|Home | About | Journals | Submit | Contact Us | Français|
The principal cause of primary non-function in orthotopic liver transplantation is thought to be preservation injury to the microvasculature. We, therefore, evaluated if effluent levels of hyaluronate, whose uptake is an endothelial cell marker, could predict early graft function and ultimate graft outcome in orthotopic liver transplantation. A total of 102 cases were studied in two phases. In the first phase, we attempted to determine if a correlation existed between effluent hyaluronate levels, early graft function and ultimate graft outcome. This phase of the study was also used to determine hypothetical cut-off values for hyaluronate which could discriminate between good and bad livers. Thirty-two livers orthotopically transplanted to randomly selected primary recipients were studied. After varying periods of static cold storage (4°C) in University of Wisconsin solution, the livers were reinfused with cold (4°C) lactated Ringer’s solution. The first 50 ml of the reperfusion effluent was collected from the infrahepatic vena cava. Effluent samples were analyzed for hyaluronate. Linear regression analysis demonstrated a significant correlation between effluent hyaluronate levels and post-operative aspartate and alanine aminotransferase levels (p<0.001 for both). Logistic regression demonstrated a highly significant correlation (p = 0.0056) between effluent hyaluronate levels and ultimate graft outcome. Generation of Receiver Characteristics Curves indicated that a level between 400 and 430 μg·l−1 could possibly discriminate between good livers and those at risk of early graft failure. The authenticity of this hyaluronate cut-off level was further confirmed in the second phase of the study where 70 consecutive primary crossmatch-negative transplants were performed. A highly significant difference was observed in peak aspartate and alanine aminotransferase levels in the first week (p<0.0006 and p<0.0005, respectively) between livers with effluent hyaluronate levels≤400 μg·l−1 and livers with hyaluronate levels higher than 400 μg·l−1 Logistic regression revealed a highly significant correlation between effluent hyaluronate levels and graft success (p=0.0001). Since hyaluronate uptake by the microvascular endothelial cell is significantly greater than production, high hyaluronate effluent levels in failed livers would be due to decreased hyaluronate uptake by the injured microvascular endothelial cell. We therefore conclude that effluent hyaluronate levels may prove to be a reliable preoperative test to assess early graft function and outcome in clinical orthotopic liver transplantation.
It is estimated that between 2–23% of human liver allografts never function, with consequent death or retransplantation (1–17). When no apparent technical or immunological causes can be identified for the early graft failure (between 1 and 14 days) and the inability to sustain life, necessitating a retransplant, the syndrome is defined as primary nonfunction (PNF). The principal cause of PNF is thought to be preservation injury to the microvascular endothelial cell (MVEC) (7, 8).
Since the critical target of preservation injury is the MVEC (9–11), we attempted to correlate effluent levels of hyaluronate (HA), an endothelial cell specific marker (12–22) with early graft function in a clinical study.
In the liver, the major sources of HA are the Ito cells (23) and the extracellular matrix (24). HA produced by these cells is taken up by the MVEC and catabolized by the action of hyaluronidase and N-acetylglucosamine-6-phosphate de acetylase (25). The rate of HA uptake by the MVEC is significantly greater than the rate of HA production by the Ito cell (23). HA uptake studies have been extensively used as a measure of MVEC integrity and function (26–30). We therefore hypothesize that preservation injury to the MVEC would reduce HA uptake by the MVEC, resulting in elevated HA levels in livers which had been injured during preservation. This hypothesis was tested in a clinical study of 102 cases.
A total of 102 cases were studied in two phases. In the first phase, 32 livers were transplanted orthotopically to consecutive adult primary recipients with negative lymphocytotoxic crossmatches; 29 were treated with FK–506 and low dose steroids (20 mg/day) and three were treated with cyclosporine and 200 mg/day methylprednisone with taper. During and following procurement, the livers were infused with University of Wisconsin (UW) solution at 4°C. After varying periods of static cold storage (13.3±2.3 SD hours), 150 cc of cold Ringer’s lactate was infused through the portal vein. The suprahepatic cava was clamped and the first 50 ml effluent was collected through the infrahepatic vena cava. These livers were then reflushed with 500 cc of cold ·UW solution. These perfusions were performed on the backtable while the livers were being prepared for implantation. The 50 ml effluent collections were centrifuged at approximately 1000 × g for 10 min to remove erythrocytes and aliquots were analyzed for HA levels. All the livers that were used in this phase of the study satisfied established preoperative criteria (4, 6, 31).
