Effluent Molecular Analysis Guides Liver Graft Allocation to Clinical Hypothermic Oxygenated Machine Perfusion

Hypothermic-oxygenated-machine-perfusion (HOPE) allows assessment/reconditioning of livers procured from high-risk donors before transplantation. Graft referral to HOPE mostly depends on surgeons’ subjective judgment, as objective criteria are still insufficient. We investigated whether analysis of effluent fluids collected upon organ flush during static-cold-storage can improve selection criteria for HOPE utilization. Effluents were analyzed to determine cytolysis enzymes, metabolites, inflammation-related mediators, and damage-associated-molecular-patterns. Molecular profiles were assessed by unsupervised cluster analysis. Differences between “machine perfusion (MP)-yes” vs. “MP-no”; “brain-death (DBD) vs. donation-after-circulatory-death (DCD)”; “early-allograft-dysfunction (EAD)-yes” vs. “EAD-no” groups, as well as correlation between effluent variables and transplantation outcome, were investigated. Livers assigned to HOPE (n = 18) showed a different molecular profile relative to grafts transplanted without this procedure (n = 21, p = 0.021). Increases in the inflammatory mediators PTX3 (p = 0.048), CXCL8/IL-8 (p = 0.017), TNF-α (p = 0.038), and ANGPTL4 (p = 0.010) were observed, whereas the anti-inflammatory cytokine IL-10 was reduced (p = 0.007). Peculiar inflammation, cell death, and coagulation signatures were observed in fluids collected from DCD livers compared to those from DBD grafts. AST (p = 0.034), ALT (p = 0.047), and LDH (p = 0.047) were higher in the “EAD-yes” compared to the “EAD-no” group. Cytolysis markers and hyaluronan correlated with recipient creatinine, AST, and ICU stay. The study demonstrates that effluent molecular analysis can provide directions about the use of HOPE.


Introduction
The adoption of expanded criteria for acceptance of organ donors broadened the pool of livers available for transplantation. However, relative to standard livers, organs pro-The grafts were allocated to our center according to the North Italian Transplant program allocation policy. Our surgical team procured all grafts, with the exception of 4 national or interregional allocated livers [43]. Alternative procedures for liver graft procurement were applied in DCD and DBD donors [44]. Normothermic regional perfusion (NRP) was applied after death declaration and before organ procurement in all DCD donors according to the Italian law. It consists of abdominal extracorporeal membrane oxygenation directly applied in the donors though the placement of femoral cannulas after death declaration [45].
All grafts underwent in situ flushing with Celsior solution (IGL) and static preservation in an ice-box.

Liver Machine Perfusion (MP)
Liver MP was applied to all DCD grafts and to DBD extended-criteria donor (ECD) organs showing one of the following characteristics: expected prolonged ischemia time (>10 h), macrovescicular steatosis >30%, or serum levels of hepatonecrosis markers exceeding 4 times the reference range.
MP was performed as back-to-base, end-ischemic dual hypothermic oxygenated machine perfusion (D-HOPE). Briefly, D-HOPE was performed using the Liver Assist device (Organ Assist, Groningen, The Netherland), and consisted in a double perfusion (4000 mL of circulating Belzer MPS-UW Machine Perfusion) through the portal vein and hepatic artery with pressures set at 4 and 25 mmHg, respectively. Grafts were actively oxygenated (FiO 2 = 100%) at a temperature of 8-10 • C.

Liver Transplantation (LT) and Classification of Post LT Complications
Liver transplantation (LT) was performed with a modified piggyback technique and bile duct anastomoses over a T-tube. Post-LT course and biochemical analysis were performed (liver and kidney function test, coagulation, blood count).
Early allograft dysfunction (EAD) was defined as: peak transaminases > 2000 IU/L (within the first week after LT), international normalized ratio > 1.7, or bilirubin > 10 mg/dL on postoperative day (POD) 7 [46]. Primary non-function (PNF) was defined as a nonrecoverable graft function needing urgent re-LT < 10 POD (AST > 5000 UI/L, INR > 3.0, and metabolic acidosis) with no sign of vascular pathology of the implanted liver [47]. Acute kidney injury (AKI) was defined and staged according to KDIGO criteria; we considered patients as affected by post-LT AKI only if assigned to ≥stage 2 [48].

