of bias

No deviations to preservation technique because of trial context. Treating physicians blinded. Intention‐to‐treat analysis used.

Low risk of bias

Complete outcome data with no loss to follow‐up.

Low risk of bias

Clear objective definition of graft loss.

Low risk of bias

This outcome was prespecified in the prepublished protocol.

Low risk of bias

We have included the prespecified overall graft loss in these graft survival analyses, rather than the post‐hoc 'liver‐specific' graft loss.

Subgroup 1.3.2 Donated following circulatory death

van Rijn 2021

Low risk of bias

Computer generated centralised randomisation. Well matched for all key factors.

Low risk of bias

Patients and procurement teams blinded. Implanting team blinding impossible. No significant deviations from intended interventions. Intention to treat.

Low risk of bias

Outcome data available for all/nearly all participants.

Low risk of bias

Death cencored graft survival with clear definition.

Low risk of bias

Prepublished protocol stating graft survival and definitions.

Low risk of bias

Low risk of bias in every domain.

Risk of bias for analysis 1.4 Ischaemic biliary complications.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Subgroup 1.4.1 Donated following brain death

Schlegel 2023

Low risk of bias

Clear description of randomisation process, concealment of block size and allocation concealment. There were baseline differences in the groups, but this was likely due to random chance and the relatively small sample size, rather than issues in the randomisation process.

Low risk of bias

No deviations to preservation technique because of trial context. Treating physicians blinded and intention‐to‐treat analysis used.

Low risk of bias

Complete outcome data with no loss to follow‐up.

Low risk of bias

Non‐anastomotic strictures at 12 months' follow‐up. Radiologists assessing MRCP were blinded to treatment group.

Low risk of bias

This outcome was prespecified in the prepublished protocol.

Low risk of bias

High quality randomised trial.

Subgroup 1.4.2 Donated following circulatory death

van Rijn 2021

Low risk of bias

Computer generated centralised randomisation. Well matched for all key factors.

Low risk of bias

Participants and procurement teams blinded. Implanting team blinding impossible. No significant deviations from intended intervention. Intention to treat.

Low risk of bias

Data available for all/nearly all participants.

Low risk of bias

Ischaemic cholangiopathy with clear definition. Review of scans by 2 blinded radiologists. Central blinded review of all clinical cases.

Low risk of bias

Prepublished protocol stating ischaemic cholangiopathy and definitions.

Low risk of bias

Clinically significant ischaemic cholangiopathy

Risk of bias for analysis 1.7 Adverse events considered non‐serious.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Czigany 2021

Low risk of bias

Centralised electronic randomisation. Well matched groups.

Low risk of bias

Open label trial due to nature of the intervention. No significant deviations. Intention to treat used.

Low risk of bias

Data available for all participants

Some concerns

Although protocol states that "Analysis of primary endpoints will be performed by an independent committee in a blinded fashion", it is likely that those actually recording all adverse events were aware of the treatment group. Knowledge of assignment to treatment group may have altered the decision to report an event as an adverse event.

Low risk of bias

Pre‐published protocol.

Some concerns

Although protocol states that "Analysis of primary endpoints will be performed by an independent committee in a blinded fashion", it is likely that those actually recording all adverse events were aware of the treatment group. Knowledge of assignment to treatment group may have altered the decision to report an event as an adverse event.

Risk of bias for analysis 2.1 Overall participant survival.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Nasralla 2018

Low risk of bias

Centralised electronic randomisation. Similar donor characteristics between randomised livers.

Low risk of bias

Open‐label trial due to nature of interventions, but no significant deviations from intended intervention. Intention to treat performed.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of patients were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Low risk of bias

Hard outcome of patient survival.

Low risk of bias

Pre‐specified Kaplan Meier estimates of 1‐year patient survival.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of patients were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Risk of bias for analysis 2.2 Graft survival.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Nasralla 2018

Low risk of bias

Centralised electronic randomisation. Similar donor characteristics between randomised livers.

Low risk of bias

Open‐label trial due to nature of interventions. No significant deviation from interventions. Intention to treat performed.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of patients were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Low risk of bias

Hard outcome of graft survival with definition.

Low risk of bias

Pre‐specified Kaplan Meier estimates of 1‐year graft survival.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of patients were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Markmann 2022

Some concerns

Recipients were randomised. However, at the time of assessment of the liver in the donor, the implanting surgeon was aware of the assigned group and could decline the liver for transplantation at that stage. When being assessed in the donor (before retrieval of the liver), the implanting surgeon was significantly more likely to accept a DCD liver for a recipient who was randomised to NMP. Therefore, the baseline risk profile of donors was different, prior to initiation of NMP or SCS (significantly more DCD in the NMP group).

High risk of bias

When being assessed in the donor (before retrieval of the liver), the implanting surgeon was aware of the intended intervention. They were significantly more likely to accept a DCD liver for a recipient who was randomised to machine perfusion. Therefore, many higher risk DCD livers which were allocated to recipients randomised to SCS were never transplanted. The only analysis of graft survival which was presented was a per‐protocol analysis.

High risk of bias

For 1 year graft survival (98% for NMP and 99% for SCS) the numbers at risk at 1 year (eFigure 2) are 104 and 105 in NMP and SCS respectively. It is not clear how patients were lost to follow‐up.

Some concerns

Graft survival is not defined in the protocol or manuscript. It is not clear whether this is death censored.

Some concerns

No pre‐specified plan for how graft survival would be analysed. eFigure 2 contains a Logrank P value.

High risk of bias

Recipients were randomised. However, at the time of assessment of the liver in the donor, the implanting surgeon was aware of the assigned group and could decline the liver for transplantation at that stage. When being assessed in the donor (before retrieval of the liver), the implanting surgeon was significantly more likely to accept a DCD liver for a recipient who was randomised to machine perfusion. Therefore, the baseline risk profile of donors was different, prior to initiation of NMP or SCS (significantly more DCD in the NMP group). The only analysis of graft survival which was presented was a per‐protocol analysis. Graft survival was not defined in the protocol or the manuscript, and it is unclear whether this was death censored.

Risk of bias for analysis 2.3 Ischaemic biliary complications (on protocol imaging).

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Nasralla 2018

Low risk of bias

Centralised electronic randomisation. Similar donor characteristics between randomised livers.

Low risk of bias

Open‐label trial due to nature of interventions. Minimal deviation from planned interventions. Intention to treat performed.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of participants were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Low risk of bias

Prespecified outcome of NAS on MRCP (> 70% of luminal diameter). Protocol MRCP. 2 independent blinded radiologists graded the MRCP.

Low risk of bias

Prespecified plan in the protocol.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of patients were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Risk of bias for analysis 2.5 Adverse events considered non‐serious.

Study

Bias

Randomisation process

Deviations from intended interventions

Missing outcome data

Measurement of the outcome

Selection of the reported results

Overall

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Authors' judgement

Support for judgement

Nasralla 2018

Low risk of bias

Centralised electronic randomisation. Similar donor characteristics between randomised livers.

Low risk of bias

Open‐label trial due to nature of interventions. No significant deviation from intended interventions. Intention to treat performed.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of patients were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Low risk of bias

2 independent blinded clinicians assessed adverse events.

Low risk of bias

Pre‐specified analysis plan in the protocol.

