Ex vivo normothermic porcine pancreas_ A ...

International Journal of Surgery 54 (2018) 206–215

Contents lists available at ScienceDirect

International Journal of Surgery journal homepage: www.elsevier.com/locate/ijsu

Original Research

Ex vivo normothermic porcine pancreas: A physiological model for preservation and transplant study

T

Rohan Kumar∗, Wen Yuan Chung, Franscois Runau, John David Isherwood, Kean Guan Kuan, Kevin West, Giuseppe Garcea, Ashley Robert Dennison Department of Hepato-Pancreato-Biliary Surgery, University Hospitals of Leicester, Leicester, LE5 4PW, United Kingdom

A R T I C LE I N FO

A B S T R A C T

Keywords: Porcine model Ex-vivo organ perfusion Machine perfusion Pancreas preservation Organ preservation and procurement

Introduction: An ex vivo normothermic porcine pancreas perfusion (ENPPP) model was established to investigate effects of machine perfusion pressures on graft preservation. Methodology: Nine porcine pancreata were perfused with autologous blood at 50 mmHg (control) pressure. Graft viability was compared against four ex-vivo porcine pancreata perfused at 20 mmHg (‘low’) pressure. Arteriovenous oxygen gas differentials, biochemistry, and graft insulin responses to glucose stimulation were compared. Immunohistochemistry stains compared the cellular viability. Results: Control pancreata were perfused for a median of 3 h (range 2–4 h) with a mean pressure 50 mmHg and graft flow 141 mL min−1. In comparison, all of the ‘low’ pressure models were perfused for 4 h, with mean perfusion pressure 20 mmHg and graft flow 40 mL.min-1. All pancreata demonstrated cellular viability with evidence of oxygen consumption with preserved endocrine and exocrine function. However, following statistical analysis, the ‘low’ pressure perfusion of porcine pancreata compared favourably in important biochemical and immunohistochemistry cellular profiles; potentially arguing for an improved method for graft preservation. Conclusion: ENPPP will facilitate whole organ preservation to be studied in further detail and avoids use of expensive live animals. ENPPP is reproducible and mimics a “donation after circulatory death” scenario.

1. Introduction Irreversible organ failure may be treated with transplantation. Over recent decades, improvements in both short and long term outcomes in transplantation have been made, such that the greatest challenge now is neither rejection nor failure but the supply of suitable donor organs. There is an ever increasing gap between supply and demand which has resulted in transplant centres exploring the use of older, ‘marginal’ and hence higher risk organs [1]. The retrieval of ‘marginal’ organs is particularly prone to injury, such that these grafts may be rendered unsuitable for clinical transplantation. The injury begins with the onset of donor death and continues during organ recovery and preservation until, and if, a successful transplantation occurs. This organ injury is due to either a total absence or inadequacy of oxygen and nutrient delivery. A reduction in oxygen and nutrient supply leads to rapid depletion of intracellular energy stores and a loss in cell membrane integrity due to osmotic swelling. In standard criteria donors (SCD), following brain death, the ischaemic

organ damage is deemed acceptable for transplantation. In extended criteria donors (ECD) and donation after circulatory death (DCD), the cellular injury may be so great that the organ is irrecoverable [1]. Conventional methods to preserve organs (following recovery until transplantation into the recipient) focus around ‘flushing’ the donor organ with specialist organ preservation solutions and cooling the organ in ‘cold static storage’ (CS) at temperature 0 °C to 4 °C. CS is the current gold standard of preservation technique across the transplant specialties. However, marginal grafts (ECD and DCD organs) are highly sensitive to preservation injury and the use of CS alone is increasingly seen as the limiting factor in graft success following transplantation [1,2]. Machine perfusion (MP) involves a device to pump preservation fluid or blood through the organ following retrieval. With MP, the organ is metabolically active during storage which potentially prevents the injury associated with CS [1–3]. MP for organ preservation has been successful in the context of human kidney and lung transplantation. In both of these fields, organs

Abbreviations: ATP, Adenosine Triphosphate; ANOVA, Analysis of Variance; CI, Cold Ischaemia; CS, Cold Static Storage; DCD, Donation After Circulatory Death; ECD, Extended Criteria Donor; ENPPP, Ex Vivo Normothermic Porcine Pancreas Perfusion; H&E, Haematoxylin and Eosin; HMP, Hypothermic Machine Perfusion; MP, Machine Perfusion; NMP, Normothermic Machine Perfusion; WI, Warm Ischaemia; SCD, Standard Criteria Donor ∗ Corresponding author. E-mail address: [email protected] (R. Kumar). https://doi.org/10.1016/j.ijsu.2018.04.057 Received 22 October 2017; Received in revised form 4 April 2018; Accepted 29 April 2018 Available online 04 May 2018 1743-9191/ Crown Copyright © 2018 Published by Elsevier Ltd on behalf of IJS Publishing Group Ltd. All rights reserved.