Effluent Processing and Molecular Analysis
Samples of effluent were processed to obtain cell-free supernatants and donor-derived leukocyte pellets (the protocol is described in detail in [49][50][51]). Effluents collected during back-table procedure before MP were analyzed to measure the concentration of cytolysis enzymes, flavin mononucleotide (FMN), caspase-cleaved keratin 18 (CK18), inflammatory mediators, metabolites, and electrolytes.

Flavin Mononucleotide (FMN)
The FMN concentration was determined by fluorescence spectroscopy. Briefly, 5 mg of Riboflavin 5 -monophosphate sodium salt hydrate (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany) were dissolved in either 0.9% sodium chloride or Celsior. Thereafter, these solutions were serially diluted to obtain different standard FMN solutions with concentration ranging from 31.25 to 1000.00 ng/mL (Supplementary Figure S2). Effluent samples were then dispensed in triplicate in black microplates, together with the standard samples. Fluorescence readings were performed using a multi-mode microplate reader (Synergy HTX, Biotek U.S, Winooski, VT, USA). A monochrome light with excitation wavelength of 460/40 nm was used, while fluorescence emission was revealed with 100% gain at 528/20 nm. Finally, the average fluorescence readings of effluent samples were plotted against the appropriate standard curve to calculate sample FMN concentration.

Soluble Caspase-Cleaved Keratin (CK) 18
Hepatocyte apoptosis was assessed by measuring the release of caspase-cleaved CK18 with a commercially available immunoassay based on the use of M30 monoclonal antibody that specifically detects the neoepitope ccK18/K18-Asp396 (M30-Apoptosense ELISA kit, PEVIVA, VLVbio-BioScientific Pty. Ltd., Stockholm, Sweden). The analysis was performed following manufacturers' instructions and the absorbance was determined using a microplate reader (Synergy HTX).

Leukocyte Count and Free Hemoglobin
Cell pellets were suspended in 0.25 mL 1X PBS solution (Sigma-Aldrich) and counted using an automated cell counter (Scepter, Millipore Corporation, Billerica, MA, USA).
Free hemoglobin concentration was assessed by applying the Allen correction to absorbance readings at 563 nm, 577 nm, 600 nm (Synergy HTX).

Statistical Analysis
Statistical analysis was performed using the JMP Pro 15 software (© SAS Institute Inc., Cary, NC, USA). Box plots and linear regression plots were created using SigmaPlot 11.0 (Systat Software Inc., San Jose, CA, USA).
Continuous variables are presented as median  or absolute values with percentage, while categorical variables were expressed in percentages. To reduce the distribution effect of the flush solution volume and to normalize data according to liver size, all the measured variables are shown as total release per gram of liver graft.
Multivariate analysis was used to explore associations across all the variables evaluated in effluent samples.
Finally, correlation between effluent variables and transplantation outcome was investigated using the Pearson's correlation coefficient.
A probability value < 0.05 was considered significant.