Some concerns

Unit of randomisation was the liver, but recipient outcomes are only available for transplanted grafts. A significantly higher proportion of patients were transplanted in the NMP group. It is unclear whether the extra grafts which were transplanted in the NMP group were lower quality grafts. This may mean grafts transplanted in the SCS were of a higher quality at baseline and therefore cause bias in favour of SCS for recipient outcomes.

Appendices

Appendix 1. Search strategies

Database	Time span	Search strategy
Cochrane Hepato‐Biliary Group Controlled Trials Register (via the Cochrane Register of Studies Web)	10 January 2023	(((organ* or machine* or regional*) and (perfusion* or preservation*)) or oxygenated rewarming or static cold storage) and ((liver or hepat*) and (transplant* or graft*))
Cochrane Central Register of Controlled Trials in the Cochrane Library	2022, Issue 5	#1 MeSH descriptor: [Perfusion] explode all trees #2 MeSH descriptor: [Organ Preservation] explode all trees #3 (((organ* or machine* or regional*) near (perfusion* or preservation*)) or oxygenated rewarming or static cold storage) #4 #1 or #2 or #3 #5 MeSH descriptor: [Liver Transplantation] explode all trees #6 ((liver or hepat*) and (transplant* or graft*)) #7 #5 or #6 #8 #4 and #7
MEDLINE Ovid	1946 to 10 January 2023	1. exp Perfusion/ 2. exp Organ Preservation/ 3. (((organ* or machine* or regional*) adj (perfusion* or preservation*)) or oxygenated rewarming or static cold storage).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 4. 1 or 2 or 3 5. exp Liver Transplantation/ 6. ((liver or hepat*) and (transplant* or graft*)).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 7. 5 or 6 8. 4 and 7 9. (randomized controlled trial or controlled clinical trial or retracted publication or retraction of publication).pt. or clinical trials as topic.sh. or trial.ti. 10. (random* or blind* or placebo* or meta‐analys*).mp. [mp=title, abstract, original title, name of substance word, subject heading word, floating sub‐heading word, keyword heading word, organism supplementary concept word, protocol supplementary concept word, rare disease supplementary concept word, unique identifier, synonyms] 11. 8 and (9 or 10)
Embase Ovid	1974 to 10 January 2023	1. exp perfusion/ 2. exp organ preservation/ 3. (((organ* or machine* or regional*) adj (perfusion* or preservation*)) or oxygenated rewarming or static cold storage).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 4. 1 or 2 or 3 5. exp liver transplantation/ 6. ((liver or hepat*) and (transplant* or graft*)).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 7. 5 or 6 8. 4 and 7 9. Randomized controlled trial/ or Controlled clinical trial/ or retracted article/ or (erratum or tombstone).pt. or trial.ti. or yes.ne. 10. (random* or blind* or placebo* or meta‐analys*).mp. [mp=title, abstract, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword heading word, floating subheading word, candidate term word] 11. 8 and (9 or 10)
LILACS BIREME	1982 to 10 January 2023	(((organ$ or machine$ or regional$) and (perfusion$ or preservation$)) or oxygenated rewarming or static cold storage) [Words] and ((liver or hepat$) and (transplant$ or graft$))
Science Citation Index Expanded (Web of Science)	1900 to 10 January 2023	#5 #4 AND #3 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=((liver or hepat*) and (transplant* or graft*)) #1 TS=(((organ* or machine* or regional*) near (perfusion* or preservation*)) or oxygenated rewarming or static cold storage)
Conference Proceedings Citation Index – Science (Web of Science)	1990 to 10 January 2023	#5 #4 AND #3 #4 TI=(random* or blind* or placebo* or meta‐analys* or trial*) OR TS=(random* or blind* or placebo* or meta‐analys*) #3 #2 AND #1 #2 TS=((liver or hepat*) and (transplant* or graft*)) #1 TS=(((organ* or machine* or regional*) near (perfusion* or preservation*)) or oxygenated rewarming or static cold storage)

Appendix 2. Proposed methods for network meta‐analysis

The following methodology explains how we plan to perform network meta‐analyses in future versions of this review, if sufficient trials are available.

Types of outcome measures

Outcomes for ranking in network meta‐analysis

We plan to rank the interventions based on their effects on our primary outcomes plus the outcomes: graft survival, ischaemic‐type biliary complications, and adverse events considered non‐serious.

Assessment of heterogeneity

We will generate tables that will allow us to evaluate the distribution of the effect modifiers, listed above, across different comparisons, and therefore, assess the validity of the transitivity assumption for indirect comparisons (see Data synthesis below). For information on how we will statistically analyse whether the transitivity assumption has been violated, see 'Inconsistency in network meta‐analysis', below (Data synthesis).

Assessment of reporting biases

For network meta‐analysis, we will construct comparison‐adjusted funnel plots (Ratnayake 2019).

Data synthesis

Network meta‐analysis

We will perform network meta‐analyses in R using the GeMTC package (R 2021; Valkenhoef 2016). This interface is powered by the GeMTC R package, which performs network meta‐analysis using a Bayesian hierarchical model (Valkenhoef 2016). We will use Markov Chain Monte Carlo (MCMC) methods to sample the posterior distribution. A key benefit of the GeMTC package is the ability to automate Bayesian hierarchical model generation, which includes the use of the supplied dataset to heuristically set priors that limit bias, and starting values that limit the chance of misdiagnosing convergence (van Valkenhoef 2012).

Priors will be vague (non‐informative) and will be generated heuristically, based on the data provided by the included clinical trials. These priors will be sufficiently vague (large variance), such that posteriors (model results) are dominated by data provided by included trials, rather than the chosen prior distributions.

For each model, we will assess convergence visually. By using the Brooks–Gelman–Rubin diagnostic, we will run the model several times in parallel, with different starting values (overdispersed starting values are automatically generated, based on the dataset, such that the parameter space is sufficiently explored); each of these chains will then be compared, generating the potential scale reduction factor (PSRF). We will use a cut‐off PSRF < 1.05, along with visual assessments of PSRF plots and time series plots, to represent acceptable convergence. By default, GeMTC sets burn in (the number of initial MCMC iterations that are discarded) at 5000, and the inference iterations (the following MCMC iterations that are actually used to draw inferences on the posterior distributions) at 20,000, We will adjust these default values as necessary, based on PSRF values, PSRF plots, and time series plots.

Once we have ensured convergence, we will assess how well the model fits the trial data. The key measure of model fit, which we will use, is the residual deviance. We will analyse both overall model fit and per‐treatment arm residual deviance, with values close to 1 representing good model fit.

Further technical details regarding the GeMTC package are available in a peer‐reviewed manuscript (van Valkenhoef 2012); further details of the GeMTC user interface are available in the user manual (Valkenhoef 2016).

We will use an intention‐to‐treat principle, and we will derive mean estimates where necessary. We will generate network maps (diagrams) for each outcome for which network meta‐analyses can be performed. We will generate forest plots to display the results of the network meta‐analyses as a series of pairwise comparisons. The network meta‐analyses output will comprise odd ratios for dichotomous data and mean differences for continuous data, accompanied by 95% credible intervals (CrI). We will then generate ranking summary tables and rankograms with 95% CrI, which will provide the probability of each intervention being ranked in a certain position (first, second, etc.) for each outcome.

