In this issue of Transplantation, Watson et al1 present the largest single-center series of human livers, evaluated during normothermic machine perfusion (NMP) before transplantation. With their vast experience using different devices, the team from Cambridge suggested their first parameters for viability assessment during NMP in 2018.2 With this excellent new work, such biomarkers are now critically analyzed with a focus on their role to predict early allograft function (EAF). Additionally, the value of suggested bile chemistry markers predicting nonanastomotic biliary strictures (NASs) is explored.1 Between May 2017 and February 2022, a total number of 203 human livers underwent NMP, and 154 (75.9%) were subsequently transplanted, with more than half being donated after circulatory death (DCD; n = 84, 54.5%).1 Based on the generally low rate of primary nonfunction (PNF) in most published series and the lack of a routine aspartate-aminotransferase measurement during posttransplant follow-up in Cambridge, the Model for EAF (MEAF) was selected as study endpoint.3 The following parameters obtained during NMP were most predictive for a good initial liver function, classified by a low MEAF score: low 2 h Lactate and alanine-aminotransferase (ALT), the limited need to supply bicarbonate (first 4 h of NMP), a high peak bile pH, and high bile/perfusate glucose ratio (Table 1). Despite such clear acceptance criteria, covering hepatocytes and cholangiocytes, 11% of implanted livers developed NAS.1 TABLE 1. - Overview of viability criteria presented by 3 leading groups within the last 6 y Category Parameters Cambridge Birmingham Groningen General information Watson et al 2018 2 Watson et al 2022 1 Mergental et al 2016 8 Mergental et al 2020 7 Van Leeuwen et al 2019 4 Van Leeuwen et al 2022 5 Perfused livers and modality 47 (35 DCD, 12 DBD), endischemic NMP 203 (123 DCD, 80 DBD), endischemic NMP 6 (4 DCD, 2 DBD), endischemic NMP 31 (14 DCD, 17 DBD), endischemic NMP 16 DCD, endischemic NMP after D-HOPE + COR 53 DCD, 1 DBD, endischemic NMP after D-HOPE Grafts transplanted 22/47 (16 DCD, 6 DBD) 154/203 (84/123 DCD, 70/80 DBD) 5/6 (4 DCD, 1 DBD) 22/31 (10 DCD, 12 DBD) 11/16 DCD 34/54 (33 DCD, 1 DBD) Source of parameter Perfusate, bile Perfusate, bile Perfusate, bile flow Perfusate, bile Perfusate, bile Perfusate, bile Decision timepoint 2 h 2 h 4 h 4 h 2.5 h 2.5 h Perfusion characteristics Hepatic artery flow — — >150 mL/min >150 mL/min — — Portal vein flow — — >500 mL/min >500 mL/min — — Perfusate chemistry pH Maintain perfusate pH >7.2 with ≤30 mmol bicarbonate supplementation Continuing requirement for bicarbonate to maintain pH >7.2 beyond 2 h. The need for bicarbonate in the first 2 h to maintain a pH>7.2 is no longer considered important >7.3 >7.3 7.35–7.45 Green: 7.35–7.45; orange: 7.25–7.35; red: <7.25 Lactate Peak lactate fall ≥4.4 mmol/L/kg/h <2.8 mmol/L <2.5 mmol/L <2.5 mmol/L ≤1.7 mmol/L Green: <1.7 mmol/L; orange: 1.7-4 mmol/L; red: >4 mmol/L Glucose Falling glucose beyond 2 h OR perfusate glucose <10 mmol/L with subsequent fall during challenge with 2.5 g glucose When glucose <10 mmol/L at 30 min → 5 g dextrose challenge, glucose rate should fall ≥1 mmol/L/h at 4 h NMP Yes vs No Yes vs No — — ALT <6000iU/L <6000iU/L — — — — Bile chemistry Bile flow a — Yes vs No (DCD livers) b Yes vs No Yes vs No >10 mL Green: ≥10 mL, ≥4 mL/last hour; orange: 5–10 mL; red: <5 mL pH Peak > 7.5 d Peak > 7.5 — >7.3 >7.45 Green: ≥7.45; orange: 7.40–7.45; red: <7.40 Delta bile value minus perfusate Glucose ≥10 mmol less bile than perfusate glucose or, if perfusate glucose <11 mmol/L, a bile glucose of ≤3 mmol/L ≥10 mmol less than in perfusate or, if perfusate glucose <11 mmol/L, a bile glucose of ≤2 mmol/L — — Absolute bile glucose: <0.67 Green: <−5; orange: −3 to −5; red: >−3 HCO3 − — — — — — Green: >0.1; orange: 0.05–0.01; red: <0.05 pH — — — — — Green: >5; orange: 3.0–5.0; red: <3 Decision-making is based on 2 h ALT, 2 h lactate, requirement for supplementary bicarbonate 2 h ALT, 2 h lactate, highest bile pH, requirement for supplementary bicarbonate in first 4 h, c bile production in DCD Lactate and 2 of the other criteria Lactate and 2 of the other criteria Based on the above-mentioned markers and thresholds Traffic light system for all parameters; parameter in the orange zone could be accepted if other criteria are green, red zone declined Livers failed despite fulfilling acceptance criteria 5/22 (1 PNF, 4 ITBL, 22.7%), median follow-up: 20 mo 1 PNF, 1 massive subcapsular hematoma at reperfusion→ ischemia, 1 NAS (graft loss at day 229; 1.95%), 3× HAT None, follow up: 7 mo 4 NAS/22 (18.2%), follow up: 6 mo 1 NAS/11 (9%), median follow up 12 mo 2/34 (1 chron rejection, 1 outflow obstruction, 5.9%), 1 NAS/34 (2.9%), median