Does the Banff Rejection Activity Index Predict Outcome in Patients With Early Acute Cellular Rejection Following Liver Transplantation?

Transcription

1 LIVER TRANSPLANTATION 12: , 2006 ORIGINAL ARTICLE Does the Banff Rejection Activity Index Predict Outcome in Patients With Early Acute Cellular Rejection Following Liver Transplantation? Barbara S. Höroldt, 1 Marco Burattin, 1 Bridget K. Gunson, 1 Simon R. Bramhall, 1 Peter Nightingale, 2 Stefan G. Hübscher, 3 and James M. Neuberger 1 1 Liver Unit, Queen Elizabeth Hospital, Birmingham, UK; 2 Wellcome Trust Clinical Research Facility, Queen Elizabeth Hospital, Birmingham, UK; and 3 Department of Pathology, Queen Elizabeth Hospital, Birmingham, UK The Banff schema incorporates a semiquantitative scoring system for grading of acute cellular rejection (ACR) of the liver allograft. The Banff rejection activity index (RAI) comprises 3 components scored from 0 to 3: venous endothelial inflammation (E); bile duct damage (B); and portal inflammation (P); the scores are combined to an overall score (the RAI). The purpose of this research was to determine the prognostic value of the Banff RAI score in predicting the response to increased immunosuppression and the long-term outcome of the graft. A retrospective study was done of patients undergoing primary liver transplantation between January 2000 and October 2004 with tacrolimus-based immunosuppression; 495 patients were included, 231 had histologically-confirmed ACR, 193 responded to 1 cycle of high-dose steroids. There was no correlation between the total RAI score and response to steroids, resistant rejection, development of chronic rejection, or graft survival. The E score was related to patient survival, a lower score being associated with a worse outcome (P 0.048). In multivariable analysis, serum bilirubin, serum aspartate aminotransferase, and E score were significant predictors of death (P 0.012). In univariable analysis, B score and bilirubin were significantly related to resistant rejection (P and 0.002, respectively), but only bilirubin was significant in multivariable analysis (logistic regression). In conclusion, although the Banff RAI score is a useful marker of the severity of rejection, neither the total RAI score nor any of the individual components correlated with response to steroids or graft survival. Liver Transpl 12: , AASLD. Received January 23, 2006; accepted February 24, Acute cellular rejection (ACR) remains a common problem following liver transplantation (LT) affecting up to 60% of patients receiving tacrolimus. 1 To overcome the problems encountered through different definitions and the varying nomenclature of ACR, a panel of international experts in LT met to agree a set of definitions regarding liver allograft rejection. 2 At the Banff Conference for Allograft Pathology, a grading system for liver allograft rejection was proposed and has been widely adopted worldwide. 3,4 The Banff schema has 2 components: a global assessment of the overall rejection grade and a semiquantitative assessment of the 3 main histological features of acute rejection: portal inflammation (P), bile duct inflammation/damage (B), and venous endothelial inflammation (E); on a scale from 0 (absent) to 3 (severe). The individual scores are added to produce the overall rejection activity index (RAI). 3 A number of studies have shown the Banff schema to be useful in the diagnosis and staging of ACR. These studies have mainly focused on the global assessment rather than the RAI. 5-7 The clinical usefulness and prognostic value of the RAI are less extensively documented and the main focus of the present study. While ACR may herald the development of chronic rejection in some liver allograft recipients, in distinction to renal allograft recipients there is no good evidence that ACR is associated with a worse outcome. Indeed, Abbreviations: ACR, acute cellular rejection; RAI, rejection activity index (of the Banff schema); P score, portal inflammation score (part of RAI); B score, bile duct damage score (part of RAI); E score, venous endothelial inflammation score (part of RAI); LT, liver transplantation; OLT, orthotopic liver transplantation. Address reprint requests to Barbara Höroldt, 17 Withens Avenue, Sheffield, S6 1WE, UK. Telephone: ; FAX: ; bsrhoeroldt@doctors.org.uk DOI /lt Published online in Wiley InterScience ( American Association for the Study of Liver Diseases.

