From Rescue to Risk: Challenges and Controversies in Pushing the Boundaries of Organ Utilization by Machine Perfusion

ABSTRACT

The ongoing shortage of donor livers has prompted the use of extended criteria donors (ECD) and the development of ex‐situ machine perfusion (MP) to expand the transplantable organ pool. MP has transformed liver transplantation (LT) by improving preservation, enabling better logistics, and allowing real‐time viability assessment that is, a process aimed at predicting liver function after transplantation based on MP parameters. Although viability assessment has been proposed to safely expand donor pool and improve LT outcomes, it remains inconsistent. This article will summarize currently available assessment protocols and recapitulate studies exploring emerging viability biomarkers, discussing the need for an operational definition of viability assessment and its implications for the design of future studies on the subject.

Despite its potential to facilitate the expansion of the donor pool, liver viability assessment by ex‐situ machine perfusion remains inconsistent. Novel biomarkers and better‐designed studies are necessary to overcome current limitations. Viability assessment should be expressed as a continuum, with decisional thresholds based on the definition of acceptable risk.

1. Introduction

The global shortage of suitable donor organs remains one of the greatest challenges in the transplantation field and has prompted efforts to expand the donor pool in the last decades. In liver transplantation (LT), utilization of liver grafts from so‐called extended criteria donors (ECD) has allowed expanding the donor pool and improving access to transplantation, but has been associated with increased risks for the recipient [1, 2].

In recent years, ex‐situ machine perfusion (MP) has emerged as a transformative technology aimed at improving organ preservation [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13], allowing viability assessment [14, 15, 16, 17, 18, 19, 20], and facilitating transplant logistics [21, 22, 23]. Although not the primary purpose, MP yielded the unexpected benefit of creating a window of opportunity to observe a liver in an isolated environment, gathering informative data about post‐LT damage and function, a process usually referred to as “viability assessment”. This marked a fundamental shift in the way organ offers are approached and allowed overcoming some limitations of previous prognostic models based on donor and recipient characteristics, as well as procedural variables [24, 25, 26]. Several studies have shown that, based on different assessment protocols, livers initially rejected based on donor features or macroscopic evaluation can be successfully recovered for transplantation, leading to a significant expansion of the donor pool with acceptable risks for the recipient [17, 18, 27, 28, 29]. In this setting, traditional donor and recipient risk factors no longer dictate organ acceptance, but instead guide the decision to use machine perfusion for viability testing.

However, as the field pushes the boundaries of organ utilization, numerous challenges have surfaced. The promise of a safe expansion of the donor pool did not come without risks, as occasional failures have been reported even for organs having been deemed transplantable based on MP parameters [18, 27, 28, 29, 30]. The risk of failure or ischemia–reperfusion‐related complications appears to be proportional to the characteristics of marginality of included grafts, with some reports suggesting caution when particularly high‐risk livers are evaluated [27, 28, 31]. Accordingly, some assessment protocols have evolved toward a more flexible approach, allowing deviations from initially strict viability criteria by incorporating recipient factors and considering waiting list pressure [29, 32, 33]. It is therefore unclear whether viability parameters during MP should overrule or rather integrate traditional risk factors. In everyday practice, the decision to accept a liver for transplantation still relies on multiple factors, with an unavoidable degree of subjectivity, and there is an urgent need for more reliable biomarkers and assessment protocols to be used especially in difficult cases.

Given this complex scenario, this narrative review explores the state of the art and key obstacles faced in maximizing organ use through ex‐situ machine perfusion, promoting a critical understanding of the value and limitations of current practice. Furthermore, it discusses potential solutions to safely and effectively advance the field.

2. What Do We Have: Current Viability Assessment Protocols

In the pre‐MP era, several prognostic models have been developed to guide organ utilization based on donor, recipient and procedural variables. The Donor Risk Index (DRI), based solely on donor factors, was among the first attempts to quantify graft risk [26]. Later, combined models such as balance of risk (BAR) [25], survival outcomes following liver transplantation (SOFT) [34], and donor age × recipient MELD (D‐MELD) [35] incorporated both donor and recipient characteristics to better predict post‐transplant outcomes. Despite their theoretical value, these models demonstrated only moderate predictive accuracy (C‐statistics often < 0.7), were inconsistently validated across centers, and lacked real‐time clinical utility. As a result, these models have not been widely adopted in routine practice, primarily due to their inability to directly assess the functional status of the organ at the time of transplantation.