R. Kumar et al.

2.3. Pancreas dissection and retrieval

deemed unsuitable to meet the criteria for SCD, have been successfully transplanted following a period of preservation with MP. The Toronto Lung Transplant Program, Canada [4] and The Heart and Lung Centre, Sweden [5] reported that rejected donor lungs, following ex vivo perfusion, performed similarly to lungs that were initially selected for transplantation. Eighteen patients have each undergone a kidney transplant from an ECD organ following preservation using MP immediately prior to transplantation [6]. Furthermore, normothermic maching perfusion (NMP) within the context of liver transplantation is currently under investigation in a multicentre randomised controlled trail comparing CS against NMP using the OrganOx metra® machine [7,8] prior to transplantation. NMP in liver transplantation has been shown to be clinically feasible [9].

The diaphragmatic attachments to the pleura were divided, suprahepatic inferior vena cava and oesophagus were divided. The thoracic (descending) aorta was cannulated and 10 mm tubing advanced to the level of the abdominal aorta, where it was secured. 1000 mL of preservative solution at 0–4 °C (Soltran® kidney perfusion fluid, Baxter Healthcare Ltd, UK) was immediately delivered to the pancreas and other abdominal viscera followed by continuous pressurised infusion. This marked the end of warm ischaemia (WI) and commencement of cold ischaemia (CI). The abdominal viscera were retracted medially and the aorta ligated at the level of the left renal artery. The pancreas and duodenum upon its aortic segment were dissected. The superior mesenteric vessels were ligated. The portal vein was divided and cannulated with 10 mm tubing at the level of the porta hepatis. The hepatic artery was divided at the level of the porta hepatis. The graft was transported in CS.

1.1. Ex vivo porcine perfusion models The porcine organ system is similar (in terms of size and density) to human organs and they are readily available and inexpensive. They may even be recovered from a commercial abattoir as a by-product of meat production, negating the ethical implications of using live animals for experimentation. The use of porcine organs dates back as far as 2500 years ago to the times of Aristotle and Erasistratus, who both performed studies on animals [10]. In 2nd century Rome, Galen dissected and studied goats and pigs [11]. Porcine ex vivo perfusion models have allowed pre-clinical organ preservation experiments to be investigated in an ethical, safe and cheap setting prior to translation into clinical practice in lung and kidney. In stark contrast, the evolution of MP techniques in pancreas and islet preservation is at a nascent stage. Despite the adverse consequences of reperfusion injury in pancreas and islet transplantation, there have been little advances in pancreas preservation techniques and CS remains the only technique in clinical use [12,13]. The purpose of this study was to develop an ex vivo porcine pancreas perfusion (ENPPP) model deemed physiological and reproducible. The ENPPP was then used to investigate the effects of one perfusion parameter on graft preservation. This is the first description in the literature of a reproducible, reliable and functional ex vivo porcine pancreas to test a null hypothesis:

2.4. Priming of organ perfusion circuit The systemic venous reservoir was filled with 1.5 L of blood with 750 mg of cefuroxime and 500 μg of epoprostenol sodium (both, GlaxoSmithKline plc. Middlesex, UK). Normothermia was achieved with a thermostatic water based heat exchanger unit at 37 °C. 2.5. Bench side preparation of the pancreas for perfusion In the laboratory, the major pancreatic duct was cannulated with a 16G (Fig. 1) and the graft flushed with normal saline in order to wash out the CS preservative solution. Once the blood had primed the circuit and reached normothermia, the ex vivo pancreas was connected. Then baseline data samples at time zero, at fifteen minutes and serially every hour, on the hour. 2.6. Data collection Data collection aimed to assess graft viability. The physiological parameters of blood perfusion pressure and flow delivery to the organ were collected in addition to blood samples. Blood gas analysis for oxygen partial pressures, pH, lactate levels flowing into the organ (via the aorta) were compared against outflow from the portal venous system. Haematological and biochemical parameters of haemoglobin, standard electrolytes, glucose and amylase were serially collected along

During ENPPP, organ viability is no different when the porcine pancreas undergoes perfusion with either a control pressure (50 mmHg) compared with a lower pressure (20 mmHg) system. 2. Materials & methods 2.1. Organ perfusion circuit Ex vivo perfusion of porcine pancreata was achieved with a Perfusion Pack Circuit connected to an Affinity® CP Centrifugal pump. Oxygen delivery was achieved with a Minimax® Plus Oxygenation System. All supplied by the Cardiovascular Division (Europe) from Medtronic Inc., Minneapolis, Minnesota, USA. 2.2. Animal procurement Porcine animal care and final exsanguinations were all compliant with statutory law. Electrocution was performed humanely by the holder of a valid license granted under the Welfare of Animals (Slaughter or Killing) Regulations of 1995 Act [14]. All porcine pancreata were retrieved from Yorkshire Landrace (45 Kg to 55 Kg). Exsanguination was achieved via division of the major neck vessels and 1.5 L of blood was collected in a heparinised sterile, non-pyogenic container. Organ access was afforded via median thoraco-laparotomy. Universally, cardiac pulseless contractions were observed, therefore ENPPP is representative of DCD transplantation.

