Primary graft nonfunction and Kupffer cell activation after liver transplantation from non-heart-beating donors in pigs



More extensive use of non-heart-beating donors (NHBD) could reduce mortality on liver transplantation waiting lists, but this is associated with more primary nonfunction (PNF). We assessed which parameters are involved in the development of PNF in livers from NHBD in a previously validated pig liver transplantation model, in which livers were transplanted after exposure to incremental periods of warm ischemia. The risk of PNF was unacceptably high (>50%) when livers were exposed to >30 minutes' warm ischemia before a short cold ischemic period. This study examined how PNF is affected by Kupffer cell activation (β-galactosidase), the generation of cytokines tumor necrosis factor alpha and interleukin 6, antioxidant mechanisms (ascorbic acid, α-tocopherol, reduced glutathione), circulating redox-active iron, and sinusoidal endothelial cell function (hyaluronic acid clearance). Kupffer cells were more activated in PNF recipients, as suggested by higher β-galactosidase levels (15 minutes after reperfusion), and secondarily, by higher production of tumor necrosis factor alpha and interleukin 6 (180 minutes after reperfusion). In addition, α-tocopherol and reduced glutathione were lower, and ascorbic acid and redox-active iron higher in PNF recipients. Finally, PNF grafts displayed progressively decreasing hyaluronic acid clearance (suggesting sinusoidal endothelial cell dysfunction) and parenchymal edema. Consequently, a reduced-flow phenomenon was documented. In grafts from NHBD that are destined to fail, β-galactosidase activity (a surrogate of Kupffer cell activation) is higher, proinflammatory cytokines are overproduced, some antioxidant mechanisms fail, and circulating redox-active iron is more rapidly released. A no-flow phenomenon is eventually observed in these failing grafts. Liver Transpl 13:239–247, 2007. © 2007 AASLD.

The growing discrepancy between available and required liver grafts for transplantation is the rationale behind the use of so-called marginal donors. The use of organs from non-heart-beating donors (NHBD)—that is, donors who have experienced cardiac arrest—is one of the means to expand the number of organs.

However, unlike in kidney transplantation, experience with the use of NHBD for liver transplantation (LT) is still limited. The most important reasons are the high risk of primary nonfunction (PNF), a complication that, unlike for the kidney, causes recipient death in the absence of a rapid retransplantation, and biliary complications.1, 2 Wider application of NHBD-LT in the future will depend on better defining the maximum period of warm ischemia (WI) that a liver can tolerate without precluding graft function, and developing strategies to improve tolerance to WI.

We previously showed in a preclinical model of NHBD-LT, by means of incremental WI periods before a short cold ischemia period, that 15 minutes' WI is well tolerated, that 30 and 45 minutes' WI induces an unacceptable high rate of PNF (50%), and that 60 minutes' WI induces 100% PNF.3 To gain a new insight into the in vivo mechanisms causing failure (PNF) of NHBD liver grafts, we determined in the present study how increasing lengths of WI, before cold ischemia, affect Kupffer cell activation, the production of proinflammatory cytokines, the function of sinusoidal endothelial cells (SEC), and arterial and venous blood flow into the graft. We also assessed the antioxidant status by analyzing blood levels of ascorbic acid, α-tocopherol, reduced glutathione (GSH), red blood cell superoxide dismutase (SOD) and glutathione peroxidase (GSH-PX) activity, and levels of redox-active iron (which has been recently recognized as an important mediator of hepatocyte injury during ischemia-reperfusion injury). A better understanding of the mechanisms involved in failure of NHBD liver grafts may allow strategies to be designed to alleviate this injury.


NHBD, non-heart-beating donors; LT, liver transplantation; PNF, primary nonfunction; WI, warm ischemia; GSH, reduced glutathione; SEC, sinusoidal endothelial cell; SOD, superoxide dismutase; GSH-PX, glutathione peroxidase; POD, postoperative day; EDTA, ethylenediaminetetraacetic acid; TNF, tumor necrosis factor; IL, interleukin; ROS, reactive oxygen species.