As brilliantly highlighted by Schneeberger in his editorial “Life of a liver awaiting transplantation” [36] on the Nasralla et al. [9] trial, MP redefines the graft from a passive object to an actively monitored organ, in which the liver is treated almost like a patient: perfused, nourished, and cared. MP thus transcends traditional preservation, creating a critical window for viability assessment and more informed transplant decisions.

In the setting of normothermic machine perfusion (NMP), various viability criteria have been proposed in recent years, with lactate clearance being one predominant factor (Table 1) [18, 29, 30, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46]. Various lactate thresholds at different timepoints have been proposed to guide transplant decisions, but a consensus is still lacking [2]. Lactate, a byproduct of anaerobic glycolysis, is produced when pyruvate is reduced by lactate dehydrogenase (LDH) under hypoxic conditions, serving as a compensatory mechanism for impaired mitochondrial ATP production. Clinically, elevated lactate is a well‐established marker of tissue hypoperfusion, and it is commonly used in liver surgery and transplantation to assess early liver remnant and graft function [47]. However, translating lactate dynamics from clinical practice to NMP is not straightforward. As Watson et al. [48] pointed out, even a small number of viable hepatocytes can clear lactate in a low‐volume perfusion circuit, potentially leading to an overestimation of the true functional capacity of the liver. Therefore, while failure to clear lactate may indicate severe injury, rapid clearance does not necessarily imply a fully functional, transplantable organ [30, 38, 49].

TABLE 1.

Viability assessment protocols during normothermic machine perfusion.

Author, year	Perfusate	PV and HA flow	Bile	Timing
Mergental et al. 2016	Lactate < 2.5 mmol/L. pH > 7.3	PV > 500 mL/min. HA > 150 mL/min	Evidence of bile production	4 h
Watson et al. 2018	Peak lactate fall ≥ 4.4 mmol/L/kg/h. pH > 7.2 with ≤ 30 mmol bicarbonate. Evidence of glucose metabolism. ALT < 6000 IU/L		pH > 7.5. Glucose: Δ ≥ 10 mmol less than perfusate or ≤ 3 mmol/L if perfusate < 11 mmol/L	2 h
Van Leeuwen et al. 2019	Lactate ≤ 1.7 mmol/L. pH 7.35–7.45		Bile production > 10 mL. pH > 7.45. Glucose < 0.67	2.5 h
Mergental et al. 2020	Lactate < 2.5 mmol/L. pH > 7.3.	PV > 500 mL/min. HA > 150 mL/min.	Evidence of bile production. pH > 7.3	4 h
Cardini et al. 2020	Lactate ≤ 2.5 mmol/L. Maintenance of a physiolgical pH. AST, ALT, LDH < 20 000 IU/L.			NA
Zhang et al. 2020	Lactate < 2.5 mmol/L. pH > 7.3	Stable flows	Evidence of bile production	4 h
Watson et al. 2022	Lactate < 2.8 mmol/L. Physiological pH. Evidence of glucose metabolism (dextrose challenge). AST/ALT < 6000 IU/L		Bile production (only for DCD livers). pH Peak > 7.5. Glucose: Δ ≥ 10 mmol less than perfusate or ≤ 2 mmol/L if perfusate < 11 mmol/L	4 h
Van Leeuwen et al. 2022	Lactate: Green: < 1.7 mmol/L; orange: 1.7–4 mmol/L; red: > 4 mmol/L. pH: Green: 7.35–7.45 Orange: 7.25–7.35 Red: < 7.25		Bile production: Green: ≥ 10 mL, ≥ 4 mL/last hour; orange: 5–10 mL; red: < 5 mL. pH green: ≥ 7.45; orange: 7.40–7.45; red: < 7.40. ⍙pH green: > 5; orange: 3.0–5.0; red: < 3. ⍙glucose green: < −5; orange: −3 to −5; red: > −3. ⍙HCO3 − green: > 0.1; orange: 0.05–0.01; red: < 0.05	2.5 h
Weissenbacher et al. 2022	Lactate < 2.5 mmol/L. pH 7.3–7.45	Stable flows	Evidence of bile production	6 h
Seidita et al. 2022	Lactate clearance normalization or at least halving of lactates at end of perfusion. pH > 7.3	Stable flows	Evidence of bile production
Quintini et al. 2022	Perfusate lactate < 4.5 mmol/L or decrease of 60% from peak in first 4 h	Stable flows	Bile production > 2 mL/h	6 h
Vogel et al. 2024	Lactate < 3.5 mmol/L	HA > 150 mL/min	Evidence of bile production. Viscous bile	4 h
Reiling et al. 2024	Lactate < 2 mmol/L within 2 h. Maintenance of physiological pH	Stable flows	Evidence of bile production	4 h

