Hypothermic Oxygenated Perfusion (HOPE) of The Pancreas

At the end of 90min DWIT all pancreas were flushed and stored in UW-solution for 5hrs. During later bench preparation another flush with 500ml UW-solution was performed and organs were separated from the liver, the bowel and the spleen with meticulous closure of small vascular branches to ensure appropriate perfusion. Both arterial vessels, supra-mesenteric and splenic artery were cannulated and used as perfusion route. Portal- and splenic veins were kept untouched for a passive outflow. The Vitasmart^Ò (Bridge to life Ltd; Medica) was used for the 2-hour HOPE procedures with UW-Machine Perfusion Solution (UW-MPS). This pressure-controlled device is CE- marked for clinical liver and kidney perfusions. The circuit with the standard disposable available for clinical applications was used and includes pressure, flow and temperature sensors and provides an active perfusate oxygenation (Oxygenator: Euroset^Ò EU5054). The device monitors pressure, flow, resistance and temperature in real-time.

Endpoints

Perfusate oxygen levels were measured through blood gas analysis. Further parameters of perfusion quality were measured by the device. HOPE-perfusates were obtained to quantify perfusate levels of Flavin-Mononucleotide-Levels (FMN) and NADH using spectroscopy. Such parameters are increasingly discussed as surrogate of mitochondrial function and injury [35-37]. Based on previous studies with other organs, perfusates were obtained within the first 30min of HOPE. Shortly, triplicates of 50ml perfusate were pipetted into standard 96 vial plates with a dilution of 1:4. Perfusate FMN levels were measured using the standard technology of spectroscopy with an excitation and emission wavelength of 485nm and 528nm, respectively. The perfusate NADH-levels were determined with an excitation and emission wavelength of 360nm and 460nm, respectively. Further details regarding the methodology can be found elsewhere [16,38,39]. At the end of cold storage or additional HOPE the pancreas weight was measured and tissues were obtained for histological assessment. Standard processing with formalin and embedding procedures were performed. Three staining procedures were done: Hematoxylin-Eosin (H&E), Toll-like-receptor-4 (TLR-4; macrophages, dendritic cells, Lifespan Bioscience: LS-B2070) and von Willebrand factor (vWF; endothelial cells, DAKO: A0082). Quantifications of TLR-4 - positive and vWF-positive cells were determined by manual counting in 20 random visual fields per experimental group. All histological analyses were performed in a blinded fashion with respect to the experimental groups. Vessels were excluded from the analysis.

Statistics, Quality Control and Ethical Approval

Completeness, plausibility, and validity of the data were independently verified (by MFC, RP and AS). Continuous variables are demonstrated as median and interquartile range (IQR). Statistical analysis was performed using the non-parametric Mann- Whitney-Wilcoxon U-test (GraphPad Prism, version 7.0, San Diego, CA, USA). P-values of <0.05 were considered significant. The experiments were carried out according to European Union (EU) directive guidelines (2010/63/EU) and Italian legislation (DLgs 26/2014) at the Centro Polyvalent Florence University (Cubo; Viale Gaetano Pieraccini, 6, 50139 Firenze FI, Italy).

Results

What Is Available In The Literature?

In 1974, Eloy et al presented the first study with Normothermic Machine Perfusion (NMP) of a canine pancreas, performed for 100min to test the secretory pancreas function [8]. This was paralleled by a first experimental comparison of 24hrs hypothermic machine perfusion (HMP) and SCS with subsequent transplantation of the canine pancreas [9]. Although the outcome of this pioneer study was not very promising, it triggered a series of experimental studies in the following years and instigated a controversial discussion regarding the best possible perfusion concepts. A total of 23 studies, 15 with HMP and 11 with NMP, were reported in the literature. The majority included pancreases from brain death donors (DBD; n=18) exposed to long CS or HMP. While no study exists today with transplantation of perfused human pancreas, five studies utilized discarded human pancreases [29,40-43]. Only Hamaoui et al have assessed IRI-features during NMP after previous HMP or standard CS in discarded human pancreases. In this study, HMP was performed with active oxygenation (HOPE) and resulted in a better organ functionality (Table 1)[41]. Dogs (n=8) and pigs (n=9) were the most frequent species in non-human studies. Since the seventies, the following perfusion systems were referenced in the literature as being used for pancreas HMP including the Gambro^Ò, the RM3 Machine Perfusion Unit, the Max-100 perfusion device (Waters Medical System^Ò), the Waves kidney perfusion device or LifePort ^Ò kidney transporter [5,26,27,29,41].

