Telephone: +47-23070000; FAX: +47-23073684;
Early bedside detection of ischemia and rejection in liver transplants by microdialysis†
Article first published online: 12 JUL 2012
Copyright © 2012 American Association for the Study of Liver Diseases
Volume 18, Issue 7, pages 839–849, July 2012
How to Cite
Haugaa, H., Thorgersen, E. B., Pharo, A., Boberg, K. M., Foss, A., Line, P. D., Sanengen, T., Almaas, R., Grindheim, G., Pischke, S. E., Mollnes, T. E. and Tønnessen, T. I. (2012), Early bedside detection of ischemia and rejection in liver transplants by microdialysis. Liver Transpl, 18: 839–849. doi: 10.1002/lt.23425
The study was financially supported by grants 2009007 and 2008104 from the South-Eastern Norwegian Health Authority.
- Issue published online: 12 JUL 2012
- Article first published online: 12 JUL 2012
- Accepted manuscript online: 12 MAR 2012 05:52AM EST
- Manuscript Accepted: 29 FEB 2012
- Manuscript Received: 9 DEC 2011
This study was performed to explore whether lactate, pyruvate, glucose, and glycerol levels sampled via microdialysis catheters in the transplanted liver could be used to detect ischemia and/or rejection. The metabolites were measured at the bedside every 1 to 2 hours after the operation for a median of 10 days. Twelve grafts with biopsy-proven rejection and 9 grafts with ischemia were compared to a reference group of 39 grafts with uneventful courses. The median lactate level was significantly higher in both the ischemia group [5.8 mM (interquartile range = 4.0-11.1 mM)] and the rejection group [2.1 mM (interquartile range = 1.9-2.4 mM)] versus the reference group [1.5 mM (interquartile range = 1.1-1.9 mM), P < 0.001 for both]. The median pyruvate level was significantly increased only in the rejection group [185 μM (interquartile range = 155-206 μM)] versus the reference group [124 μM (interquartile range = 102-150 μM), P < 0.001], whereas the median lactate/pyruvate ratio and the median glycerol level were increased only in the ischemia group [66.1 (interquartile range = 23.9-156.7) and 138 μM (interquartile range = 26-260 μM)] versus the reference group [11.8 (interquartile range = 10.6-13.6), P < 0.001, and 9 μM (interquartile range = 9-24 μM), P = 0.002]. Ischemia was detected with 100% sensitivity and greater than 90% specificity when a positive test was repeated after 1 hour. In 3 cases of hepatic artery thrombosis, ischemia was detected despite normal blood lactate levels. Consecutive pathological measurements for 6 hours were used to diagnose rejection with greater than 80% sensitivity and specificity at a median of 4 days before the activity of alanine aminotransferase, the concentration of bilirubin in serum, or both increased. In conclusion, bedside measurements of intrahepatic lactate and pyruvate levels were used to detect ischemia and rejection earlier than current standard methods could. Discrimination from an uneventful patient course was achieved. Consequently, intrahepatic graft monitoring with microdialysis may lead to the earlier initiation of graft-saving treatment. Liver Transpl, 2012. © 2012 AASLD.
As many as 20% of transplanted liver grafts are lost within the first year despite considerable therapeutic improvements in treatment during the last decades.1 Most grafts are lost within the first week or weeks. As a result, attempts to improve overall graft survival rates should include the early detection and adequate treatment of complications such as acute rejection2 and vascular occlusion with subsequent ischemia.3 The latter is an especially major contributor to the loss of grafts. Biliary complications and infections are other organ-threatening complications.4-6
Vascular complications such as hepatic artery thrombosis usually occur during the first week or weeks after transplantation and are the most frequent reasons for urgent retransplantation.5, 7, 8 Ischemia entails increased anaerobic glycolysis along with the increased production of lactate and an unchanged or decreased concentration of pyruvate.9, 10 Because the relatively high portal vein flow can conceal local ischemia, the blood lactate levels may be normal.11 Increased glycerol levels reflect a breakdown of phospholipids due to cell membrane degradation.12
The reported incidence of acute rejection in the literature is 30% to 60%.2 Acute rejection is suspected when the activity of circulating liver transaminases, the concentration of bilirubin, or both increase, but graft biopsy is required to confirm or refute the suspicion.13 However, because early biopsy findings may be inconclusive, rebiopsy is frequently required.14 Because of contraindications such as coagulopathy and anticoagulant treatments13, 15 and the particular need of children for general anesthesia during the procedure, some patients are given high doses of corticosteroids when there is clinical suspicion of cell-mediated inflammation rather than a definite diagnosis. Most episodes of acute rejection are cellular rather than antibody-mediated, and T lymphocytes are central players.2, 16 Although increased leukocyte activity implies increased aerobic glycolysis with simultaneously increased production of lactate and pyruvate,16 the potential benefits of metabolism monitoring are unresolved.