follow-up: 38 mo The table describes the evolution of viability criteria during NMP of human livers. Most criteria were adapted throughout an increasing number of cases and with more experience, clinically and with perfusion devices.aPhysiological bile production of a healthy liver is 100 mL/1 h.bBile production was 18 and 24.4 mL within the first 4 h of NMP in transplanted DCD and DBD livers, respectively.cParameters considered for decision making but not mutually exclusive, team experience and individual decision were applied.dIn 16 of 22 transplanted livers, bile pH was measured; livers that produced bile and showed a bile pH of >7.5 in the first 4 h of NMP did not present histological features of relevant stromal necrosis found in the major septal ducts or develop later NAS after transplantation.ALT, alanine-aminotransferase; COR, controlled oxygenated rewarming; DCD, donated after circulatory death; HOPE, hypothermic oxygenated perfusion; NAS, nonanastomotic biliary strictures; NMP, normothermic machine perfusion; PNF, primary nonfunction. WHICH POSTTRANSPLANT OUTCOME MEASURES ARE OF INTEREST FOR PREDICTION DURING MACHINE PERFUSION? There is a general lack of consensus regarding relevant posttransplant outcome parameters, which should be predicted during preservation. To reliably forecast dichotomous parameters, including graft survival and specific complications, much larger cohorts, which include livers with different risk profiles and longer follow-up, are required. Together with the limited caseload in most series, this might be one reason why currently used biomarker combinations are not consistent in their prediction of posttransplant NAS.4,5 Parameters and score models, which are expressed continuously, are more often selected. Some variables required to calculate recently developed models of EAF may, however, be lacking, because they are not everywhere routinely measured after transplantation. Additionally, most EAF models were developed in the context of cold storage and will require modifications when used with machine perfusion. Transaminases are released during any type of reperfusion on devices and in the recipient, where plasma values contribute to most EAF scores. At the same time, the perfusate ALT concentration is used by the Watson group as a parameter for decision making during NMP. The double ALT release from the same graft, during NMP and after implantation, may impact both the predictive value of the here used viability parameters and the role of the MEAF score. LIVER FUNCTION OR INJURY: WHAT DO WE REALLY MEASURE IN THE PERFUSION CIRCUIT? Understandably, most viability markers used during perfusion were derived from clinical practice, where they are routinely measured in patients with liver failure or after liver transplantation.6 Various challenges to assess the enormous metabolic liver capacity imposed by an ex situ circuit with a limited perfusate volume should be considered as very different from any physiologic scenario. One direct consequence appears with controversial study results and the proposal of various parameter combinations and thresholds claimed to be of value as surrogate of liver viability.2,5,7 The best example is lactate, where absolute perfusate values were suggested with at least 3 different thresholds.1,5,7,8 To achieve lactate clearance on an isolated circuit, only a few metabolic intact hepatocytes are required. This is further supported by the PNF, described after implantation of livers, which cleared lactate quickly during NMP.2,9 In contrast, perfusate lactate levels may be exceptionally high because of ischemic liver segments, which may lead to liver decline due to a presumed too high risk. The rate of lactate fall per liver weight, rather than an absolute perfusate concentration, was therefore nominated to better correlate with the “real” parenchymal injury. Authors further suggested to test the hepatocellular lactate clearance capacity in an experimental NMP circuit, preloaded with lactate.2 Despite their importance to evaluate liver injury, Transaminase levels are rarely considered of impact during perfusion.1 With the additional release from circulating erythrocytes, the parameter AST has obvious disadvantages.2 In contrast, in their newest series, Watson et al declined 21 livers based on perfusate ALT values, found beyond their established threshold of 6000iU/L at 2 h of NMP.1 Although high or steadily increasing transaminases are naturally recognized as a surrogate of advanced liver injury, the perfusate marker ALT alone demonstrated only weak correlations with clinically relevant outcomes.10 A higher predictive accuracy may be achieved when the liver weight is considered. Despite the common knowledge that livers of 3 kg weight will naturally