Abbreviations: HA, hepatic artery; PV, portal vein.

Alongside lactate, other parameters such as perfusion flows, macroscopic appearance, and bile production are frequently used. Additionally, bile composition, especially bile pH, bicarbonate, and glucose, has gained recognition for prediction of cholangiocyte viability and biliary epithelial integrity. A bile pH above 7.4 and high bicarbonate concentrations suggest preserved secretory function of the peribiliary glands and intact cholangiocyte mitochondrial activity. Conversely, elevated bile glucose indicates leakage from plasma and impaired reabsorption, signaling biliary epithelial damage [29, 50, 51].

The Groningen group proposed a comprehensive composite system, dividing thresholds for organ acceptance into green, orange, and red zones [29]. Notably, these criteria were proposed in livers having been treated by hypothermic oxygenated perfusion (HOPE) and controlled oxygenated rewarming (COR) prior to NMP. This model combined perfusate and biliary parameters to assign a color code, ultimately guiding graft acceptance. However, the final decision to proceed with transplantation was left to the discretion of the transplanting team, taking into account the recipient's medical condition and urgency.

In the setting of HOPE, viability assessment is less established (Table 2) but shows promise through mitochondrial biomarkers such as flavin mononucleotide (FMN) and nicotinamide adenine dinucleotide (NADH) [14, 19, 52, 53]. FMN is a cofactor tightly bound to complex I of the mitochondrial electron transport chain and plays a key role in oxidative phosphorylation. During ischemia–reperfusion injury (IRI), mitochondrial damage leads to the release of FMN into the perfusate, reflecting the extent of mitochondrial dysfunction [54]. Elevated FMN levels correlate with oxidative stress and impaired mitochondrial respiration, making FMN a sensitive and early marker of graft viability during HOPE [14]. This has been validated in a recent multicenter study, which demonstrated that FMN outperforms traditional donor and recipient risk factors in predicting graft loss after transplant [15]. More recently, FMN quantification has also been described in NMP perfusate and bile, opening the door to validation in normothermic settings and potentially at the donor side during normothermic regional perfusion [55].

TABLE 2.

Viability assessment during hypothermic oxygenated machine perfusion.

Author, year	Biomarkers	Timing
Muller 2019	FMN	30 min
Schlegel 2020	FMN (< 8000 AU). NADH (< 10 000 AU)	30 min
Patrono 2020	Perfusate lactate, AST, ALT, LDH, glucose, and pH during DHOPE	30 min
Eden 2023	FMN (< 6000 AU). NADH (< 8.000 AU)	30 min

Abbreviation: AU, arbitrary units.

Thanks to the possibility of assessing liver viability, MP has been used to recover previously discarded (“orphan”) livers, predominantly from DCD donors, with utilization rates ranging from 47% to 100% (Table 3), highlighting its significant potential to expand the donor pool if more widely adopted [18, 29, 30, 38, 39, 40, 49, 50, 56, 57].

TABLE 3.