In the second phase of the study, 70 consecutive primary crossmatch-negative adult orthotopic livers were transplanted in 70 adult recipients. All the preservation, flushing and effluent collection techniques were identical to those used in phase I. All the patients received 1 g of methylprednisone in the operating room and were then maintained on FK-506 and 20 mg/day of methylprednisone. All 102 recipients were monitored daily for liver function. Ultrasound, biopsy and other invasive procedures were performed as clinically indicated.
A separate group of 14 adult primary recipients with negative lymphocytotoxic crossmatches were studied. In this group of patients the donor organ was flushed with 150 cc of cold UW solution instead of the Ringer’s lactate used in the remaining 102 cases. As before, the suprahepatic cava was clamped and the first 50 ml effluent was collected through the infrahepatic vena cava. The livers were subsequently reflushed with 500 cc of cold UW solution.
The effluent was assayed for HA values with a radiometric assay, using kits marketed by Pharmacia Diagnostics (Uppsala, Sweden) (12). All samples, including standards, were run in duplicate. Percent variation of results never exceeded 2.2%. The study was conducted in a double-blind fashion with the clinical care physicians in the intensive care unit or wards being unaware of the HA levels, and the laboratory personnel (PNR, JS) not being involved in or cognizant of the clinical monitoring, diagnostic tests, including biopsies, or treatment of the patients.
Graft function after transplantation was stratified by the criteria of Makowka et al. Based on a retrospective statistical analysis of 219 human liver transplants, the study defines peak serum aspartate aminotransferase (AST) levels of < 1500 IU/l and peak serum alanine aminotransferase (ALT) levels < 1000 IU/l at any time during the first postoperative week as indications of good grafts versus poor grafts if the levels were higher. These criteria are now extensively used by workers in this field (1,31,32).
Linear regression analysis was used to study the correlation between preoperative effluent levels of HA and post-operative AST and ALT levels. The correlation between effluent HA levels and graft survival was studied by logistic regression (33). Hypothetical pre-operative cut-off levels of HA which could discriminate between grafts exhibiting either good or poor post-operative function were determined by the generation of Receiver Characteristic Curves (ROC) (34). The accuracy of these cut-off levels was confirmed in the second phase of the study. The Mann-Whitney V-test, a nonparametric equivalent to the standard two sample t-test, was used to compare HA and liver function between groups. Univariant logistic regression was used to predict PNF based on HA levels.
A strong positive correlation was observed between effluent HA levels and postoperative AST (R2=37·30%, p<0.001) and ALT levels (R2=43·20%, p<0.001). (Fig 1).
Stratification of the 30 grafts with available AST and ALT levels by the criteria of Makowka et al. (1, 31, 32) resulted in 19 grafts being classified as good and 11 as poor grafts on the basis of AST levels. The mean effluent HA level in good grafts was 256.0±132.0 μg·l−1 and was significantly lower (p=0.0003) than the mean HA levels in the poor grafts (405.0±39.0 μg·l−1).
When the data were stratified on the basis of post-operative ALT levels, 17 grafts were classified as good and 13 as poor. Mean HA levels in the good grafts were 256.0±140.0 μg·l−1 and were significantly lower (p=0.0017) than the mean effluent HA levels of 485.0±39.0 μg·l−1 in the poor grafts
Logistic regression analysis of the data demonstrated a highly significant (p=0.0056) correlation between effluent HA levels and graft outcome up to 90 days (Fig. 2).
Generation of ROC indicated that an HA level between 400 and 430 μg·l−1 could possibly discriminate between good livers and those at risk of early graft failure (Table 1)
Table 2 shows the HA levels in the five livers that failed from PNF in this phase. All livers had HA levels higher than the 400–430 μg·l−1 cut-off. However, two of the six livers had an HA level dose to the 430 μg·l−1 cut-off level. We, therefore, hypothesized that an HA level of 400 μg·l−1 would be a more appropriate value at which to discriminate between good livers and those at risk of early graft failure. The authenticity of this effluent HA cut-off level was confirmed in Phase II of the study.