Fig. 1. Ex Vivo Normothermic Pancreas Perfusion, complete with arterial inflow, portal venous outflow. The pancreatic duct has been cannulated to collect pancreatic juice. The duodenal contents have been controlled with plastic tubing to prevent autologous blood contamination. 207


R. Kumar et al.

3.2. Perfusion parameters

with pancreatic juice hourly. Endocrine function measured insulin levels using Enzyme Linked Immuno-Sorbent Assay from blood samples collected pre and post stimulation with glucose. The grafts were biopsied at baseline (first fifteen minutes) then hourly and immunohistochemical staining was achieved; haematoxylin and eosin (H&E), anti-Caspase 3 antibody, M30 CytoDEATH and for Adenosine Triphosphate (ATP) Synthetase. The biopsy samples were always retrieved from the tail of the pancreas.

Pancreatic graft blood flow and blood pressure remained stable throughout each pressure perfusion experiment (Fig. 3). The control pancreata grafts were perfused with a mean arterial perfusion pressure of 50.5 (50.2–50.8) mmHg. Mean arterial blood flow to the control grafts was 141.3 (122.4–160) mL.min−1. The low pressure grafts were perfused with a mean arterial perfusion of 21.1 (20.5–21.7) with a significantly lower graft blood flow of 40 (31–48) mL.min−1 (p < 0.01) (Fig. 3).

2.7. Statistical analysis

3.3. Arterio-venous oxygen differential

Comparisons of results were subject to statistical analysis with relevant parametric tests following confirmation of normality distributions. The Student t-test was applied within the same experiment, whilst analysis of variance (ANOVA) was reserved for repeated experiments. Probability values of less than 0.05 were considered significant. Analysis was with GraphPad Prism version 6.00 for Windows (GraphPad Software, La Jolla California USA, www.graphpad.com).

All thirteen grafts demonstrated oxygen uptake with significant arterial to venous oxygen partial pressure differentials upon blood gas analysis (Fig. 4). All grafts had a mean arterial (aortic) partial pressure of oxygen of 76.7 (69.7–83.6) kPa. All grafts had a mean (portal) venous partial pressure of oxygen of 6.2 (5.4–7.1) kPa. This differential reached significance, p < 0.0001. The magnitude of arterial to venous oxygen differential did not differ significantly when the control group was compared with the low pressure system, (p = NS) (Fig. 4).

3. Results

3.4. Acid-base (pH), lactate and electrolytes

A total of thirteen porcine pancreata were retrieved and perfused with autologous normothermic blood, ex vivo. The control group of grafts were perfused at 50 mmHg (n = 9). The control group was compared to a second series of ‘low pressure’ grafts perfused at 20 mmHg (n = 4). All control pancreata were viable for a median of three hours (range two to four hours) whereas all low pressure grafts were viable for at least four hours. All thirteen (n = 9 + 4) pancreata demonstrated evidence of oxygen absorption and cellular viability confirmed with immunohistochemistry. Biochemical parameters supported preservation of endocrine and exocrine functionality.

As perfusion time progressed, all thirteen pancreata demonstrated an increase in blood perfusate acidity, evaluated by a decrease in arterial blood gas pH. Mean pH 7.32 (7.23–7.40). This was associated with an accumulation in lactate anions and potassium. In contrast the sodium and chloride ions were relatively stable during the perfusion and similar to porcine normal range, in vivo (Figs. 5 and 6). 3.5. Exocrine function The exocrine component was assessed with the collection of serum amylase and volume of pancreatic juice (rate). The control group values were compared with the low pressure system (Table 1 & Fig. 7). Amylase levels were significantly lower in the low pressure system in comparison with control group (p < 0.03). The rate of pancreatic juice production, however, was similar in both the control and low pressure systems (p = NS).

3.1. Graft retrieval ischaemic times The median WI time for the control group was 4 (3:30 to 7:15) minutes and similar to low pressure system of 5 (4:30 to 7:45) minutes (p = NS). The median CI time for the controls was 127 (112–145) and similar to the low pressure system of 136 (124–147) minutes (p = NS) (Fig. 2).

3.6. Endocrine function A comparison of the endocrine functionality between the control

Fig. 2. Box and whisker plots demonstrating the frequency distribution of warm ischaemic time (right) & cold ischaemic time (left). There was no significant difference between the control and low pressure system. 208


R. Kumar et al.

Fig. 3. Mean Arterial Blood Flow (+/− standard deviation) in mL.min−1 for the low pressure system perfused at a continuous pressure of 20 mmHg. Both blood flow and pressure have been plotted with respect to time of perfusion in minutes (left). The arterial blood flow (mL.min−1) in the control grafts was compared with the low pressure group (right). The arterial blood flow was higher in the control grafts when compared with the low pressure perfusion system (p < 0.01).