Studies in which machine perfusion has been used to recover discarded or “orphan” liver grafts.

Author, year	N	MP technique	Device	Viability assessment	Use rate	PNF	1‐year graft survival	IC
Mergental et al. 2016	6	NMP	Liver Assist; OrganOx Metra	Yes	5/6 (83.3%)	0%	—	NA
Watson et al. 2017	12	NMP	Liver Assist	Yes	NA	1/12 (8.3%)	—	3/12 (25%)
Watson et al. 2018	47	NMP	Liver Assist	No	22/47 (46.8%)	1/22 (4.5%)	NA	4/22 (18.2%)
De Vries et al. 2019	7	DHOPE‐COR‐NMP	Liver Assist	Yes	5/7 (71.4%)	0%	NA	0%
Matton et al. 2019	6	NMP	Liver Assist	Yes	4/6 (66.7%)	0%	NA	0%
Mergental et al. 2020	31	NMP	Organox	Yes	22/31 (71%)	0%	86%	4/22 (18.2%)
Reiling et al.2020	10	NMP	Organox	Yes	10/10 (100%)	0%	NA	0%
Quintini et al. 2022	21	NMP	Homegrown device	Yes	15/21 (71.5%)	15/21 (71.5%)	NA	1/15 (6.7%)
Van Leeuwen et al. 2022	54	DHOPE‐COR‐NMP	Liver Assist	Yes	34/54 (63%)	34/54 (63%)	94%	1/34 (2.9%)
Morawski et al. 2025	5	NMP	Liver Assist	Yes	5/8 (62.5%)	1/5 (20%)	NA	NA

Abbreviations: COR, controlled oxygenated rewarming; DHOPE, dual hypothermic oxygenated machine perfusion; IC, ischemic cholangiopathy; MP, machine perfusion; NMP, normothermic machine perfusion; PNF, primary non‐function.

Despite these promising preliminary results, however, current viability protocols have multiple limitations. Many include subjective elements such as liver appearance and perfusion homogeneity [30, 38]. Core parameters like lactate clearance and bile production suffer from limited specificity. Most published protocols are single‐center studies with small sample sizes and lack external validation, limiting reproducibility. Proposed timing and threshold values appear to be device‐specific, further complicating comparisons between different studies. Additionally, despite careful assessment, major complications such as ischemic cholangiopathy, primary non function (PNF), and graft loss have not been completely avoided [18, 43, 44]. These limitations underscore the need for standardized, externally validated, and reliable protocols for assessing liver viability.

3. What Are We Searching for: Novel Biomarkers

Recent studies have explored novel potential biomarkers measurable during MP, with the aim of improving graft viability assessment and enhancing prediction of post‐transplant outcomes (Table 4). These efforts have encompassed a wide array of biological targets, including metabolic and inflammatory markers, transcriptomic signatures, and mitochondrial function parameters. With the exception of three studies based on data from randomized controlled trials [58, 59, 60], most studies are observational and characterized by small sample sizes and the use of weak clinical endpoints [45, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70].

TABLE 4.

Studies investigating new biomarkers during machine perfusion in human liver allografts.