Donor Characteristics and Liver Graft Allocation to the MP Procedure
Donor data and procurement timings are shown in Table 1. Ten (20%) grafts were derived from DCD donors (7 category III and 3 category II) and 39 (80%) from DBD donors. The total preservation time lasted 565 min (510-660). All DCD liver grafts showed a highrisk profile, according to the UK-DCD score [18]. Consistently, their median DRI was 2.21 (1.85-2.37), which was higher than the DRI of 1.6 observed in DBD livers. In addition, six grafts showed a macrovescicular steatosis of >30%.
Eighteen (47%) grafts were subjected to MP, where 10 were procured from DCD and 8 from DBD-ECD donors meeting the criteria specified above, except for livers 18 and 19, which were subjected to MP to extend the preservation time because of logistic reasons (change of recipient) [36,45,52]. Table 1. Donor and recipient characteristics. Data are presented as median  or n (%), n = 49. Abbreviations: BMI, body mass index; ICU, Intensive care unit stay; DCD, cardiac death donors; DBD, brain death donors; WIT, warm ischemia time; cDCD, controlled or type III DCD; uDCD, uncontrolled or type II DCD; DHOPE, dual hypothermic oxygenated machine perfusion; NRP, normothermic machine perfusion; ECD, extended criteria donors; MELD, model for end-stage liver disease; HCC, hepatocellular carcinoma; PNF, primary non function; EAD, early allograft dysfunction; AKI, acute kidney injury; AST, aspartate aminotransferase; ALT, alanine transferase; re-LT, liver re-transplantation; * only for DCD donor.

Recipients' Characteristics and Outcome of Transplantation
The mean recipient model for end stage liver disease (MELD) score was 13 (9-17) points, while 11% of the patients received an urgent allocated graft (Table 1). Pre-LT chronic renal failure was diagnosed in four (11%) cases. One-year graft survival was 90% (44/49) with a patient survival of 92% (45/49). Two recipients died (sepsis, intracranial hemorrhage) within 10 days after transplantation. One patient was re-transplanted because of a hepatic artery thrombosis.
Despite an overall high-risk population in our cohort, graft survival, censored for tumor-related death, remained comparable to the levels seen in an ideal LT cohort presented in a recent benchmark study ( Table 1, no cases of PNF, 1 case of ischemic type biliary lesion in a DBD graft) [53]. EAD was diagnosed in 7 cases and 14 patients developed AKI. Finally, a strong association was observed between FMN and free hemoglobin (r = 0.781, p < 0.001).

Differences in the Bio-Molecular Profiles of Effluent Fluids Collected from DCD and DBD Livers
Data reported in Table 2 indicate that livers procured from DCD and DBD donors show different effluent molecular profiles. Among differentially expressed mediators, there were molecules relevant to inflammation (CCL2/MCP-1, CXCL8/IL-8, CXCL9/MIG, TNF-α, ANGPTL4), hepatocyte apoptosis (CK18), and coagulation (Protein C, D-Dimer). Of note, concentration of the anti-inflammatory cytokine IL-10 was lower in effluent fluids of DCD compared to those obtained from DBD livers. Differences in BUN and K+ levels were likewise observed.
Finally, no relationship was found between the concentration of lactate, AST, and ALT assessed in NRP perfusates and the release of these factors in effluent fluids.  Table 2 indicate that livers procured from DCD and DBD donors show different effluent molecular profiles. Among differentially expressed mediators, there were molecules relevant to inflammation (CCL2/MCP-1, CXCL8/IL-8, CXCL9/MIG, TNF-α, ANGPTL4), hepatocyte apoptosis (CK18), and coagulation (Protein C, D-Dimer). Of note, concentration of the anti-inflammatory cytokine IL-10 was lower in effluent fluids