Normothermic Pancreas Perfusion

The known advantage of Normothermic Machine Perfusion (NMP) as seen in other solid organs is the opportunity to assess viability. To perform the best possible NMP with near-physiologic conditions, an optimized equipment is needed. Organ positioning is one key feature to avoid pressure necroses particularly during prolonged perfusion as recently described with full and partial human livers [44,45]. The first experimental pancreas study with NMP was done as early as in 1974 [8]. Until today, two studies describe NMP in discarded human pancreas (Table 1) and nine experimental non-transplant studies were published with pancreases from pigs, dogs or rodents (Supplementary Table 2). Unfortunately, the studies with human organs lack a comparator or baseline control group [42,43]. Barlow et al performed a short (1- 2hrs) endischemic NMP of discarded DBD pancreases after 13hrs cold storage and demonstrated the link between a higher donor risk and perfusate amylase and lipase levels. This study also showed higher levels of insulin secretion from younger pancreas[42]. Three of five organs developed focal acinar necrosis and one showed extensive fat necrosis in histology. In contrast, Nassar et al performed NMP for 6hrs after shorter cold storage of only 4hrs with better results. Such human pancreas showed healthier acini at the end of 6hrs NMP with normal chromogranin staining (Table 1)[43]. Such findings parallel recent clinical studies with NMP in livers. Posttransplant results were superior with NMP when performed instead of cold storage compared to an endischemic approach [46]. NMP in the recipient center after transport led to a comparably high rate of non-anastomotic strictures in livers with higher risk, e.g., from DCD donors [47,48]. Animal studies further supported the recent findings with perfusion of human pancreases. Shorter endischemic NMP of 90min to 5hrs resulted in well-maintained tissue and limited inflammation[8,30,31,49-54]. Most studies included DBD pancreas and lack a high risk donor model. In-house devices and ECMO-based equipment were used by many. Various perfusate compositions were reported in the literature with autologous whole blood or washed erythrocytes as oxygen carriers in most recent studies [30,31,49]. Particularly in older studies the cold ischemia time before NMP is frequently not reported (Supplementary Table 2) [8,52-55].

What are Optimal Perfusion Conditions for Hypothermic Machine Perfusion?

Five transplant studies explored the effect of non-actively oxygenated, prolonged HMP (24-48hrs) with high perfusion pressures (30-60mmHg) [9,23,24,56]. Three studies showed an equal or even better recipient survival and pancreas function after 24-48hrs SCS compared to such high-pressure-HMP [9,23,24]. Already in 1975, Tersigni et al described improved posttransplant results with lower perfusion pressures of 10-25mmHg[25]. While a perfusion at too low pressures (e.g., 10mmHg) led to incomplete pancreas perfusion with limited protective effects, high pressures between 30 and 70 mmHg caused pancreatic edema [5,7,24,27]. Such findings were paralleled by Karcz et al, who showed good outcomes and a minimal weight gain with a pressure between 15- 23mmHg [57]. Mild or moderate edema may be even beneficial for islet digestion and isolation [49]. Most authors would probably agree with a peak perfusion pressure of 15-25mmHg during HMP [1,5,7]. Next, in addition to organ temperature and tissue quality, perfusion flow is related to the set pressures. Toledo-Pereyra et al demonstrated superiority of a pulsatile flow for arterial systems, to maintain sheer-stress regulating the inflammatory response of endothelial cells; an enhanced expression of Kruppel-like- factor-2 is discussed as relevant for the microcirculation with potential antithrombotic properties [5,58]. Flow- and pressure related effects are also linked to the perfusion duration. Most older studies aimed to replace SCS with HMP and the subsequently long HMP-duration contributed to pancreas edema with often no superiority compared to SCS, particularly with high pressures. Shorter HMP, e.g., 12-24hrs, was found superior compared to 48hrs HMP, where posttransplant survivals were seen inferior as demonstrated by Florack et al [5,23,58]. The addition of mannitol to the perfusate can help to minimize graft edema [1,41,57]. In 1992, Kenmochi et al explored the impact of a short 1hr endischemic HMP demonstrating more pancreas edema, inflammation and dysfunction with prolonged Donor Warm Ischemia Times (DWIT) of 30-60min [56]. Despite this important step towards an endischemic HMP-approach, the perfusion was probably too short, with high pressures of 50mmHg (Table 1)[56]. Such findings were paralleled by perfusion studies in other organs. Results from HOPE-treatment of livers demonstrated that too high pressures induce damage already during the first perfusion hour and diminish the protective effect achieved with high perfusate oxygen levels [22].