2 OUTCOME OF ACR AND THE BANFF SCORE 1145 TABLE 1. Baseline Characteristics of Patients All patients ACR No ACR Age (yr) (16-73) (16-73) (16-72) Male, female (number) 277/ / /105 Diagnosis (number) ALD AIH Biliary disease FHF Metabolic Viral Other HCC (number) MELD* 15 (6-50) 15 (6-50) 14 (6-44) BMI ( ) ( ) ( ) Blood tests at OLT Bilirubin (micromoles/l) 60 (4-1261) 76 (5-1291) 55 (4-958) INR 1.3 ( ) 1.3 ( ) 1.3 (0.9-13) Creatinine (micromoles/l) 90 (30-852) 92 (49-852) 90.5 (56-885) Creatinine 150 (Number of patients) Abbreviations: MELD, Model of End-Stage Liver Disease; FHF, fulminant hepatic failure; OLT, orthotopic liver transplantation; ALD, alcoholic liver disease; AIH, autoimmune hepatitis; HCC, hepatocellular carcinoma; BMI, body mass index; INR, international normalized ratio. NOTE: Unless otherwise specified all values are given as median (range). *The MELD score is not adjusted for the presence of cancer. There is no significant difference in the incidence of ACR between the various indications for OLT. some studies have shown that early ACR is associated with better graft and patient survival, and that more severe rejection is associated with a better outcome than those with mild rejection While the Banff schema represents a useful and validated method of assessing the presence and severity of acute cellular rejection, the clinical implications (particularly of the RAI) are poorly defined. We therefore undertook a retrospective study to assess whether the aggregate Banff score 2 (RAI) or its subscores is associated with a response to increased immunosuppression, with the occurrence of further episodes of rejection, and with patient and graft outcome. PATIENTS AND METHODS Patients Between January 2000 and October 2004, 559 adult patients received a primary orthotopic liver transplant (OLT) at the Liver Transplant Unit at the Queen Elizabeth Hospital, Birmingham, UK, from deceased donors. This time frame was selected because all recipients had a standard immunosuppressive regime (see below). Follow-up was censored at 31 October Patients were included in the study if they fulfilled all the following criteria: primary transplant, primary immunosuppression with tacrolimus, and survived for more than 5 days following LT without graft failure. Exclusion criteria were primary immunosuppression other than tacrolimus, treatment for suspected rejection in the absence of liver biopsy, or treatment for suspected rejection that was not confirmed by histology. A total of 495 patients were included and 116 were excluded: 52 had a previous graft, 15 died within 5 days of transplantation, 6 developed graft failure within 5 days of transplantation, 9 had ACR not confirmed on liver biopsy, 10 were treated for ACR but no liver biopsy performed, and 24 did not have tacrolimus-based primary immunosuppression. Details of the underlying liver disease and transplant variables are summarized in Table 1. All biopsies were scored by dedicated liver histopathologists according to the Banff schema. The primary study outcome criteria were patient and graft survival, secondary outcome measures were response to treatment (need for 1 course of high-dose steroids) and the development of chronic rejection. Immunosuppression All patients were treated with triple immunosuppression postoperatively (usually starting within 6 hours of transplantation). Corticosteroids were given as hydrocortisone 100 mg twice daily intravenously until patients were converted to oral prednisolone starting at 20 mg/day. This dose was reduced at fortnightly intervals and discontinued at 3 months (except for patients transplanted for autoimmune hepatitis, in which corticosteroids were continued indefinitely at a dose of 5-7.5

3 1146 HÖROLDT ET AL. mg/day). Azathioprine was given at a dose of 1-2 mg/kg of body weight/day orally or via a nasogastric tube, and reduced in the presence of neutropenia or thrombocytopenia. Patients who were intolerant of azathioprine and considered to be at risk of rejection were converted to oral mycophenolate at an oral dose of 2 gm daily. Tacrolimus was given orally or via a nasogastric tube at a daily dose 0.1 mg/kg of bodyweight in 2 divided doses. Dose adjustments were according to trough levels (to maintain trough whole blood levels between 10 and 15 ng/ml for the first 3 months and then reduced to 5-10 ng/ml) and according to clinical and laboratory parameters (such as renal function, infection, rejection, or signs of toxicity). Indications for Biopsy Liver biopsies were obtained if there was a clinical suspicion of ACR or in the presence of unexplained graft dysfunction; only 