Telephone: +47-23070000, FAX: +47-23073684
Inflammatory markers sampled by microdialysis catheters distinguish rejection from ischemia in liver grafts
Article first published online: 26 SEP 2012
Copyright © 2012 American Association for the Study of Liver Diseases
Volume 18, Issue 12, pages 1421–1429, December 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., Wælgaard, L., Pischke, S. E., Mollnes, T. E. and Inge Tønnessen, T. (2012), Inflammatory markers sampled by microdialysis catheters distinguish rejection from ischemia in liver grafts. Liver Transpl, 18: 1421–1429. doi: 10.1002/lt.23503
- Issue published online: 7 DEC 2012
- Article first published online: 26 SEP 2012
- Accepted manuscript online: 5 JUL 2012 04:53PM EST
- Manuscript Accepted: 13 JUN 2012
- Manuscript Received: 30 MAR 2012
- South-Eastern Norwegian Health Authority (Helse Sør-Øst) and the Research Council of Norway. Grant Numbers: 2009007, 2008104
Rejection and ischemia are serious complications after liver transplantation. Early detection is mandatory, but specific markers are largely missing, particularly for rejection. The objective of this study was to explore the ability of microdialysis catheters inserted in liver grafts to detect and discriminate rejection and ischemia through postoperative measurements of inflammatory mediators. Microdialysis catheters with a 100-kDa pore size were inserted into 73 transplants after reperfusion. After the study's completion, complement activation product 5a (C5a), C-X-C motif chemokine 8 (CXCL8), CXCL10, interleukin-1 (IL-1) receptor antagonist, IL-6, IL-10, and macrophage inflammatory protein 1β were analyzed en bloc in all grafts with biopsy-confirmed rejection (n = 12), in grafts with vascular occlusion/ischemia (n = 4), and in reference grafts with a normal postoperative course of circulating transaminase and bilirubin levels (n = 17). The inflammatory mediators were elevated immediately after graft reperfusion and decreased toward low, stable values during the first 24 hours in nonischemic grafts. In grafts suffering from rejection, CXCL10 increased significantly (P = 0.008 versus the reference group and P = 0.002 versus the ischemia group) 2 to 5 days before increases in circulating alanine aminotransferase and bilirubin levels. The area under the receiver operating characteristic curve was 0.81. Grafts with ischemia displayed increased levels of C5a (P = 0.002 versus the reference group and P = 0.008 versus the rejection group). The area under the curve was 0.99. IL-6 and CXCL8 increased with both ischemia and rejection. In conclusion, CXCL10 and C5a were found to be selective markers for rejection and ischemia, respectively. Liver Transpl, 2012. © 2012 AASLD.
Most rejections of liver transplants are cellular rather than antibody-mediated, and T lymphocytes are central players.1 Acute cellular rejection (ACR) typically occurs within the first 6 weeks after transplantation and is a relatively frequent complication with an incidence in the range of 30% to 60%.2 The detection and subsequent treatment of ACR are important because the condition may have a negative impact on outcomes, particularly for patients undergoing transplantation for hepatitis C cirrhosis and primary sclerosing cholangitis.2-7 ACR is suspected when the activity of circulating transaminases and/or concentrations of bilirubin increase, and although biopsy is not routinely performed at all liver transplant centers,8 it is needed to confirm or refute the suspicion.9 However, repeated biopsy procedures may be required because biopsy samples (particularly those taken at early stages) may be negative,10 and ACR is thus regularly detected with a considerable time delay. Because biopsy may not be performed on account of contraindications, some patients are administered high doses of glucocorticosteroids for an uncertain clinical indication.9, 11 Thus, there is a need for early markers of ACR.