Transitivity is a key assumption of network meta‐analysis. This is the assumption that effect modifiers (such as donor and recipient characteristics) are similar in all included trials (Salanti 2012). This means that any participant could theoretically have been randomised to any of the treatment options (termed joint randomisability). Transitivity also requires that treatments grouped into a single node are sufficiently similar. As described above, we will consider the following potential effect modifiers when evaluating transitivity: DCD versus DBD, expanded criteria versus standard criteria donors, steatotic grafts, age of donors, duration of cold ischaemic time or perfusion time, type of perfusion fluid used, and recipient factors (age, indication, retransplants). To achieve a connected network, we foresee having to group the static cold storage arms of multiple trials into a single node, even though different cold storage solutions may be used.

Incoherence in network meta‐analysis

Measures of incoherence (sometimes referred to as inconsistency) allow statistical testing to determine whether the transitivity assumption has been violated. The null hypothesis for these analyses is that the network displays transitivity. It may not be possible to perform statistical tests of incoherence, as we predict all trials will have two arms and use static cold storage as their control arm. Therefore, networks will not have any complete loops (a star network), and we will not be able to compare treatment effects calculated from different loops.

If we are able to perform incoherence testing, we will test for global incoherence in the entire network, as well as analysing local incoherence, by comparing indirect with direct effect estimates (also referred to as the node‐splitting method). We will follow methodology described by Dias and colleagues, and outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Dias 2010; Higgins 2021a).

Sensitivity analysis

Sensitivity analyses in the network meta‐analysis

For the network meta‐analyses, we will undertake a sensitivity analysis by sequentially removing single trials to review the resulting discrepancies in the ranking data.

Summary of findings and assessment of the certainty of evidence

Network meta‐analysis

We will assess the confidence in outputs from the network meta‐analyses (certainty of evidence) using the CINeMA (Confidence In Network Meta‐Analysis) online tool (Salanti 2014; CINeMA 2017). We will assess the confidence in each individual comparison by considering the following six domains: within‐study bias, indirectness, imprecision, heterogeneity, incoherence, and reporting bias. This will allow us to generate summary tables on the confidence in the results from each individual comparison in the network meta‐analyses. We will generate summary of findings tables for network meta‐analyses using the approach detailed by Yepes‐Nuñez 2019.

Appendix 3. Descriptions of the bias domains in RoB 2 tool for randomised trials with a summary of the issues addressed

Bias domain	Issues addressed
Risk of bias arising from the randomisation process	Was the allocation sequence random? Was the allocation sequence concealed until participants were enrolled and assigned to interventions? Did baseline differences between intervention groups suggest a problem with the randomisation process?
Risk of bias due to deviation from the intended intervention (effect of assignment to intervention)	Whether: participants were aware of their assigned intervention during the trial; carers and people delivering the interventions were aware of participants' assigned intervention during the trial. When interest is in the effect of assignment to intervention: (if applicable) deviations from the intended intervention arose because of the experimental context, and if so, whether they were unbalanced between groups and likely to have affected the outcome; an appropriate analysis was used to estimate the effect of assignment to intervention.
Bias due to missing outcome data	Whether: data for this outcome were available for all, or nearly all, randomised participants; (if applicable) there was evidence that the result was not biased by missing outcome data; (if applicable) missingness in the outcome was likely dependent on its true value (e.g. proportion of missing outcome data, or reasons for missing outcome data, differ between intervention groups).
Bias in measurement of the outcome	Whether: the method of measuring the outcome was inappropriate; measurement or ascertainment of the outcome could have differed between intervention groups; outcome assessors were aware of the intervention received by study participants; assessment of the outcome was likely to have been influenced by knowledge of intervention received.
Bias in selection of the reported results	Whether: trial was analysed in accordance with a prespecified plan that was finalised before unblinded outcome data were available for analysis; the numerical result being assessed is likely to have been selected, on the basis of the results, from multiple outcome measurements (e.g. scales, definitions, time points) within the outcome domain; the numerical result being assessed is likely to have been selected, on the basis of the results, from multiple analyses of the data.

Data and analyses

Comparison 1. Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Overall participant survival	4		Hazard Ratio (IV, Random, 95% CI)	0.91 [0.42, 1.98]
1.1.1 Donated following brain death	3		Hazard Ratio (IV, Random, 95% CI)	0.69 [0.26, 1.83]
1.1.2 Donated following circulatory death	1		Hazard Ratio (IV, Random, 95% CI)	1.46 [0.41, 5.20]
1.2 Serious adverse events (90‐day follow‐up)	2	156	Odds Ratio (M‐H, Random, 95% CI)	0.45 [0.22, 0.91]
1.3 Graft survival	4		Hazard Ratio (IV, Random, 95% CI)	0.45 [0.23, 0.87]
1.3.1 Donated following brain death	3		Hazard Ratio (IV, Random, 95% CI)	0.39 [0.18, 0.85]
1.3.2 Donated following circulatory death	1		Hazard Ratio (IV, Random, 95% CI)	0.65 [0.18, 2.33]
1.4 Ischaemic biliary complications	2	326	Odds Ratio (M‐H, Random, 95% CI)	0.32 [0.12, 0.83]
1.4.1 Donated following brain death	1	170	Odds Ratio (M‐H, Random, 95% CI)	0.33 [0.03, 3.19]
1.4.2 Donated following circulatory death	1	156	Odds Ratio (M‐H, Random, 95% CI)	0.31 [0.11, 0.92]
1.5 Primary non‐function	4	482	Odds Ratio (M‐H, Random, 95% CI)	0.32 [0.07, 1.43]
1.5.1 Donated following brain death	3	326	Odds Ratio (M‐H, Random, 95% CI)	0.31 [0.06, 1.72]
1.5.2 Donated following circulatory death	1	156	Odds Ratio (M‐H, Random, 95% CI)	0.33 [0.01, 8.20]
1.6 Early allograft dysfunction	4	482	Odds Ratio (M‐H, Random, 95% CI)	0.35 [0.23, 0.53]
1.6.1 Donated following brain death	3	326	Odds Ratio (M‐H, Random, 95% CI)	0.27 [0.16, 0.46]
1.6.2 Donated following circulatory death	1	156	Odds Ratio (M‐H, Random, 95% CI)	0.52 [0.26, 1.03]
1.7 Adverse events considered non‐serious	1	46	Odds Ratio (M‐H, Random, 95% CI)	0.18 [0.01, 4.03]

1.7. Analysis.

Comparison 1: Hypothermic oxygenated machine perfusion (HOPE) versus static cold storage (SCS), Outcome 7: Adverse events considered non‐serious

Comparison 2. Normothermic machine perfusion (NMP) versus static cold storage (SCS).

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Overall participant survival	1		Hazard Ratio (IV, Random, 95% CI)	1.08 [0.31, 3.80]
2.2 Graft survival	2		Hazard Ratio (IV, Random, 95% CI)	1.20 [0.44, 3.29]
2.3 Ischaemic biliary complications (on protocol imaging)	1	155	Odds Ratio (M‐H, Random, 95% CI)	0.78 [0.27, 2.27]
2.4 Early allograft dysfunction	3	540	Odds Ratio (M‐H, Random, 95% CI)	0.40 [0.22, 0.74]
2.5 Adverse events considered non‐serious	1	222	Odds Ratio (M‐H, Random, 95% CI)	0.92 [0.54, 1.57]

2.5. Analysis.

Comparison 2: Normothermic machine perfusion (NMP) versus static cold storage (SCS), Outcome 5: Adverse events considered non‐serious

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Czigany 2021.

