Characterization of the Withdrawal Phase in a Porcine Donation after the Cardiac Death Model


Corresponding author: Richard B. Freeman,


Transplantation of donation after cardiac death (DCD) livers has higher rates of organ failure and complications, specifically ischemic biliary injuries. Reported large animal DCD models all employ active means to halt circulation, contrary to human DCD protocol. We report a DCD porcine model in which the animal passively progresses to cardiac death, thereby more closely mimicking human DCD scenario. Sixteen Yorkshire pigs (10 females, 6 males, 30–45 kg) had a mean time of 26:19 min ± 14:14 from withdrawal of ventilatory support (WVS) to circulatory arrest and 44:38 min ± 16:37 from WVS to electrical standstill. Cessation of hepatic flow (HF) occurred well before electrical standstill (22:15 min ± 10:09), previously not described in human or animal DCD. Histologically comparing livers from our DCD model demonstrated a dramatic increase in hepatocyte vacuolization, disorganization of endoplasmic reticulum, formation of mitochondrial inclusions and apoptosis compared with control specimens. Subtle changes were also evident in biliary epithelial cells (BEC). This results in severe cellular changes before reperfusion. Early histologic evidence suggests that there is severe hepatocyte and biliary cell disruption in our DCD model. Further research using this model may provide a deeper understanding of the pathophysiology of the DCD liver.


biliary epithelial cells


caspase 3


cold preservation


donation after brain death


donation after cardiac death


endoplasmic reticulum


electron microscopy


hepatic artery flow


hepatic artery buffer response


hepatic flow


portal vein flow


systolic blood pressure


warm ischemia time


withdrawal of ventilatory support


With increasing pressure on transplant centers to meet the demand of waiting lists, transplant programs have expanded the range of acceptable donor organs. Responding to this pressure, centers are increasingly using livers from donors expiring from cardiac causes, the so-called donation after cardiac death (DCD) protocol. From 2000–2007, the US has experienced a more than eightfold increase in the number of livers procured from DCD donors (1). The Netherlands and the United Kingdom have also reported similar increases in the rates of DCD organ donation (2,3).

In contrast to the more widely practised donation after brain death (DBD) where death is certified using neurologic criteria while circulation is maintained prior to and during organ recovery, DCD donors progress to death with a gradual reduction and ultimate irreversible cessation of circulation prior to certifying death, complying with the dead donor rule (4). As experience has accumulated, it has become increasingly clear that organs recovered from DCD donors do not function as well as organs transplanted from brain dead donors. Livers transplanted from DCD donors have an adjusted hazard ratio of 1.85 for graft failure at 4 years compared with non-DCD recipients (5). Recipients of DCD livers also experience an increased rate of ischemic bile duct strictures and biliary necrosis, often leading to retransplantation or recipient death (6).

In light of the inferior outcomes with DCD liver transplantation, many investigators have attempted to better understand the physiologic events of DCD, and in particular, how these events influence DCD liver function after transplant. Several large animal DCD models have been developed; however, none of these replicate the human DCD scenario very closely. In all published studies, some form of active instigation of cardiac arrest, such as defibrillation or use of cardioplegic medications (7–9), is utilized, the antithesis of the human case where active euthanasia is strictly prohibited. As a result, the slow reduction in circulatory activity that is a constant in the human DCD experience does not occur in the animal models where cardiopulmonary function is abruptly and actively halted.

To address this gap and investigate more closely the DCD events occurring during the time after life sustaining therapy is withdrawn but before death is declared, we developed a large animal model of liver donation that more accurately mimics the withdrawal phase in the human DCD scenario. Using this model, we have better characterized changes in the systemic and hepatic circulation during the withdrawal phase and provide preliminary histological evidence for its the effects on the liver and biliary epithelial cells (BEC) after cold storage of these livers.


In accordance with the Institutional Animal Care and Use Committee guidelines, 16 Yorkshire pigs (35–50 kg) were used for this study. All animals were premedicated with a intramuscular injection of Telazol (8 mg/kg), Ketamine (4 mg/kg) and Xylazine (4 mg/kg) prior to being intubated and anesthetized with isoflurane (1–2% in 100% oxygen). All animals were further monitored with an oxygen saturation probe on their earlobe, 3-lead electrocardiograph and a femoral arterial line.

Two animals were used as controls. Following sedation and intubation, a midline laparotomy was made and the suprahepatic aorta and suprahepatic inferior vena cava were isolated and clamped immediately prior to the administration of euthanasia solution (Beuthanasia-D 1 mg/10l bs, Schering-Plough Animal Health Corp., Union, NJ, USA) to prevent circulation into the liver. Immediately following this, the livers were then flushed through the portal vein (PV) and infrarenal aorta with 2 L of University of Wisconsin solution, recovered and cold stored as outlined below.

Study animals: through a midline laparotomy, the porta hepatis was dissected and flow probes (Transonic, Ithaca, NY, USA) were placed around the hepatic artery and portal vein. Hepatic flow (HF) was defined as the sum of Hepatic arterial (HA) flow and PV flow. Flow rates, arterial waveforms and oxygen saturation were all digitally recorded through LabChart 6 (ADInstruments, Colorado Springs, CO, USA).

Once all probes were placed and the animal was adequately sedated, the study began by discontinuing both isoflurane and mechanical ventilation. Thiopental (3–6 mg/kg/h) was administered intravenously and titrated to maintain adequate sedation. Respiration rate and heart rate were used as markers for depth of sedation. The animal was then allowed to progress to death as defined below. A total of 10 000 units of heparin were given intravenously after withdrawal of ventilatory support (WVS).

Circulatory arrest time was determined when a pulse pressure of zero, determined by arterial line, was observed. Declaration of death was determined by an absence of organized electrical activity on EKG monitor for 5 min, electrical standstill. Warm ischemia time (WIT) was calculated as the time from WVS to the time death was declared.

Once the animal was declared dead, the liver was flushed with 2 L of University of Wisconsin solution through both the PV and aorta. The liver was then recovered and stored in two plastic bags in an ice-slush bath for 4 h. As this study did not involve transplantation of the liver, flushing of the common bile duct was not performed in the ice-slush bath.

Liver wedge biopsies were taken immediately prior to the WVS (time 0), every 10–15 min thereafter until electrical standstill and hourly while in cold storage. Control animals had biopsies taken prior to aortic cross clamping and hourly during cold storage.

Light microscopy

Liver biopsies were fixed in 10% formalin, embedded in paraffin, and sectioned at 5 μm. Subsequently the sections were deparaffinized, dehydrated and stained with hematoxylin and eosin (H&E). The following parameters were examined: extent of hepatocyte vacuolization, sinusoidal dilation and vascular congestion.

Electron microscopy

Additional liver biopsy samples of 5 mm in diameter samples were fixed in Trump's fixative, postfixed in 1% osmium tetroxide with sodium cacodylate buffer for 3 h at 20°C, and stained en bloc with 5% aqueous uranyl acetate. Thick sections (1 μm) were then stained with toluidine blue. Thin sections of 50–70 nm were cut from the larger section using a LKB8801 ultramicrotome (LKB, Stockholm, Sweden), stained with uranyl acetate and lead citrate. All samples were examined and photographed with transmission electron microscope (Philips EM 201).


Liver biopsy sections (5 μm) were heated at 60°C for 30 min, deparaffinized and hydrated through a series of xylene and alcohol solutions. Immunohistochemistry for caspase 3 (Casp3) and CK19 was prerformed on a Techmate 1000 automated immunostainer (Ventana Medical Systems, Tucson, AZ, USA) using avidin/biotin complex (ABC) staining procedure after antigen recovery, and a DBA kit (Ventana Medical Systems) for the color forming reaction. Anti-Casp3 (SantaCruz Biotechnology) is a mouse monoclonal antibody and anti-CK19 (Ventana) is a mouse monoclonal antibody.

Histologic analysis was performed by a veterinary pathologist (JA).


Following WVS and initiation of the thiopental infusion, all animals progressed to death (range 24–76 min) (Table 1). The mean time from WVS to electrical standstill was 44:38 min (± 16:37 min). In all animals, circulatory arrest occurred prior to electrical standstill. The mean time from WVS to circulatory arrest was 26:19 min (± 14:14). An additional 19:40 min (± 10:53) elapsed after circulatory arrest to electrical standstill.

Table 1.  Times withdrawal of ventilatory support (WVS); circulatory arrest; electrical standstill
 Mean timeSDRange
WVS to circulatory arrest26:1914:1412:30–57:30
WVS to electrical standstill44:3816:3724:01–76:00
Circulatory arrest to electrical standstill19:4010:531:14–41:00
WVS to no hepatic flow21:1713:449:45–50:50
No hepatic flow to circulatory arrest2:352:250:00–7:26
No hepatic flow to electrical standstill22:1510:098:40–44:00

HF ceased 21:17 min (± 13:44) following WVS. In all animals, HF ceased at the time of, or prior to circulatory arrest, at a mean of 2:35 min (± 2:25) prior to circulatory arrest. The resultant mean time from cessation of HF, where the liver was without perfusion, to electrical standstill was 22:15 min (± 10:09) (Figure 1). There was no correlation between thiopental dose and time to arrest.

Figure 1.

Graphic representation of the relationship of hepatic circulation, renal circulation with oxygen saturation and mean arterial pressure from the time of withdrawal of ventilatory support from one study animal. Red arrows represent mean time to cessation of hepatic flow (A), circulatory arrest (B) and electrical standstill (C).


In control animals where the liver was recovered immediately prior to sacrificing the animal, biopsies taken at time 0 appeared very similar on H&E staining to those taken after 4 h of cold storage. In contrast, significant histological changes are evident minutes after WVS. Livers from animals subjected to our DCD protocol and the resultant warm ischemia demonstrated hepatocyte vacuolization, sinusoidal swelling and congestion (Figure 2). After 4 h of cold storage, specimens from these DCD livers demonstrated more prominent vacuolization and sinusoidal swelling and early signs of endothelial sloughing were observed. Disorganization of hepatic cords also became more evident after cold storage of the DCD livers. No evidence of microthombus formation was present at any time point in either the sinusoids or biliary capillaries.

Figure 2.

(A) Normal porcine liver revealing central vein and well-organized hepatic cords with normal appearing sinusoids; (B) control liver after 4 h cold preservation (CP) illustrating slightly dilated central vein and sinusoids with very mild congestion; (C) DCD liver at time of electrical asystole: disorganized hepatic cords with moderate amount of vacuolization of hepatocytes; (D) DCD liver after 4 h CP: further disintegration of hepatic cords and extensive vacuolization of hepatocytes (hematoxylin and eosin, 200×).

In the absence of reperfusion, a might be expected, no evidence of invasion of inflammatory cells or ischemic cholangitis was appreciated in either control or experimental biopsies at any time point. In addition, no areas of necrosis were evident at any time points for both control and DCD biopsies.

Electron microscopy

At baseline, hepatocytes are seen with well-organized endoplasmic reticulum (ER) and rounded mitochondria (Figure 3A and B). However, at the time of electrical standstill, as seen on H&E, extensive vacuolization becomes dramatically evident in each hepatocyte, with many of the vacuoles containing low-density proteinaceous material. Lamellated membrane structures and mitochondria can also be seen in the vacuoles (Figure 3C and D). Distribution and size of vacuoles were variable throughout the hepatocytes. In the control livers, the integrity of the ER was maintained after 4 h of cold preservation. (Figure 3E and F).

Figure 3.

(A) Normal porcine liver: several hepatocytes with sinusoids lined by endothelial and Kupffer cells and a few erythrocytes (electromicroscopy [EM], ×2500). (B) Higher magnification revealing cytoplasm containing normal appearing mitochondria, smooth and rough endoplasmic reticulum and a few peroxisomes (EM, ×10 000). (C) Control liver at time of electrical asystole: hepatocyte with numerous large membrane bound vacuoles containing loosely arranged laminated membrane structures; increased in size of the space of Disse; dilated sinusoids lined by endothelial, Ito cells and Kupffer cells (EM, 2500). (D) Higher magnification of the vacuoles which contain lamellated membrane structures and one of which contains a mitochondria (EM, ×10 000). Porcine liver after 4 h of cold preservation. (E) Control porcine liver: several hepatocytes exhibiting crescentic artifact and minimal vacuolization (EM, ×2500). (F) Higher magnification revealing round mitochondria and well-organized endoplasmic reticulum (EM, ×10 000). (G) Cardiac death liver demonstrating extensive vacuolization with proteinaceous content (EM, ×2500). (H) Higher magnification of cardiac death liver demonstrating mitochondrial inclusions and disorganized endoplasmic reticulum (EM, ×10 000).

After 4 h of cold storage of the DCD livers, hepatocyte vacuolization was extensive and closer examination revealed a disorganization of ER. The mitochondria did not demonstrate swelling, which is often seen in early phases of apoptosis; however, inclusions began to appear within them (Figure 3G and H).


Sporadic Casp3 stained hepatocytes can be seen in the control animal after 4 h of cold preservation; however, this may represent naturally occurring cell turnover of the liver (Figure 4). In contrast, after 4 h of cold preservation the liver tissue had many more cells stain positive for Casp3. The pattern and distribution of Casp3 staining was variable across different DCD animals; however, we did observe a trend toward more cells staining positive as the length of WIT increased. However, this was observational only. No BEC stained positive for Casp3 in either model at any time point.

Figure 4.

Immunohistochemical staining with caspase-3. (A) Control liver after 4 h of cold preservation revealing sporadic uptake of caspase-3. (B) Cardiac death liver after 4 h of cold preservation revealing more diffuse uptake of caspase-3 (hematoxylin and eosin, 200×).

Biliary epithelium

No evidence of biliary epithelial shrinkage or pyknotic nuclei was evident as observed with H&E on either control or DCD biopsies at any time point. Immunohistochemical staining with CK-19 also did not demonstrate any appreciable changes at any time point (Figure 5). However, after 4 h of cold storage in the DCD model, a slight separation between each biliary epithelium and with the extracellular matrix became evident. The contour of the nuclei also became more irregular when compared to the control (Figure 6). Neither of these findings was evident in control biopsies after 4 h of cold storage.

Figure 5.

Immunohistochemical staining with CK-19. (A) Control liver after 4 h of cold preservation illustrating one large biliary duct and several smaller biliary ducts. (B) Cardiac death liver after 4 h of cold preservation illustrating a few biliary ducts without change in CK-19 uptake.

Figure 6.

(A) Cross-section of normal porcine biliary ductule lined with biliary epithelial cells (BEC). (B) Control biliary duct after 4 h of cold preservation revealing minimal change. (C) Cardiac death biliary ductule after 4 h of cold preservation demonstrating an increased separation between each BEC and from extracellular matrix. Increased irregularities in the contour of nuclear membranes of BEC also seen (electromicroscopy, ×2500).


DCD requires the withdrawal of life-sustaining therapy and the confirmation of death prior to initiation of any organ recovery efforts. This process, known as the withdrawal phase, induces a slow reduction of oxygen delivery to the organs of interest, before death is declared. In contrast, in DBD protocols, the donor is declared dead by neurological criteria prior to organ recovery and therefore circulation can be stopped abruptly at the time of organ removal.

The time in which it takes for the DCD donor to progress to death has untoward effects on graft viability, in both human (10,11) and animal studies (7,8). However, characterizing WIT solely based upon the time from withdrawal of life-sustaining therapy to declaration of death is incomplete; furthermore, until this study, no one has addressed the distinction between systemic circulatory arrest and blood flow to the liver. Systemic and organ-specific hemodynamic profiles and oxygen delivery during this period seem to play a critical role in the overall viability of organs. In a study by Ho et al. (12), the duration of systolic blood pressure (SBP) <50 mmHg directly correlated with outcome. Numerous other animal studies have also reported the inverse relationship of WIT and graft survival (7,8). In a hypoperfusion porcine model (13), sustained systemic hypoperfusion resulted in the exhaustion of the hepatic artery buffer response (HABR) and subsequent impairment of hepatic function. The HABR has also been described in human studies (14).

In the current large animal DCD models, WIT is simulated after the cardiac arrest is actively induced either by causing ventricular fibrillation (15) or by administering high concentrations of potassium intravenously (16). Warm ischemia is then simulated by waiting for a period of time before recovering the organ and cold storing. However, these methods do not simulate the slow reduction in oxygen delivery to the liver, did not record blood flow in the hepatic artery and PV as we did here, and do not incorporate the sequential loss of visceral perfusion seen so clearly in our results. Whether this sequence also occurs in the human situation is unknown and, because continuous flow measurements of the hepatic artery and PV are not likely to be available in the human DCD scenario, it may be difficult to confirm that our model does mimic human DCD perfectly. However, by taking a more passive approach where cardiac arrest is allowed to occur without intervention, we believe that our large animal model more closely mimics the clinical scenario. In our model, and in contrast to others in the literature, we recreate a slow reduction of oxygen delivery and perfusion to the viscera. From our model, we observed that hepatic circulation ceases before circulatory arrest, indicating that total hepatic ischemic time may in fact be longer than previously appreciated. Our results also highlight that using circulatory arrest compared with electrical standstill as definition of death imposes significantly different additional ischemia time on the organs of interest. Based on these findings, broader investigation into whether these declaration criteria have an impact on outcome is warranted.

Our histologic preliminary findings of sinusoidal swelling, hepatocyte swelling, vacuolization and apoptosis with increasing WIT in our DCD animals are consistent with previous animal studies (8,17). It should, however, be emphasized that the changes we observed at the cellular and subcellular level all occurred prior to reperfusion, indicating a significant amount of ischemia induced disruption prior to any reperfusion. Further transplantation studies using this model may allow for calculation of the semiquantitative histologic score as described by Monbaliu et al. (8) to help more accurately establish the correlation between the duration of warm ischemic time to degree of hepatocellular disruption in our model.

Immunohistochemical staining of Casp3 was utilized in our study to demonstrate evidence of apoptosis, as it functions as the final step in of apoptosis, initiating such events as cell shrinkage, surface blebbing and nuclear lobulation (18). However, as not all forms of apoptosis require Casp3 activation, additional studies including other proinflammatory markers and cytokines, such as TNF-α and IL-1, may provide a fuller understanding on the mechanisms of cell death in our model.

We also examined the effects of the withdrawal phase on BEC, antecedent to reperfusion. Immunohistochemically staining for CK-19 demonstrated intact BECs on light microscopy; however, upon higher magnification using electron microscopy, we were able to observe subtle differences in BEC morphology which may represent the initial inciting insults that lead to future cholangiopathy after reperfusion. The increased separation between each BEC and from the basement and extracellular matrix may be evidence of anoikis (19), which is commonly seen in primary biliary cirrhosis. The progressive irregularities in contour of nuclear membrane observed may also represent early signs of cell shrinkage and pyknotic nuclei, both signs of apoptosis. Again all of these findings are evident prior to reperfusion and likely a result of the ischemic injury induced by the DCD process, which may account for the relative absence of Kupffer cells and lymphocytes.

Our findings might shed light on the conflicting DCD liver outcomes reported to date. In a recent retrospective review (20), total ischemic time ≥9 h in donors >50 years old was shown to be predictive of developing ischemic cholangiopathy. When further deconstructing the total ischemic time, WIT, duration of hypotension (SBP ≤50 mmHg and ≤30 mmHg) and oxygen saturation <70% were not predictive with development of ischemic cholangiopathy. Another review of DCD liver transplants (21) also found several prolonged WIT (>50 min) DCD livers with little to no biliary complications. In contrast, guidelines by the American Society of Transplant Surgeons recommend WIT to be under 30 min for better results (22). The results of our study suggest that differences in DCD LT results might be explained by differences in criteria used to declare death. That is, our results suggest that livers procured from donors declared dead by EKG criteria are subjected to more WIT than those from donors declared dead by cessation of arterial circulation. This possibility underscores the importance of a DCD model that more closely follows the human protocol such as we describe here. Other models in which cardiac arrest is induced actively cannot account for this possibility.

By allowing the animals to passively progress to death with the use of a thiopental infusion, our design does not allow for a high degree of reproducibility (again similar to the human setting), which should be taken into account when extrapolating from our data. However, it does highlight the unpredictable nature faced by transplant centers regarding human DCD. We will need to expand our sample size to demonstrate a correlation between the dose of thiopental and time to circulatory arrest and cessation of cardiac electrical activity. The use of this porcine DCD model in subsequent research studies may provide a deeper understanding of the critical sentinel events of the withdrawal phase of DCD and point to potential areas for intervention.

Establishment of this model provides the platform for dissecting all of the events influencing outcome of DCD liver transplantation. Our next step is to transplant these organs and measure ischemia reperfusion injury parameters. With increasing samples size we can begin to quantify the degree of liver injury before and after transplantation and make correlations with the physiologic and molecular parameters. We also have a model in which interventions can be tested.


The authors would like to thank the New England Organ Bank for their support, and Annette Shepard-Barry and Inna Lomakin for their technical assistance.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.