8 patients underwent protocol (day 7) biopsies (6 of whom had an elevated alkaline phosphatase level at the time of biopsy). In the majority, biopsies were obtained prior to treatment of ACR; however, in some cases treatment was started before the biopsy or its result was available. In these cases, the date of diagnosis of ACR was the day treatment was instituted. Two patients had 2 distinct episodes of ACR within 100 days of OLT with normalization of laboratory values and histology between episodes (subsequent analysis includes only the first episode). Of the 495 included patients, 231 had biopsy-proven ACR within 100 days of LT. Of these, 177 developed ACR within 14 days of transplantation and 54 between days. Grading of ACR The histological findings were graded according to the Banff schema. 2 The global assessment of the histological changes was categorized as absent, indeterminate, mild, moderate, or severe according to the Banff schema. The RAI was calculated from the 3 individual scores (B, P, and E; the E score did not differentiate between portal and centrilobular endothelial inflammation). For the statistical analysis the RAI score was subsequently grouped as follows: RAI of below 2 were classified as absent, 2 as indeterminate rejection, 3 to 4 as mild, of 5 to 6 as moderate and of greater than 6 as severe. The RAI categories do not equate to the global assessment categories and are not interchangeable, so that, for example, for a given biopsy, the overall global assessment may grade the rejection as moderate whereas the total RAI score is 7 and would therefore be classified as severe. To avoid confusion the global assessment will be referred to by the grade (mild, moderate, severe) whereas the RAI score will be referred to by grade with the numeral score in parentheses, e.g., severe (RAI 7). Treatment of ACR The decision whether and how to treat patients lay with the senior responsible clinicians, who would decide whether an rejection episode required steroid boluses, increase in baseline immunosuppression, addition of new drugs or no change in treatment at all. First-line treatment consisted of administration of steroid boluses either prednisolone (200 mg orally for 3 days) or intravenous methylprednisolone 1,000 mg for 3 days; thereafter the steroid dose was reduced to 20 mg prednisolone per day. In those with mild rejection, the only treatment was increase of the tacrolimus dose when a low trough level was regarded as contributory to the ACR; azathioprine was restarted or changed to mycophenolate where appropriate. Response to Treatment of ACR An episode of ACR was defined as responsive when 2 out of 3 laboratory parameters (serum transaminases, alkaline phosphatase, or bilirubin) fell to values less than twice the upper limits of normal within 2 weeks of the start of the treatment or diagnosis. Histological evidence of resolution was evaluated where available, but routine follow-up biopsies were not carried out to confirm histological resolution. Nonresponsive or partially-responsive ACR was treated with a further bolus treatment of steroids or with OKT 3 (1 patient), with or without other adjustment of the immunosuppressive treatment. Statistical Analysis Statistical analysis was performed with SPSS for Windows (SPSS, Chicago, IL). Categorical data were summarized as frequencies and percentages, continuous data as medians and ranges. Survival data were analyzed by the Kaplan-Meier method and the Tarone- Ware test or by Cox regression. Dichotomous outcome variables were analyzed by logistic regression. Stepwise Cox and logistic regression were used for the multivariable analyses. A P value less than 0.05 was deemed significant. RESULTS Grading of ACR and Treatment at Time of ACR A total of 495 patients were included in the study, of whom 231 developed histologically-confirmed ACR within 100 days following OLT. The baseline characteristics of all patients are given in Table 1 and the details at the time of diagnosis of ACR in Table 2 (In Table 2a by the global assessment, and in Table 2b by RAI grouping). Rejection was classified as indeterminate (RAI 2) in 4 patients, mild (RAI 3-4) in 41 patients, moderate (RAI 5-6) in 83 patients, and severe (RAI 7-9) in 103 patients. The global assessment graded 7 biopsies as indeterminate, 52 as mild, 94 as