Author, year	MP	Model	n	Endpoint	Technology used	Readiness	Observations
(A) Metabolic biomarkers
Meszaros et al. 2022	NMP	Observational prospective	50 (DBD, n = 37; DCD, n = 13)	L‐GrAFT	High‐resolution respirometry	Promising	Mitochondrial respiration assessment during the early phase (first 6 h) of NMP correlated with L‐GrAFT score
Patrono et al. 2022	DHOPE	Observational prospective	10 (DBD, n = 9; DCD, n = 1)	L‐GrAFT	Microdialysis	Promising	Microdialysate 2 h‐glucose and lactate levels correlated with L‐GrAFT score
Guo et al. 2023	IFLT	Observational prospective	29 (DBD, n = 29)		Liquid chromatography‐mass spectrometry (UHPLC system)	Exploratory	Identification of different metabolic profiles of standard and extended‐criteria DBD
Zhylko et al. 2025	DHOPE	Randomized prospective	26 (DBD, n = 26)	EAD	Basic laboratory equipment	Clinically available	Perfusate lactate correlates with EAD, post‐LT ALT peak and hospital length of stay
(B) Inflammatory/immunological biomarkers
Basta et al. 2023	NRP	Randomized prospective	19 (DCD, n = 19)		ELISA, fluorescence spectroscopy, RT‐qPCR, immunoassays (MILLIPLEX)	Promising	Identification of different profiles of liver injury, inflammation, regeneration and metabolism between type 2 and 3 DCD livers
Mathis et al. 2024	NMP	Observational prospective	79 (DBD, n = 62; DCD, n = 15)	PRS	ECL assay	Promising	IL‐6 at the end of NMP correlated with catecholamine requirement after reperfusion
Irsara et al. 2025	NMP	Observational retrospective	16 (DBD, n = 12; DCD, n = 4)		ELISA	Promising	sPD‐L1 levels in NMP perfusate correlated with acute‐phase proteins and liver injury markers
Pezzati et al. 2025	NMP	Observational retrospective	30 (DBD, n = 16; DCD, n = 14)	Graft loss	ELISA	Promising	IL‐6 correlated with graft loss and bilirubin peak within POD 7, TNF‐α correlated with INR peak within POD 7, IL‐10 correlated with ICU stay
(C) Cellular injury biomarkers
Weissenbacher et al. 2022	NMP	Observational prospective	55 (DBD, n = 50; DCD, n = 5)	EAD, MEAF	Basic laboratory equipment	Clinically available	ALP, AST, ALT, and LDH trends correlated with MEAF score. vWF trend correlated with EAD
Sorbini et al. 2024	DHOPE	Observational prospective	45 (DBD, n = 41; DCD, n = 4)	EAD	Digital droplet PCR, fluorescence spectroscopy	Promising	Mt‐cfDNA levels in DHOPE perfusate correlated with FMN levels and predicted EAD and severe in‐hospital complications
Maeda et al. 2025	NMP	Observational retrospective	23 (DBD, n = 5; DCD, n = 18)	EAD	Basic laboratory equipment	Clinically available	Perfusate hemoglobin levels correlated with EAD and cytolysis markers
Rauter et al. 2025	DHOPE	Observational retrospective	40 (DBD, n = 38; DCD, n = 2)	EAD	ELISA	Promising	Syndecan‐1 (marker of glycocalyx degradation) correlated with EAD
Basta et al. 2025	NRP + NMP/HOPE	Randomized prospective	57 (DCD, n = 57)	EAD	Immunoassays (MILLIPLEX)	Promising	Arginine‐1 levels at the end of MP correlated with EAD and were higher in discarded grafts
(D) Transcriptomic biomarkers
Hautz et al. 2024	NMP	Observational prospective	50 (DBD, n = 33; DCD, n = 12)	EAD	RT‐qPCR	Promising	Different transcriptomic signatures during NMP in transplanted vs. discarded livers, and in EAD vs. non‐EAD cases
(E) Imaging‐based biomarkers
Fodor et al. 2022	NMP	Observational prospective	21 (DBD, n = 15; DCD, n = 6)		Contactless and non‐ionizing radiation imaging system (TIVITA)	Exploratory	Near‐infrared perfusion index and tissue water index correlated with perfusate lactate and pH

Abbreviations: DBD, donation after determination of death by neurological criteria; DCD, determination of death by circulatory criteria; DHOPE, dual hypothermic oxygenated machine perfusion; EAD, early allograft dysfunction; ECL, electrochemiluminescence; ELISA, enzyme‐linked immunosorbent assay; IFLT, ischemia‐free liver transplantation; L‐GrAFT, liver graft assessment following transplantation score; LT, liver transplantation; MEAF, model for early allograft failure; MP, machine perfusion; Mt‐cfDNA, mitochondrial cell‐free DNA; NMP, normothermic machine perfusion; NRP, normothermic regional perfusion; PCR, polymerase chain reaction; PD‐L1, programmed death‐ligand 1; POD, postoperative day; PRS, post‐reperfusion syndrome; UHPLC, ultra‐high pressure liquid chromatography; vWF, Von Willebrand factor.