Porcine Model of NHBD-LT

This study was approved by the local animal care committee and carried out in accordance with the local guidelines on the care and the use of laboratory animals.4 We used our previously described preclinical porcine model of NHBD-LT with livers originating from controlled NHBD.3 Briefly, in each experimental group, containing 6 animals, donor livers were exposed to a period of in situ WI (0, 15, 30, 45, or 60 minutes in group 1, 2, 3, 4, and 5, respectively), flushed with cold (4°C) histidine tryptophan ketoglutarate solution, procured, cold stored for 4 hours, and transplanted following standard techniques.3 After graft reperfusion, flow probes were implanted around the hepatic artery and the portal vein (Transonic, Ithaca, NY) without kinking or narrowing these vessels. The blood flow was monitored postoperatively with a dual-channel ultrasonic transit time volume flow meter (Transonic). Arterial blood pressure was kept stable by adequate fluid management throughout the first 3 hours after reperfusion. Bile was drained into a collecting bag and readministered in the proximal jejunum through a feeding jejunostomy.

In this model, PNF was defined as irreversible in the following instances: (1) persistent encephalopathy, (2) metabolic acidosis and hypoglycemia, (3) severe coagulopathy, and (4) reduced or absent bile production and increased serum aspartate aminotransferase levels. No PNF was observed in groups 1 and 2 (0 and 15 minutes WI, respectively) (Table 1). In groups 3 and 4 (30 and 45 minutes' WI) PNF rate was 50%, whereas 60 minutes' WI (group 5) provoked 100% PNF.

Table 1. Characteristics of Pig Livers Transplanted After Exposure to Incremental Periods of Warm Ischemia
CharacteristicGroup 1Group 2Group 3Group 4Group 5P value
  • *

    One death occurred at day 1 after transplantation (pulmonary edema).

  • One death occurred at day 1 after transplantation (no clear cause of death found).

  • One death occurred at day 1 after transplantation (inability to wean from ventilator, possibly because of brainstem injury).

Warm ischemia time0 min15 min30 min45 min60 min
Primary nonfunction0/6 (0%)0/6 (0%)3/6 (50%)3/6 (50%)6/6 (100%)
Day 4 survival6/6 (100%)5/6 (83%)*2/6 (33%)2/6 (33%)0/6 (0%) 
Aspartate aminotransferase 15 minutes after reperfusion (U/L)129 ± 56574 ± 5041,861 ± 4911,958 ± 1,1663,622 ± 1,564<0.05
Factor V 3 hours after reperfusion (%)49 ± 1348 ± 2411 ± 817 ± 104 ± 20.002
Bile production 3 hours after reperfusion (mL)30 ± 7.918.8 ± 6.914.6 ± 9.211.2 ± 6.92.3 ± 20.106


Blood Sample Collection

Blood samples were taken from a central venous catheter placed in the internal jugular vein. To better reflect changes occurring within the liver itself, the central catheter was placed into the outflow tract of the suprahepatic inferior vena cava. Blood samples were sequentially collected at the following time points: before laparotomy (baseline), at the end of the anhepatic phase (0 minutes), at 5, 15, 60, and 180 minutes after reperfusion, and at postoperative day (POD) 1, 2, 3, and 4. Blood was collected into ethylenediaminetetraacetic acid (EDTA) tubes (tumor necrosis factor alpha [TNF-α], interleukin 6 [IL-6], and hyaluronic acid), into trace-element-free Vacutainers (β-galactosidase, redox-active iron). After centrifugation of whole blood (10 minutes, 3,000g), plasma or serum was frozen at −25°C and stored until analysis, unless stated otherwise.

Kupffer Cell Activation

β-Galactosidase levels were measured fluorometrically in serum by using a specific substrate, 4-methylumbelliferyl-galactoside, and by detecting the formation of 4-methylumbelliferone, as previously described.5


TNF-α and IL-6 were measured in plasma with commercially available pig-specific enzyme-linked immunosorbent assay kits (R&D Systems, Abingdon, UK). All samples, including standard and control solution, were assayed in duplicate.

Antioxidant Status

Whole EDTA blood was centrifuged (15 minutes, 900g). For the determination of ascorbic acid, EDTA plasma was stabilized with 10% metaphosphoric acid and frozen at −80°C until analysis. Plasma concentrations of total ascorbic acid were determined spectrophotometrically by the reduction of 2,6-dichloro-phenol-indophenol (Merck, VWR, Belgium) according to Omaye et al.6

Plasma concentrations of α-tocopherol were determined by high-pressure liquid chromatography on reverse-phase columns (C18 120, 100 × 45) with an isocratic elution (methanol-water 98:2) and ultraviolet detection at 280 nm as described by Bieri et al.7 Intra- and interassay coefficients of variation (CV) were <3.7%.