4 OUTCOME OF ACR AND THE BANFF SCORE 1147 TABLE 2. Parameters at Time of ACR No ACR (N 264) ACR (N 231) Indeterminate/ mild (N 59) Moderate (N 94) Severe (N 78) a. Banff grade - global assessment Blood tests (day ACR diagnosis or Day 8 if no ACR) AST (Iu/L) 98 (50-410) 96 ( ) 99 (23-852) 97 ( ) 96 ( ) ALP (Iu/L) 552 ( ) 1060 ( ) 993 ( ) 1098 ( ) 866 ( ) Bilirubin ( mol/l) 36 (5-347) 110 (12-785) 88 (12-409) 111 (12-498) 130 (19-785) Creatinine (day 86 (50-409) 101 (12-446) 99 (55-446) (38-437) 96 (54-343) before, mol/l) INR (day before) 1.1 ( ) 1.1 ( ) 1.1 ( ) 1.1 ( ) 1.2 (1-2.5) Immunosuppression Steroids (all patients are on steroids at Day 8 and at the time of ACR) Azathioprine Number Median dose 125 mg 125 mg 125 mg 125 mg 125 mg Median dose* 1.64 mg/kg 1.63 mg/kg 1.67 mg/kg 1.60 mg/kg 1.60 mg/kg MMF Tacrolimus Patients All All All All All Dose 6 mg (1-20) 7 mg (2-17) 7 mg 2-10) 7 mg (2-17) 7 mg (2-16) Level (day before) 7.95 ng/ml ( ) 7.65 ng/ml (1.2-30) 7.9 ng/ml ( ) 7.4 ng/ml (2-18.5) 7.4 ng/ml (1.5-30) b. Banff RAI scores Blood tests (day ACR diagnosis or Day 8 if no ACR) AST (IU/L) 98 (50-410) 96 ( ) 98 (35-852) 93 ( ) 96 ( ) ALP (IU/L) 553 ( ) 1061 ( ) 993 ( ) 1115 ( ) 931 ( ) Bilirubin ( mol/l) 37 (5-347) 111 (12-785) 67 (12-409) 112 (16-498) 125 (12-785) Creatinine (day 87 (50-409) 101 (12-446) 99 (55-446) 104 (55-444) 96 (38-343) before, mol/l) INR (day before) 1.1 ( ) 1.1 ( ) 1.1 ( ) 1.1 ( ) 1.2 (1-2.5) Immunosuppression Steroids (all patients are on steroids at Day 8 and all at the time of ACR) Azathioprine Patients Median dose 125 mg 125 mg 125 mg 125 mg 125 mg Median dose* 1.64 mg/kg 1.63 mg/kg 1.54 mg/kg 1.61 mg/kg 1.60 mg/kg MMF Tacrolimus Patients All All All All All Dose 6 mg (1-20) 7 mg (2-17) 6 mg (2-10) 7 mg (2-14) 7 mg (2-16) Level (day before) 7.95 ng/ml ( ) 7.65 ng/ml (1.2-30) 8.25 ng/ml ( ) 7.4 ng/ml ( ) 7.4 ng/ml (1.5-30) Abbreviations: INR, International normalized ratio; AST, aspartate aminotransferase; ALP, alkaline phosphatase; MMF, mycophenolate mofetil; Patients, number of patients. *Median dose mg/kg bodyweight. NOTE: The results are given as medians (range). The median time of ACR was Day 9, results for patients who did not experience ACR refer to Day 9 post-olt, apart from tacrolimus level, tacrolimus dose, and INR: these are based on Day 8. moderate, and 78 as severe ACR. The different grades of the rejection severity obtained by the 2 methods (subjective assessment and RAI grouping) showed good agreement with each other (kappa 0.70). At the time of diagnosis of ACR, all patients were on corticosteroids and tacrolimus, the median daily dose of tacrolimus was 6 mg and the median whole blood trough level was 7.65 ng/ml (range 1.2 to 19 ng/ml). A total of 174 patients were on azathioprine with a median dose of 1.69 mg/day/kg bodyweight, 17 patients were on mycophenolate mofetil (2 gm/day); the remainder were on dual immunosuppression with tacrolimus and corticosteroids. In addition, 15 patients experienced further episodes of ACR more than 100 days post-olt Correlation of Outcome with the Banff Scores Response to Treatment Of 231 patients diagnosed with ACR, 162 patients were treated with high dose steroids (151 prednisolone 200 mg for 3 days, the remainder with methylprednisolone IV). In 163 patients, the tacrolimus dose was increased and 45 patients were either restarted on azathioprine or mycophenolate mofetil was introduced (5 of whom received no additional courses of high-dose steroids). One patient each was switched to sirolimus and to cyclosporin A. Overall, in 32 patients no treatment change was instigated. In these patients the rejection episode was graded indeterminate or mild in the majority of pa-

5 1148 HÖROLDT ET AL. TABLE 3. Treatment and Outcome of ACR Episodes Within 100 Post-OLT Indeterminate/ Moderate Severe a. Banff grade global assessment All ACR mild ACR ACR ACR Number Increase dose of tacrolimus New azathioprine/mmf Other treatment CyA 1 Sirolimus 1 - OKT 3 High-dose steroids course Further high-dose steroids No treatment Response* No response Death CCR Regraft Death/regraft for CCR b. Banff RAI scores All ACR RAI 2-4 RAI 5-6 RAI 7-9 Number Increase dose of tacrolimus New azathioprine/mmf Other treatment 3 1-CyA 1-Sirolimus 1-OKT 3 High-dose steroids course Further high-dose steroids No treatment Response* No response Death CCR Regraft Death/regraft for CCR Abbreviations: CCR, chronic cellular rejection; MMF, mycophenolate mofetil; AST, aspartate aminotransferase; ALP, alkaline phosphatase. *Response to adjustment of baseline immunosuppression plus/minus single course of highdose steroids. Where available, this is the histological response, otherwise it is the biochemical response (normalization of 2 out of 3 biochemical markers bilirubin, AST, ALP within 2 weeks of treatment initiation). Two patients were regrafted and survived, 1 patient died. tients. Grading by global assessment: 1 