Microdialysis is a technique that enables monitoring of the tissues and organs of interest9 (Fig. 1). So far, the detection of brain ischemia is the best validated application of this method.12, 17-19 Most other studies have been performed in animals; however, more than 2000 clinical investigations have been published. In the United States, its clinical application is presently restricted to neurointensive care units because only the brain catheter (CMA 70, M Dialysis AB, Stockholm, Sweden) is approved by the Food and Drug Administration for clinical use. In Europe, it is also Conformité Européenne–marked for reconstructive surgery with muscle or subcutaneous flaps,20 gastrointestinal surgery for the detection of anastomotic leaks or infections,21, 22 liver surgery,23 and surgery in other organs. A membrane with a pore size of 20 kDa can detect metabolic substances (lactate, pyruvate, glucose, and glycerol), whereas a pore size of 100 kDa allows measurements of mediators of inflammation (cytokines, chemokines, and complement) without an increase in the outer diameter of the catheter. The catheters have mainly been used to monitor tissues at risk for ischemia,12, 17, 18, 20 including liver grafts.24-26 A pilot study from our group has suggested that the microdialysis method may detect rejection as well.26
We hypothesized that ischemia could be detected as increased anaerobic glycolysis causing increases in lactate levels (but not pyruvate levels) and thus an increased lactate/pyruvate (LP) ratio. Furthermore, we hypothesized that rejection could be detected as increased aerobic glycolysis with simultaneous increases in lactate and pyruvate levels and a stable LP ratio. The primary objective of the present study was to explore whether the monitoring of substances of glucose metabolism in liver grafts with microdialysis catheters could be used to detect ischemia and rejection. Subsequently, we explored whether this method could enable earlier detection than current clinical monitoring practices.
PATIENTS AND METHODS
All liver grafts transplanted over the course of 1 year were included in this open-label, prospective, observational study. The study protocol was approved by the regional ethics committee. Detailed oral and written information was provided, and an informed consent form was signed before participation.
During the course of 1 year (beginning in October 2008), 69 patients received 73 liver grafts. Each transplanted graft was defined as 1 study subject. Seven patients were given antirejection treatment without histological confirmation of the diagnosis and were excluded from the analyses because the diagnosis was regarded as uncertain in these cases. The donor, patient, and graft characteristics for the 66 investigated liver transplants are presented in Table 1.
|Donors (n = 60)|
|Age (years)||55.0 (2.2-82.6)|
|Graft recipients (n = 62)|
|Age (years)||51.3 (0.5-69.6)|
|MELD score||14 (5-39)|
|Liver transplantation (n = 66)|
|Transplantation before study||5|
|Transplantation twice during study||4|
|Split liver transplantation||14|
|Liver grafts (n = 66)|
|Weight (g)||1350 (141-2340)|
|Total ischemia time (hours)||8.0 (2.9-16.1)|
|Hepatic artery||210 (13-940)|
|Portal vein||1450 (210-3100)|
Study Groups and Diagnostic Criteria
The endpoints were defined as follows:
Ischemia. Grafts with vascular occlusion/stenosis or infarction confirmed by Doppler ultrasound and/or liver computed tomography (CT).
Rejection. Grafts with biopsy-confirmed acute cellular rejection (ACR).