release higher levels of any biomarker during perfusion, when compared with an 800 g graft, this variable is not routinely considered yet. Based on the clinical need to limit NAS, bile production and bile chemistry are measured to assess cholangiocytes. The Cambridge group did not see a correlation between bile production and outcome after transplantation in their first series presented in 2018. Livers, which developed NAS, were among those with the highest and lowest bile production.2 In the recent study and with increasing experience, authors have now declined DCD livers based on the lack of bile production and the subsequent inability to assess cholangiocyte integrity.1 Although canalicular bile has similar glucose levels as plasma, glucose is reabsorbed by healthy cholangiocytes, leading to low bile glucose levels. Together with their additional role to maintain bile alkalic through bicarbonate secretion, the absolute bile values, and bile/perfusate deltas of glucose, pH and bicarbonate were recently suggested by the Groningen group as the best parameter combination to explore the cholangiocyte performance.5 Of note, in the recent Cambridge series, 16 DCD livers were rejected based on a low peak bile pH and/or the bile/perfusate glucose. Another 17 recipients developed NAS (11%) despite fulfilling bile chemistry criteria with a high peak bile pH of >7.5. And some livers, which failed bile chemistry criteria, were transplanted without development of NAS.1 This underlines the complexity of viability testing with the use of such rather unspecific parameters. The strict application of bile chemistry criteria developed in 1 center may lead to a false low liver acceptance rate when applied in another center or country. The importance to gradually modify viability parameters became obvious along with increasing experience and based on the development of NAS in livers, which passed previous viability criteria.4,5 The group from Groningen, therefore, repetitively presented new series of livers accepted during NMP based on slightly modified criteria (Table 1).5 Most parameters claimed to represent liver viability, appear rather peripheral to the site, where ischemia-reperfusion injury (IRI) occurs.11,12 When oxygen is reintroduced into ischemic tissue, reactive oxygen species are immediately released from mitochondrial complex I. Surrogate markers for mitochondrial function (and injury) should therefore be more frequently explored, not only during hypothermic oxygenated perfusion (HOPE) but also during NMP.11,13,14 For example, flavin mononucleotide (FMN), a marker, whose release from complex I is known as the primary lesion during IRI was later rediscovered with a clear link to liver quality and predicted liver function, recipient complications, and graft survival.10 WHAT CRITERIA SHOULD BE FULFILLED BY CLINICALLY RELEVANT BIOMARKERS? To define the best possible set of biomarkers, a large set of experiments with animal models and the confirmation in good quality human livers is required. Relevant viability parameters confirm that healthy livers are maintained viable during perfusion. This approach is routine when a new experimental model is established. This step is, however, frequently omitted. The ideal biomarker should also achieve a high sensitivity and positive predictive value to identify livers, which lead to impaired results once implanted. More importantly, a high specificity is required to avoid transplanting livers, which fail despite passing all set viability criteria. The here presented series of 203 perfused grafts is an important step. Overall large registry data of perfused livers with transplantation are required to enable a statistically sound calculation. WHAT ARE THE CURRENT CHALLENGES TO DEFINE THE THRESHOLD OF A BIOMARKER? The remarkable hepatic capacity to regenerate despite significant injury confounds the identification of a clear and sustainable biomarker threshold, applicable in most clinical scenarios, particularly during perfusion on an isolated circuit, which lacks the physiological liver position and the communication with other organs. Various perfusion parameters, including perfusate components and temperatures, may certainly impact further. The technique of NMP is highlighted as near physiological by many. Perfusion criteria, frequently claimed as signs for organ viability, include, for example, a target portal vein flow of >500 mL/min or a bile flow of >10 mL/h (Table 1). Both parameter thresholds appear, however, quite far away from physiological human values. Next, suggested biomarkers show specific variations of their baseline concentration in healthy livers, which is, however, not considered as a reference to define a threshold yet. Higher caseloads are therefore required