Fig. 4. Partial pressures of Oxygen sampled from the arterial inflow at the aortic segment (PaO2) and the venous outflow, sampled at the portal vein (PvO2). Error bars represent the 95% confidence interval at each time point during the perfusion. The low pressure system demonstrated a significant arterial to venous oxygen differential (p < 0.0001). The arterial to venous differential did not significantly differ when the control group was compared with the low pressure system (right).

mean concentration of glucose rose to 23.2 (21.0–25.6) mM (p < 0.0001). During the ‘basal’ unstimulated state, the mean concentration of insulin in the control group was 227 (182–272) picomoles per litre; the equivalent unstimulated values in the low pressure system was 29 (17–42) picomoles per Litre. Following the glucose bolus stimulation, at 125 min, the mean

and low pressure grafts was assessed by quantification of insulin production in response to an appropriate stimulus of a glucose bolus (Fig. 8). Both groups demonstrated a significant rise in insulin production following glucose stimulation (p < 0.0001) indicative of an intact endocrine system. The mean perfusate glucose concentration was 4.9 (4.3–5.3) mM. Following the delivery of a glucose bolus administered at 125 min, the

Fig. 5. The pH decreased with increased time of perfusion, indicating an increase in acidity (left). This decrease in pH of the blood perfusate was associated with an increase in lactate anion concentration (right). Error bars represent the 95% confidence interval at each time point during the perfusion. Both control and low pressure experiments have been plotted. There was no significant difference in the trends observed. 209


R. Kumar et al.

Fig. 6. The potassium cation concentration increased with increased time of perfusion, there was no significant difference when the control was compared to the low pressure system (p > 0.05) (left). The sodium and chloride concentrations remained relatively stable in contrast (right) for both the controls and low pressure system, there was no significant difference (p > 0.05). Error bars represent the ± 1SD at each time point during the perfusion.

cellular viability (Figs. 9–11). Anti-Caspase 3 and M30 CytoDEATH antibodies were used to quantify the presence of cellular death (Table 2 & Fig. 12). These stains demonstrated an improved cell death profile in the low pressure models. The mean number of Anti-Caspase 3 positive cells in the low pressure group were 6 (95% confidence interval 3 to 9) compared with 34 (95% confidence interval 11 to 57) cells per 3000 cells in the control (p < 0.008). The mean number of M30 CytoDEATH positive cells in the low pressure group were 5 (95% confidence interval 3 to 8) compared with 45 (95% confidence interval 7 to 84) cells per 3000 cells in the control (p < 0.02). The activity of ATP synthetase was observed and graded (Table 2 & Fig. 13). Grade I (best), represented the presence of > 95% of the slide section area staining positive for ATP synthetase complex V, grade II (> 90 but < = 95) %, and grade III (> 85 ≤ 90) %. The scores assigned to the low pressure models reflected an improved profile for ATP Synthetase activity in comparison with the control group (p < 0.016). H&E grades were assigned to each section (Table 2 & Fig. 13). The potential grades ranged from I (best) to IV (worst). The scores assigned were similar when the control was compared against the low pressure models (p = NS).

Table 1 The control exocrine parameters have been compared with the low pressure system during perfusion with respect to time. All values are means with value of one standard deviation in parentheses. Time (mins)

[Amylase] (U.L−1) Control

[Amylase] (U.L−1) Low Pressure

Pancreatic Juice rate (mL.hr−1) Control

Pancreatic Juice rate (mL.hr−1) Low Pressure

15

3500 (300) n=9 7100 (700) n=9 9500 (1200) n=9 12000 (2000) n=5 18000 (500) n=4

2200 (500) n=4 2900 (1900) n=4 3300 (2700) n=4 4000 (2300) n=4 4000 (2000) n=4

–

–

2.7 (1.1) n = 9

3.5 (1.0) n = 4

3.6 (3.3) n = 9

7.0 (3.6) n = 4

6.3 (7.2) n = 5

4.8 (2.4) n = 4

5.6 (5.5) n = 4

6.5 (1.3) n = 4

60 120 180 240

insulin concentration in the control group increased significantly to 798 (665–764) picomoles per litre; the equivalent stimulated values in the low pressure system were 356 (324–388) picomoles per Litre (p < 0.0001) (Fig. 8). When the pre and post stimulation insulin levels were compared between the control and low pressure groups; the controls were found to have higher baseline and post glucose stimulation insulin level during perfusion (p < 0.03).

4. Discussion 4.1. Ex vivo normothermic porcine pancreas perfusion as a model This is the first report of an established porcine pancreas perfusion model that facilitates whole organ pancreas preservation study within the context of MP. The focus of this study was initially to establish an ENPPP model. The ENPPP model was then utilised to test one

3.7. Immunohistochemistry for cellular viability All pancreata were subject to immunohistochemical analysis for

Fig. 7. The exocrine function of the control group was compared against the low pressure models. The amylase levels were higher in the control group (p < 0.03) (left). The rate of pancreatic juice production in millilitre per hour was not significantly different when the control group was compared with the low pressure models (p > 0.4) (right). The error bars are ( ± 1 SD). 210


R. Kumar et al.

Fig. 8. Both the control group and low pressure models demonstrated significant insulin response to an appropriate glucose stimulation (right). The glucose stimulation was administered at 125 min (left). Both the ‘steady state’ and the increased glucose levels during stimulation were statistically comparable in both the control and low pressure models. However both the mean ‘steady state’ and stimulated state insulin levels were higher in the control group. The error bars are 95% confidence intervals.