indeterminate, 15 mild, 14 moderate, 2 severe; by RAI grouping: 2 indeterminate (score: 2), 16 mild (3-4), 12 moderate (5-6), and 2 severe (7-9). Those patients had tacrolimus levels that were regarded as adequate. The main reason for not treating was that the rejection episode was regarded not being severe enough to justify steroid bolus treatment, but in 10 patients active sepsis contraindicated the use of high-dose steroid treatment. Of those 32 patients, 5 died within 4 weeks of diagnosing ACR, all from septic complication (none from rejection, all who died had only mild rejection); 1 patient with severe rejection was already listed for regrafting for hepatic artery thrombosis, the other patient with severe rejection had sepsis precluding increased immunosuppression. In 193 patients, ACR settled either spontaneously or with the initial adjustment of immunosuppression including a single course of high-dose steroids. A total of 38 patients failed to respond to a single course of high-dose steroids (this was classified as resistant rejection); 35 of these patients were retreated with a further steroid bolus, 4 patients received a third course of steroids, and 1 received OKT 3 infusion. Further treatment was deemed inappropriate for a variety of reasons in 3 patients. Details about the treatment received and the outcome of patients are summarized in Table 3a and b. There was no statistically significant difference in either the total RAI score or the global assessment for those who developed resistant rejection. There was a trend for patients requiring additional courses of steroids to have a higher E value indicative of an increase in zone 3 necroinflammatory lesions in these patients; however, this did not reach statistical significance (P 0.127). A high B score (severe biliary inflammation or damage) was predictive of the need for further bolus treatment with corticosteroids (P 0.015; odds ratio for a B

6 OUTCOME OF ACR AND THE BANFF SCORE 1149 Figure 1. Kaplan Meier curve for survival by E score of the RAI. E score, endothelial damage and inflammation grade: mild (1); moderate (2); or severe (3); Censored includes deaths that were not liver related. score of 3 vs. a B score of [95% confidence interval, ] and for a B score of 3 vs. a B score of [95% confidence interval, ]). While B score and serum bilirubin were significantly related to resistant rejection in univariable analysis (P and respectively), only serum bilirubin was significant in multivariable analysis (logistic regression). Patient Survival Overall 86 of 495 patients died in the study period. Eleven patients died of causes unrelated to either their underlying liver disease or their graft or to any medication/treatment they received for either (e.g., traffic accident, myocardial infarction). In 75 patients, the death was classified as liver-related (death due to graft failure or complications of their liver disease, graft, or treatment). The median time to liver related death was 52 days (range 7-1,216 days). Of the 75 liver related deaths, 41 occurred in patients with ACR within 100 days of transplantation. The majority of these were related to sepsis and/or multiorgan failure. No single RAI score or category (mild, moderate, or severe rejection) was predictive of death. There was trend of patients with a lower RAI score having an increased risk of liver-related death; however, this did not reach statistical significance (P 0.115; hazard ratio for RAI score lower by ; 95% confidence interval, ). Of the individual components, only a low E score was significantly associated with liver death (P 0.049; Tarone-Ware test). In the multivariable analysis, serum bilirubin, aspartate aminotransferase (at time of biopsy), and the E score were significantly associated with liver death (P 0.012, the hazard ratio for an E score of 1 vs. an E score of 2 was (95% confidence interval, ) and for an E score of 1 vs. an E score of 3 was (95% confidence interval, ) (see Fig. 1). When liver death was subdivided into liver death due to graft failure from chronic rejection and other causes of liver death (e.g., hepatic artery thrombosis), the low E score remained significantly related to other causes of liver death, but not to those due to graft failure from chronic rejection. There was no statistically significant difference between patients with E score 1 and those with E score of 2 or 3 with respect to gender distribution (52% and 52% male patients, respectively), their median body mass index was similar (body mass index of 30.3 kg/m 2 and 29.7 kg/m 2, respectively), this remained so when choosing a body mass index of