Compared to rejection, ischemia due to vascular occlusions is less common, but it is more severe. For instance, thrombosis of the hepatic artery is a major contributor to the loss of grafts and represents the most frequent cause of urgent retransplantation.12-15 Vascular occlusion may be suspected because of increased concentrations of lactate or increased activity of transaminases in blood, but a Doppler ultrasound examination by trained radiologists represents the current standard investigation.16, 17 However, the clinical utility of Doppler ultrasound is restricted because continuous monitoring using microprobes, for example, is currently not available for clinical use.18
Microdialysis is a technique that enables the monitoring of tissues and organs of interest.19 Microdialysis catheters have mainly been used to monitor metabolic parameters in tissues that are at risk for ischemia,17, 20, 21 including liver grafts.22-24 In a recent clinical study,24 we found that the bedside monitoring of the metabolic mediators lactate and pyruvate in grafts may be a useful tool for detecting rejection and ischemic vascular complications. However, the study revealed that the specificity in detecting rejection was highly dependent on the number of performed measurements, and this suggested that additional monitoring of inflammatory parameters might be a better approach than isolated monitoring of metabolic mediators.
Inflammatory mediators have traditionally been linked to exogenous stimuli such as bacteria, viruses, and donor tissues and are termed pathogen-associated molecular patterns.25 At present, it is well known that the immune system can also become activated by the recognition of intracellular proteins (termed alarmins) that are released upon cell injury due to, for example, ischemia/reperfusion injury.26-29 Both pathogen-associated molecular patterns and alarmins lead to a distinct and complex inflammatory response that includes complement activation with the release of the very potent complement activation product 5a (C5a) fragment and the production of cytokines responsible for intercellular signaling and chemokines inducing chemotaxis.
The main objective of the present study was to investigate through the use of microdialysis catheters in liver grafts whether the inflammatory response in ACR could be distinguished from the inflammation associated with ischemia.
PATIENTS AND METHODS
Definitions of Clinical Endpoints
Rejection required histologically confirmed ACR in biopsy samples. The onset of rejection was clinically defined as the time at which either the circulating alanine transaminase (ALT) level or the bilirubin level increased from the previous day. All increases in ALT were considered pathological because the normal course involves decreasing values during the first 1 to 2 weeks. Compared to ALT, bilirubin is a biologically more complex parameter.30 For several years, our liver transplant unit's indication for performing graft biopsy on the basis of bilirubin increases has been an increase of 25% or more to values ≥ 25 μM. Accordingly, we used the same quantities to identify pathological values in this study. Ischemia required grafts with vessel occlusion diagnosed by ultrasound, computed tomography angiography, or surgical exploration. The references were those grafts with an uneventful clinical and biochemical course. In those patients, the activity of liver transaminases or the concentration of bilirubin in blood did not increase at any time during the period of monitoring.
The study protocol was approved by the regional ethics committee. Detailed oral and written information was provided, and written informed consent was obtained before a patient's inclusion.
During a 1-year period (starting in October 2008), 73 liver grafts were monitored postoperatively with microdialysis catheters. Because of limited personnel and economic resources, we were unable to include the whole population in this study. For this report, all grafts with biopsy-confirmed ACR (n = 12) and all grafts fulfilling the aforementioned criteria for references (n = 17) were included. We were able to include 4 of the 6 grafts that fulfilled the criteria for ischemia. In 2 episodes of hepatic artery occlusion detected at the bedside as increases in lactate and lactate/pyruvate, surgical revascularization was achieved before a sufficient amount of microdialysate was collected. According to a definition of ischemia broader than that described previously, ischemia occurred in 3 additional grafts [hepatic tissue infarction (2) and hepatic artery stenosis (1)]. In those grafts, inflammatory mediators could not be analyzed because the relevant samples accidentally were not frozen for later analyses. An overview of the 33 included grafts and the 40 excluded grafts is shown in Fig. 1. The baseline characteristics of the 33 included grafts are shown in Table 1.