Association with Transplantation-Related Complications and Recipient Outcome
Effluent concentration of cytolysis markers AST, ALT, and LDH was significantly higher in the "EAD-yes" group ( Figure 3). Conversely, there was an increased release of K+ in the samples from the "EAD-no" group (0.012 (0.010-0.012) mmol/g vs. 0.015 (0.012-0.016) mmol/g, p = 0.016). With regard to the other biomarkers assessed in effluent fluids, no significant difference was observed between the "EAD-yes" and the "EAD-no" groups (Supplementary Table S2 A further analysis investigated any potential association between the variables measured in the effluent fluids and clinical data related to transplantation outcome, including days of recipient hospitalization in intensive care unit (ICU) and biochemical parameters of liver and kidney function assessed in recipients' plasma samples (peak). Effluent biomarkers of liver damage correlated with recipient serum creatinine (LDH: r = 0.517, p = 0.001; AST: r = 0.538, p = 0.0007; ALT: r = 0.570, p = 0.0003), recipient serum AST (LDH: r = 0.745, p < 0.0001; AST: r = 0.726, p < 0.0001; ALT: r = 0.698, p < 0.0001), and ICU stay (LDH: r = 0.459, p = 0.001; AST: r = 0.400, p < 0.0001; ALT: r = 0.381, p = 0.029). Of interest, strong or moderate correlations were found between effluent HA and recipient creatinine, AST, and ICU stay (Figure 4). Finally, there were weaker positive associations between ICU stay and effluent concentration of ANGPTL4 (r = 0.532, p = 0.009) and CXCL8 (r = 0.329, p = 0.048).

Discussion
The present proof-of-concept study indicates that the analysis of effluent fluids collected from liver grafts during SCS can provide valuable information to guide surgeons in the decision making on D-HOPE application to marginal organs. In fact, the clinical decision to allocate specific livers to the MP procedure was retrospectively supported by a peculiar molecular profile in the liver effluents. In addition, a donor-specific signature was observed, with increased concentration of inflammation-and cell death-related factors in the effluent fluids collected from DCD grafts compared to those seen in DBD livers. Finally, an interesting correlation between the effluent concentrations of certain molecules and EAD as well as recipient clinical data was observed.
Higher numbers and quality of donor organs can be lifesaving for many patients waiting for a transplant. The MP technique allows a possible evaluation of viability and function as well as reconditioning of suboptimal grafts before transplantation. However, there are no reliable criteria to determine whether a marginal liver is immediately suitable for transplantation or if it rather might benefit from further treatment with MP. Therefore, the decision to perfuse a specific liver or not is currently based on a few objective criteria, on the center-specific policy, and on surgeons' individual "gut feeling" [13].
The assessment of specific parameters in effluent fluids could reduce the uncertainty about liver viability and would facilitate the preoperative estimation of graft quality. In fact, such biological material has a huge diagnostic potential as it contains molecules and cells from the graft, challenged by the noxious stimuli inherent in the transplantation procedure, from the donor treatment withdrawal and surgery throughout the entire preservation period until implantation. Of note, the effluent analysis provides information on the entire liver parenchyma rather than on a small part of the hepatic tissue, which is a clear advantage compared to liver biopsies. Consistently, several researchers demonstrated that the effluent composition can reveal both pre-existing injury, including brain death-and steatosis-related inflammation, and the damage acquired during procurement and preservation [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36].
We therefore hypothesized, that the molecular analysis of effluents could help with the selection of marginal livers, that may require a more accurate assessment and