censored, which includes deaths that were not liver related 30 kg/m 2 as cutoff (37/67 and 77/164, respectively; P Fisher s exact test). The percentage of patients being transplanted for hepatitis C virus cirrhosis was slightly higher for patients with E 1; however, this did not reach statistical significance (16.42% vs. 9.76%; P , Fisher s exact test). More patients with E 1 had been grafted with a liver noted to be steatotic at transplantation (14/67 vs. 13/ 164), P (Fisher s exact test Censored, which includes deaths that were not liver related). Early ACR (less than 14 days) was associated with a better prognosis than relatively late ACR ( 14 days) (P for death and graft failure, Tarone-Ware test). The median RAI score was similar for both groups; (6.00 for ACR 15 days and 5.75 for ACR between days 15 and 100). Graft Failure During the study periods, 24 of 495 patients developed irreversible graft failure and were considered for regrafting. Of those 24 patients, 21 had documented ACR within 100 days of transplantation, including 3 with severe or established chronic rejection. The graft failure occurred 7 to 800 days (median 124) following transplantation. One patient was deemed to be too ill to survive regrafting (recurrent hepatitis C virus) and 1 patient refused (hepatic artery thrombosis). Of the remaining 22 patients, 2 patients died prior to regrafting; 1 from chronic rejection and 1 from hepatic artery thrombosis; 20 patients were regrafted. The indications of these 20 patients were chronic rejection (2), hepatitis C virus recurrence (4), and hepatic artery thrombosis (14). Nine patients died following regrafting (median time to death 35 days; range days). Neither the total RAI nor global assessment of rejection (mild, moderate, or severe rejection) nor any of the RAI subscores (or its respective value) was found to be significantly in predicting graft failure. Of the liver tests on the day of biopsy only a high aspartate aminotransferase value (at biopsy) was predictive of a negative outcome (combined outcome death or graft failure) by Cox regression analysis. Chronic Rejection Of 495 patients, 31 showed features of chronic rejection on follow-up biopsies. Of these 21 had experienced an

7 1150 HÖROLDT ET AL. episode of acute rejection within 100 days of transplantation (9.1% of the 231 patients who developed ACR during this period). Overall, 37 patients experienced ACR more than 100 days post-olt; 13 of these 37 patients (35.1%) developed features of chronic rejection. Of those 13 patients, 7 had also suffered ACR less than 100 days post-olt. Three out of 31 patients with chronic rejections had no documented ACR prior to the diagnosis of chronic rejection. The median time to the histological diagnosis of chronic rejection was 256 days (range days) post-olt. Overall, 4 patients developed clinical and pathological features suggestive of advanced chronic rejection, 3 of whom had ACR documented within 100 days of transplantation and 1 patient had no prior ACR documented. Of these, 2 were regrafted and survived with no recurrence of chronic rejection in the second graft, 1 patient died from chronic rejection being unsuitable for regrafting, and 1 patient recovered from chronic rejection regaining adequate graft function. Neither the RAI nor the subjective assessment nor any of the subscores was found to be significantly related to chronic rejection or bile duct loss. DISCUSSION Acute and chronic rejection remain important complications of LT. The incidence of both acute and chronic rejection (defined as graft failure due to chronic rejection) has declined over the last few decades, from 60 to 75% to 40 to 50% and 10 to 15% to around 2%, 3,12 respectively. Many factors are associated both with the risk of rejection and graft outcome. These include patient variables (such as age and preexisting liver disease), donor variables (such as length of intensive therapy unit stay, age, cause of death, and the degree of steatosis), donor-recipient interactions (mismatch of human leukocyte antigen or ABO blood groups, or gender mismatch 13 ), intraoperative variables (including warm and cold ischemia time, and requirement for intraoperative transfusion 14 ) and posttransplantation variables (e.g., immunosuppression, timing of rejection). Several aspects regarding the predictive value of early biopsies in ACR are of importance. A number of studies have emphasized the importance of inflammatory lesions affecting the hepatic venules and the surrounding liver parenchyma in identifying a more severe form of acute rejection. This form is less likely to respond to immunosuppression and more likely to progress to chronic rejection. 