|Donors (n = 33)|
|Sex: male/female (n/n)||21/12|
|Age (years)*||57 (2-83)|
|Graft recipients (n = 30)|
|Sex: male/female (n/n)||18/12|
|Age (years)*||47 (1-69)|
|Model for End-Stage Liver Disease score*||14 (5-39)|
|Primary diagnosis (n)|
|Primary sclerosing cholangitis||6|
|Bile duct atresia||4|
|Acute liver failure||1|
|Liver transplantation (n = 33)|
|Transplantation before study||3|
|Transplantation twice during study||3|
|Split liver transplantation (n)||5|
|Peak AST (U/L)*†||869 (117-3628)|
|Peak ALT (U/L)*†||654 (54-2695)|
|Liver grafts (n = 33)|
|Weight (g)*||1275 (260-2020)|
|Total ischemia time (hours)*||8.1 (3.4-16.1)|
|Hepatic artery||210 (13-500)|
|Portal vein||1300 (400-3100)|
The method has been described elsewhere.24 Briefly, 1 microdialysis catheter with a 100-kDa pore size (CMA 65, M Dialysis AB, Stockholm, Sweden) was inserted into each liver lobe by a split-needle technique after graft reperfusion. For split transplants, 1 catheter was used. The catheters were perfused with a fluid containing dextran and electrolytes (Plasmodex, Meda AB, Solna, Sweden) at a velocity of 1 μL/minute by microinjection pumps (CMA 107, M Dialysis AB). The samples were collected in microvials (M Dialysis AB). Catheter perfusion was started before insertion in the liver tissue, and the first microvials were discarded after approximately 30 minutes of liver perfusion to prevent the analysis of fluid that had not been perfused through the liver.
After each patient's arrival at the critical care unit, samples were collected and immediately frozen to −70°C at the following time points: 0, 4, 8, 12, and 24 hours. Thereafter, samples were collected and frozen twice daily. Freezing samples at an exact time point after graft reperfusion was not feasible, and all reported time points are related to the time of the reperfusion of the portal vein. When 2 catheters were used, samples from the catheter that sampled the microdialysate for the longest period were chosen for analysis. The complement activation product C5a was analyzed en bloc with a human C5a enzyme-linked immunosorbent assay (Kit II, BD Biosciences, San Jose, CA). After pilot analyses of 3 patients suffering from rejection with the Bio-Plex human cytokine 27-plex assay (Bio-Rad Laboratories, Inc., Hercules, CA), the following 6 mediators were analyzed en bloc: interleukin-1 receptor antagonist (IL-1ra), C-X-C motif chemokine 8 (CXCL8), CXCL10, interleukin-6 (IL-6), IL-10, and macrophage inflammatory protein 1β (MIP1β).
Groups were compared with the Mann-Whitney U test, and repeated measurements were compared with Friedman's test and/or the Wilcoxon signed-rank test. P values resulting from analyses involving the reference, rejection, and ischemia groups were Bonferroni-adjusted for comparisons of the 3 groups. Receiver operating characteristic curves were created to determine the ability of the parameters to discriminate grafts suffering rejection and ischemia from the reference group. The area under the receiver operating characteristic curve was calculated, and the null hypothesis was an area under the curve (AUC) of 0.5. The optimal cutoff value was defined as the value closest to the top left corner. All presented P values are 2-tailed (PASW 18.0, IBM, Chicago, IL).
There was a tendency toward a younger age for the recipients in the rejection and ischemia groups versus the reference group. Likewise, patients with the specified complications tended to stay longer in the hospital and were examined more frequently with Doppler ultrasound postoperatively (Table 2). However, all patients survived and were discharged from the hospital except for 1 patient who died after an ischemic vascular complication.