to validate biomarkers to confirm the suggested cutoffs, ideally through multicenter validation studies, where posttransplant outcomes are available after liver perfusion with the same technical modality, for example, NMP with the same device, perfusate composition, and a comparable overall donor-recipient risk.6 The newly introduced traffic light system and delta of bile/perfusate for bicarbonate and pH by the Groningen group appears as a great example here.5 Various combinations of parameters identified in the green, orange, or red zone are generally possible and require a high number of perfusions and transplantations before all subcategories achieve a sufficient caseload to reliably predict specific outcomes. Extended criteria donor livers, which are rejected in 1 center based on national guidelines or experience, are not automatically considered to be too high risk in all other countries.5 In the Netherlands, the donor age cutoff (≤60 y) was reported as the main reason to allocate DCD livers to a combined preservation protocol of D-HOPE plus NMP.5 DCD livers with a donor age >70 y are, however, transplanted with cold storage alone in experienced centers in other countries, provided the overall risk is well controlled,15 and old DCD livers with longer donor warm ischemia times are also transplanted with NRP ± HOPE.5,17 Once the most decisive parameter is identified, the value should be tested prospectively, which will ultimately lead to necessary modifications of applied parameter constellations and thresholds, as demonstrated by the recent VITTAL study and the present work from Cambridge.1,7 Predefined parameters should then also be challenged on the bench in experimental studies using a transplant model of comprehensive injury. Although the suggested pathway is well known, the immediate need for clinically useful markers pushes us to skip some of the required steps. In addition, the transplant community is only at the beginning of a routine use of machine perfusion with subsequently slow increase in experience and number of perfused livers with assessment during perfusion. The best examples are the 3 thresholds (<1.7 mmol/L, <2.5 mmol/L, and <2.8 mmol/L) suggested for perfusate lactate levels to support the decision making at 2, 2.5, or 4 h of NMP.1,4,5,7,8 In contrast, the suggested FMN threshold was systematically established through real-time spectroscopy of perfusates from >50 human HOPE perfusions.6,10,13 The advantage here is that enough graft loss events (both PNF and NAS) were present in this cohort to clearly link them with high perfusate FMN levels based on a very high mitochondrial injury and subsequent downstream tissue inflammation.6,10,13 The metabolic response of mitochondria to different injury levels, as measured by FMN release from complex I, was also tested in systematic experimental studies.13 Today, the role of FMN is increasingly explored under various perfusion conditions.10,13-15 The multicenter validation study with >650 perfusate samples is currently underway and will bridge earlier clinical results and experimental findings.13,15 Similarly to the ample system presented by the Groningen team, perfusate FMN values are allocated to different complication categories in recipients (green, orange, and red). The most valuable set of future biomarkers will not just predict the occurrence of PNF or NAS in a dichotomous “yes or no mode” but also enable a more specific forecast of different complication categories. Similar to the gradual increase of IRI-associated mitochondrial injury and tissue inflammation, separate perfusate FMN thresholds are developed for early allograft dysfunction, kidney injury, NAS, and PNF. The liver acceptance will then depend on national regulations, the center policy, and the willingness of a transplant unit to cope with different levels of predicted complications and subsequent increased costs. Outcomes after transplantation are however also dependent on the disease severity and the medical (and anesthesiological) managment of the liver recipient; a very sick candidate requiring high inotrop support may well induce an EAF in a graft that previously met all viability criteria on the machine with high predictive accuracy. WHAT ARE FUTURE DIRECTIONS? In this edition, Transplantation presents the currently largest series of perfused human livers with viability assessment and transplantation. Despite this important new contribution, the lack of consensus in terms of risk factors, perfusion modalities, surrogate markers for viability, and targeted outcome parameters remains. Further larger studies, though challenging to design, should emerge from a fast cooperation with a focus on the real value of current biomarkers in context of a higher utilization rate.