For the determination of GSH, blood samples were snap-frozen and stored at −80°C until analysis. GSH was determined spectrophotometrically by the Bioxytech kit (Oxis International, Portland, OR). Intra- and interassay CV for GSH was 1 and 3%.

SOD, GSH-PX, and hemoglobin were determined from EDTA blood, kept on ice, and analyzed within 8 hours. After centrifugation (15 minutes, 900g), erythrocytes were hemolyzed and analyzed spectrophotometrically with Ransod and Ransel kits (Randox Laboratories, Crumlin, United Kingdom) respectively. Intra- and interassay CVs were lower than 5 and 7%, respectively.

Redox-Active Iron

The bleomycin assay for detecting redox-active iron in pig serum was carried out according to a previously described method, adopted for small sample volumes.8 The assay had a detection limit of 0.1 μM.

SEC Function

Hyaluronic acid levels were measured in plasma with a commercially available sandwich protein binding assay in a microplate format (Corgenix, Peteresborough, UK). The assay uses microwells coated with a highly specific hyaluronic acid binding protein from bovine cartilage to capture hyaluronic acid, and an enzyme-conjugated version of hyaluronic acid binding protein to detect and measure hyaluronic acid in samples. Reference solutions (prepared from rooster comb hyaluronic acid) were used to calculate test results (ng/mL). All samples, standards, and control solution were assayed in duplicate.

Assessment of Hepatic Flow

The average blood flow in the hepatic artery and portal vein was measured by separate flow probes through transit-time ultrasound measurements. The blood flow was continuously recorded during the first 3 hours after graft reperfusion and daily thereafter in surviving animals.

Statistical Analysis

Data are presented as mean ± SD. Two types of analysis were performed. A first analysis was performed that compared differences at different time points between experimental groups, either in function of the WI period or in function of outcome (PNF vs. non-PNF). Therefore, a 1-way analysis of variance was conducted by pairwise multiple comparison procedures (different WI periods) or t test (PNF vs. non-PNF). In a second analysis, we studied the evolution over time of parameters after reperfusion compared with baseline values in PNF vs. non-PNF recipients by a Friedman repeated-measures analysis of variance on ranks. A P value < 0.05 was considered significant. The analysis was performed by SPSS 12 and Sigmastat 2.0 for Windows (SPSS, Chicago, IL).


Kupffer Cell Activation

In the control group (no WI), serum β-galactosidase activity remained below the limit of detection (50 U/mL) in 5 of 6 recipients before reperfusion and at 15, 60, and 180 minutes after reperfusion. In groups 2, 3, 4, and 5 (with WI), β-galactosidase levels increased after reperfusion and peaked early at 15 minutes (397 ± 242, 371 ± 144, 399 ± 163, and 874 ± 296 U/L, respectively) and gradually decreased thereafter (at 60 and 180 minutes after reperfusion).

Serum β-galactosidase activity was significantly higher in PNF vs. non-PNF recipients at 15 minutes (664 ± 157 vs. 220 ± 93 U/L; P = 0.03), at 60 minutes (328 ± 69 vs. 155 ± 44 U/L; P = 0.027), and at 180 minutes (281 ± 84 vs. 107 ± 25 U/L; P = 0.05) after reperfusion (Fig. 1).

Figure 1.

Activity of β-galactosidase in serum of PNF and non-PNF recipients before laparotomy (baseline), at end of anhepatic phase (0 minutes), at 15, 60, and 180 minutes after reperfusion, and at POD 1, 2, 3, and 4. Data are mean ± SD. *P < 0.05 between PNF and non-PNF recipients.


TNF-α concentrations gradually increased after reperfusion in all groups (P < 0.05 at 180 minutes compared with baseline), but TNF-α concentrations were higher in PNF than in non-PNF recipients and peaked at 180 minutes after reperfusion in both groups (512 ± 151 pg/mL vs. 283 ± 79 pg/mL, respectively; P = 0.086) (Fig. 2A).

Figure 2.

Concentrations of (A) TNF-α and (B) IL-6 in plasma of PNF and non-PNF recipients before laparotomy (baseline), at end of anhepatic phase (0 minutes), at 15, 60, and 180 minutes after reperfusion, and at POD 1, 2, 3, and 4. Data are mean ± SD. *P < 0.05 between PNF and non-PNF recipients.