Fig. 9. Anti-Caspase 3 (left) and M30 CytoDEATH (right) immunohistochemical staining, shown at x 10 magnification, at 2 h. An increase in positive cells were noted with increased perfusion time. A positive (dead) cell is marked with arrow.

Fig. 10. An example of preserved islets (white arrow) with surrounding parenchyma, 10× magnification (left). An example of well-preserved exocrine ducts (blue arrow), 20× magnification (right). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

All pancreata were viable macroscopically and the duodenal segment showed peristalsis. Considering that all grafts demonstrated an arterial to venous oxygen differential and in the presence of positive cells stained for active form of an enzyme involved with ATP production it is reasonable to argue that the oxygen uptake by grafts was being utilised at a cellular level, adding further support to the belief that this is a viable and physiological graft.

parameter of MP. For a sufficient length of time post mortem, the model was viable at a cellular level with evidence of preserved endocrine and exocrine function. Furthermore, we have demonstrated that this technique is feasible and reproducible with statistical consistency to assess graft viability (dependent variable) at two different perfusion pressures (independent variable). 211


R. Kumar et al.

the closed system. A rough estimation of the graft resistance is possible by calculating it from the pressure divided by flow. Taking mean values: for low pressure grafts, 20 mmHg/40 mL min−1 = 0.5 mmHg min.mL−1. For the control grafts, 50 mmHg/140 mL min−1 = 0.35 mmHg min.mL−1. Perhaps thrombosis at a microvascular level or differences in vasomotor constriction within the graft could be possible explanations for this apparent increase in graft resistance seen with perfusion at low pressure. However, irrespective of potential differences in resistance within the grafts between low and high pressure; It is important to highlight that this reduction in graft blood flow was not associated with a change in the arterial to venous oxygen differential when compared to the control experiments. This understanding was reflected in the choice of a continuous perfusion pressure of 50 mmHg, in our controls, following a preliminary study (data not shown, n = 8). 50 mmHg was a much lower pressure in comparison to mean aortic pressure of circa 90 mmHg, taken from physiological study in vivo [15,16]. The choice of using 50 mmHg was based on our preliminary experiments [unpublished]. Perfusion of grafts at initially a physiological in vivo pressure of 90 mmHg caused profound oedema, thus grafts were perfused with by decreasing stepwise to 80 mmHg and then 70 mmHg and so forth. It was found that at 50 mmHg was the first optimal pressure preventing visible oedema in the graft and resultant several hours of perfusion. The remainder of the perfusion parameters such as acidity, lactate and electrolytes did not vary significantly. The mean values of lactate and potassium are outside the normal range in vivo, and these may represent an area for further optimisation. However, given that the two groups had similar (statistically) values; then any differences observed in the graft preservation parameters are meaningfully interpretable. All four pancreata at low pressure were viable for four hours. All pancreata produced pancreatic juice, representative of exocrine functionality. The volume of pancreatic juice was not significantly different in comparison with controls. The amylase level was significantly lower at all time points during the lower pressure perfusion in comparison with controls (p < 0.03). The absolute value of amylase may be relevant at the time of graft connection to circuitry, given that this is a closed system further rises with time may need careful interpretation. A reduction in the amylase levels in the low pressure models may represent an improvement in graft preservation and signify a reduction in parenchymal/acinar damage. The endocrine function was confirmed by a rise in insulin in the portal venous blood following glucose stimulation. This observation was maintained in the low pressure system too. Given that both the controls and low pressure models had similar levels of glucose stimulation from ‘steady state’ to ‘stimulated state’ (p = NS); it may have been expected for both sets of experiments to also exhibit similar levels of insulin secretion before and after glucose stimulation? However, following stimulation the absolute value of insulin production in the low pressure models was lower in comparison with the controls (p < 0.0001). The pancreas is a low pressure organ [13,17,18] and perhaps the use of a very low perfusion pressure of 20 mmHg in comparison with 50 mmHg may have led to a reduction in endothelial sheer

Fig. 11. The presence of positive staining for active form of ATP Synthetase was demonstrated in the majority of fields in the majority of biopsy samples. This ATP Synthetase stain is positive during active ATP production.