Because leukocyte activation could imply lactate production, cases of infection (which were defined as an elevation of C-reactive protein, concomitant antimicrobial therapy, and a focus in the graft or close to the graft while the microdialysis catheter was in the liver) were analyzed separately. Cases with increases in circulating bilirubin and/or transaminase levels and histologically confirmed cholestasis in the graft were also analyzed separately because the biochemical picture of cholestasis may be similar to the biochemical picture of rejection.
The data for 1 case of acute rejection were confounded by preceding ischemia (thrombosis of the hepatic artery) and thus were not included in the statistical analyses. Another patient was given antirejection therapy for 2 separate episodes of rejection, and because biopsy was not performed the second time, these data were also excluded from analyses.
The microdialysis system consisted of a small battery-driven syringe pump coupled to a double-tubular catheter with a semipermeable membrane at the tip (Fig. 1). The catheter membranes used in this study had a pore size of 100 kDa, an outer diameter of 0.6 mm, and a length of 30 mm. A secure thread was positioned 60 mm from the tip (CMA 65, M Dialysis AB). After the vessel and bile anastomoses were finished, the left abdominal wall was punctured from the inside by a hypodermic needle (14G × 3 ½″, Sterican, B. Braun AG, Melsungen, Germany) through which the catheters were led (each catheter entered the abdomen through a separate hole; the left side of the abdominal wall was chosen to avoid the access for postoperative ultrasound examinations). The caudal part of each liver lobe was punctured 1 to 3 cm laterally from the falciform ligament in the cranial-lateral direction with a splittable introducer (1 mm Ø). In split transplants, only 1 catheter was used. After the removal of the splittable introducer, any possible bleeding was stopped by manual compression and, if necessary, by an absorbable blood clot–inducing material (Surgicel, Ethicon, Cornelia, GA). The catheters were secured in the falciform ligament or hepatic capsule by a 6.0 Vicryl suture (at which each catheter's secure thread was fixed). Finally, a skin secure procedure was performed with a 4.0 thread. The catheters were perfused with a fluid containing dextran and electrolytes (Plasmodex, Meda AB, Stockholm, Sweden) at a velocity of 1 μL/minute by microinjection pumps (CMA 107, M Dialysis AB). The lactate, pyruvate, glucose, and glycerol levels and the LP ratio were analyzed at the bedside every 1 to 2 hours (Iscus, M Dialysis AB) after the patient's arrival at the critical care unit. Samples were also frozen for later analyses of inflammatory mediators (which will be discussed in a separate publication). The catheters were kept in situ for as long as they were able to sample microdialysates, and then they were removed transcutaneously.
Increases from one day to the next in circulating bilirubin or alanine aminotransferase (ALT) levels of at least 25% were considered pathological. With respect to lactate levels in microdialysates, increases of at least 50% from stable baseline values were considered pathological. However, particularly to verify our rejection hypothesis in a mathematical, researcher-independent way, the initial exploration of the data was performed with linear mixed models. Except for ischemic events that could clearly be defined by radiological examinations, data from other grafts were coded according to the endpoints of the study: rejection, infection, and cholestasis. Thus, the time point of the occurrence (eg, rejection) was not regarded at this stage in the data analyses. We searched for potential group differences, time dependencies, and possible effects of donor, recipient, and graft characteristics. The measured metabolic mediators were dependent variables. A random effects model was used, and we performed model selection for each variable by choosing the model achieving the lowest information criteria. The random effects of patients and groups were included in the model. Between-groups comparisons were further performed with the Mann-Whitney U test, and the Wilcoxon signed-rank test was used for repeated measurements. Receiver operating characteristic (ROC) curves were constructed to explore the ability of the investigated mediators to discriminate ischemia and rejection from uneventful conditions. The area under the curve (AUC) was calculated. The optimal cutoff value for each variable was defined as the value closest to the top left corner, and these values were used for contingency table analyses to determine sensitivity and specificity. Values from the right and left lobes of whole transplants and values from split grafts were handled separately. One split graft that was lost because of ischemia in the early postoperative period was not examined for rejection. Positive and negative predictive values are not reported because the results may have been influenced by the exclusion of 7 possible rejections. All P values are 2-tailed and have been Bonferroni-adjusted according to the number of compared groups. Statistical analyses were performed with PASW 18.0 (IBM, Chicago, IL).