A substantial increase in serum IL-6 was observed at 180 minutes after reperfusion, reaching significantly higher concentrations in PNF vs. non-PNF recipients (3,982 ± 763 vs. 2,120 ± 300 pg/mL, respectively; P = 0.034) (Fig. 2B).

Antioxidant Status

Plasma ascorbic acid significantly increased after reperfusion in both PNF and non-PNF recipients compared with baseline levels (P < 0.001; Fig. 3A). In PNF recipients, ascorbic acid was higher than in non-PNF recipients at 60 minutes (21.39 ± 1.7 vs. 16.28 ± 1.33 μg/mL, P = 0.017) and at 180 minutes after reperfusion (19.74 ± 2.1 vs. 14.56 ± 1.17 μg/mL, P = 0.072).

Figure 3.

Concentrations of (A) plasma ascorbic acid, (B) α-tocopherol, and (C) GSH before laparotomy (baseline), and at 5, 60, and 180 minutes after reperfusion. Solid lines, PNF recipients; dashed lines, non-PNF recipients. Data are mean ± SD.*P < 0.05 between PNF and non-PNF recipients.

At 60 and 180 minutes after reperfusion, plasma α-tocopherol levels were significantly (P < 0.05) lower in both PNF recipients (0.980 ± 0.108 and 0.745 ± 0.08 μg/mL, respectively) and non-PNF recipients (1.352 ± 0.111 and 0.745 ± 0.08 μg/mL, respectively), when compared with baseline levels (1.802 ± 0.16 for PNF, and 2.136 ± 0.182 μg/mL and non-PNF) (Fig. 3B). At 60 minutes after reperfusion, levels of α-tocopherol were significantly lower in PNF recipients vs. non-PNF recipients (P = 0.028) and remained lower at 180 minutes after reperfusion (P = 0.08).

Blood levels of GSH were lower in PNF compared with non-PNF recipients at 5 minutes (421.8 ± 19.7 vs. 493.2 ± 27.8 μmol/L; P = 0.05), at 60 minutes (363.5 ± 20.9 vs. 445.1 ± 27.2 μmol/L; P = 0.06), and at 180 minutes (416.1 ± 43.4 vs. 489.5 ± 33.8 μmol/L; P = 0.19) (Fig. 3C). In PNF recipients, GSH after reperfusion was lower than baseline levels (421.8 ± 18.6 and 363.5 ± 20.9 μmol/L at 5 and 60 minutes vs. 486.3 ± 31.1 μmol/L; P = 0.09, and P = 0.003, respectively). In non-PNF recipients, GSH was not significantly different from baseline levels (493.2 ± 27.8 and 445.1 ± 27.2 at 5 and 60 minutes vs. 413.5 ± 44.3 μmol/L; P = 0.143, and P = 0.548, respectively).

At different time points after reperfusion, no differences in erythrocyte SOD and GSH-PX were observed among grafts exposed to different WI periods or according to PNF vs. non-PNF.

Redox-Active Iron

Levels of redox-active iron in serum increased both in PNF and non-PNF recipients after reperfusion (P < 0.05 at 15, 60, and 180 minutes compared with baseline). Redox-active iron was higher in PNF recipients at 15 and 60 minutes (0.19 ± 0.021 and 0.21 ± 0.03 μmol/L) compared with non-PNF recipients (0.077 ± 0.018 and 0.11 ± 0.028 μmol/L) (P < 0.0001 and P = 0.054, respectively) (Fig. 4).

Figure 4.

Redox-active iron in serum of PNF and non-PNF recipients before laparotomy (baseline), at 5, 60, and 180 minutes after reperfusion. Solid lines, PNF recipients; dashed lines, non-PNF recipients. Data are mean ± SD. *P < 0.05 between PNF and non-PNF recipients. Redox-active iron levels increased both in PNF and non-PNF recipients after reperfusion and are markedly higher in PNF recipients at 15 and 60 minutes.