Unsurprisingly, NMP has emerged as the preferred platform for investigating liver graft metabolism and function. The analysis of the inflammatory response during NMP has revealed correlations between cytokine release and postoperative outcomes, suggesting that characterizing the inflammatory response to ischemia–reperfusion injury (IRI) may provide a more accurate and earlier assessment of graft viability. In particular, IL‐6 [66, 69] and its downstream or cross‐regulated effectors, such as IL‐10, TNFα [69], and PD‐L1 [68], have been shown to correlate with markers of injury and may predict clinically relevant outcomes such as post‐reperfusion syndrome (PRS), graft loss, and hospital length of stay. Pezzati et al. [69] suggested that such predictive information could be obtained as early as during the second hour of NMP. Additionally, through transcriptomic profiling or immunoassays, it may be possible to define a “metabolic signature” that significantly distinguishes high‐quality grafts from those ultimately discarded, thereby enabling risk stratification [60, 64, 65].

Despite a strong mechanistic rationale, the main limitation of the aforementioned studies is the reliance on specialized methodologies (e.g., high‐resolution respirometry, RT‐qPCR, electrochemiluminescence) that can be implemented only in a minority of centers. As such, they fail to offer real‐time, actionable data that can be readily obtained and interpreted during routine clinical perfusion and be easily implemented across transplant centers.

Significantly fewer studies have explored novel biomarkers during HOPE [58, 63, 67, 70], likely due to the metabolically inactive state of the liver during cold perfusion, which substantially limits the spectrum of biomarkers that can be analyzed. Our group has proposed the use of microdialysis catheters to sample the extracellular space during HOPE, observing that glucose and lactate levels measured in the dialysate correlated with L‐GrAFT score postoperatively [63]. However, this technique may be subject to the same criticism outlined above—namely, that it does not constitute a “user‐friendly” or easily accessible method in a standard operating room. As a more readily implementable approach, we reported that mitochondrial cell‐free DNA measured in HOPE perfusate could serve as a predictor of postoperative graft function and complications [67]. More recently, Zhylko et al. [58] proposed perfusate lactate, a widely accepted viability biomarker during NMP, as an indicator of graft function and damage also during HOPE. However, a convincing mechanistic explanation backing their observations is lacking, and more robust data are required to validate these findings.

Despite the increasing interest in viability assessment, it is important to emphasize that none of the reported novel biomarkers has yet transitioned from a research focus to widespread adoption in clinical practice.

4. Discussion

Despite its limitations, MP technology has already revolutionized liver transplantation by expanding the boundaries of organ utilization. The increase in usable organs is no longer a distant prospect but a clinical reality. Given the increased utilization rates worldwide, it is likely that many potentially transplantable organs were historically discarded due to limitations in preservation and assessment methods. The possibility to rely on better preservation modalities and to obtain data on liver damage and function has increased the propensity to use more marginal organs, a paradigm shift that would have been inconceivable without advanced technologies enabling real‐time organ evaluation. The revolution has been so profound that nowadays many transplant professionals would consider it unethical to transplant a high‐risk liver without any form of viability assessment, even if this assessment remains imperfect and it is still evolving.