Nine liver transplant recipients (4 males and 5 females) had ischemia, and 12 (5 males and 7 females) experienced rejection. In 39 recipients (19 males and 20 females), no major events occurred during the period of metabolic substance sampling (the reference group). The recipients in both the ischemia and rejection groups were younger than the recipients in the reference group. No differences in the Model for End-Stage Liver Disease (MELD) scores were found between the groups. The total graft ischemia time was somewhat longer in the ischemia group. The peak concentrations of circulating transaminases measured within 24 hours after graft reperfusion did not differ between the groups (Table 2). During the observation period, we identified 3 patients with an infection in the graft or close to the graft, and 3 had cholestasis confirmed by graft biopsy. Two of the investigated patients died: one because of complications related to ischemia on the first postoperative day, and another in the reference group because of pulmonary aspergillosis 16 days after transplantation.
|Reference Group (n = 39)||Rejection Group (n = 12)||P Value||Ischemia Group (n = 9)||P Value|
|Age (years)||57.0 (0.5-69.6)||44.7 (1.9-60.8)||0.02||36.7 (0.5-61.3)||0.07|
|Body mass index (kg/m2)||24 (14-34)||22 (15-24)||0.01||20 (9-30)||0.08|
|MELD score||14 (5-39)||14 (5-33)||>0.99||12 (6-22)||0.96|
|Graft ischemia time (hours)||7.6 (2.9-12.3)||6.6 (4.1-12.1)||>0.99||9.0 (6.9-16.1)||0.03|
|Peak AST (U/L)*||1112 (116-10,574)||570 (117-1848)||0.09||1466 (38-3628)||0.90|
|Peak ALT (U/L)*||841 (52-7304)||506 (54-1013)||0.07||1206 (32-2445)||>0.99|
Five of the 9 cases of ischemia had hepatic artery thrombosis: in 2 cases, the blood flow was successfully reestablished by surgery; 2 patients successfully underwent retransplantation; and in 1 case, the blood flow was reestablished spontaneously. One graft had thrombosis in both the portal vein and the hepatic artery, and the patient succumbed during an attempt at retransplantation. In 1 graft, stenosis of the right hepatic artery was detected by CT angiography. In 2 grafts, an infarction was detected; 1 patient successfully underwent retransplantation, and the other patient's graft recovered after surgical removal of a hematoma causing pressure and thereby a graft infarction located close to the split surface. In 11 of the 12 ACR cases, circulating ALT levels increased. The increase started at a median of 6 days (range = 3-11 days) after graft reperfusion, and an increase in the median value from 148 (range = 28-410 U/L) to 450 U/L (range = 166-1056 U/L, P = 0.003) was observed. In 9 cases, the bilirubin level increased. The increase started 7 days (range = 4-15 days) after reperfusion, and an increase from 18 (range = 7-100 μM) to 72 μM (range = 28-470, P = 0.01) was observed. Antirejection treatment was initiated at a median of 6 days (range = 4-13 days) after transplantation, and the median rejection activity index score was 5 (range = 3-8). All rejections were successfully treated with pulses of methylprednisolone. Within 1 year after transplantation, 3 more patients were treated for episodes of rejection, and 2 of these patients also experienced rejection during the first weeks.
The microdialysis catheters collected dialysates for a median of 10 days (range = 1-26 days). No episodes of major bleeding occurred with the insertion of the catheters. Minor episodes of bleeding at the site of puncture were registered in 4 cases. These were handled by manual compression; in 2 cases, Surgicel was also used. All catheters were successfully removed without indications of significant bleeding. No episodes of infection occurred that could be related to the catheters. No significant difference was found between the microdialysis data from right and left lobes or between the mean values for the 2 lobes and the data from split grafts. Thus, in further analyses (except for contingency table analyses), the mean values of the 2 liver lobes and the values from split grafts were used.
During the first 24 hours after the operation, linear mixed model analyses revealed that the concentrations of all mediators showed a time-dependent decline in the 51 nonischemic grafts of the reference and rejection groups. Lactate decreased at 0.2 mM/hour, pyruvate decreased at 5.6 μM/hour, glucose decreased at 0.3 mM/hour, and glycerol decreased at 1.4 μM/hour (P < 0.001 for all comparisons). No statistically significant differences were found between the 2 groups. Neither the total ischemia time nor the concentration of circulating ALT measured within 24 hours after the operation had any statistically significant influence on the metabolic parameters (data not shown).
Ischemia (n = 9)
In the linear mixed model analyses, the estimated mean values for the lactate levels (5.2 mM, P < 0.001), the LP ratio (156.4, P = 0.02), and the glycerol levels (225 μM, P = 0.01) were significantly higher in the ischemia group versus the reference group (1.5 mM, 13.6, and 35 μM, respectively). Similar results were found when the median values were compared (Fig. 2 and Table 3). In the graft with hepatic artery stenosis that was diagnosed by CT angiography, pathological values were detected only for the affected side with an increased lactate level and an increased LP ratio; the glycerol level was normal. ROC curves (Fig. 3) revealed that lactate could be used to discriminate the ischemia group from the reference group with an AUC of 1.00 [95% confidence interval (CI) = 1.00-1.00] and with an optimal cutoff value of 3.0 mM. Pyruvate could not be used to discriminate ischemia (AUC = 0.44), but the LP ratio could be used to discriminate ischemia with an AUC of 0.99 (95% CI = 0.98-1.00) and with an optimal cutoff value of 20. Glycerol could be used to discriminate ischemia with an AUC of 0.85 (95% CI = 0.69-1.00) and with an optimal cutoff value of 29 μM. In the contingency table analyses using a lactate level > 3 mM and an LP ratio > 20 as the ischemia criteria (Table 4), measurements at a single time point detected ischemia with 100% sensitivity, regardless of the number of monitored lobes or the number of measurements. As for specificity, requiring positive results from both liver lobes was beneficial for whole liver transplants. When a new positive measurement was required after 1 hour, specificity values greater than 90% were achieved even with only 1 catheter. No further clinically relevant improvements were observed when 3 consecutive positive measurements were required. With the aforementioned criteria for ischemia, 6 episodes of vascular occlusion were detected with microdialysis catheters at a median of 30 hours (range = 4-142 hours) after graft reperfusion. Lactate levels in arterial blood increased in only 3 of these cases; in the other 3 cases of hepatic artery thrombosis, the arterial lactate levels remained unchanged. In the patients whose arterial lactate levels increased, the increases were observed 0.3, 6, and 57 hours after hepatic artery thrombosis was detected with microdialysis catheters. Accordingly, pathological intrahepatic metabolism was detected with microdialysis catheters before the radiological confirmation of vascular occlusion in all cases. Although the patients were monitored with microdialysis catheters, Doppler ultrasound examinations were performed a median of 0.44 times per day (range = 0.28-3.36 times per day) in the ischemia group and a median of 0.29 times per day (range = 0.00-2.15 times per day) in the reference group (P = 0.03).
|Reference Group (n = 39)||Rejection Group (n = 12)||P Value||Ischemia Group (n = 9)||P Value|
|Lactate (mM)||1.5 (1.1-1.9)||2.1 (1.9-2.4)||<0.001||5.8 (4.0-11.1)||<0.001|
|Pyruvate (μM)||124 (102-150)||185 (155-206)||<0.001||98 (73-163)||>0.99|
|LP ratio||11.8 (10.6-13.6)||11.6 (9.8-13.1)||>0.99||66.1 (23.9-156.7)||<0.001|
|Glucose (mM)||6.0 (4.9-7.6)||7.4 (6.4-7.9)||0.19||5.4 (3.4-7.8)||0.84|
|Glycerol (μM)||9 (9-24)||13 (9-17)||>0.99||138 (26-260)||0.002|
|Yes (n)||No (n)|
|Measurement at a single time point|
|Right and left lobes||2||42||100||95|
|Two consecutive measurements|
|Right and left lobes||2||42||100||100|
|Three consecutive measurements|
|Right and left lobes||2||42||100||100|
Rejection (n = 12)
In the linear mixed model analyses (using all data after metabolic stabilization), the estimated mean pyruvate value was significantly higher in rejection group grafts versus reference group grafts (177 versus 125 μM, P = 0.009). When the group-time interaction was analyzed, time-dependent increases in both the lactate level (with an estimated fixed effect of 0.003 mM/hour, P < 0.001) and the pyruvate level (with an estimated fixed effect of 0.03 μM/hour, P = 0.05) were revealed for grafts in the rejection group. The opposite effect was observed in the reference group, in which the lactate level decreased with an estimated fixed effect of 0.0004 mM/hour (P = 0.01) and the pyruvate level decreased at 0.007 μM/hour (P < 0.001). These results support our hypothesis that the lactate level increases during rejection. Thus, when the ROC curves were being constructed (Fig. 3), the values sampled from the first time point at which lactate had increased at least 50% for a minimum of 6 hours were coded as rejection. Lactate could be used to discriminate the rejection group from the reference group with an AUC of 0.87 (95% CI = 0.77-0.97) and with an optimal cutoff value of 2 mM. Pyruvate could be used to discriminate rejection with an AUC of 0.88 (95% CI = 0.78-0.97) and with an optimal cutoff value of 170 μM. The LP ratio and glycerol could not be used to discriminate rejection (AUC = 0.45 for both). In the contingency table analyses using a lactate level > 2 mM, a pyruvate level > 170 μM, and an LP ratio < 20 as the rejection criteria (Table 5), measurements at a single time point detected rejection with 100% sensitivity regardless of the number of monitored lobes or the number of measurements. However, when measurements at only a single time point were required, a maximum specificity not greater than 42% was achieved (2 lobes). A specificity of 83% was achieved with consecutive positive measurements for at least 6 hours from both lobes at the cost of a reduction of the sensitivity to 89% (this was caused by 1 patient who had positive measurements for only 5 hours). With the aforementioned criteria, rejection was detected at a median of 3.6 days (range = 1.2-6.4 days) after graft reperfusion before any pathological (>25%) increases in ALT and bilirubin, and this was statistically significant. Rejection was detected at a median of 4 days before the ALT levels (P = 0.02) and the bilirubin levels (P = 0.04) increased and at a median of 4 days (range = 0-10 days) before biopsy was performed (P = 0.01). In 4 cases, metabolic normalization after treatment was observed (Fig. 4); in the other cases, the catheters ceased to function during treatment. The measured concentrations of lactate were significantly higher in the rejection group versus the reference group (P < 0.001), but they were significantly lower in the ischemia group (P < 0.001; Fig. 2 and Table 3). In comparison with the reference group, the pyruvate level was significantly higher in the rejection group (P < 0.001). The baseline values (ie, the values sampled between metabolic stabilization and the onset of rejection) did not differ from those of the reference group. Although the patients were monitored with microdialysis catheters, Doppler ultrasound examinations were performed a median of 0.37 times per day (range = 0.19-0.51 times per day) in the rejection group and a median of 0.29 times per day (range = 0.00-2.15 times per day, P = 0.21) in the reference group.
|Yes (n)||No (n)|
|Measurement at a single time point|
|Right and left lobes||9||36||100||42|
|Consecutive measurements (>6 hours)|
|Right and left lobes||9||36||89||83|
Infection (n = 3) and Cholestasis (n = 3)
No differences in the estimated means of the infection and cholestasis groups and the reference group were detected. As in the rejection group, there was a positive group-time interaction for lactate in the infection group (estimated fixed effect = 0.001 mM/hour, P = 0.003), but for pyruvate, the increase did not reach a level of statistical significance (estimated fixed effect = 0.04 mM/hour, P = 0.15). In the cholestasis group, no effect of time was observed.
Our results suggest that the bedside sampling of metabolic mediators with microdialysis catheters may be used to detect episodes of ischemia and rejection in liver grafts and that these conditions may be identified earlier than is possible with current standards of monitoring and care. The method's ability to detect ischemia and thus alert care providers to possible vessel occlusion is particularly important. Prolonged ischemia is detrimental for the graft. Consequently, early detection by microdialysis may be graft-saving and improve patient survival. The levels of sensitivity and specificity for the detection of rejection indicate that this method might also be a useful clinical tool in cases of suspected allograft rejection. Thus, this method might act as an alert for performing biopsy for verification, particularly in the early stages of rejection.
Measuring intrahepatic pyruvate levels in addition to lactate levels enables a distinction to be made between aerobic and anaerobic glycolysis and is thus important for the accuracy of diagnostics. With the reduced oxygen supply in ischemia, the condition is dominated by anaerobic glycolysis; nicotinamide adenine dinucleotide dehydrogenase, a cofactor of the enzyme pyruvate dehydrogenase, is increased, and the balance between pyruvate and lactate is moved toward lactate.27 Thus, the LP ratio increases in patients with anaerobic glycolysis/ischemia. In rejection, the simultaneous increases in lactate and pyruvate levels with a stable LP ratio most likely reflect increased aerobic glycolysis due to lymphocyte activation.16
In cases of vessel occlusion, clinical and biochemical signs are initially frequently absent. The arterial line is usually removed shortly after extubation, and blood lactate levels are not measured routinely thereafter. Despite the excellence of Doppler ultrasound for detecting, for example, an occluded hepatic artery, it is not a suitable tool for continuous monitoring.28 In 3 of the 6 episodes of vessel occlusion, biochemical signs of ischemia were detected only with microdialysis catheters even though circulating lactate levels were measured frequently by blood sampling as part of our clinical routine; this underscores the importance of intraorgan monitoring. Furthermore, the findings in 1 graft with hepatic artery stenosis indicate that microdialysis not only detects total occlusion of the hepatic artery but also partially reduces perfusion, and differentiation between pathologies in the 2 lobes may be possible. When an obviously pathological measurement is being assessed, it should be kept in mind that a catheter positioned in a well-circulated organ certainly will detect a global pathology. For example, hyperglycemia will be detected as an increased concentration of glucose in the liver. To confirm or refute that the pathology is due to focally generated markers, blood sampling can be used: a blood sample showing similar or higher values in the blood versus the organ indicates a global pathology, whereas high values in the organ and low values in the blood strongly indicate a focal pathology. In certain cases, an additional catheter in, for example, the subcutaneous tissue may be helpful in distinguishing a focal pathology from a global pathology.26 Measuring glycerol might be useful for distinguishing between total and partial vessel occlusion because the cell membrane degradation product is released in cases of cell death.12 In pediatric liver transplantation with small vessels, close monitoring of intrahepatic parameters is likely to be particularly beneficial. In order to reduce the number of negative examinations, we suggest that a single positive measurement from 1 catheter should be repeated within 1 hour to confirm a pathological value before a Doppler ultrasound examination is performed (Table 4).
In cases of suspected rejection, the time aspect is less urgent in comparison with suspected ischemia. Although the end stage of rejection is fibrosis with a need for retransplantation, there is no general agreement about the significance of initiating antirejection treatment at an early time point.2, 29 However, we consider the early treatment of rejection important. Because the blood flow in the portal vein is diminished with rejection, the risk of portal vein occlusion is increased.30 Patients with primary sclerosing cholangitis (PSC) have a higher incidence of both acute and chronic rejection. In these patients, acute rejection is associated with a higher incidence of chronic rejection and PSC relapse in the graft, which again is associated with worse long-term outcomes.31 In the large population of patients undergoing transplantation for hepatitis C virus cirrhosis, antirejection treatment with intravenous corticosteroids results in massive viral replication and thus recurrence of the disease.32 It is theoretically conceivable that the early detection of rejection and subsequent tailored treatment regimens guided by continuous microdialysis monitoring to establish the lowest effective dose of immunosuppression would be beneficial in these patients. Individualized regimens could theoretically also be advantageous in patients infected with Epstein-Barr virus and in patients already diagnosed with a malignancy; aggressive antirejection treatment is associated with a higher incidence of posttransplant lymphoproliferative disease,33 and immunosuppressive medications may trigger the growth of malignant cells.34 Accordingly, patients with PSC, hepatitis C virus cirrhosis, Epstein-Barr virus infections, and malignancies may represent subgroups of patients for whom close intrahepatic monitoring might be particularly useful. When the use of microdialysis is being considered for diagnosing rejection, the variation of the sensitivity and specificity with the number of repeated measurements should be considered; relying on measurements at a single time point for decisions about the initiation of antirejection therapy would imply that a number of patients would unnecessarily be administered potentially harmful high doses of corticosteroids. By requiring consecutive positive measurements for 6 hours before the initiation of antirejection therapy, we achieved clinically acceptable levels of both sensitivity and specificity in this study (Table 4). Thus, we argue that this approach would be clinically beneficial without negative effects on graft outcomes.
Bleeding may theoretically occur with both the insertion and removal of catheters. In addition, catheters can become contaminated and can be sources of bacterial and fungal infections just as other foreign bodies can be. One hundred twenty-eight catheters were inserted into livers in this study, and there were 4 episodes of minor bleeding with catheter insertion that were easily halted within minutes by manual compression or Surgicel. Apart from this, no bleeding or infections related to the catheters were observed. On the basis of our previous experience26 and this study, with almost 200 catheters implanted in livers without any adverse events such as bleeding or infection, microdialysis may be implemented as part of routine monitoring after liver transplantation as far as safety is concerned. The technique allows the full mobilization of patients and does not have a major negative impact on the activities of daily life in the postoperative period. The bedside analysis of metabolic parameters is feasible with only a marginally increased work load. As expected, the frequencies of Doppler ultrasound examinations were higher in the groups with pathologies versus the reference group. In this study, the lack of a control group without catheters limits our ability to draw any conclusions about whether the catheter measurements contributed to the significant differences in Doppler ultrasound examinations.
The use of microdialysis for long-time monitoring is limited because the catheters become clotted by cellular debris and proteins.35 According to our experience, it is difficult to predict the time of catheter clotting and thus avoid it. The current study also has several specific limitations. Although more than 70 liver grafts were monitored, the number of grafts with ischemia or biopsy-proven rejection was restricted. Thus, future large studies confirming our results are needed before the implementation of this method in routine clinical practice. Better validation of the data would most likely be achieved by the implementation of biopsy in the study protocol. Importantly, this study was not designed to detect or elucidate the clinical course of infection and cholestasis. In the literature, the reported incidence of infection is approximately 30%,6 and biliary complications occur in approximately 15%.4 In this study, we describe only cases of infection occurring in the liver or close to the liver while the microdialysis catheters were in the patients (median = 10 days, range = 1-26 days). Accordingly, several patients had infections at other sites and in the period after the catheters were removed. In order to present data that are as validated as possible, we have also provided data only from grafts with cholestasis that was proven by biopsy. Accordingly, there certainly may have been cases of subclinical cholestasis in the reference group. Hence, the results for infection and cholestasis should be interpreted cautiously.
In addition to studies designed to confirm our results, an outcome study monitoring liver transplants with microdialysis-guided treatment is encouraged for the future. Although the detection of rejection with greater than 80% sensitivity and specificity may be considered clinically acceptable, investigators are also encouraged to study whether the monitoring of inflammatory mediators is a better idea. Furthermore, studies of other organ transplants would indeed be interesting. In principle, all transplanted organs could be monitored. Implementing graft biopsy in study protocols should be considered, although this would increase the risk of bleeding due to anticoagulant treatments and/or coagulopathy.13, 15 After liver transplantation (eg, 7 days), we propose performing protocol biopsy for all grafts.
In conclusion, measurements made by microdialysis in liver transplants detected ischemia and rejection with highly clinically acceptable levels of sensitivity and specificity. Episodes of vessel occlusion were detected while arterial lactate levels remained unchanged, and signs of rejection were detected several days before significant changes were observed in routine blood samples. If future studies confirm our results, this method may offer the advantage of early detection of ischemia, which might save grafts. It may further provide clinically valuable information about rejection, particularly at early stages.
The authors thank the many patients, nurses, and physicians involved in this study. Janne Gripheim and Per Ohlin provided excellent technical support during the period of patient enrollment.
- 1Organ Procurement and Transplantation Network. Liver graft survival rates in the United States. http://optn. transplant.hrsa.gov/. Accessed March 2011.
- 15Evaluation of risk factors for bleeding after liver biopsy in children. J Pediatr Gastroenterol Nutr. In press., , , , .