SEC Function

A first increase in plasma hyaluronic acid was observed in all experimental groups at the end of the anhepatic phase, and levels further increased at 15 minutes after reperfusion in all experimental groups. Highest levels were monitored in group 5 (1,450 ± 99 ng/mL) compared with groups 1, 2, 3, and 4 (628 ± 68, 830 ± 240, 959 ± 168, and 1,250 ± 145 ng/mL; P = 0.017) at 60 minutes after reperfusion. At 180 minutes after reperfusion, hyaluronic acid levels in group 5 (1,420 ± 180 ng/mL) remained increased, whereas levels decreased in all other groups (292 ± 61, 455 ± 82, 620 ± 141, and 955 ± 165 ng/mL in groups 1, 2, 3, and 4, respectively; P < 0.0001).

At 60 and 180 minutes after reperfusion, hyaluronic acid was higher in PNF compared with non-PNF recipients (1,295 ± 111 vs. 782 ± 81 ng/mL; P = 0.013 and 1,216 ± 137 vs. 448 ± 62 ng/mL; P < 0.0001, respectively) (Fig. 5).

Figure 5.

Concentrations of hyaluronic acid in plasma of PNF and non-PNF recipients before laparotomy (baseline), at end of anhepatic phase (0 minutes), at 15, 60, and 180 minutes after reperfusion, and at POD 1, 2, 3, and 4. Data are mean ± SD. *P < 0.05 between PNF and non-PNF recipients. Hyaluronic acid was higher in PNF compared with non-PNF recipients at 60 and 180 minutes after reperfusion.

Hepatic Blood Flow

At 60 minutes after reperfusion, arterial blood flow was similar among PNF and non-PNF grafts (137 ± 57 and 125 ± 92 mL/min, respectively; P = 0.883) (Fig. 6). At 180 minutes after reperfusion, arterial blood flow in PNF grafts decreased and became lower compared with non-PNF grafts (87 ± 55 and 220 ± 75 mL/min, respectively; P = 0.004). Similarly, portal blood flow at 60 minutes after reperfusion was similar among PNF and non-PNF grafts (466 ± 178 and 486 ± 168 mL/min, respectively; P = 0.425), whereas at 180 minutes, it became lower in PNF than in non-PNF grafts (328 ± 144 vs. 480 ± 118 mL/minutes; P = 0.031).

Figure 6.

Blood flow in PNF and non-PNF recipients was measured 60 and 180 minutes after reperfusion in hepatic artery and in portal vein. Open triangle, arterial blood flow in PNF recipients; solid triangle, arterial blood flow in non-PNF recipients; open circle, portal blood flow in PNF recipients; solid circle, portal blood flow in non-PNF recipients. Data are mean ± SD. *P < 0.05 between PNF and non-PNF recipients. At 180 minutes after reperfusion, blood flow is lower in hepatic artery and in portal vein in PNF grafts compared with non-PNF recipients.


We previously showed in a preclinical porcine model of NHBD-LT that exposure of the liver to increasing lengths of WI before a short fixed period of cold ischemia is associated with more hepatocellular injury, an increasing risk of PNF, and subsequently worse recipient survival.3 In the present study, we investigated in vivo mechanisms involved in causing failure (PNF) of NHBD liver grafts by studying how increasing lengths of WI affect Kupffer cell activation, the balance between pro- and anti-inflammatory cytokines, antioxidant status and redox-active iron in blood, and the function of SEC. For these parameters, circulating surrogate markers were studied because the preclinical model used precludes the performance of repeated liver biopsies after transplantation. The arterial and venous blood flow into the graft was also studied.

β-Galactosidase is a lysosomal glycohydrolase primarily released on activation of Kupffer cells.9, 10 Kupffer cells are known to play a major role in ischemia-reperfusion injury, but to our knowledge, the effect of WI on Kupffer cell activation in a LT model has not been studied in detail before. In our NHBD-LT model, we found that (1) exposure to WI induces an increase in plasma levels of β-galactosidase; (2) this increase is the highest in recipients of grafts exposed to the longest WI; (3) this phenomenon of Kupffer cell activation occurs rapidly (15 minutes) after reperfusion; and (4) β-galactosidase discriminates between recipients with fatal injury (PNF) from nonfatal liver injury (non-PNF).