|Reference Group (n = 17)||Rejection Group (n = 12)||P Value||Ischemia Group (n = 4)||P Value|
|Age (years)||50 (1-69)||44 (2-61)||0.26||27 (1-54)||0.55|
|Model for End-Stage Liver Disease score||14 (5-39)||14 (5-33)||>0.99||15 (6-22)||>0.99|
|Graft ischemia time (hours)||8.0 (3.4-11.9)||6.6 (4.1-12.1)||>0.99||9.0 (8.0-16.1)||0.24|
|Doppler ultrasound/day (n)*||0.2 (0.1-0.9)||0.4 (0.1-0.5)||0.33||0.7 0.3-3.4||0.13|
|Peak AST (U/L)†||1212 (116-10,574)||570 (117-1848)||0.78||1466 (1276-3628)||0.14|
|Peak ALT (U/L)†||915 (69-7304)||506 (54-1013)||0.74||1206 (654-2445)||0.36|
|Hospital stay (days)||24 (18-46)||31 (23-40)||0.21||36 (7-69)‡||0.63|
Minor bleeding related to the insertion of the catheters was registered in 3 cases. This bleeding was rapidly halted by manual compression and a hemostatic agent (Surgicel, Ethicon, Cornelia, GA). The catheters collected the dialysate for a median of 10 days (range = 1-21 days). No significant bleeding related to the withdrawal of the catheters was registered, and no episodes of infection could be related to the catheters.
Inflammatory Mediators for 24 Hours After Graft Reperfusion
The concentrations of inflammatory mediators in the 29 grafts without vascular complications are reported for the following periods after reperfusion of the portal vein: 0 to 6, 6 to 10, 10 to 16, and 16 to 24 hours. From initially relatively high concentrations, there were time-dependent decreases for C5a, IL-1ra, IL-6, IL-10, and MIP1β (P < 0.001 for all analyses), whereas CXCL8 and CXCL10 decreased nonsignificantly (Fig. 2). No differences in any of the investigated periods were found between the reference group and the grafts that later developed rejection.
Rejection (n = 12)
Rejection was suspected when the activities of circulating liver enzymes or concentrations of bilirubin increased. Pathological increases were observed at a median of 3.9 days (range = 0.9-8.5) after reperfusion for ALT and at a median of 7.4 days (range = 4.2-14.6 days) after reperfusion for bilirubin. The median ALT level increased from 73 U/L (range = 14-539 U/L) to 273 U/L (range = 69-989 U/L, P = 0.002), and the bilirubin level increased from 18 μM (range = 7-100 μM) to 72 μM (range = 28-470 μM, P = 0.01). ACR was confirmed by liver biopsy and a histological evaluation before a methyl prednisolone pulse treatment was initiated at a median of 6 days (range = 4-13 days) after transplantation. This therapy was successfully carried out for all patients.
With increasing time, the concentrations of CXCL10 increased in both the rejection group and the reference group, but it increased significantly earlier and to significantly higher values in the rejection group (Fig. 3). In comparison with the first day after transplantation, significantly higher median values were detected on day 3 in the rejection group (65 versus 561 pg/mL, P = 0.02) and on day 9 in the reference group (88 versus 800 pg/mL, P = 0.03). On days 3 and 4 (ie, in samples collected before any patient had been given antirejection treatment), the values were significantly higher in the rejection group versus the reference group (P = 0.05 and P = 0.02, respectively). Intrahepatic CXCL10 increased earlier than circulating ALT (Fig. 4) and bilirubin. After the initial mandatory peak in ALT following graft reperfusion, increases in CXCL10 were observed during the period of the physiological decrease in ALT. Increases in CXCL10 greater than 100% were observed after a median of 2.2 days (range = 1.2-3.8 days). This was 1.7 days before there was any increase in ALT (P = 0.02) and 5.2 days before bilirubin increased by 25% or more (P = 0.008). The peak concentrations of CXCL10 were significantly higher in the rejection group versus both the reference group (P = 0.008) and the ischemia group (P = 0.004; Table 3 and Fig. 5). CXCL10 discriminated rejection from references with an AUC of 0.81 [95% confidence interval (CI) = 0.65-0.97, P = 0.005; Fig. 6] and from ischemia with an AUC of 1.00 (95% CI = 1.00-1.00, P = 0.004). The optimal cutoff value for discriminating episodes of rejection from the reference group was 2322 pg/mL. For the other mediators, the increments during rejection were not statistically significant (Table 3).