Two alternative approaches to viability assessment are predominant nowadays that is, monoparametric and multiparametric models. Multiparametric models, like Birmingham [17, 18, 32] and Groningen [29, 42] protocols, integrate functional (e.g., lactate clearance, bile production) and structural (e.g., bile pH, bicarbonate) criteria, reflecting the complexity of organ viability. These criteria are sometimes hierarchically organized in major and minor criteria. While this multiparametric strategy increases flexibility, it comes at the cost of specificity and simplicity, often leaving clinicians uncertain about how to proceed. These models also lack robust multicenter, prospective validation, and may not be universally transferable across different perfusion devices. Importantly, complications can still arise even when all viability criteria are met [17, 18, 28], highlighting the need for more reliable markers. Notably, van Leeuwen et al. [29] acknowledged that final transplant decisions are often based not only on viability scores but also on clinical judgment and case urgency, an approach that reintroduces the subjectivity these models wanted to eliminate. In contrast, monoparametric models, like FMN‐based assessment during HOPE, provide clear and reproducible thresholds that can reduce personal variability and improve decision‐making. However, they are also highly sensitive: a single elevated FMN value could lead to the unnecessary non‐use of an otherwise viable organ. Moreover, FMN protocols have thus far only been validated retrospectively and in a small number of centers, lacking prospective multicenter confirmation [14, 15, 19]. A further concern is that in the event of technical failure or inability to measure FMN, clinicians are left with no alternative parameters, potentially reverting to traditional subjective assessments, which could be particularly difficult for teams trained to rely on this biomarker.

Regardless of the approach, two intertwined concepts—namely, the definition of viability and that of acceptable risk—need to be clarified to effectively advance the field. Until now, viability has been identified as a binary (yes or no) feature, which was likely linked to the necessity to come to a binary decision (to transplant or not to transplant). Therefore, viable has been used as a synonym for transplantable. However, a clear definition of a viable liver is currently lacking. If we intend viability as the capacity of a liver to function in the recipient and to avoid ischemia–reperfusion injury‐related complication (e.g., ischemic cholangiopathy), then viability is probably a continuous gradient rather than a binary characteristic. Furthermore, liver transplantation inherently carries risk, and clinical outcomes also heavily depend on recipient factors. Striving to develop a perfectly sensitive and specific protocols would probably represent an elusive aim. To establish a viability threshold, one possible approach would be to define as “viable” any liver whose MP parameters are associated with a risk comparable to that of a standard liver. Alternatively, a viable graft may be defined as one characterized by a risk estimate that falls within the range of outcomes deemed acceptable by current transplantation benchmarks. This approach would not be new, as it was the concept behind the development of Milan criteria for LT for HCC [71]. These criteria, which were successively refined by applying a so‐called metroticket paradigm [72], were not defined aiming for zero HCC recurrence but rather to allow a comparable survival as for others non‐HCC LT indications. Additionally, acceptable risk thresholds could vary in different healthcare systems according to resources, organ shortage, patient population, and waiting list pressure. Applying a similar framework to develop new viability criteria would set realistic goals and avoid too restrictive criteria, possibly leading to the non‐use of potentially transplantable livers [73].

The role of viability assessment also requires clarification regarding whether it should overrule all the other considerations or be integrated into a comprehensive risk profile that includes recipient factors, possibly with the help of machine learning algorithms. A precious lesson learned from prognostic models [25, 34, 35] is that the success of LT depends on the interplay between donor, recipient, and procedural variables. If we intend viability as a continuum, it is expected that recipient factors, like the severity of liver disease, influence clinical outcomes. Therefore, a clinically useful viability model should define whether a particular liver is likely to function (or be “viable”) for a particular recipient. In this view, MP parameters would add significant information to the process of donor‐recipient matching but should be interpreted in the light of recipient status and procedural complexity.

In summary, MP has fundamentally shifted the landscape of LT, but properly designed studies will be essential to refine viability criteria, balance complexity with practicality, and establish meaningful, patient‐centered endpoints (Figure 1). Reliable biomarkers are urgently needed to refine current assessment protocols, as reflected by the burgeoning of studies on the subject. Although many of these exploratory studies rely on advanced technologies that are not currently feasible for broad clinical implementation due to cost, complexity, or required expertise, the rapid evolution of biomedical technology holds the potential to significantly change this landscape. As tools become more accessible, automated, and integrated into clinical workflows, some of these biomarkers may eventually become routine components of graft evaluation during machine perfusion in the near future.

FIGURE 1.

A proposed framework to overcome limitations of current liver viability assessment protocols. [Color figure can be viewed at wileyonlinelibrary.com]

Author Contributions

D.P.: concept and design, article drafting and revision; R.P. and N.D.S.: literature review and article drafting; R.R.: critical revision of the article.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.