β-Galactosidase was already proposed as a marker of ischemia-reperfusion injury after ex vivo normothermic reperfusion of liver grafts subjected to 24 hours' cold preservation time.11 In our study, however, β-galactosidase levels remained low when livers had been exposed to cold ischemia only. A shorter period of cold ischemia (4 hours) and the use of an in vivo LT model may account for this difference between our experimental results and those of St Peter et al.11

The highest increase of β-galactosidase in liver recipients exposed to the longest WI in our model is in line with previous observations in a porcine model of in situ hepatic ischemia-reperfusion.12 Whereas in the latter study β-galactosidase peaked within 3 hours after reperfusion, the release of β-galactosidase in liver recipients exposed to WI was even more rapid. This suggests that in livers exposed to WI, Kupffer cells became activated early after reperfusion, or were already activated before reperfusion. The latter possibility is supported by recent electron microscopic analysis of biopsy samples obtained after WI (but before cold storage and reperfusion) and showing, in contrast with baseline conditions, unequivocal signs of Kupffer cell activation: enlargement, presence of many large heterogeneous lysosomes, and fragments of phagocytized red blood cells (data not shown). Finally, the levels of β-galactosidase discriminated recipients with fatal injury (PNF) from those with functioning grafts. This is consistent with findings of Liu et al.12 who reported highest β-galactosidase levels after irreversible liver ischemia-reperfusion injury caused by 90 minutes in situ WI.

Altogether, these data suggest that Kupffer cells are rapidly activated by WI and play an important role in initiating the cascade of inflammatory events leading to failure (PNF) of NHBD liver grafts. Therefore, inhibition or modulation of Kupffer cell activation appears critical in attenuating the reperfusion injury after NHBD-LT. This is in agreement with the results of others.13–17 These groups showed in different models of NHBD-LT (pig and rat, and ex vivo LT models) that the elimination of Kupffer cells in the donor by liposome-encapsulated dichloroethylene diphosphanate, 42 hours before obtaining the liver, had a beneficial effect on graft function and cytokine production. Our study complements these earlier findings by demonstrating the dependence of Kupffer cell activation and the following graft failure on the duration of WI.

To further investigate the effect of WI on Kupffer cells in NHBD-LT, we monitored the circulating levels of TNF-α and IL-6, both of which are proinflammatory cytokines mainly released by activated Kupffer cells. In PNF recipients, the early activation of Kupffer cells was followed (at 3 hours after reperfusion) by higher TNF-α and IL-6 levels compared with non-PNF recipients. This late production of proinflammatory cytokines may have been initiated earlier, during WI. Indeed, biopsy samples taken before reperfusion had higher expression of TNF-α mRNA (by reverse transcriptase–polymerase chain reaction) in livers exposed to longer WI times (data not shown). This suggests once more that Kupffer cells may be “preactivated” as a direct effect of WI and that this can lead to abundant proinflammatory cytokine secretion after reperfusion. Although IL-6 has been found to be hepatocyte protective in a rodent model,18 it is considered proinflammatory and a marker of increased mortality and morbidity.19 This is consistent with our observations, in which the dramatic 70-fold increase of IL-6 was associated with a fatal outcome (PNF) after transplantation of livers exposed to prolonged WI.

Apart from the generation of proinflammatory cytokines, Kupffer cells have been repeatedly shown to play a prominent role in generation of reactive oxygen species (ROS).20 We therefore measured certain antioxidant mechanisms that normally counteract the production of ROS. After reperfusion, we observed an increase in ascorbic acid, a decrease in α-tocopherol and GSH, and no change in SOD and GSH-PX activity in the recipients.

Ascorbic acid, an important general reducing agent, increased after reperfusion, as observed after lung and brain ischemia21, 22; this perhaps reflects an adaptive response to ROS during ischemia-reperfusion. Higher levels of ascorbic acid in PNF compared with non-PNF recipients may reflect an increased demand as an adaptive response to a higher generation of ROS.

α-Tocopherol declined after reperfusion in all animals. A similar decrease has also been previously noted after experimental and clinical LT; it probably reflects an increased consumption as a result of ROS production.23, 24 The new finding in our experiments is that this decrease in α-tocopherol was more pronounced in PNF grafts, where we noted an ongoing consumption of α-tocopherol between 60 and 180 minutes after reperfusion.

GSH represents another endogenous, intracellular antioxidant mechanism. In PNF recipients, we observed after reperfusion decreasing GSH concentrations, which were greatly lowered at 5 and 60 minutes compared with non-PNF recipients in which GSH concentrations initially increased and remained stable thereafter. In failing grafts, this could reflect an enhanced oxidation or depletion of intracellular GSH observable in plasma and/or an impaired or insufficient endogenous antioxidant defense as a direct result of substantial liver damage.

It has been shown that iron catalyzes the production of ROS and might be implicated in the pathophysiology of ischemia-reperfusion injury in general.25, 26 During ischemia, chelatable redox-active iron is released from intracellular stores and thus is made available to catalyze the subsequent formation of ROS on reoxygenation.27 In postischemic liver tissue, extracellular transferrin-bound iron can provide an additional source (through the transferrin receptor) of intracellular redox-active iron levels.28 Iron chelators have been recently proven to ameliorate WI reperfusion injury in the kidney.29 The liver has the largest iron storage capacity in the body, and WI is likely to induce a major iron release. For these reasons, we were interested in studying the evolution of redox-active iron in our NHBD model and particularly in determining whether a differential pattern would discriminate PNF grafts from functioning grafts. We observed a higher systemic release of redox-active iron in PNF recipients as early as 15 minutes after reperfusion. It is of note that we showed earlier that sensitive markers of hepatocellular damage, such as aspartate aminotransferase and liver fatty acid binding protein,3, 30 also appear early in the circulation, suggesting that hepatocyte necrosis may account for the initial increase of redox-active iron.

Redox-active iron in circulation has been increasingly recognized as an important mediator of hepatocyte injury, probably through an oxidative stress mechanism.31 Interestingly, excess iron has been shown to induce NF-κB activation and subsequent TNF-α release by Kupffer cells,32 and this could account for the higher TNF-α production that we see in PNF grafts. On the basis of these data, the use of iron chelators should be considered when designing strategies to ameliorate graft viability after NHBD-LT.

To assess the influence of WI on the functional integrity of SEC, hyaluronic acid was measured. This mucopolysaccharide is selectively and rapidly cleared (half-life 5-6 minutes) from the circulation by hepatic SEC. In porcine models of LT from heart-beating donors, hyaluronic acid clearance is regarded as an early and reliable predictor of SEC damage and allograft viability.33, 34 So far, to our knowledge, no studies have looked at the hyaluronic acid clearance after NHBD-LT. We observed a first increase of hyaluronic acid at the end of the anhepatic phase, independent of the duration of WI. This is a direct consequence of absent hyaluronic acid clearance during the anhepatic phase. After graft reperfusion, SEC function recovered faster in livers subjected to no or only a short period of WI. In PNF recipients, SEC function remained severely impaired compared with non-PNF recipients; this was most noticeable at 60 minutes and 180 minutes after reperfusion. Histology and electron microscopy revealed that signs of SEC damage were present but discrete after exposure to WI only, but they evolved into massive zones of coagulation necrosis and complete disruption of SEC lining 60 minutes after reperfusion in PNF grafts (Monbaliu et al., Transplantation 2005; and data not shown). Altogether, these data suggest that failure of the microcirculation, although not immediate, represents an important final event in the pathogenesis of PNF. This is also supported by our hepatic blood flow measurements. Indeed, both the arterial and the portal blood flow were substantially reduced 3 hours after reperfusion in failing grafts, an observation that directly demonstrates the progressive development of a “no-flow phenomenon” in PNF grafts.

In conclusion, we found, in a preclinical model of NHBD-LT, that in grafts destined to fail, β-galactosidase (a surrogate of Kupffer cell activation) is higher and is followed by the overproduction of proinflammatory cytokines (TNF-α and IL-6). PNF recipients are also characterized by the failure of certain antioxidant mechanisms and release of large quantities of redox-active iron. Finally, failing grafts display progressive microcirculatory disturbances and establishment of a progressive no-flow phenomenon.

Biological interventions aimed at improving outcome of NHBD-LT should be multifactorial in nature and focus on modulation of Kupffer cell activation, inhibition of proinflammatory cytokines, optimization of antioxidant mechanisms, chelating circulating redox-active iron, and ultimately in maintaining adequate flow and viability to the graft.


Presented in part at the American Transplant Congress, Seattle, Washington, May 22, 2005. We thank André Vanderputten (Tramedico, St. Niklaas, Belgium) and Dr. E. Schaffner (Köhler Chemie, Alsbach-Hahnlein, Germany) for providing histidine tryptophan ketoglutarate solution; Lydia Coolen for help with article's preparation; Hedwig Bogaerts for the advice on the statistical analysis; and Jean Paul Cheramy, Veerle Heedfeld, Christel Dubuisson, and Marleen Vanrusselt for the expert technical assistance.