All pancreata produced pancreatic juice representative of exocrine function. Endocrine function was further confirmed by a statistically significant rise in insulin following stimulation with glucose. Insulin secretion following pancreatic stimulation with glucose occurs via ATP dependent pathways in β islets of Langerhans, which is additional argument for a metabolically active graft. Relevant immunohistochemistry revealed minimal physiological cellular death in biopsy samples. Data from the control group were compared with models perfused at a lower pressure revealed there were some significant differences; graft blood flow, exocrine assessment with amylase levels, the magnitude of insulin production and immunohistochemistry. These outcome differences in the data findings, per se, make it possible to argue that evidence has been provided to reject the null hypothesis. The feasibility of using ENPPP to investigate a research hypothesis has further rationalised it as both ‘reliable and reproducible’. 4.2. Ex vivo normothermic porcine pancreas perfusion as a model to investigate graft preservation The low pressure was 20 mmHg and these grafts were compared with the control group, which had been perfused at a pressure of 50 mmHg. The WI, CI, acidity, lactate, and electrolytes were similar in both groups. The only parameter that was altered was the perfusion pressure. This pressure reduction manifested in reduced mean graft organ blood flow from 141 mL min−1 to 40 mL min−1 (p < 0.01). In vascular physiology, blood flow is dependent upon the dot product of the pressure difference and inverse of vascular resistance. Therefore, given that the resistance throughout the ex vivo circuitry remained unchanged between the two groups, a reduction in arterial blood pressure from 50 to 20 mmHg translated into a reduced blood flow in

Table 2 Comparative immunohistochemistry biopsy scores during perfusion with respect to time. Anti-Caspase 3 and M30 CytoDEATH values are mean positive cells per 3000 cells counted, with one standard deviation in parentheses. H&E and ATP Synthetase values are median scored grade, with the range in parentheses. Time (mins)

Anti-Caspase 3 (+ve Cells/3000) Control

Anti-Caspase 3 (+ve Cells/3000) Low Pressure

M30 CytoDEATH (+ve Cells/3000) Control

M30 CytoDEATH (+ve Cells/3000) Low Pressure

H&E (I - IV) Control

H&E (I - IV) Low Pressure

ATP Synthetase (I - III) Control

ATP Synthetase (I III) Low Pressure

15 60 120 180 240

14 19 39 38 60

2 6 8 7 7

10 21 44 70 82

3 4 5 8 7

II (I - II) II (II - III) II (II - IV) II (II - III) III (II - IV)

I (I – II) II (I – II) II (I – II) II (I – III) II (II – IV)

I (I - II) II (I - II) II (II - III) III (II - III) III (III - III)

I (I – II) I (I – I) I (I – II) I (I – II) II (I – II)

(12) (13) (29) (37) (54)

(1) (3) (2) (5) (3)

(6) (10) (24) (54) (34)

(1) (2) (2) (4) (5)

212


R. Kumar et al.

Fig. 12. The low pressure pancreata also demonstrated evidence of cellular viability. An increase in number of cells staining positive for Anti-Caspase III (left) and M30 CytoDeath (right) was noted with increase in perfusion time in both low pressure and control groups. However, the low pressure models had a lower mean number of cells staining positive for anti-caspase III (p < 0.008) and M30 CytoDeath (p < 0.02). The mean number of cells (+/− 1SD) are plotted with respect to time.

response to glucose stimulation and were histologically viable, a high amylase level cannot simply be solely attributable to pancreatic damage. Potentially, any acinar damage sustained during operative retrieval and CS may have been responsible for the initially raised amylase levels. Given that ENPPP occurs within a closed circuit, amylase along with other exocrine enzymes are free to circulate and cause further graft (hydrolytic/proteolytic) damage. Graft lysis would then lead to progressively increased amylase levels with time. Therefore, one explanation may be that the initial absolute amylase level may help to quantify the level of acinar graft damage sustained during operative retrieval and CS? However, further rises in amylase have to be interpreted with caution, given that this is a closed circuit. The addition of a porcine kidney in parallel to our pancreas circuit, may confer the advantage of stabilising the biochemical parameters. However, in order to remove amylase (large protein molecule) would require filtration with a plasmapheresis machine. Then the use of serial amylase levels in our model may become meaningful and help quantify the acinar damage and this would effectively ‘open’ a closed circuit? Addition of a kidney would not be dissimilar to current clinical transplantation practice whereby patients with type I diabetes and end stage renal failure undergo a simultaneous pancreas and kidney transplant as curative treatment. The WI time is a limitation in the ex vivo model with the median WI was 4–5 min across all models with range of 3 min and 30 s to 7 min and 45 s. The WI time in this porcine DCD model was short in comparison with human pancreas DCD transplantation scenario. Data from the University of Wisconsin which has been collected over 29 years showed

stress sustained by the grafts. A reduction in endothelial trauma would have meant a reduction in parenchymal/acinar damage. A reduction in acinar and parenchymal damage would account for not only the reduction in amylase levels but also the lower insulin levels observed. The observation that there was a statistically significant rise in insulin following similar glucose stimulation argues favourably for an intact endocrine feedback system in both the controls and low pressure models. There are a number of possible explanations for these findings and the lower insulin secretion values may represent an additional reduction in islet damage in the low pressure models. However, when islets are damaged they degranulate and release insulin and it is not possible to differentiate between these two mechanisms in the early phase. The immunohistochemical stains for markers of cell death also demonstrated an improvement in addition to an increase in the presence of ATP synthetase in the low pressure models when compared to controls. 4.3. Potential limitations of the ENPPP perfusion model The mean serum amylase ranged from 3500 to 18300 U L−1. In humans, amylase levels that are greater than the upper limit of normal range are considered diagnostic for pancreatitis although paradoxically the absolute value does not predict the severity of an episode of acute pancreatitis. Instead, well-established physiological scoring systems are used to determine severity (e.g. Glasgow score and Ranson's Criteria). In pancreatic transplantation, higher amylase levels may be related to poor outcomes in donor grafts [19,20]. In ENPPP, high amylase levels may also be representative of graft dysfunction, but this has to be interpreted with care. Given that all pancreata demonstrated insulin