12,15-17 Also, these zone 3 necroinflammatory lesions could be interpreted as being an initial feature of chronic rejection before other features such as bile duct loss become evident. Additionally the zone 3 lesions are an important component in defining severe rejection as part of the global assessment according to the Banff schema. 2 It is therefore not surprising that severe rejection (as graded by the global assessment) has been shown to be predictive of the development of chronic rejection. 5-7 Severe venous endothelial inflammation according to the E component of the RAI score is also defined on the basis of zone 3 necroinflammatory changes. This was shown to be predictive of graft outcome in the study by Demetris et al. 5 We found the severity of biliary inflammation correlates with need for repeated pulses of steroid as well as subsequent bile duct loss. Both of these correlated significantly with serum bilirubin at the time of the biopsy; the magnitude of bilirubin elevation, although of poor specificity, can serve as a marker of the severity of bile duct injury. A low E value (E1) was found to correlate with an unfavorable outcome; increased risk of liver death, but not of liver death related to chronic rejection. The reason for this association is not clear and our outcome is at variance with the observations made in an earlier study by Demetris et al. 5 ; in which severe endothelial inflammation (E3) was found was found to be a risk factor for progression to chronic rejection. The reason for these apparently conflicting results is unclear. To find an explanation for this discrepancy, we scrutinized the pre-, peri-, and posttransplantation variables of patients with ACR, comparing those with E score 1 with patients with an E score of 2 or 3. There was no significant difference regarding the patients age, gender, Model for End-Stage Liver Disease score, and percentage of patients with hepatitis C virus cirrhosis between the 2 groups. Only a steatotic graft was found significantly more often in patients with a low E score. Whether this statistical significance is clinically significant will need to be addressed in further studies. Steatosis in the graft at the time of transplantation was not associated with an increased risk of death from liver failure. One possible explanation for the apparent beneficial effect of rejection on outcome can be found in the concept of window of opportunity of immunological engagement, first introduced by Calne 18 and Dresske et al. 19 Calne 18 postulated that immunological engagement is an important step in achieving immunological tolerance. A window of opportunity for immunological engagement, in the early posttransplantation period, was suggested during which immunological interaction between donor and recipient immunogenicity may allow for the development of tolerance and so lead to improved graft survival. It has to be emphasized that while an E score 1 was significantly related to liver death, it was not significantly related with liver death due to complications of rejection and/or treatment of rejection. The outcome of an episode of ACR has be distinguished from the outcome of LT overall as many factors can lead to an adverse outcome of LT independently from ACR. There are drawbacks to our study; it is retrospective and treatment has evolved slightly with time, the patient population is inevitably diverse and the indication for transplantation correlates with the probability of acute rejection, although we did not find significant difference among the different indications. Our findings are in contrast to those Demetris et al., 5 who reported that the Banff schema predicts graft loss from rejection and also found good correlation between severity and

8 OUTCOME OF ACR AND THE BANFF SCORE 1151 transaminases. While the incidence of ACR was comparable to those seen in this study, the majority of ACR episodes was graded mild (73%), whereas in our study only 20% had mild ACR. They also found a high E score to be predictive of graft failure and cellular rejection. In our study a high B value was predictive of need for repeated steroid pulses and of ductopenia; a low E score was predictive of liver death. It is however not the only study contrasting with Demetris et al. 5 Doussets et al. s 9 as well Tippners et al. s 20 studies also failed to demonstrate a correlation between the severity of ACR and clinical outcome. There is no readily available explanation for the apparently conflicting findings in studies assessing the clinical value of grading acute