How Much Perfusate Oxygen is Beneficial During Hypothermic Perfusion?

In recent studies, perfusion pressure and duration were reduced to 25mmHg and 6hrs, respectively. The low perfusate oxygen levels may be an additional reason for the limited HMP- effect in earlier studies. Authors often claim to use an oxygenated perfusate with however no further details on oxygenator type or size and the lack of perfusate oxygen partial pressures (Table 1). Only two recent studies present perfusate oxygen levels [41,59]. Lemkuil et al from the UK have demonstrated the perfusate oxygen-dependent ATP-reloading of the pancreas, similar as seen with other solid organs. Authors describe less inflammation and better endo- and exocrine pancreas function during later evaluation on the normothermic perfusion device (Tables 2,3)[40]. Similar results are known from kidneys and livers; perfusate oxygen levels of 21% or a pO2 of <20kPa (8-18kPa) is not enough to trigger metabolic changes and rebuild ATP [60-64]. Subsequently, HMP without active oxygenation as done with the LifePort for many years, has limited effects. A recent RCT demonstrated clear superiority of oxygenated HMP in kidney transplantation [6]. Similar findings are seen with livers; deoxygenated HOPE induced the same IRI as seen with unperfused, cold stored controls [22,65]. Such results are paralleled by the observed tissue protection through the simple addition of oxygen to cold stored organs with persufflation (bubbling) techniques. Higher tissue ATP levels were seen after oxygen persufflation through the graft vasculature. Similarly, as with above described oxygenated HMP such pancreases demonstrated lower levels of IRI and less tissue edema after normothermic reperfusion [66]. The high relevance of oxygen was also demonstrated with the development of preservation concepts where oxygen can be added to standard cold storage [66]. Kuroda et al. from Japan presented in 1988 a new “Two-layer Method”. Perfluorocarbon was added to the conventional cold storage flush solution [67]. These molecules have the unique ability to reversibly bind much more oxygen with a 20-fold higher oxygen concentration compared to human blood. As shown in experiments, Perfluorocarbon forms the lowest layer below the cold storage solution and the oxygen diffuses passively into the pancreas with subsequent ATP recovery. A few early experimental studies demonstrated a prolonged viability of organs. An experimental transplant study showed that slightly more recipients of two-layer method preserved pancreases achieved insulin independency compared to standard cold storage [68]. Although these studies confirm the high relevance of oxygen, the effective penetration of oxygen into deep tissue layers without perfusion remains unproven. Of interest appears also another perfusion parameter: the perfusion duration. Several authors advocate for a 2hrs cold perfusion to achieve the required metabolic switch of mitochondria [16,21,22]. Within the first minutes of HOPE, oxygen reactivates mitochondrial complex proteins with normally directed electron flow and metabolism of previously accumulated toxic metabolites, (e.g., succinate and NADH). Such metabolic changes lead to lower ROS-levels and are key protective effects, also described with HOPE in livers, hearts and kidneys [16,18,19,22,69]. The ATP-reloading during HOPE in pancreas as demonstrated by Leemkuil et al parallels such findings [40].

Series of Experimental HOPE-Perfusions of The DCD Pancreas

DCD pancreas underwent 5hrs SCS and subsequent 2hrs HOPE treatment (n=6) with the following conditions: pressure controlled system with a perfusion pressure of 15-20mmHg; the perfusion flow was pulsatile and ranged between 80-110ml/min/g; the observed resistance index was 0.7-0.8; the measured perfusion temperature was 6-10^°C; perfusate: 1.5l UW-Machine perfusion solution. The median pancreas weight was 324g (IQR: 52g) with a weight gain of 4.2-14.8% during perfusion. Perfusate oxygenation was kept above a pO₂ of 60kPa, confirmed with blood gas analysis; pO₂ pressures of 69-83kPa were achieved within the first 30min of HOPE and maintained thereafter. Three DCD pancreases underwent HOPE with fluoresceine and demonstrated a homogenous fluoresceine staining throughout the entire gland within the first 15min of perfusion. The histological assessment confirmed the fluoresceine distribution (Figure 1).