Early graft function following OLTX was monitored by AST and ALT. AST and ALT levels in the first 7 post-operative days were both significantly elevated in the group with high effluent levels >400 μg·l−1) as compared to the group with HA evels ≤400 μg·l−1(p<0.0006 for AST, p<0.0005 for ALT. Table 3).
Table 4 shows an overall 90-day graft survival of 87.1% (61 of 70). PNF was only noted in livers with effluent HA levels >400 μg·l−1 None of the 58 livers with effluent levels ≤400 μg·l−1 developed PNF, while 6 of 12 livers with effluent levels >400 μg·l−1 failed from PNF (p<0.005). Graft survival beyond 90 days was achieved in 55 of 58 livers with low effluent HA levels as compared to 6 of 12 livers with effluent HA values >400 μg·l−1 (p<0.001). Three livers with low HA levels failed between the seventh and ninth days post-operatively. These livers were lost to hepatic artery thrombosis in two cases and death from pneumonia in the third.
Mean HA levels of successful grafts were 218.4±128.0 μg·l−1 and were significantly lower than the effluent HA levels in the six grafts that failed from PNF, of 549.3± 187.3 μg·l−1 (p=0.0002) As in the first phase, all the PNF grafts had effluent HA levels >400 μg·l−1.
False-positive livers were defined as those that were successfully transplanted despite effluent HA levels >400 μg·l−1 The livers, however, had an enzyme release pattern similar to the six livers which experienced PNF (Table 5).
Logistic regression demonstrated a highly significant correlation between effluent HA levels, incidence of PNF and ultimate graft outcome (p=0.0001), improved Chi-square=22.13) (Fig. 3).
Two of the 14 livers studied failed from PNF HA levels in these two livers were 692.50±576.25 μg·l−1 Mean HA levels in the remaining 12 livers were 524.25±334.60 μg·l−1 The HA levels in the two groups were similar, alhough no statistical inference can be drawn because of the limited number of cases.
The results of this study suggest that effluent levels of HA reflect HA uptake by the MVEC and can be used to evaluate graft viability with a significant degree of accuracy. The washout has to be with lactated Ringer solution. In pilot experiments with animal and human livers, the HA test lost all discrimination if the reperfusion flush was performed with UW solution. Similar results of the non-discriminating properties of the HA test. if performed with non-Ringers lactate washout, have also been reported in a recent study (35) where the backtable flush ‘was performed with 5% dextrose. The reason for this selective requirement of a lactated Ringer’s flush solution is currently not known. The Ringer’s lactate flush did not have a deleterious effect on ultimate graft outcome since the incidence of PNF in this study (11/102, 10.78%) was not significantly different from that reported earlier (1, 4, 6).
All 102 livers used in the study satisfied established preoperative criteria (4, 6, 31) regarding their suitability as donor organs and should not have failed from PNF All the 11 livers (5 in the first phase and 6 in the second) which failed from PNF (Table 3 and Fig. 3), had effluent levels >400 μg·l−1 Despite satisfying preoperative criteria, all these livers subsequently demonstrated histological evidence of severe preservation injury. Conversely, none of the 79 livers (21 in phase 1, 58 in phase 2) with effluent HA levels s400 μg·l−1 failed from PNF.
Twelve (6 in each phase, 11.76%) of the 102 livers are classified as false-positive since they ultimately survived despite effluent HA levels which were >400 μg·l−1 However, 9 of these 12 livers (3 in the first phase+6 in the second phase) exhibited a severe injury pattern which was just short of catastrophic. The remaining three livers were unremarkable.
PNF is the leading cause of early graft loss and is responsible for enormous morbidity, mortality and expense. We believe that systematic measurement of effluent HA levels will permit detection of livers that have been severely injured and are, therefore, at high risk of developing PNF, or at the very least exhibit a severely troubled post-operative course (false positive). In the presence of a shrinking donor pool, this may not be sufficient to prompt a transplant surgeon to discard a liver. The utility of the test, we believe, lies in the fact that a low HA level guarantees a good liver, while the presence of a high HA level should forewarn the surgeon of the increased risk of PNF or a significantly deranged early graft function. The HA test is easy to perform and can be rapidly assayed within 2 h.
We therefore, conclude that:
The authors are grateful to Ion Deaciuc Ph. D., Department of Physiology. LSU Medical Center. New Orleans for a critical review of the manuscript.