Fig. 13. The low pressure pancreata demonstrated evidence of cellular viability with similar score grades for the Haematoxylin and eosin staining (p > 0.28) (left); however an improved score grade for ATP Synthetase Complex V stain (p < 0.016) (right). The median grade with the error bars representing the inter-quartile range are plotted with respect to time. 213


R. Kumar et al.

5. Conclusion

that the mean WI time for a pancreas transplantation was 20.8 min ( ± 10.9) with an outlier at 64 min [21]. In order to make the model more representative of the DCD setting, a longer WI time could be easily implemented. By allowing more time to elapse between the visualisation of circulatory arrest and the cold flush of the aorta, during retrieval the WI could be prolonged to better reflect current clinical practice. The ‘mode of dying’ in ENPPP is different when compared with clinical practice and this may represent a potential limitation for arguing a similarity with a DCD setting, viz. During the agonal phase immediately prior to death, the animal may release vasoconstrictor catecholamines in response to the hypovolaemia and pulseless electrical activity that occurs from the current method of pancreas retrieval in this thesis. This is in contrast to current clinical practice where organs may be recovered in a ‘controlled’ DCD setting with the withdrawal of care from a donor suffering with failed intensive care. Although, on the intensive care unit, some of the differences in neurohumoral responses between ENPPP and clinical practice may be offset in those donors being infused with inotropes and vasopressors. The perfusate in ENPPP was autologous blood with the addition of heparin to reduce coagulation. Citrate to chelate the calcium ions and thrombolytic agents (e.g. urokinase, streptokinase) are potential alternatives that may have been considered. The perfusate could be further optimised and be more representative of clinical practice by using ‘packed red cells’ only (i.e. exclude the platelets and leukocytes), which may have led to reduction in graft thrombosis. During the H&E immunohistochemistry staining analysis, none of the slides demonstrated the presence of (inflammatory) white cells. One explanation for this is that, perhaps the leukocytes had ‘marginated’ secondary to laminar flow within the circuitry. Thus the removal of leukocytes with cell salvage may not be an imperative step. Furthermore, this explanation of ‘margination’ of white cells would also help account for the absence of leukocytes from the stained sections. It may also be worth considering the use of a different preservation solution. The preservation solution used in these experiments was Soltran® kidney perfusion fluid, (Baxter Healthcare Ltd, Renal Division, Hospital Equipment and Supplies, Northampton, UK) and the potassium in this solution is composed with citrate. UW is the ‘gold standard’ for CS in pancreas preservation and the potassium is instead conjugated with lactobionate. These differences may have contributed to potassium accumulation within the closed circuit. Arguably, a potential limitation of ENPPP would be the small numbers in each group. A prospective power calculation was impossible because this is the first study of its kind and an accurate estimation of the ‘size of effects’ between the groups unknown. Following the literature review, universally all porcine ex vivo studies, hitherto, contain very few numbers in each group [22]. This is in stark contrast with the numbers used in population studies. A power calculation is important in avoiding a type II error. However, given that most of the thesis p values are highly significant rather than borderline; it makes the chances of a type II error highly unlikely. Nevertheless, in order to satisfy a fuller understanding, a retrospective power calculation has been provided using the insulin data as a worked example. The ‘size of effect’ may be approximated using a Glass's delta test: difference in the means between the low pressure and control grafts would be ‘delta’ (285.6 picomoles per litre) divided by the SD of the control group (79.4 picomoles per litre). Using a significance level set at 0.05 with number of replicates (n = 9), this gave an adequate power, calculated at 0.967. Notwithstanding the statistical discussion above, the ability to have tested a research null hypothesis successfully, per se, validates the pancreas perfusion system as a ‘reproducible and reliable’ model.