|Reference Group (n = 17)*||Rejection Group (n = 12)*||P Value||Ischemia Group (n = 4)†||P Value|
|C5a (pg/mL)||1000 (800-1200)||1400 (700-4000)||0.74||21,400 (14,800-25,700)||0.002|
|CXCL10 (pg/mL)||772 (141-2287)||3938 (1955-11,147)||0.008||152 (114-217)||0.41|
|CXCL8 (pg/mL)||105 (22-180)||203 (82-390)||0.36||609 (455-1693)||0.03|
|IL-1ra (pg/mL)||0 (0-75)||17 (0-72)||>0.99||851 (68-4491)||0.06|
|IL-6 (pg/mL)||10 (0-48)||88 (16-177)||0.11||133 (57-385)||0.04|
|MIP1β (pg/mL)||86 (23-114)||96 (40-206)||>0.99||25 (11-88)||0.82|
Ischemia (n = 4)
Three of the 4 episodes of vascular occlusion involved hepatic artery thrombosis that occurred without preceding warning in otherwise stable patients. In 1 case, the flow was successfully reestablished by surgery, whereas retransplantation was successful for the other 2 cases. One episode of combined thrombosis of the hepatic artery and the portal vein occurred 12 hours after graft reperfusion in a critically ill patient requiring blood transfusions as well as vasopressors both during and after transplantation. This patient succumbed shortly thereafter.
The peak concentrations of C5a, CXCL8, and IL-6 were significantly higher in grafts with ischemia versus the references (Table 3 and Fig. 5). C5a was the only mediator that was significantly higher in the ischemia group versus the rejection group (P = 0.008). C5a discriminated grafts with ischemia from references with an AUC of 0.96 (95% CI = 0.87-1.00, P = 0.005; Fig. 6) and from ACR grafts with an AUC of 0.88 (95% CI = 0.69-1.00, P = 0.03). The optimal cutoff value for discriminating episodes of ischemia from the reference group was 2365 pg/mL. CXCL8, IL-1ra, and IL-6 also discriminated well between grafts with ischemia and references, but not as well as C5a did (data not shown). The reported peak concentrations were sampled during the period spanning from 9 hours before the Doppler ultrasound examination confirming vessel occlusion to 11 hours afterward (ie, during the period of ongoing ischemia).
Our results suggest that microdialysis catheters in liver grafts are able to detect significant changes in inflammatory mediators during both rejection and ischemia, and specific parameters may be used to discriminate these conditions from both uneventful courses and each other. CXCL10 was the most prominent marker of rejection, whereas C5a was significantly elevated exclusively in ischemic grafts. Because rejection was detected 2 to 5 days earlier in comparison with the current standard methods, CXCL10 may be an interesting candidate as an early specific marker of rejection.
Despite CXCL10 being produced by a variety of cells (eg, mononuclear cells, activated stellate cells, endothelial cells, and hepatocytes), reports indicate that it may be a specific marker of the rejection of liver grafts31-34 as well as other solid organ transplants.35-39 CXCL10 expression is induced by interferon-γ and, through the stimulation of its receptor chemokine (C-X-C motif) receptor 3, contributes to the migration of T lymphocytes into the inflamed tissue.31-34, 40 Thus, interferon-γ and CXCL10 are early mediators of the cell-mediated adaptive immune system. Our results support the idea that CXCL10 may be independent of the innate immune system; the levels did not increase with ischemia and seemed to be independent of ischemia/reperfusion injury (Fig. 2). A biological explanation for the selective increase in CXCL10 being related to rejection and not ischemia may be the precursor interferon-γ, which probably is independent of complement activation.41, 42 The complement system represents the innate immune system, and the biologically highly potent activation product C5a is induced by the activation of all initial activation pathways. C5a has a number of effector functions and is, for example, a powerful chemoattractant causing the accumulation of neutrophil granulocytes and an important inducer of inflammation.43, 44 Our data support studies showing that the complement system plays a key role in ischemia/reperfusion injury.26, 45 In contrast to C5a, IL-1ra has potent anti-inflammatory activity.46 Interestingly, these 2 inflammatory mediators, which counteract each other, closely reflect ischemia/reperfusion, and this indicates that the inflammatory network functions to balance the detrimental effects of this condition. CXCL8, a chemokine,47 and IL-6, a proinflammatory cytokine,48 increased with ischemia, but a positive trend was also detected with rejection. This indicates that these mediators are expressed by a general activation of the immune system and not specifically by the adaptive (cell-mediated) or innate immune system. Consequently, neither IL-6 nor CXCL8 seems to be an attractive candidate in the search for specific markers of rejection.
The slow but significant increase in CXCL10 in the reference group is consistent with the results from our earlier study.23 The current study, using a derived cohort of transplant patients as controls (the reference group), does not formally answer whether these peaks represent alloreactivity/subclinical rejection or reflect the normal course after liver transplantation. However, on the basis of the results of the current study (significantly higher peak concentrations of CXCL10 in the rejection group versus the reference group), we assume that the somewhat increased CXCL10 levels in the reference group represent cell-mediated inflammation. A certain degree of alloreactivity after liver transplantation is probably common and can be transient. Minor and/or transient discharges of transaminases or bilirubin may not always be detected in peripheral blood samples, or the increases can be too small to justify liver biopsy. But contrary to the previous study,23 C5a did not increase with rejection in the current study. This may have been due to the more complicated course of the 3 rejection cases in the former study; 2 patients were critically ill before transplantation and were treated with the molecular adsorbent recirculating system, and 1 of these patients had massive hepatitis and several bowel perforations needing surgical correction. In addition to rejection and stenosis of the choledochus anastomosis, the last graft also had evidence of vascular graft ischemia. Accordingly, hepatitis and ischemia probably account for increases in C5a concurrently with rejection.
We have recently shown that by using microdialysis catheters to monitor metabolic parameters, we can detect ischemia close to real time as increased lactate levels and lactate/pyruvate ratios with an AUC of 1.00; rejection was detected several days before increases in circulating transaminases and bilirubin as simultaneous increases in lactate and pyruvate with an AUC of 0.88, but the values for sensitivity and specificity were highly dependent on the number of measurements.24 Accordingly, we consider the most interesting finding of the present study to be the detection of CXCL10 as a possible specific marker of rejection.
The present study has several important limitations. In contrast to metabolic mediators, which can be analyzed relatively cheaply and feasibly at the bedside, the analysis of inflammatory parameters is still time- and cost-consuming laboratory work. Therefore, we were compelled to reduce the number of analyses. Inflammatory responses to other important conditions such as cholestasis and infection were not investigated. Not investigating consecutive patients precluded us from reporting sensitivity and specificity values and the appurtenant positive and negative predictive values. Furthermore, the number of grafts with ischemia was restricted, and these results should, therefore, be interpreted with caution. Because pilot analyses with the 27-plex kit were exclusively performed for grafts suffering from rejection, mediators likely to be important in ischemia and ischemia/reperfusion injury (eg, tumor necrosis factor α) were not investigated. The present study also reflects some of the limitations of microdialysis catheters. Their use is time-limited because the membranes become clotted by, for example, cellular debris without the possibility of predicting the time of catheter malfunction.49 As shown in a previous study by our group,50 the ability to sample larger proteins is restricted.
In conclusion, we have shown that CXCL10 can be used to selectively detect the rejection of liver grafts before circulating ALT or bilirubin levels increase. Future studies confirming our results are needed before the detection of CXCL10 by microdialysis catheters can be applied clinically as an early marker of liver graft rejection. C5a can be used to detect ischemia and is thus a promising marker for discriminating rejection from ischemia.
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, and Julie Katrine Lindstad helped with the laboratory analyses.
- 33T cell-mediated biliary epithelial-to-mesenchymal transition in liver allograft rejection. Liver Transpl 2010; 16: 567-576., , , , , , .