Study characteristics
Methods	Study design: open label multicentre RCT Country: Europe (4 centres in Germany and Czech Republic) Duration of follow‐up: 1 year
Participants	Number of participants: 46 (23 HOPE, 23 SCS) Median donor age: 73 years in HOPE group, 71 years in SCS group Median recipient age: 60 years in HOPE group, 63 years in SCS group Number of DCD: 0 Number of ECD: 46 (100%) Median preservation times Median warm ischaemic time: N/A Median cold ischaemic time: 375 in HOPE group, 503 in SCS group Median machine perfusion time: 145 (IQR 101–203) min (HOPE group) Median total preservation time: 495 min in HOPE group, 502 min in SCS group
Interventions	End‐ischaemic HOPE with the Liver Assist (Organ Assist). Portal vein only, and a minimum duration of 1 hour. Perfusate was 3–4 L Belzer UW machine‐perfusion solution (Bridge to Life) SCS with HTK according to local policy
Outcomes	Primary Peak ALT in the first week post‐transplant. Also attempted to allow for the wash‐out effect by normalising to the first post‐transplant ALT value (within 3 hours of transplant) Secondary Postoperative complications (Clavien‐Dindo classification) Comprehensive Complication Index Length of intensive care unit stay Length of hospital stay Incidence of early allograft dysfunction (Olthoff – day 7) 1‐year recipient and graft survival
Notes	Funding source: START‐Program (#136/17 to GL and #23/19 to ZC) and the Clinician Scientist Program (to ZC) of the Faculty of Medicine, RWTH Aachen University and by the Excellence Initiative of the German federal and state governments (G:(DE‐82) ZUK2‐SF‐OPSF443 to G.L.) PS was supported by the German Research Foundation grant STR1095/6–1 (Heisenberg professorship). Study author's key conclusion: HOPE improves early outcome in ECD‐DBD.

Ghinolfi 2019.

Study characteristics
Methods	Study design: pilot, open‐label RCT Country: Italy Duration of follow‐up: 6 months
Participants	Number of participants: 20 transplant recipients Median donor age: 81 years in NMP group, 80 years in SCS group Median recipient age: 57 years in NMP group, 55 years in SCS group Number of DCD: 0 Number of ECD: 20 (all donors aged over 70 years) Median preservation times Median warm ischaemic time: N/A Median cold ischaemic time: 280 min in NMP group, 394 min in SCS group Median machine perfusion time: 250 min (NMP group) Median total preservation time: 522 min in NMP group, 463 min in SCS group
Interventions	End ischaemic NMP with Liver Assist (Organ Assist) with blood‐based perfusate. Initial perfusion at 20 °C, then raised 1 °C every 2 min until 37 °C. This was done using a back‐to‐base method SCS using Celsior solution
Outcomes	Primary 6‐month graft and patient survival Secondary Peak transaminases in first week Biliary complications at 6 months – definitions given in the methods. Histopathological manifestations of ischaemia reperfusion injury Early allograft dysfunction Post‐reperfusion syndrome
Notes	Funding source: Tuscan Regional Transplant Authority Study author's key conclusion: suggests improved histology, larger studies are needed for outcomes.

Markmann 2022.

Study characteristics
Methods	Study design: multicentre RCT Country: US Duration of follow‐up: 24 months planned (first 12 months in the current study)
Participants	Number of participants: 300 livers transplanted (153 in OCS group, 147 in SCS group). 'Per‐protocol population' were 151 in OCS group and 142 in SCS group Participant details – used livers with moderate risk profile, and ≥ 1 of: DBD > 40 years Expected cold ischaemic time > 6 hours Steatotic livers (but macrosteatosis ≤ 40%) DCD (but aged ≤ 55 years) Median donor age: 47.5 years in NMP group, 45.8 years in SCS group Median recipient age: 59.2 years in NMP group, 61.4 years in SCS group Number of DCD: 28 in NMP group, 11 in SCS group Number of ECD: not defined Median preservation times Median warm ischaemic time: not reported Mean cold ischaemic time: 175.4 (SD 43.5) min in NMP group, 338.8 (SD 91.5) min in SCS group Mean machine perfusion time: 276.6 (SD 117.4) min (NMP group) Mean total preservation time: 454.9 (SD 133.9) min in NMP group, 338.8 (SD 91.5) min in SCS group
Interventions	NMP with Portable Organ Care System Liver device, initiated at the donor hospital. Oxygenated normothermic pulsatile flow in hepatic artery and low‐pressure flow in portal vein. Blood‐based perfusate. SCS solution used was not reported.
Outcomes	Primary Early allograft dysfunction Secondary Extent of reperfusion syndrome Incidence of ischaemic biliary at 6 and 12 months Overall recipient survival after transplant Graft survival Need for renal replacement therapy Intensive care unit and hospital stay Serious adverse events (definition given in their protocol)
Notes	Funding source: Transmedics – led trial design, and was responsible for data collection and generating the study report. Study author's key conclusion: OCS reduces early injury

Nasralla 2018.

Study characteristics
Methods	Study design: multicentre open‐label RCT Country: Europe (UK, Belgium, Spain, Germany) Duration of follow‐up: 1 year
Participants	Number of participants: 335 livers randomised. Included (after exclusions such as non‐proceeding DCDs) 137 in NMP group and 133 in SCS group. Transplanted 121 in NMP group and 101 in SCS group Median donor age: 56 years in NMP, 56 years in SCS group Median recipient age: 55 years in NMP group, 55 years in SCS groups Number of DCD: 63/170 randomised and 34/121 transplanted in NMP group, 60/164 randomised and 21/101 transplanted in SCS group Number of ECD: not specified/defined Median preservation times Median functional warm ischaemic time in DCD: 21 min in NMP group, 16 min in SCS group (P = 0.003) Median cold ischaemic time: 126 min in NMP group, 465 min in SCS group Median machine perfusion time: 547.5 min (NMP group) Median total preservation time: 714 min in NMP group, 465 min in SCS group
Interventions	NMP with OrganOx Metra. Minimum duration 4 hours and maximum duration 24 hours. 500 mL gelofusine (B. Braun Ltd) and 3 units of donor‐matched packed red blood cells. Antibiotics were given at the outset and heparin, insulin, prostacyclin, bile salts, and fat‐free parenteral nutrition were infused during the perfusion. SCS according to local practice. Protocol suggests UW or HTK; no data on number of recipients receiving each.
Outcomes	Primary Peak AST in the first week post‐transplant Secondary Organ discard rate Post‐reperfusion syndrome (> 30% drop in mean arterial pressure, persisting > 1 min, within 5 min of reperfusion) Primary non‐function Early allograft dysfunction (Olthoff, 7 days) Length of intensive care unit and hospital stay Need for renal replacement therapy Ischaemic cholangiopathy on 6‐month magnetic resonance cholangiopancreatography (2 blinded radiologists) Graft survival Participant survival
Notes	Funding source: European Commission Seventh Framework Programme (FP7) Grant (No 305934) Study author's key conclusion: NMP reduces early graft injury and improves utilisation.

Ravaioli 2022.

Study characteristics
Methods	Study design: monocentric open label RCT Country: Italy Duration of follow‐up: 6‐months. Median 473 (IQR 236–618) days
Participants	Number of participants: 135 randomised, 110 transplanted (55 in each group) Median donor age: 76 years in HOPE group, 72 years in SCS group Median recipient age: 57 years in HOPE group, 60 years in SCS group Number of DCD: 0 Number of ECD: all ECD Median preservation times Median warm ischaemic time: N/A Median cold ischaemic time: 255 min in HOPE group, 420 min in SCS group Median machine perfusion time: 145 (IQR 120–185) min (HOPE group) Median total preservation time: 400 min in HOPE group, 420 min in SCS group
Interventions	End‐ischaemic HOPE with Vitasmart (Bridge to Life) with Belzer UW machine perfusion solution. From arrival at recipient centre until implant. SCS with Celsior solution
Outcomes	Primary Early allograft dysfunction Secondary Graft survival Participant survival Early Allograft Failure Simplified Estimation risk score Surgical complications Primary non‐function Post‐reperfusion syndrome Length of hospital stay Biliary and vascular complications at 6 months
Notes	Funding source: National Health System Research Study author's key conclusion: HOPE improves early allograft dysfunction and graft survival. Note: clarification on mortality and adverse event data was provided by the corresponding author following email correspondence.

Schlegel 2023.

Study characteristics
Methods	Study design: multicentre controlled trial Country: Europe (Birmingham, Leeds, London (UK); Ghent, Leuven (Belgium); Groningen (the Netherlands); Lyon, Paris (France); Vienna (Austria); Zurich (Switzerland)) Duration of follow‐up: 1 year
Participants	Number of participants: 177 randomised, 170 transplanted (85 per group) Median donor age: 60.5 (IQR 47.0–72.0) years (overall) Median recipient age: 59.0 (IQR 50.2–64.0) years (overall) Number of DCD: 0 Number of ECD: not reported Median preservation times Median warm ischaemic time: N/A Median cold ischaemic time: 373.0 (IQR 299.2–471.8) min in HOPE group, 427.0 (IQR 356.0–487.0) min in SCS group Median machine perfusion time: 95.5 (IQR 73.0–137.0) min (HOPE group) Median total preservation time: 474.0 (IQR 403.5–588.0) min in HOPE group, 427.0 (IQR 356.0–487.0) in SCS group
Interventions	End‐ischaemic HOPE with Liver Assist (Organ Assist). Portal vein only at 3 mmHg for 1–2 hours (specified 1 hour minimum). 3 L recirculating Belzer machine perfusion solution (Bridge to Life Ltd.). The SCS prior to HOPE was performed with: 3 HTK, 26 UW, 55 IGL. SCS based on institutions standard. 1 HTK, 27 UW, 57 IGL.
Outcomes	Primary Number of recipients with post‐transplant complications Secondary Comprehensive Complication Index Laboratory parameters Duration of hospital and intensive care unit stay Graft survival Participant survival Biliary complications
Notes	Several outcomes were added post‐hoc, and not predefined in the protocol. Due to the risk of bias that this introduces, we discussed these post‐hoc outcomes narratively and they are not included in any meta‐analyses. Funding source: entirely funded by the Swiss National Science Foundation (33IC30_166909, 32003B_153012), including the perfusate, the perfusion machine disposables, and the monitoring. The funding party played no role in study design, performance, analysis, or the decision to publish. The participating centres provided the perfusion device (Liver Assist, Organ Assist, now XVIVO), and the training for machine perfusion for each centre was supervised by the study principal investigator.

van Rijn 2021.

Study characteristics
Methods	Study design: multicentre RCT Country: Europe (6 centres; main centre in the Netherlands) Duration of follow‐up: 1 year
Participants	Number of participants: 160 randomised, 156 transplanted (3 in SCS group and 1 in HOPE group cancelled, before any trial procedures started) Median donor age: 52 years in HOPE group, 49 years in SCS group Median recipient age: 60 years in both groups Number of DCD: all DCD Number of ECD: N/A Median preservation times Median warm ischaemic time: 29 min in HOPE group, 27 min in SCS group Median cold ischaemic time: 6 hours 11 min in HOPE group, 6 hours 49 min in SCS group Median machine perfusion time: 2 hours 12 min (HOPE group) Median total preservation time: 8 hours 44 min in HOPE group, 6 hours 49 min in SCS group
Interventions	Dual HOPE with Liver Assist (Organ Assist). End‐ischaemic for minimum 2 hours, using 4 L Belzer UW machine‐perfusion solution (Bridge to Life) with 3 mmol/L glutathione. SCS with Belzer UW CS
Outcomes	Primary Symptomatic cholangiopathy at 6 months. Imaging diagnosis (2 blinded independent radiologists) plus jaundice, recurrent cholangitis or cholestatic liver blood tests. Secondary Non‐symptomatic ischaemic cholangiopathy on 6‐month protocol magnetic resonance cholangiopancreatography Post‐reperfusion syndrome Primary non‐function Early allograft dysfunction (Olthoff criteria) Intensive care unit stay Hospital stay Hepatic artery or portal vein thrombosis Anastomotic leak or stricture Use of renal replacement therapy at 6 months Liver blood tests at 1 week, 1 month, 3 months, and 6 months Graft survival Patient survival
Notes	Funding source: Fonds NutsOhra, Bridge to Life provided UW, Organ Assist provided training only (units bought the machine). Funders had no role in trial. Study author's key conclusion: HOPE reduces symptomatic ischaemic cholangiopathy

ALT: alanine transaminase; AST: aspartate aminotransferase; DBD: donated following brain death; DCD: donated following circulatory death; ECD: extended criteria donor; HOPE: hypothermic oxygenated machine perfusion; HTK: histidine‐tryptophan‐ketoglutarate solution; IGL: Institut Georges Lopez solution; IQR: interquartile range; min: minute; N/A: not applicable/available; NMP: normothermic machine perfusion; RCT: randomised clinical trial; SCS: static cold storage; UW: University of Wisconsin solution.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Boteon 2018	Preclinical study
ChiCTR1800014529	Non‐randomised study
De Goeij 2022	Non‐randomised study
Fedaruk 2019	Preclinical study where none of the grafts proceeded to transplant.
Fodor 2021	Non‐randomised study
Guarrera 2007	Non‐randomised study
Guarrera 2011	No control group
Hefler 2022	Non‐randomised study
ISRCTN89667087	Non‐randomised study
Jarchum 2018	Review article
Kosmoliaptsis 2017	Non‐randomised study
Krdzalic 2019	Non‐randomised study
Maroni 2021	Retrospective study
Mehta 2018	Non‐randomised study
Meszaros 2021	No control group
Mohkam 2021	Retrospective, non‐randomised study
Muller 2022	No primary data
Oniscu 2018	Non‐randomised study
Patrono 2019	Retrospective, non‐randomised study
Perin 2021	Non‐randomised study
Rayar 2021	Non‐randomised study with historical control group
Watson 2018	Non‐randomised study

Characteristics of studies awaiting classification [ordered by study ID]

Minor 2022.

Methods	Randomised clinical trial
Participants	People listed for liver transplant
Interventions	Experimental group: livers will be put on a CE‐certified organ perfusion machine (Liver Assist, Fa. Organ Assist, The Netherlands) and machine perfusion started via the hepatic artery and hepatic portal vein in a closed circuit using Belzer machine perfusion solution. Perfusate will be oxygenated to a partial pressure of oxygen > 500 mmHg via 2 oxygenators included in the arterial and portal circuits. The temperature of the perfusate will be increased slowly over time to reach a steady state of 20 °C after 60 min. Total perfusion time will be 90 min or slightly longer, if the recipient preparation time exceeds the minimum perfusion time of 90 min (for further details see: Hoyer et al. Controlled oxygenated rewarming of cold stored livers prior to transplantation: first clinical application of a new concept. Transplantation 2016;100:147‐52). Control group: static cold storage
Outcomes	Primary Serum peak aspartate aminotransferase during first 3 days after transplantation Secondary Death assessed using patient medical records from admission to 3 months after transplantation Early allograft dysfunction according to Olthoff (bilirubin > 10 mg/dL or international normalised ratio > 1.6 at postoperative day 7, or peak aspartate aminotransferase > 2000 U/L during first week after transplantation, based on patient medical records from admission to 1 week after transplantation) Retransplantation assessed using patient medical records from admission to 3 months after transplantation Duration of intensive care unit stay assessed using patient medical records Ischaemia reperfusion injury based on laboratory tests assessed using patient medical records from admission to 1 week after transplantation Liver maximum function capacity assessed using LiMAx test 1 day after transplantation Evaluation of correlation of machine perfusion parameters (obtained during ex vivo machine perfusion) with ulterior graft function after transplantation assessed using patient medical records from admission to 3 months after transplantation. All recipients were observed for 7 days following transplantation on a daily basis. Follow‐up included additional observations on the day of discharge and 3 months after transplantation. Recipients were followed until 3 months after the last participant was randomised for this trial and were asked to attend clinical routine follow‐up subsequent to termination of the study.
Notes

Characteristics of ongoing studies [ordered by study ID]

Huang 2020.

Study name	Prospective, single‐centre, randomised controlled trial to evaluate the efficacy and safety of ischaemia‐free liver transplantation (IFLT) in the treatment of end‐stage liver disease.
Methods	Single‐centre RCT
Participants	People with end‐stage liver disease.
Interventions	Experimental: ischaemia‐free liver transplantation Control: static cold storage
Outcomes	Primary end point is the incidence of early allograft dysfunction after liver transplantation. Intraoperative and postoperative parameters of donor livers and recipients will be observed and recorded, and postoperative liver graft function, complications, and recipient and graft survival will be evaluated.
Starting date	Not available
Contact information	Professor Xiaoshun He; gdtrc@163.com; Professor Zhiyong Guo; rockyucsf1981@126.com
Notes

ISRCTN15686690.

Study name	Controlled oxygenated rewarming of liver grafts by ex‐situ machine perfusion prior to transplantation
Methods	RCT
Participants	People aged > 18 years with end‐stage liver disease who are listed for liver transplantation
Interventions	Experimental: donor livers are subjected to 2 hours of temperature‐controlled oxygenated machine perfusion before implantation, starting with a perfusion temperature of 10 °C which is slowly increased up to 20 °C. Control: donor livers do not receive any experimental treatment before implantation.
Outcomes	Primary Serum peak value of AST during the first 3 days after transplantation Secondary Death/6‐month graft survival Retransplantation within 6 months after implantation Time of stay in intensive care unit Early allograft dysfunction according to Olthoff (bilirubin > 10 mg/dL or INR > 1.6 at postoperative day 7, or peak AST > 2000 U/L during first week after transplantation) Hepatic tissue perfusion measured 1 hour after revascularisation
Starting date	2016
Contact information	Professor Andreas Paul
Notes

NCT02775162.

Study name	A multicenter randomized controlled trial to compare the efficacy of ex‐vivo normothermic machine perfusion with static cold storage in human liver transplantation
Methods	RCT
Participants	Recipients of liver transplantation
Interventions	Experimental: normothermic machine perfusion with OrganOx Metra Control: standard of care (ice)
Outcomes	Primary Percentage of participants with early allograft dysfunction at 7 days Secondary Number of participants with primary non‐function at 10 days Number of participants with graft survival at 6 months Number of participants with subject survival at 6 months Number of participants with post‐reperfusion syndrome at 1 day Biochemical liver function via bilirubin at 6 months Biochemical liver function via gamma‐glutamyl transferase at 6 months Biochemical liver function via alanine aminotransferase at 6 months Biochemical liver function via AST at 6 months Biochemical liver function via alkaline phosphatase at 6 months Biochemical liver function via INR at 6 months Biochemical liver function via lactate at days 1–7 Number of participants with evidence of ischaemia‐reperfusion injury via comparison of pre‐reperfusion biopsies to post‐reperfusion biopsies at 1 day Number of participants with biliary investigations and biliary interventions at 6 months Number of livers randomised but not transplanted at 1 day Number of livers that did not experience a safety event at 10 days Health economic implications via length of intensive care unit stay and length of hospital stay at 6 months Quality of life via the EQ‐5D‐5L Quality of Life Questionnaire at 6 months
Starting date	17 May 2016
Contact information	Stuart Knechtle, MD, Duke University
Notes

NCT03484455.

Study name	Study to evaluate performance of the Organ Recovery Systems LifePort® Liver Transporter System, a machine perfusion system, for liver transplant (PILOT™)
Methods	RCT
Participants	Recipients of liver transplantation
Interventions	Experimental: hypothermic machine perfusion with Organ Recovery Systems LLT system Control: static cold storage
Outcomes	Primary Number of participants with early allograft dysfunction at 7 days Early allograft dysfunction defined as total bilirubin ≥ 10 mg/dL or INR ≥ 1.6 or AST > 2000 IU/L or alanine aminotransferase > 2000 IU/L
Starting date	3 April 2019
Contact information	Study Director: Stan Harris Organ Recovery Systems
Notes	Ongoing study, none of the data published so far can be included in the review. Study authors contacted, but due to US Food and Drug Administration regulations are not at liberty to share further results at this stage.

NCT03929523.

Study name	Hypothermic oxygenated perfusion for extended criteria donors in liver transplantation (HOPExt)
Methods	RCT
Participants	Recipients of extended criteria donors
Interventions	Experimental: end‐ischaemic HOPE Control: classic static cold storage
Outcomes	Primary Early allograft dysfunction according to Olthoff criteria during first postoperative week: defined by the presence of ≥ 1 of: bilirubin level > 10 mg/dL (i.e. 171 µmol/L) on postoperative day 7 INR > 1.6 on postoperative day 7 AST or alanine aminotransferase levels > 2000 IU/L within the first 7 postoperative day Secondary Early allograft dysfunction using MEAF score during first 3 postoperative days (includes bilirubin, alanine aminotransferase maximum and INR maximum at postoperative day 3; range 0 (better outcome) to 10 (worse outcome)) Early allograft dysfunction using L‐GrAFT during first 10 postoperative days (includes AST, INR, total bilirubin, and platelets every day until postoperative day 10; range −6 (better outcome) to +6 (worse outcome)) Untargeted liver graft metabolic profiling on day of liver transplantation (day 0) (profiling by high‐resolution nuclear magnetic resonance – 1‐hour nuclear magnetic resonance spectrometer) on liver graft biopsies on the back‐table before and after liver machine perfusion Occurrence of post‐reperfusion syndrome on day of liver transplantation (day 0) defined as a 50% decrease in median arterial pressure during the 5 minutes following the graft revascularisation 90‐day morbidity and mortality during first 90 days after surgery Severe postoperative complications (Dindo‐Clavien ≥ 3) or death Duration of intermediate care unit stay (days) from randomisation until intermediate care unit discharge, estimated up to 7 days Duration of hospital stay in days from randomisation until hospital discharge, estimated up to 21 days Assessment of intra‐ and extrahepatic biliary complications (except for patients who underwent a retransplanted during the study using liver contrast‐enhanced magnetic resonance imaging including a magnetic resonance cholangiopancreatography within 1 year after liver transplantation) 3‐month and 1‐year recipient and graft survivals within 1 year after liver transplantation Hospital costs (Euros) of liver transplantation at 1 year after liver transplantation
Starting date	10 September 2019
Contact information	Mickael Lesurtel; +33 1 40 87 58 95; email: mickael.lesurtel@aphp.fr
Notes

NCT03930459.

Study name	Efficacy of ex‐situ normothermic perfusion versus cold storage in the transplant with steatotic liver graft (ORGANOXLAFE)
Methods	RCT
Participants	Recipients of steatotic liver graft
Interventions	Experimental: normothermic machine perfusion Control: static cold storage
Outcomes	Primary Peak ALT and AST at days 1, 3, 5, 7 post‐transplant Secondary Primary graft failure at day 10 post‐transplant (i.e. irreversible graft dysfunction that requires emergency hepatic replacement during the first 10 days after liver transplantation, in the absence of technical or immunological causes) Graft survival at day 30, month 6, month 12 post‐transplant Patient survival at day 30, month 6, post‐transplant, month 12 post‐transplant Post‐reperfusion syndrome, measured by mean arterial pressure levels during the first 5 minutes after reperfusion defined as a decrease in mean arterial pressure > 30% of the baseline value for > 1 minute during the first 5 minutes after reperfusion. This will be evaluated in the context of the use of vasopressors. Biochemical function of the liver measured by bilirubin post‐transplant levels at day 1, day 3, day 5, day 7, day 30, month 6, month 12 post‐transplant Biochemical function of the liver measured by gamma‐glutamyl transferase post‐transplant levels at day 1, day 3, day 5, day 7, day 30, month 6, month 12 post‐transplant Biochemical function of the liver measured by AST post‐transplant levels at day 1, day 3, day 5, day 7, day 30, month 6, month 12 post‐transplant Biochemical function of the liver measured by ALT post‐transplant levels at day 1, day 3, day 5, day 7, day 30, month 6, month 12 post‐transplant Biochemical function of the liver measured by INR post‐transplant levels at day 1, day 3, day 5, day 7, day 30, month 6, month 12 post‐transplant Early graft dysfunction at 7 days post‐transplant defined by: bilirubin > 10 mg/dL daily 7 after transplant, INR > 1.6 on day 7 after transplantation, peak ALT and ART > 2000 IU/L in the first 7 days after transplantation Intensive care stay duration at day 30 Hospital stay duration at day 30 Renal replacement therapy need at day 30, month 6, month 12 post‐transplant Intraoperative thromboelastogram result during transplant surgery Histological evidence of reperfusion injury during transplant surgery. Post‐reperfusion biopsies will be compared with baseline pre‐reperfusion biopsies and classified according to standard histological criteria (blind comparison to third parties) Evidence of biliary stenosis in magnetic resonance cholangiography at 6 months after transplantation
Starting date	2019
Contact information	Maria Cortell; 0034961246711; investigacion_clinica@iislafe.es
Notes

NCT04203004.

Study name	Hypothermic oxygenated perfusion with cytokine filtration in clinical liver transplantation: a randomised controlled trial
Methods	RCT
Participants	Liver transplant recipients
Interventions	Experimental: HOPE with cytokine filtration by CytoSorb; cytokine filtration during HOPE Control: HOPE‐standard; patients transplanted with livers preserved by HOPE without cytokine filtration
Outcomes	Primary Incidence of post‐reperfusion syndrome intraoperatively, during the first 5 minutes after reperfusion of the liver graft. Aggarwal definition: a decrease in mean arterial pressure > 30% below the baseline value, for ≥ 1 minute, occurring during the first 5 minutes after reperfusion of the liver graft Secondary Entity of ischaemia‐reperfusion injury 2 hours after reperfusion of the liver graft. Assessment of liver biopsy according to Suzuki histological grading system modified by UCLA group (Sosa RA et al. JCI Insight 2016;1(20):e89679) Incidence of early allograft dysfunction at day 7 post‐transplant using Olthoff definition: presence of ≥ 1 of: bilirubin ≥ 10 mg/dL on postoperative day 7, INR ≥ 1.6 on postoperative day 7, ALT or AST > 2000 UI/mL within first 7 postoperative days
Starting date	1 March 2020
Contact information	Contact: Stefania Camagni, MD; 0352674771 ext 0039; scamagni@asst‐pg23.it
Notes

NCT04644744.

Study name	Hypothermic oxygenated (HOPE) versus normothermic machine perfusion (NMP) in human liver transplantation (HOPE‐NMP)
Methods	RCT
Participants	Recipients of extended criteria livers
Interventions	Experimental: end‐ischaemic HOPE for a minimum of 2 hours (until hepatectomy) Experimental: end‐ischaemic normothermic machine perfusion for a minimum of 4 hours (up to 24 hours) Control: conventional cold storage
Outcomes	Primary Postoperative complications using Comprehensive Complication Index after first 90‐days postoperatively Secondary Peak alanine aminotransferase during first week postoperatively Peak AST during first week postoperatively Early allograft dysfunction during first week postoperatively using Olthoff criteria (bilirubin 10 mg/dL on day 7; INR 1.6 on day 7; ALT or AST > 2000 IU/L) Primary non‐function during first week postoperatively defined as graft with poor function requiring retransplantation or leading to death within 7 days after the primary procedure without any identifiable cause of graft failure Biliary complications at 6 months postoperatively assessed by magnetic resonance imaging or magnetic resonance cholangiopancreatography Organ utilisation rate during the first week postoperatively as rate of donor‐allograft offers that result in liver transplantation Total organ preservation time before preservation after liver implantation (0–3 hours) Duration and costs of initial intensive care unit stay as days of admission following liver transplantation (participants will be followed for 6 months postoperatively) Duration of hospital stay in days of hospital admission after discharge and up to 6 months after liver transplantation Cost of hospital stay calculated in days of hospital admission after discharge and up to 6 months after liver transplantation Postoperative complications using the Comprehensive Complication Index for 1‐year postoperatively Postoperative major complications using the Clavien‐Dindo Complication score for 1‐year postoperatively 1‐year recipient‐ and graft survival for 1‐year postoperatively
Starting date	14 January 2021
Contact information	Contact: Georg Lurje, MD; +4930450652339; georg.lurje@charite.de
Notes

NCT04744389.

Study name	Comparison of hypothermic versus normothermic ex‐vivo preservation (DCDNet)
Methods	RCT
Participants	Recipients of donation following circulatory death and extended criteria donor following brain death liver transplants
Interventions	Experimental: hypothermic machine perfusion Experimental: normothermic machine perfusion
Outcomes	Primary Rate of graft loss at 6 months postoperatively Death of participant, relisting, or retransplantation (composite outcome) Rate of ischaemic‐type biliary lesions assessed by magnetic resonance imaging/magnetic resonance cholangiopancreatography at 6 months postoperatively (composite outcome) Secondary 1‐year graft survival 1‐year recipient survival Level of BCL‐2/BAX at the liver histology after 2 hours of perfusion (BCL‐2/BAX is members of the Bcl‐2 family of regulator proteins that regulate cell death and correlates with graft loss level of Soluble Keratin 18 in the perfusate. Soluble Keratin 18 is a marker of necrosis and apoptosis and correlates with graft loss) Level of HMGB1 in the perfusate after 2 hours of perfusion (HMGB1 acts as danger‐associated molecular pattern (DAMP) molecule that amplifies immune responses during tissue injury and correlates with graft loss)
Starting date	15 December 2020
Contact information	Davide Ghinolfi, MD, PhD; 00393282185278; d.ghinolfi@ao‐pisa.toscana.it
Notes

NCT04812054.

Study name	Dual hypothermic oxygenated machine perfusion in liver transplantation using allografts from donors after brain death
Methods	RCT
Participants	Recipients of donation following brain death liver transplants
Interventions	Experimental: HOPE Control: simple cold storage
Outcomes	Primary Model for early graft dysfunction score at 3 days Continuous measure of allograft function in the early period after liver transplantation, calculated using serum ALT activity, INR for prothrombin time, and serum bilirubin concentration within 3 first postoperative days Secondary Recipient survival at 2‐year follow‐up Graft survival (retransplantation or death) at 2‐year follow‐up Incidence of biliary complications (bile leaks, strictures) at 2‐year follow‐up Postoperative complications in the 90‐day postoperative period classified by type and their severity according to the Clavien‐Dindo grading system Cytokine release in the perioperative period: serum levels of interleukin‐2, interleukin‐10, tumour necrosis factor‐alpha, high mobility group box 1, and 8‐hydroxy‐2'‐deoxyguanosine before reperfusion, 90 minutes after reperfusion, and 24 hours after reperfusion Allograft ischaemia‐reperfusion injury using Suzuki score in wedge allograft biopsy 90 minutes after reperfusion Hepatocyte apoptosis in the allograft using TUNEL assay in wedge allograft biopsy 90 minutes after reperfusion Endothelial activation in the allograft using Von Willebrand factor and P‐selectin staining in wedge allograft biopsy 90 minutes after reperfusion Activation of innate immunity in the allograft using Toll‐like receptor 4 staining in wedge allograft biopsy 90 minutes after reperfusion Oxidative injury of the allograft using malondialdehyde assay in wedge allograft biopsy 90 minutes after reperfusion Allograft adenosine triphosphate content in wedge allograft biopsy 90 minutes after reperfusion
Starting date	5 April 2021
Contact information	Contact: Michal Grat, MD, PhD; 0048225992545; michal.grat@wum.edu.pl
Notes

NCT05045794.

Study name	Bridge to HOPE: hypothermic oxygenated perfusion versus cold storage prior to liver transplantation
Methods	RCT
Participants	Liver transplant recipients
Interventions	Experimental: VitaSmart Liver Machine Perfusion System Control: static cold storage
Outcomes	Primary Proportion of recipients with early allograft dysfunction at day 7 post‐transplant Secondary MEAF score within 3 days post‐transplant (based on definition by Olthoff KM et al. Validation of a current definition of early allograft dysfunction in liver transplant recipients and analysis of risk factors. Liver Transplantation 2010;16(8):943‐9) Proportion of patients with primary non‐function within 7 days post‐transplant (based on definition by Pareja et al. A score model for the continuous grading of early allograft dysfunction severity. Liver Transplantation 2015;21(1):38‐46) Length of hospital stay up to study participation ends at 1‐year follow‐up Length of intensive care unit stay up to study participation ends at 1‐year follow‐up Duration on dialysis up to study participation ends at 1‐year follow‐up (from establishment of graft reperfusion until discontinuation of dialysis (measured in days)) Donor liver utilisation rate up to study participation ends at 1‐year follow‐up Patient survival rate at 30 days, 6 months, 1‐year post‐transplant Graft survival rate at 30 days, 6 months, 1‐year post‐transplant Incidence of adverse events by 1‐year post‐transplant Incidence of serious adverse events by 1‐year post‐transplant Incidence of unanticipated adverse device effects by 1‐year post‐transplant Incidence of ischaemic cholangiopathy at 6 months, 1‐year post‐transplant Incidence of biopsy‐confirmed liver rejection at 6 months, 1‐year post‐transplant
Starting date	16 December 2021
Contact information	Chris Stanford, MA; 720‐201‐1640; c.stanford@b2ll.com
Notes

ALT: alanine transaminase; AST: aspartate aminotransferase; HOPE: hypothermic oxygenated machine perfusion; INR: international normalised ratio; L‐GRaFT: Liver Graft Assessment Following Transplantation risk factor; MEAF: Model for Early Allograft Function; RCT: randomised clinical trial.

Differences between protocol and review

In the protocol (Tingle 2021), we discussed methodology for network meta‐analysis. However, we identified insufficient trials and the only comparisons were HOPE versus SCS and NMP versus SCS. In this instance, the advice from Cochrane is to perform simple indirect comparisons (Bucher 1997), and to include the methodology for network meta‐analysis as an appendix, so that it can be used in updates of the review should sufficient studies be published (Higgins 2021a). Appendix 2 is provided in line with this guidance.

In the protocol, we stipulated that we may use Trial Sequential Analysis (Brok 2008; Brok 2009; Thorlund 2010; Wetterslev 2008; Wetterslev 2017). However, the Cochrane Handbook for Systematic Reviews of Interventions advises against using Trial Sequential Analysis except when used as a sensitivity analysis of imprecision (Higgins 2021a); this decision follows on from a Cochrane Expert panel consensus statement on the issue which recommended against the use of sequential methods in Cochrane Reviews (Cochrane Scientific Committee 2018). In updates of this review we will continue to follow the most up‐to‐date Cochrane guidance.

Contributions of authors

• Conception of the review: ST

• Design of the review: ST, JD, ET, RF, BM, SP, CW

• Co‐ordination of the review: ST

• Drafted the review: ST

• Editing of review: ST, JD, ET, RF, BM, SP, CW

• Approved the final version: ST, JD, ET, RF, BM, SP, CW

Sources of support

Internal sources

• No internal sources of support, UK

  N/A

External sources

• NIHR Blood and Transplant Research Unit, UK

  This study was supported by the National Institute for Health Research (NIHR) Blood and Transplant Research Unit in Organ Donation and Transplantation at the University of Cambridge, in collaboration with Newcastle University and in partnership with National Health Service Blood and Transplant (NHSBT). The views expressed are those of the review authors and not necessarily those of the National Health Service, the NIHR, the Department of Health or NHSBT.

• Cochrane Hepato‐Biliary Group, Copenhagen Trial Unit, Centre for Clinical Intervention Research, Capital Region, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark

  Help with review preparation and editorial process

Declarations of interest

ST: none.

JD: none.

ET: none.

RF: none.

BM: none.

SP: none.

CW: none.

New