In the literature there are several models, irrespective of the organ focus, which argue that ex vivo porcine models are a suitable tool for study in translational transplant surgery [22–24]. With the exception of pancreas organ, several studies have endeavoured to make comparisons of preservation methods with the use of ex vivo porcine perfusion models. Porcine ex vivo perfusion models have allowed organ preservation experiments to be investigated in an ethical, safe and inexpensive setting before being translated into clinical practice in kidney and lung transplantation. Our ENPPP is now one such reliable and reproducible model for further optimisation and use to investigate pancreas whole organ and islet preservation techniques. Ethical approval Not applicable. Sources of funding None. Author contribution RK (design, performed research, collected data, analysed and wrote paper). WYC (design, collected and analysis) FR (collected data). JDI (collected data), KGK (collected data), KW (analysis and design), GG (analysis and assisted writing), ARD (contributed to design, analysis and writing of paper, head of department). Conflicts of interest None. Research registration number Not applicable. Guarantor All authors are in agreement with the entire content and all jointly take full responsibility for the work and conduct of the study: RK, WYC, FR, JDI, KGK, KW, GG, ARD. References [1] R.J. Ploeg, P.J. Friend, New strategies in organ preservation: current and future role of machine perfusion in organ transplantation, Transpl. Int. 28 (6) (2015) 633. [2] A. Nassar, et al., Ex vivo normothermic machine perfusion is safe, simple, and reliable: results from a large animal model, Surg. Innovat. 22 (1) (2015) 61–69. [3] D. Balfoussia, et al., Advances in machine perfusion graft viability assessment in kidney, liver, pancreas, lung, and heart transplant, Exp Clin Transplant 10 (2) (2012) 87–100. [4] M. Cypel, et al., Normothermic ex vivo lung perfusion in clinical lung transplantation, N. Engl. J. Med. 364 (15) (2011) 1431–1440. [5] S. Lindstedt, et al., Comparative outcome of double lung transplantation using conventional donor lungs and non-acceptable donor lungs reconditioned ex vivo, Interact. Cardiovasc. Thorac. Surg. 12 (2) (2011) 162–165. [6] M.L. Nicholson, S.A. Hosgood, Renal transplantation after ex vivo normothermic perfusion: the first clinical study, Am. J. Transplant. 13 (5) (2013) 1246–1252. [7] OrganOx, OrganOx Living Organs for Life, (2018) [cited 2018 4th April 2018]; Available from: http://www.organox.com/. [8] HRA, RCT of Normothermic Perfusion Vs Cold Storage in Liver Transplant V1, 2014 [cited 2018 4th April 2018]; Available from: https://www.hra.nhs.uk/planningand-improving-research/application-summaries/research-summaries/rct-ofnormothermic-perfusion-vs-cold-storage-in-liver-transplant-v1/. [9] R. Ravikumar, et al., Liver transplantation after ex vivo normothermic machine preservation: a phase 1 (First-in-Man) clinical trial, Am. J. Transplant. 16 (6) (2016) 1779–1787. [10] L.G. Wilson, Erasistratus, galen, and the pneuma, Bull. Hist. Med. 33 (1959) 293–314.

214


R. Kumar et al.

[11] J.S. Prendergast, The background of Galen's life and activities, and its influence on his achievements, Proc. Roy. Soc. Med. 23 (8) (1930) 1131–1148. [12] A.D. Barlow, S.A. Hosgood, M.L. Nicholson, Current state of pancreas preservation and implications for DCD pancreas transplantation, Transplantation 95 (12) (2013) 1419–1424. [13] A.D. Barlow, et al., Use of ex vivo normothermic perfusion for quality assessment of discarded human donor pancreases, Am. J. Transplant. 15 (9) (2015) 2475–2482. [14] legislation.gov.uk, The Welfare of Animals (Slaughter or Killing) Regulations, (1995) [cited 2016 8th March 2016]; Available from: http://www.legislation.gov. uk/uksi/1995/731/contents/made. [15] T. Xanthos, et al., Baseline hemodynamics in anesthetized landrace-large white swine: reference values for research in cardiac arrest and cardiopulmonary resuscitation models, J Am Assoc Lab Anim Sci 46 (5) (2007) 21–25. [16] T. Xanthos, et al., Cardiopulmonary arrest and resuscitation in Landrace/Large White swine: a research model, Lab Anim 41 (3) (2007) 353–362. [17] D. Ridgway, et al., Preservation of the donor pancreas for whole pancreas and islet transplantation, Clin. Transplant. 24 (1) (2010) 1–19. [18] K.G. Kuan, et al., Extracorporeal machine perfusion of the pancreas: technical

[19] [20]

[21] [22]

[23]

[24]

215

aspects and its clinical implications–a systematic review of experimental models, Transplant. Rev. 30 (1) (2016) 31–47. D.A. Axelrod, et al., Systematic evaluation of pancreas allograft quality, outcomes and geographic variation in utilization, Am. J. Transplant. 10 (4) (2010) 837–845. K.P. Drewitz, et al., Predictors of non-transplantation of adult donor organs–an observational study using routine data from Eurotransplant, BMC Health Serv. Res. 14 (2014) 584. J.M. Bellingham, et al., Donation after cardiac death: a 29-year experience, Surgery 150 (4) (2011) 692–702. R. Kumar, et al., Ex vivo porcine organ perfusion models as a suitable platform for translational transplant research, Artif. Organs 41 (9) (2017 Sep) E69–E79 Epub 2017 Mar 6. C.M. Meers, et al., A model of ex vivo perfusion of porcine donor lungs injured by gastric aspiration: a step towards pretransplant reconditioning, J. Surg. Res. 170 (1) (2011) p. e159-67. R.W. Jamieson, et al., Hepatic steatosis and normothermic perfusion-preliminary experiments in a porcine model, Transplantation 92 (3) (2011) 289–295.