Figure 1: Pancreas preparation and Hypothermic oxygenated pancreas perfusion. Bench preparation of the porcine DCD pancreas (A) and cannulation of both arterial inflow vessels, e.g., SMA and SA to connect the pancreas with the device for hypothermic oxygenated perfusion (HOPE)(B). Pancreas perfusion parameters obtained during HOPE (C). Fluoresceine stained pancreas after 15min of HOPE, macroscopic and histological image after fluoresceine staining (D). HOPE: Hypothermic oxygenated perfusion; PV: portal vein; SA: splenic artery; SMA: supra-mesenteric artery; SV: splenic vein;

As described with different solid organs and first in hearts in 1969, mitochondrial complex-I releases flavin-mononucleotide (FMN) and NADH during re-oxygenation [16,22,39,69,70]. Such molecules have auto-fluorescent characteristics and are increasingly used for liver viability testing before transplantation. Expectedly, both molecules were released from pancreas mitochondria during HOPE with median perfusate concentrations of: FMN: 5461 A.U. (IQR: 4481-5926) and NADH: 7867 (IQR: 7066-8319) A.U..Compared to HOPE in livers and kidneys, the amount of released FMN in pancreases appears 10- and 3-fold lower, respectively, also based on the smaller organ size, the different mitochondrial density and metabolic activity of the pancreas.

The histological assessment demonstrated no signs of ischemic necrosis or congestion, minimal macrophage activation and inflammation at the end of 2hrs HOPE compared to low risk, baseline control DBD organs (Figure 2). The known dilatation of the peripheral vasculature was also seen here after HOPE of the pancreas (Figure 2A). This phenomenon was described earlier in other organs, e.g., livers or kidneys. With the need to completely preserve the entire periphery of this specific gland to reduce IRI-associated inflammation, this finding is of particular interest also in context of the here applied lower perfusion pressure of maximal 20mmHg during HOPE. No significant differences were observed regarding the number of proinflammatory tissue macrophages and endothelial cells throughout the study groups (Figure 2D).

Figure 2: Immunohistochemistry of Pancreas. Images obtained after pancreas staining for Hematoxylin & Eosin (A), TLR-4 (B) and vWF-positive tissue (C). The number of positive cells were quantified per HPF (D); CIT: cold ischemia time; DBD: Donation after brain death; DCD: donation after circulatory death; HOPE: Hypothermic oxygenated perfusion; TLR-4: toll-like receptor 4; vWF: von Willebrand factor;

Discussion

Despite the early testing in 1974, pancreas perfusion remains in its infancy and the routine clinical implementation lacks behind other abdominal organs [71]. Only a few research programs exist to overcome described hurdles, which include the quality of existing studies (e.g., small case load with various, suboptimal perfusion conditions; a few, old studies with pancreas transplantation), the lack of human transplant studies and the small number of active pancreas transplant programs [5,71]. With the low vascularization, the pancreas is particularly sensitive to barotrauma, caused by prolonged perfusion and high pressures. Knowing the various benefits of dynamic organ preservation, including re-oxygenation, toxin-elimination, viability assessment, therapeutic interventions and endothelial cell protection, the pancreas should benefit most [41,72,73].

This literature review and cases series of HOPE in pancreas confirms what is reported with other organs. HMP should be performed at limited systolic pressures of £ 25mmHg for at least 2hrs, but ideally not more than 6-12hrs [5,7,74]. Once the oxygen has converted mitochondrial metabolisms and rebuilt energy in the entire gland, the perfusion pressure could potentially be reduced in cases, where longer perfusion is needed to bridge logistical challenges and to avoid relevant edema and necrosis. Such findings were further paralleled by a greater islet viability isolated from discarded human pancreas after HOPE as shown by Doppenberg et al in 2021[74]. More studies are required to identify such timings and to explore the acceptable SCS-duration before and after HOPE-treatment. The increasing interest in Normothermic Machine Perfusion (NMP) for all solid organs refers to the opportunity to quantify released IRI-associated molecules for viability assessment, which is particularly interesting in steatotic pancreas and organs from extended criteria donors. With regard to the preservation quality, comparative studies demonstrated however superiority of hypothermic perfusion techniques, thus none included a model of pancreas transplantation [5]. Hamaoui et al reported higher porcine and human pancreas damage with more tissue lesions after normothermic perfusion compared to HOPE [41]. Similar results were shown for livers. In addition, the injury observed during NMP was more pronounced with relevant SCS before perfusion. Recent clinical transplant studies showed this for livers, paralleled by two recent experimental studies with NMP in human pancreas [47,48,75,76]. While, Barlow et al demonstrated severe pancreas damage after endischemic NMP with previous 13hrs SCS, Nassar et al found healthy acini at the end of NMP performed after only 4-6hrs SCS, in a however small number of perfusions (n=3)[42,43]. As with HMP, perfusion pressures during NMP should not exceed 20mmHg[49]. A few groups with increasing expertise in pancreas perfusion are currently developing better circuits to further explore the role of NMP in pancreas. Barlow et al. demonstrated the feasibility of ex-situ pancreas NMP, where perfusate amylase levels correlated well with pancreas fat infiltration and exocrine function [5].

Most viability parameters assessed during NMP appear however quite peripheral to the known IRI-instigator, mitochondria. Pancreas injury and secretory function is frequently assessed during NMP through perfusate amylase and lipase levels; lactates, lactate dehydrogenase, insulin, glucagon, glucose, and C-peptide are further markers of injury and B-cell function [5]. The histology and the perfusion resistance indices may serve as additional parameters to evaluate the pancreas.

In contrast to such markers of downstream injury and function, molecules released from complex-I, such as FMN, were identified by many during the reperfusion of kidneys, livers, hearts and even brain under various conditions, and seem also evident in pancreas as shown here [38,39,70,77]. Perfusate FMN-levels are currently used to accept high-risk DCD livers for transplantation [10,50,52].

The experimental part of this study has some limitations. We are aware that this study does not present functional, molecular or dynamic parameters, which will require larger, prospective trials comparing most recent perfusion concepts in the same donor and recipient risk categories, as currently under preparation.

To increase the utilization of DCD pancreas, where in addition to exocrine dysfunction a low islet yield and beta cell function is known, organs would benefit from HOPE treatment and assessment, explored in upcoming studies [78-80].

The best perfusion conditions suggested for the pancreas based on current literature include: a pulsatile, hypothermic perfusion at 6-10 ^°C with pressure between 15-25mmHg and

related flows for a duration of 2-6hrs (max. 12hrs), using a highly oxygenated (pO₂ >60kPa) perfusion solution (e.g., UW-MPS). Devices in current clinical use for livers and kidneys before transplantation can provide safe pancreas perfusion, given they are pressure controlled and administer the required perfusate oxygen needed to recondition mitochondria.

Author Contributions

Data curation, Fabio Staderini, Janina Eden, Nadia Navari and Andrea Schlegel; Formal analysis, Mauricio Flores Carvalho; Funding acquisition, Paolo Muiesan and Andrea Schlegel; Methodology, Mauricio Flores Carvalho and Andrea Schlegel; Resources, Mattia Dimitri, Fabio Cianchi, Adriano Peris, Paolo Muiesan, Philipp Dutkowski, Fabio Marra and Andrea Schlegel; Supervision, Andrea Schlegel; Writing - original draft, Mauricio Flores Carvalho and Andrea Schlegel; Writing - review & editing, Andrea Corvi, Philipp Dutkowski and Fabio Marra.

Funding

Funding was provided by the Swiss National Science Foundation grant no: 32003B-140776/1, 3200B-153012/1, 320030-189055/1, and 31IC30-166909 to P.D. and A.S. This study was supported by University Careggi grant no 32003B-140776/1 dedicated to P.M. This work was further supported by the OTT grant No.: DRGT641/2019 (cod.prog. 19CT03). The authors confirm that all funders played no role in study design, data collection, data analysis, interpretation, or writing of the report.