Discussion
The present proof-of-concept study indicates that the analysis of effluent fluids collected from liver grafts during SCS can provide valuable information to guide surgeons in the decision making on D-HOPE application to marginal organs. In fact, the clinical decision to allocate specific livers to the MP procedure was retrospectively supported by a peculiar molecular profile in the liver effluents. In addition, a donor-specific signature was observed, with increased concentration of inflammation-and cell death-related factors in the effluent fluids collected from DCD grafts compared to those seen in DBD livers. Finally, an interesting correlation between the effluent concentrations of certain molecules and EAD as well as recipient clinical data was observed.
Higher numbers and quality of donor organs can be lifesaving for many patients waiting for a transplant. The MP technique allows a possible evaluation of viability and function as well as reconditioning of suboptimal grafts before transplantation. However, there are no reliable criteria to determine whether a marginal liver is immediately suitable for transplantation or if it rather might benefit from further treatment with MP. Therefore, the decision to perfuse a specific liver or not is currently based on a few objective criteria, on the center-specific policy, and on surgeons' individual "gut feeling" [13].
The assessment of specific parameters in effluent fluids could reduce the uncertainty about liver viability and would facilitate the preoperative estimation of graft quality. In fact, such biological material has a huge diagnostic potential as it contains molecules and cells from the graft, challenged by the noxious stimuli inherent in the transplantation procedure, from the donor treatment withdrawal and surgery throughout the entire preservation period until implantation. Of note, the effluent analysis provides information on the entire liver parenchyma rather than on a small part of the hepatic tissue, which is a clear advantage compared to liver biopsies. Consistently, several researchers demonstrated that the effluent composition can reveal both pre-existing injury, including brain death-and steatosis-related inflammation, and the damage acquired during procurement and preservation [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35].
We therefore hypothesized, that the molecular analysis of effluents could help with the selection of marginal livers, that may require a more accurate assessment and reconditioning with MP. To test this idea, we performed an analysis of effluent samples collected from grafts either subjected to MP based on the surgeon's judgment or transplanted without this procedure. In addition to cytolysis enzymes and metabolites, we investigated the concentration of inflammation-related mediators and DAMPs, previously shown to be accurate markers of brain death-induced inflammation and damage, liver IRI, and poor transplantation outcome [38][39][40].
Our study discloses a distinctive molecular signature in the effluents of livers referred to the D-HOPE procedure. The most remarkable observation is the reduced concentration of the anti-inflammatory mediator IL-10 in fluids of grafts, which underwent MP, compared to those collected from livers immediately used for transplantation. IL-10 is a pleiotropic anti-inflammatory cytokine expressed in different liver cell types, including hepatocytes, Kupffer cells, sinusoidal endothelial cells, type 2 T-helper (Th2) cells, and lymphocytes [54]. Of note, recent evidence indicated IL-10 as a master regulator of macrophage efferocytosis and phenotypic conversion in sterile liver injury, such as IRI [55]. In particular, this molecule induces macrophage switching from a proinflammatory to a restorative phenotype and a low concentration of IL-10 was associated with failure of debris uptake, prolonged liver damage, aberrant hepatocyte proliferation and fibrosis [55,56]. Another interesting result of our research concerns the higher release of inflammation-related mediators, such as PTX3 and CXCL8/IL-8, from livers subsequently subjected to MP. Beside activated leukocytes, DAMPs and inflammatory mediators are released by stressed and dying hepatocytes. IL-8/CXCL-8 promotes neutrophil and macrophage migration to the liver and regulates hepatocyte survival and proliferation upon IR injury [57]. Of note, increased serum IL-8/CXCL-8 concentration was associated with higher serum transaminases in patients who received a liver transplantation [27]. PTX3 is a pattern-recognizing protein produced by macrophages and hepatic stellate cells in response to proinflammatory signals and Toll-like receptor engagement [58]. In liver pathology, the detection of an elevated blood concentration of PTX3 is an index of hepatic disease [38,59]. These observations collectively show that in our case series, livers subjected to MP were actually associated with a proinflammatory and non-resolutive effluent profile compared to grafts not referred to D-HOPE. Interestingly, clinical [8] and preclinical [60] studies showed that the application of D-HOPE can efficiently modulate such inflammatory and cell death pathways activated by perimortem and donation events and exacerbated by high-risk donor characteristics. The beneficial effect of D-HOPE is based on the induction of mitochondrial Complex I and II reprogramming [10,61], with subsequent improved function of the entire respiratory chain, improved Succinate metabolism, and enhanced ATP production. This ultimately leads to reduced reperfusion injury and lower risk of post-LT complications [10].