rejection. However, they suggest, that at present the Banff schema (RAI) can not be used accurately to predict the outcome of acute rejection. Given the conflicting outcomes of several studies addressing the suitability of the Banff score and particularly the RAI score for predicting the outcome of early ACR following LT, prospective studies addressing this issue would be able to clarify matters. Assessing the severity of acute rejection through the Banff score (RAI) may nevertheless still be of clinical importance in guiding the treatment of rejection episodes. For those patients with only mild ACR the use of high dose steroids may be avoided (with all its inherent potential complications and side effects). Conversely, for patients who have moderate or severe rejection, prompt treatment with corticosteroids may prevent progression to chronic rejection. This is supported by the low incidence of graft failure from chronic rejection (3/495 patients 0.06%) in the present study. REFERENCES 1. Margarit C, Rimola A, Gonzalez-Pinto I, Cuervas-Mons V, Edo A, Andreu H, et al. Efficacy and safety of oral low-dose tacrolimus treatment in liver transplantation. Transpl Int 1998;11(Suppl 1):S International Working Party. Terminology for hepatic allograft rejection. Hepatology 1995;22: Anonymous. Banff schema for grading liver allograft rejection: an international consensus document. Hepatology 1997;25: Update of the International Banff Schema for Liver allograft Rejection: Working recommendations for histopathological staging and reporting of chronic rejection. Hepatology 2000;31: Demetris A, Rupert K, Dvorchik I, Jain A, Minvervini M, Nalesnik MA, et al. Realtime monitoring of acute liverallograft rejection using the Banff schema. Transplantation 2002;9: Ormonde DG, de Boer WB, Kierath A, Bell R, Shilkin KB, House AK, et al. Banff schema for grading liver allograft rejection: utility in clinical practice. Liver Transpl Surg 1999;5: Blakolmer K, Jain A, Ruppert K, Gray E, Duquesnoy R, Murase N, et al. Chronic liver allograft rejection in a population treated primarily with tacrolimus as baseline immunosuppression: long-term follow-up and evaluation of features for histopathological staging. Transplantation 2000;69: Neuberger J, Adams D. What is the significance of acute liver allograft rejection? J Hepatology 1998;29: Dousset B, Conti F, Cherruau B, Louvel A, Soubrane O, Houssin D, Calmus Y. Is acute rejection deleterious to long-term liver allograft function? J Hepatology 1998;29: Bartlett AS, Ramadas R, Furness S, Gane E, McCall JL. The natural history of acute histologic rejection without biochemical graft dysfunction in orthotopic liver transplantation: asystematicreview. LiverTranspl2002;8: Farges O, Nocci KA, Sebagh M, Reynes M, Bismuth H. Low incidence of chronic rejection in patients experiencing histological acute rejection without simultaneous impairment in liver function tests. Transplant Proc 1995;27: Gouv ASH, van den Heuvel MC, van den Berg AP, Sloff MJH, dejong KP, Poppema S. Significance of parenchemymal changes of acute cellular rejection in predicting chronic liver graft rejection. Transplantation 2002:73: Candinas D, Gunson BK, Nightingale P, Hubscher S, Mc- Master P, Neuberger JM. Sex mismatch as a risk factor for chronic rejection of liver allografts. Lancet 1995;346: Paloma JC, Jiminez C, Moreno E, Garcia MA, Brewed J, Loinaz C, et al. Effects of intraoperative blood transfusion on rejection and survival after Orthotopic liver transplantation. Transplant Proceeding 1995;27: Neil DA, Hübscher SG. Histologic and biochemical changes during the evolution of chronic rejection of liver allografts. Hepatology 2002;35: Lovell MO, Speeg KV, Halff GA, Molina DK, Sharkey FE. Acute hepatic allograft rejection: a comparison of patients with and without centrilobular alterations during first rejection episode. Liver Transpl 2004;10: Hassoun Z, Shah V, Lohse CM, Pankratz VS, Petrovic LM. Centrilobular necrosis after orthotopic liver transplantation: Association with acute cellular rejection and Impact on Outcome. Liver Transpl 2004;10: Calne R. WOFIE hypothesis: some thoughts on an approach toward allograft tolerance. Transplant Proc 1996; 28: Dresske B, Zavazava N, Huang DS, Lin X, Kremer B, Fandrich F. WOFIE augments the immunosuppressive potency of FK-506. Window of opportunity for immunological engagement. Transpl Immunol 1998;6: Tippner C, Nashan B, Hoshino K, Schmidt-Sandte E, Akimaru K, Boker KH, et al. Clinical and subclinical acute rejection early after liver transplantation: contributing factors and relevance for the long-term course. Transplantation 2001;72: