Vascular complications after liver transplantation are rare but potentially fatal,1-3 and up to 50% of patients with vascular complications require retransplantation.3 Arterial thrombosis is the most common complication after liver transplantation, with a reported incident of up to 12% in adult recipients and up to 40% in pediatric recipients, and it constitutes up to 60% of all posttransplant vascular complications.4 Hepatic arterial thrombosis occurs about as often early after liver transplantation (within 30 days after the operation, 46.7%) as late (after 30 days, 53.3%).3 However, severe allograft dysfunction occurs predominantly after early hepatic arterial thrombosis, which also is associated with increased mortality.3 Late hepatic arterial thrombosis, on the other hand, results in biliary tract complications. Other vascular complications are inferior caval and hepatic vein thrombosis (1%),4 portal thrombosis (4%), and arterial and portal stenosis and aneurysm.5 Acute portal vein thrombosis or stenosis during the early course after liver transplantation may result in graft failure and necessitate retransplantation in up to 15% of the patients.2 Early detection of impaired liver graft function is thus vital for efficient treatment. The microdialysis technique6, 7 may be a suitable monitoring technique for vascular complications in the liver, as it enables continuous bedside monitoring of liver cell metabolism through a microdialysis catheter placed in the liver tissue.8, 9 The catheter has a tubular semipermeable dialysis membrane on its tip and is slowly perfused with Ringer solution, which equilibrates over the membrane with the molecules in the surrounding interstitial fluid (see Rosdahl et al.10). The dialysis fluid is collected, and glucose, lactate, pyruvate, and glycerol are analyzed by a bedside analyzer every 20 minutes.
Fluctuations in glucose may reflect changes in the blood glucose level or cellular glucose uptake,10 glycogenolysis, or changes in blood flow and thus altered glucose delivery to the tissue.11 Glycogenolysis is normally triggered in order to maintain a steady blood glucose level12 but can also be triggered by liver ischemia.13-16 During ischemia, large quantities of lactate are produced. However, increased lactate production can also be the result of hypermetabolism.17 To discriminate ischemia from hypermetabolism, pyruvate, which is the precursor of lactate, is analyzed, and the lactate/pyruvate ratio is calculated. In case of hypermetabolism, both lactate and pyruvate increase, and therefore the ratio remains unchanged, whereas during hypoxia/ischemia, pyruvate decreases and lactate increases; this leads to an increased lactate/pyruvate ratio.18 Glycerol is released primarily from fatty acid and phospholipid metabolism. As phospholipids are the main compounds of the cell membrane, glycerol release may also indicate cell membrane damage.19-22 In the present study, glycerol was considered as a marker for (hepatic) cellular membrane disintegration.
There are several publications based on the clinical use of microdialysis in transplanted livers that report significant correlations between increased interstitial lactate in the liver with sinusoidal endothelial cell injury, initial poor liver function,23 and reperfusion injury24 after liver transplantation. Other studies have described the normal course of glucose, lactate, glycerol, and pyruvate postoperatively after liver transplantation.8 Although clinical investigations are mainly observational, an experimental preclinical setting offers the possibility of inducing and studying liver ischemia under controlled conditions. The present study was undertaken to assess microdialysis as a monitoring technique for detection of vascular complications following liver surgery through the clamping of either the hepatic arteries or the portal vein. Warm liver ischemia in the pig is well tolerated for 120 minutes.25, 26 To ensure successful reperfusion and recovery, the hepatic artery was occluded for only 2 hours. Portal occlusion, on the other hand, is better tolerated by the liver27 and was therefore extended to 3 hours. In this group, portocaval shunts were made for splanchnic decompression.28, 29
MATERIALS AND METHODS
The principle of microdialysis is to mimic the passive function of a capillary blood vessel by the perfusion of a tubular, semipermeable membrane introduced into a tissue. The microdialysis probe has previously been described in detail.6, 7, 10 In general, it is a thin, double-lumen, concentric plastic tube with a 30-mm, semipermeable, tubular membrane (cutoff of 20,000 Da) at its distal end (see Rosdahl et al.10). A perfusion fluid is pumped through the outer tube and flows underneath the membrane, where the exchange between the interstitial fluid and the perfusion fluid takes place. At the tip, the fluid enters a small hole in the inner tube and flows backwards to finally be collected into a microvial (CMA Microdialysis AB, Stockholm, Sweden).
CMA 70 microdialysis catheters (CMA Microdialysis AB) with a 60-mm shaft and a 30-mm membrane were used in the liver, and CMA 60 microdialysis catheters (CMA Microdialysis AB) with a 20-mm shaft and a 30-mm membrane were used as subcutaneous reference catheters. They were all perfused with a perfusion fluid (T1 perfusion fluid, CMA Microdialysis AB; 147 mM Na+, 4 mM K+, 2.3 mM Ca2+, and 156 mM Cl−), which was pumped at a speed of 0.3 μL/minute with a CMA 106 pump (CMA Microdialysis AB).
In all groups, a CMA 60 catheter was placed subcutaneously as a reference catheter to distinguish local ischemic reactions in the liver from systemic changes.30, 31 The sampling frequency was 20 minutes.
Throughout the experiment, the vial holder was kept at the same level as or lower than the tip of the microdialysis catheters, as there was a risk of getting too small sample volumes per vial if the pump had to work against the hydrostatic pressure.32
All microdialysis catheters have an intrinsic delay proportional to the time that it takes to pump the perfusion fluid from the tip of the catheter to the vial. The intrinsic delay has been estimated to be around 20 minutes for the types of catheters used in this study (CMA Microdialysis AB, unpublished data, 2008).
Samples were analyzed with a CMA 600 microdialysis analyzer (CMA Microdialysis AB), a clinical chemistry analyzer using enzymatic reagents and colorimetric measurements. Substrate-specific reagents for glucose, lactate, pyruvate, and glycerol were used when the samples were being analyzed.
Anesthesia and Surgical Procedure
Eighteen female crossbred (Swedish Landrace/Yorkshire/Hampshire) littermate pigs, with a body weight of 30 to 34 kg, were divided into 3 groups of 6 pigs: an arterial occlusion group, a portal occlusion group, and a control group.
Before the operation, all animals were fasted for 26 hours with free access to water. Anesthesia was induced by an intramuscular premedication of 12 mg/kg ketamine hydrochloride (Ketaminol Vet, Veterinaria AG, Zurich, Switzerland), 5 mg/kg azaperone (Stresnil Vet, Janssen-Cilag Pharma, Wien, Austria), and 0.05 mg/kg atropine (Atropin, NM Pharma AB, Stockholm, Sweden).
Midazolam (1-4 mg/kg; Alpharma AS, Oslo, Norway) was given intravenously prior to intubation (Mallinckrodt Medical, Ireland; inner diameter = 7.0-7.5). The anesthesia was maintained by inhalation of 1:2 O2/N2O plus 0.5% to 1.5% halothane, which was complemented with fentanyl (Alpharma AS, Oslo, Norway) if necessary. A Ringer acetate intravenous infusion (7-10 mL/kg/hour) was given at 39°C. Blood gas analysis (i-STAT, Abbott, Abbott Park, IL; EG7+ cartridges), electrocardiogram, body temperature, and urine production were continuously recorded during the experiment. The body temperature was maintained at 38°C to 39°C with an external heating device when needed.
Before the abdominal incision, a reference catheter (CMA 60) was inserted into the subcutaneous tissue over the left pectoral area. Midline laparotomy was performed; afterward, CMA 70 catheters were introduced into the liver with a steel cannula with a split catheter according to the method described by Nowak et al.9 Each catheter was introduced into the lower margin in the middle of a large lobe.
After the placement of the catheters in the liver, the hepatic arteries were dissected and prepared for occlusion in the arterial group. In the portal group, the portal vein was dissected and prepared for occlusion, and an H-shunt vascular graft (polytetrafluoroethylene; Impra Carboflo, Tempe, AZ) with a 6-mm diameter and a 4- to 6-cm length was sutured between the portal vein and the inferior caval vein. The H-shunt was held closed until occlusion of the portal vein.
After 2 hours of baseline collection, the hepatic arteries or the portal vein were occluded, and in the portal group, the portal/inferior caval vein H-shunt was opened in order to provide venous outflow from the intestines. The occlusion was maintained for 2 hours in the arterial group and for 3 hours in the portal group. At the time of unclamping, the portocaval H-shunt was ligated. After reperfusion, the livers were monitored for another 3 hours in the arterial group and for another 2 hours in the portal group before the pigs were sacrificed. Control group animals were monitored for a total time of 7 hours without dissection or occlusion of any vessels.
The operations were carried out in a randomized order. Two operations were made at the same time in the same operation theater: arterial and portal group operations, arterial and reference group operations, or portal and reference group operations. All operations and all handling of the microdialysis were done under clean but not sterile conditions. The study was approved by the local animal ethics committee. Animals received care according to the Guide for the Care and Use of Laboratory Animals (NIH publication 86-23, 1985).
The data are presented as the mean ± the standard error of the mean. Differences between the arterial or portal occlusion group and the control group were analyzed with repeated-measures analysis of variance. Repeated-measures analysis of variance was also used for analysis of the differences between the liver and subcutaneous reference catheters (within a study group). The analyzed time points were 1 hour before clamping, 1 hour after the beginning of occlusion, and 2 hours after reperfusion. In the case of significant differences over time, post hoc tests with Bonferroni correction were applied in order to investigate when significance was reached. In the case of a missing value, it was replaced by the mean of the before and after values if possible (interpolation); otherwise, the before value was used. Significance was set to the level of P < 0.05.
The measured metabolites are glucose, glycerol, lactate, and pyruvate. Pyruvate, however, is used only for calculating the lactate/pyruvate ratio, as presented below.
Arterial Occlusion Group
In comparison with the control group, concentrations of all measured metabolites initially increased during arterial occlusion (P ≤ 0.005; Figs. 1A–4A). Significance was reached 60 minutes after occlusion in the lactate/pyruvate ratio, glucose, and glycerol, and lactate had increased significantly already after 40 minutes of occlusion. After peak levels at 1 hour, glucose and lactate started to decrease, and at the time of reperfusion, glucose had reached the baseline level. Glycerol and the lactate/pyruvate ratio remained stable at a high level throughout the occlusion period. During arterial occlusion, all measured metabolites were significantly higher in the liver than in the subcutaneous reference (P ≤ 0.05).
At the start of reperfusion, glucose in the arterial group had already returned to baseline levels. No further changes in glucose were observed during reperfusion (Fig. 1A).
Lactate, which started to normalize already during arterial occlusion, continued to decrease after reperfusion (P ≤ 0.001) to baseline levels. However, lactate decreased during the reperfusion period in the liver and in the subcutaneous tissue in all groups (P ≤ 0.001; Fig. 2A,B).
The lactate/pyruvate ratio (P ≤ 0.001) and glycerol (P = 0.01) started to decrease at reperfusion in the liver and reached baseline values after about 1.5 hours (Figs. 3A and 4A).
Portal Occlusion Group
Levels of the measured metabolites were stable during portal occlusion. However, the lactate/pyruvate ratio changed significantly over time (P ≤ 0.01) in comparison with the control group, mainly because of a decrease in the lactate/pyruvate ratio in the control group. Lactate levels tended to increase in both the liver and the subcutaneous tissue, but the increase was not significant.
In the portal group, all measured levels were stable after reperfusion, except for lactate, which showed a continuous decrease over time (P ≤ 0.001; Fig. 2A,B).
Control Group and Subcutaneous Reference
All measured metabolite levels in the subcutaneous reference were stable during the occlusion, except for the lactate/pyruvate ratio, which increased in the arterial group and in the portal group (both P < 0.001; Fig. 4B). Throughout the study, liver glucose was significantly higher in the control group than in the arterial and portal groups (Fig. 1A). Similarly, glucose levels were significantly higher in the control group subcutaneous reference catheter versus the subcutaneous reference catheters of the arterial and portal groups (Fig. 1B). There were no significant differences in the control group between the subcutaneous reference and liver catheters in the lactate, glycerol, or lactate/pyruvate ratio.
In the present study, we show that hepatic artery occlusion causes ischemia and cell damage that is detectable with microdialysis, as indicated by increased glucose, lactate, and lactate/pyruvate ratio as ischemic markers and intrahepatic glycerol as a marker of cell damage. Portal occlusion, however, did not result in any significant changes in the measured metabolites.
During the first hour of arterial occlusion, interstitial liver glucose showed a significant increase, and this was followed by a decrease to baseline levels during the second hour of occlusion. The initial increase was most likely due to glycogenolysis triggered by the hypoxia caused by arterial occlusion.13-16 Glucose levels normalized before reperfusion, and this may indicate that the liver starts to adjust to the impaired arterial blood flow, decreasing the rate of glycogenolysis. As the portal vein continuously perfused the liver, it is unlikely that the decrease in glucose levels could be explained by impaired blood glucose delivery.
During arterial occlusion, there was an initial increase in lactate, peaking at 1 hour, and subsequently, there was a slow decrease. This decrease in lactate coincided with the decrease of glucose levels, and furthermore, the lactate/pyruvate ratio increased significantly upon clamping, after which levels stabilized throughout occlusion, indicating a steady state of anaerobic metabolism. After reperfusion, the liver metabolism normalized, and all metabolites returned to baseline levels.
Elevated levels of glycerol are seen during arterial occlusion but not during portal occlusion. This indicates that the liver cell damage is more pronounced by impaired arterial flow than by portal vein blockage. The increased glycerol could also be caused by the increased fat metabolism that has been shown to occur during liver ischemia.33 During reperfusion, glycerol levels normalized, and this indicated that the liver cell injury was not progressive.
Occlusion of the portal vein resulted in no major metabolic changes. We interpret the lack of any major effect on liver metabolism as due to the hepatic arterial buffer response,34 an intrinsic property of the hepatic artery, which dilates upon restricted portal vein flow. The hepatic arterial buffer response has been reported to have the capacity to increase hepatic arterial flow by 25%,34, 35 which may be enough to support the liver with oxygen, because the liver has an extraordinary ability to increase its oxygen extraction.36 The liver is, however, unable to reciprocally compensate for an arterial occlusion by dilating the portal vein.34 One limitation of the current study is that portal occlusion was performed under splanchnic decompression by the performance of an H-shunt between the portal vein and the inferior caval vein. However, for a controlled and stable porcine model, this is necessary in order to avoid cardiac stress or even death because the pig is particularly susceptible to portal venous obstruction.28, 29 The tendency for increased lactate in the liver was similar to that in the subcutaneous reference catheter. This may be the result of the limited liver capacity for lactate uptake during portal occlusion as well as the use of an H-shunt because the lactate from the intestines enters the systemic blood system and then goes back to the liver through the hepatic artery, hence increasing interstitial lactate levels in the liver. The lack of significant changes during portal occlusion may be explained by the use of the H-shunt because physiological changes due to intestinal stasis did not add to the pathophysiological picture. Further studies are needed on other species, shorter ischemia times, partial portal occlusion, and the occlusion of portal vein branches.
Our findings may be relevant in liver transplantation, where reperfusion traditionally is performed through the portal vein first. In a previous study of porcine liver transplantation, in which the portal vein was reperfused before the hepatic artery, we demonstrated that the lactate/pyruvate ratio did not decrease until after arterial reperfusion.9 This indicates that hepatic ischemia continues until arterial reperfusion, and areas such as the biliary tract, mainly supplied by the hepatic artery,37 are exposed to warm ischemia until arterial reperfusion. Although clinical reports have not demonstrated any differences in reperfusion injury, graft function, or outcome between initial arterial reperfusion and initial portal reperfusion38, 39 and no major differences in liver function,40 preclinical studies have shown that initial arterial reperfusion may cause less reperfusion injury to the liver.41, 42 Additionally, the majority of clinical studies are based on plasma levels of serum alanine aminotransferase, serum aspartate aminotransferase, and serum bilirubin and do not focus on biliary complications. The biliary tract depends entirely on an arterial blood supply.37 The importance of early arterial reperfusion is further emphasized by a lower rate of biliary complications when simultaneous arterial and portal revascularization is used in humans.43
The differences between the arterial and portal groups may be explained by different surgical preparations of the liver hilus between the groups. This is, however, not likely because it has been shown that microdialysis data quickly normalize after preparation to the same level within 120 minutes, with or without preparation.44 A limitation of the study may be that only 1 microdialysis catheter was used in the liver, which was assumed to reflect changes in the whole liver; however, according to Kannerup et al.,45 there are no differences in metabolic changes during ischemia between liver lobes, and they emphasized that 1 microdialysis catheter in the liver is enough for postoperative liver monitoring. Our results demonstrate furthermore that measurements from the subcutaneous reference catheters did not differ from the control group liver catheter measurements. Thus, catheters placed subcutaneously may be used as a reference to the liver.
In conclusion, the present study shows that arterial occlusion causes ischemia and cell damage that is detectable with microdialysis, as indicated by increased glucose, lactate, lactate/pyruvate ratio, and glycerol. The present experimental conditions could not reveal whether portal occlusion can be detected by microdialysis, as there were no ischemic changes in the portal occlusion group. The lack of metabolic changes upon portal occlusion may be explained by the arterial buffer response, which increases the hepatic arterial flow and oxygen delivery to compensate for the restricted portal flow. Another possible interpretation of these results is that arterial occlusion is more harmful to the liver than portal occlusion.