A Call for Randomization in Clinical Trials of Liver Machine Perfusion Preservation

Liver transplantation (LT) is a life-saving operation for patients with advanced liver failure. Although mortality from liver disease continues to increase, the number of available grafts remains relatively unchanged, leading to the death of ~15%-30% patients on the waiting list. Therefore, transplant teams have resorted to using marginal organs with often poorer outcomes. In order to improve the quality of “marginal” organs and reduce the severity of liver injury following transplantation, extracorporeal techniques such as ex situ normothermic machine perfusion (NMP) have been explored, but data whether these approaches improve outcome are still pending.⁽¹⁾

The authors of the VITTAL study address these questions by suggesting that evaluating marginal livers using NMP could avoid discarding many organs on subjective assessment only.⁽²⁾ This study had two primary objectives: (1) to establish the feasibility of NMP to assess graft function and increase the number of transplantable livers; (2) to achieve successful transplantation of these previously rejected livers evaluating 90-day patient survival. In their work, Mergental et al. assessed 31 livers discarded by all UK centers (n = 7) using NMP and were able to transplant 22 (71%) of them with a 100% 1-year patient and 86.4% 1-year graft survival. In our opinion, the main message of the study is that NMP can support the decision-making process by providing the surgeons with additional information on liver function. This study did not compare NMP preservation with standard cold preservation, but its design and choice of the endpoints would not be able to provide convincing evidence of NMP superiority to the standard static cold storage (SCS) in terms of preservation quality. The capacity of NMP to maintain grafts in a near-physiological state allows the evaluation of both liver injuries and synthetic capacities. However, an ideal biomarker, specifically able to predict clinically relevant outcomes, has not been identified yet.⁽¹⁾ Mergental et al.’s work represents a milestone in the pursuit of establishing reliable graft viability criteria, and the adaptive trial design in this study lays the groundwork for a future randomized trial where NMP and SCS in a high-risk donor cohort are compared head to head. Unfortunately, the parameters chosen to assess graft quality are all focused on hepatocellular function/damage alone (no qualitative bile parameter was considered, such as pH, lactate dehydrogenase, glucose, or bicarbonate), and, even if their adoption may reduce the risk of primary nonfunction or severe graft dysfunction, they are unable to predict or prevent ischemic cholangiopathy. As a matter of fact, 6 patients (27%) developed biliary complications at 6 months, 4 (18%) needed retransplantation after a median follow up of 542 days, and 4 (18%) more patients had radiological findings compatible with ischemic-type biliary lesions. This highlights that hepatocellular viability is likely physiologically separate from biliary/cholangiocyte viability and should be included in future trials to check viability.⁽³⁾

Although such studies are important and welcome, it is necessary to point out the limitations: (1) lack of randomization; (2) lack of objective, universally reproducible inclusion and exclusion criteria (e.g., liver biopsy); and (3) inappropriate control group. We must disagree with the authors’ statement: “… the study design had to be non-randomized because to conduct a similar study using previously declined livers in a randomized way would be ethically unacceptable.” NMP is not considered experimental practice in LT anymore, given that its safety and feasibility have been already shown⁽⁴⁾; however, its clinical superiority to SCS is not available yet,^{(5, 6)} and a large randomized trial did not show differences in relevant clinical outcomes like patient and graft survival, biliary complications, use of blood products, overall complication rate, and hospital or intensive care unit (ICU) stay.⁽³⁾ A randomized comparison to SCS would have been unethical only if NMP was undisputedly associated with significantly better transplant outcomes.

Achieving good results with perfused grafts that were previously declined by other centers does not necessarily mean that NMP was responsible for graft “rescue” or quality improvement. Several other single-center experiences reported good or excellent outcomes when using discarded livers without using ex situ machine perfusion (MP) (Table 1). Indeed, the same center that performed the VITTAL trial recently reported good outcomes of 206 previously discarded grafts after SCS compared with a matched low-risk cohort of primarily accepted livers, showing that even without MP, an appropriate recipient selection may grant comparable outcomes in terms of incidence of primary nonfunction (2.4% vs. 1.7%; P = 0.5483), in-hospital mortality (6.3% vs. 4.1%; P = 0.2293), and 3-year graft (82.5% vs. 84.1%; P = 0.6872) and patient (85.4% vs. 87.6%; P = 0.8623) survival.⁽⁷⁾

Randomization is crucial because data on organ utilization with MP are difficult to interpret given that the decision to use or not a graft is based on a subjective decision, as acknowledged by the authors. No definitive objective viability criteria are available, and the decision of whether to transplant or discard a liver rather depends on the peculiar practice of the respective transplant centers. Organ acceptance has been shown to be influenced by logistics, day of the week and time of the day, costs, center volume and performance, and regulatory boards. MacConmara et al. recently observed, in a retrospective study of 228 livers preserved with MP, that the number of discarded grafts in USA decreased from 13.3% to 3.5% when a MP strategy was used, with no difference in 1-year patient and graft survivals.⁽⁸⁾ We have challenged this enthusiasm in view of the use of MP restricted to few high-volume centers, thus providing significant biases in their conclusion. For example, whereas moderate-severe macrosteatosis and advanced donor age are absolute reasons to discard a graft in smaller centers, those grafts are routinely transplanted by several large and experienced programs.⁽⁹⁾

In the VITTAL trial, only 22 of 126 discarded grafts (17.4%) that qualified for NMP were eventually transplanted. This may suggest a strict selection of grafts with favorable characteristics for inclusion in the MP arm of the study. Most of the grafts considered for the VITTAL study were previously discarded based on a subjective evaluation of steatosis (64 grafts; 35%), but only three of those eventually perfused had macrosteatosis >30%. Furthermore, no objective histological characteristic (e.g., steatosis, necrosis, and fibrosis) has been taken into consideration to define these organs as marginal. A protocol biopsy of all discarded livers would have been extremely helpful in promoting a gradual expansion of acceptability criteria using objective reproducible parameters and a comparable control group.

Analyzing the 31 perfused organs in this study, there were two main reasons for considering them as high-risk grafts: (1) a Donor Risk Index (DRI) >2.0 (23 of 31; 74%) and (2) an expected prolonged cold ischemia time (CIT; 10 of 33; 32%). Prolonged CIT has been widely regarded as a risk for post-LT complications. It might be postulated that NMP has the potential to improve transplant logistics, delay LT, and counterbalance longer CIT. However, to appraise this effect, timing of pre-SCS-NMP and duration of NMP should have been matched as well for a fair comparison. In general, matching patients on prespecified objective and reproducible criteria is important to mitigate the impact of any bias.

The International Liver Transplantation Society (ILTS) through the Special Interest Group (SIG) “DCD, Preservation and Machine Perfusion” established a working group to evaluate and discuss the relevant literature and establish consensus statements and suggestions on how to design future clinical trials in liver perfusion during the “DCD, Liver Preservation, and Machine Perfusion” consensus conference held in Venice, Italy on January 31, 2020. Recommendations privileged randomized trials using clinically relevant endpoints, such as 1-year graft/patient survival, ischemic cholangiopathy, or other relevant complications (https://ilts.org/education/lectures/machine-perfusion-and-clinical-trials-session-special-considerations-and-pitfalls-in-clinical-trials-using-machine-perfusion/).⁽⁶⁾

Randomized studies to validate viability criteria are therefore not only ethically acceptable, but also necessary, given that they can minimize bias commonly involved in liver graft selection and allocation. To design well-powered studies, a large number of transplants and multicenter collaborations may be necessary. The main question that remains unsolved after this study is whether and how MP will lead to superior outcomes of these higher-risk “declined” grafts when compared to nonperfused organs to justify the increased costs and complexity in logistics and extra personnel requirement associated with MP. It is important to establish whether MP really improves graft quality and reliably predicts outcomes or whether it is just another element subjectively used to increase surgeons’ confidence to accept a higher-risk liver graft. The burden of proof required to do “more” using a more-complex approach should probably be higher than to do “less,” particularly if doing more is likely to increase overall health care costs.

Author Contributions

P.N.M. conceived and wrote the first draft. P.-A.C., R.J. and D.G. drafted the manuscript and reviewed it critically for important intellectual content.