With regard to traditional markers of liver injury, there was no significant difference in effluent concentration of AST, ALT, and LDH of livers referred to MP, compared to grafts not subjected to ex situ perfusion. Of note, the present study includes the evaluation of FMN, a novel biomarker of cell viability, whose concentration is measured in perfusate samples to assess liver graft mitochondrial function throughout HOPE [62]. Although increasing data support its robustness and reliability in the prediction of organ viability during oxygenated ex situ perfusion [62,63], our results suggest that the evaluation of effluent FMN during SCS could provide different information, probably due to the absence of an active oxygenation and dynamic perfusion. Of interest, the multivariate analysis indicated a positive association between FMN and free hemoglobin concentration. Targeted studies are needed to determine the biological significance of FMN release during donor management and the preservation phase.
Additional analysis documented a peculiar inflammation, cell death, and coagulation profile in fluids collected from DCD livers, compared to those obtained from DBD grafts. This condition could reflect the detrimental influence of prolonged WIT secondary to the 20 min no-touch period, which is mandatory in the Italian setting [45]. Moreover, the marked inflammatory signature observed in our DCD grafts could be a consequence of the NRP procedure, consisting of an oxygenated in situ perfusion, applied following cardiac arrest in DCD organ donation. In fact, in other case-series, where livers were procured immediately after death declaration without NRP [64,65], DCD grafts showed increased necrosis-related biomarkers compared to DBD organs, but no sign of enhanced inflammation. Our data further support the diagnostic power of effluent analysis in liver transplantation, suggesting that it could be used to assess both the degree and the nature of injury in each graft type.
Finally, correlations between recipient short-term outcome and effluent concentration of either cytolysis biomarkers or HA were observed. Along with biliary complications and AKI, EAD is a post-LT complication [66] mainly consequent to IRI-induced hepatocellular injury, endothelial damage, and inflammatory cell activation [67]. In our case series, all evaluated markers of hepatocellular damage correlated with EAD, consistent with previous reports [19,20,[23][24][25]29,31,33]. In addition, effluent enzyme levels were positively associated with recipient parameters of liver and kidney function and hospitalization duration. Concerning HA, significant correlations were found between its effluent concentration and recipient blood creatinine, AST and ICU stay. HA is normally taken up from the circulation and metabolized by hepatic microvascular endothelial cells. Therefore, increased effluent HA concentration can denote a loss of integrity or functional failure of sinusoidal cells. Our results are consistent with previous studies, where HA uptake measured in the caval effluent during recipient operation [22,37] or in back-table effluents [29] was lower in patients with poorer early graft function following LT.
A significant limitation in the present study resides in the small number of liver transplantations performed in our center, which limits the validation of each of the variables assessed in effluent fluids. In addition, based on the excellent outcomes with a low number of complications, despite the high donor risk, we were not able to correlate effluent molecular analysis with recipient outcomes. Targeted studies are required to better characterize the prognostic ability of effluent biomarkers to predict the occurrence of post-transplant complications.

Conclusions
The present proof-of-concept study shows that molecular analysis of effluent samples can provide objective criteria reflecting the cumulative effects of noxious stimuli affecting liver graft quality during the organ donation process. We suggest that bio-molecular stratification of livers would significantly contribute to implement the clinical strategy with or without application of the MP procedure, leading to improved, cost-saving, and tailored healthcare resource utilization based on the specific injury experienced by each graft. In addition, marginal organs that are currently deemed unsuitable because of the presumed high-risk to develop complications could be successfully used in case a "healthy" biomarker profile is found. Of interest, novel technologies can enable surgeons to perform real-time evaluation of effluent fluids during back table preparation [68,69], with results available in time to support the decision making of the surgeon and the entire surgical team. This perspective could pave the way for the adoption of a personalized medicine-based approach in the field of organ transplantation, which would ultimately lead to improved donor-recipient matching and to better post-transplant results. Informed Consent Statement: Patients signed an informed consent form for data usage, while they did not receive any exam and treatment different from clinical routine or international guidelines.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Acknowledgments:
The authors would like to thank all surgeons, nurses, and transplant staff for their valuable support and Andrea Carlin for the technical support.

Conflicts of Interest:
The authors have no conflict of interest to declare.