Identifying the risk factors associated with recurrence of primary biliary cirrhosis after liver transplantation may affect immunosuppression and increase understanding of the pathogenesis. Four hundred eighty-five patients with PBC were followed for a median of 79 months after transplantation; histological evidence of recurrence was found in 23%. On multivariate analysis, the only risk factor identified with recurrence was the type of calcineurin inhibitor used. The odds ratio for recurrence on tacrolimus was 2.73 (95% confidence interval: 1.84–4.10) compared with cyclosporine. For those receiving cyclosporine, the median time to recurrence was 123 months and for those on tacrolimus 62 months (P < .001). Reasons for this difference between the 2 calcineurin inhibitors are not clear. (Liver Transpl 2004;10:488–491.)
Primary biliary cirrhosis (PBC) is a chronic, progressive disease characterized by progressive destruction of the small- to medium-sized intrahepatic bile ducts. The cause of PBC remains uncertain. PBC recurs in some patients after liver transplantation1, 2; defining the risk factors for recurrence may help not only in allowing the introduction of measures to reduce the effect of recurrence, but may give further insight into the pathogenesis of this condition.
Reported rates and severity of recurrence vary between centers1, 3: there are many possible explanations for the differences observed but the type of immunosuppression may be important. Although recurrence of PBC was initially reported in patients receiving cyclosporine-based therapy, early recurrence of diseases reported in patients receiving tacrolimus, both in grafts from cadaveric and living donors.4, 5 Haagsma suggested that the lack of corticosteroids use may be an important factor.3 We reported that the choice of calcineurin inhibitor (CNI) may be an important factor6 based on analysis at 2 years of patients in a randomized prospective study comparing tacrolimus and cyclosporine-based immunosuppression. However, several other studies have reported that the choice of CNI has no effect on recurrence.7–10 We, therefore, analyzed pre- and post-transplant factors that are associated with PBC recurrence.
Between 1982 and December 2002, 485 patients were transplanted for PBC. The diagnosis of PBC was made on the basis of the history, examination, serology and histology (Table 1). Eighty-eight percent of the patients were female, and the median age at transplant was 55 years. More than 93% were positive for antimitochondrial antibodies (as detected by immunofluorescence). All but 11 were Caucasoid; of the non-Caucasoids, 8 were of Asian origin, 2 Oriental, and 1 Afro-Caribbean origin. Twenty-one received a split graft and 2 received a reduced graft, the rest a cadaveric whole graft. The median follow-up was 79 months.
Table 1. Characteristic of Patients Grafted for PBC and Probability for Recurrence of Disease
Abbreviations: INR, international normalized ratio; IMS, immunosuppression.
Ratio of PBC recurrence rates for presence of characteristic versus absence or, for continuous variables, for values at or above median versus values below median.
Calculated from actual values of continuous variables rather than dichotomous classifications used for hazard ratios.
Serum bilirubin (μmol/L)
Serum creatinine (mmol/l)
Tacrolimus as initial IMS
Azathioprine at 1 year
Warm ischaemic time (min)
Cold ischaemic time (min)
Donor/recipient gender mismatch
Patients were started on triple immunosuppression with corticosteroids, azathioprine, and a CNI. Until 1990, cyclosporine was the main immunosuppressive agent; tacrolimus was introduced in 1990 and after 1999 became the major, first-line CNI. Corticosteroids were withdrawn by 3 months except in those instances of recurrent rejection. Six patients treated initially with cyclosporine were switched to tacrolimus in the first year and 4 from tacrolimus to cyclosporine. None of the patients were taking ursodeoxycholic acid before the diagnosis of recurrent PBC.
The diagnosis of recurrent PBC was made on the basis of a transplant for PBC and compatible histology and exclusion of other causes of bile duct damage, such as features of acute or chronic rejection, biliary outflow obstruction, and vascular abnormalities. Biliary outflow obstruction was assessed using ultrasound.
Protocol liver biopsies were done annually until 1997; thereafter, a biopsy was done at year 1 and 3-yearly intervals unless there were clinical indications to change. The biopsies were assessed by 1 histopathologist (S.G.H.). Histological criteria for the diagnosis of recurrent PBC are as described in detail elsewhere11 and included portal granulomas, inflammatory bile duct lesions, and bile duct loss, in the absence of features suggestive of rejection, biliary obstruction, or infection.
The disease free survival time was defined for those with recurrence as the time from transplantation to the first biopsy showing evidence of PBC and for those with no recurrence as the time from transplantation to the most recent biopsy, the latter times being treated as censored observations. Where necessary continuous variables were transformed (by taking logarithms or reciprocals) so that they were approximately normally distributed. Both univariate and multivariate (stepwise) analyses used Cox regression analysis to determine the factors associated with recurrence of PBC and survival curves were estimated by the Kaplan-Meier method. All statistical analyses were performed using SPSS for Windows 10.0.7. Survival curves were compared using the Breslow tests because, of the available tests, this places least reliance on time points with smaller numbers.
The characteristics of the patients and donors are shown in Table 1. Of the patients who had at least 1 post-transplant biopsy, 266 had cyclosporine as the initial immunosuppressive agent and the remainder (n = 90) started with tacrolimus as the main immunosuppressive agent. In addition, 14% were also receiving corticosteroids at 6 months and 13% at 2 years. None of the patients was treated with OKT3 or anti-thymocyte globulin. Seven were given mycophenolate but none was given sirolimus before the diagnosis of recurrent PBC was made on liver histology. Acute cellular rejection occurred in 56% of those receiving cyclosporine and in 46% of those receiving tacrolimus as the calcineurin inhibitor; chronic rejection developed in 8.7% of those receiving cyclosporine and in 3.7% of those receiving tacrolimus.
Histological evidence of disease recurrence was found overall in 114 patients (23%), of whom 78 were receiving cyclosporine and 36 tacrolimus as the primary CNI immunosuppressive drug. The rate of recurrence and the number at risk is shown in Fig. 1.
In the univariate analysis, only the use of immunosuppression was associated with recurrence of disease (tacrolimus as initial immunosuppression, hazard ratio 2.73, 95% confidence interval: 1.82–4.10, P < .001; azathioprine, hazard ratio 0.67, (95% CI: 0.46–0.97, P = .035). However, in the multivariate analysis, only the CNI was associated with recurrence. The median time to recurrence was 123 months for patients taking cyclosporine and 62 months for those taking tacrolimus (P < .001) (Fig. 1). Although azathioprine use at 1 year was associated with a lower risk of recurrence, while significant in the univariate model (median time to recurrence in those taking azathioprine at 1 year 120 months compared to 93 months in those who had discontinued azathioprine by 1 year, (P = .03)) the effect was small and did not appear in the multivariate model.
In this study, the strongest factor associated with disease recurrence is the calcineurin inhibitor used. While it is possible that tacrolimus use is a surrogate marker for another and possibly time-dependent factor such as donor age, this is unlikely as during the overlap period between 1992 and 1996, when both agents were used, this difference is still seen and rates of recurrence are similar in patients grafted between 1991-5 and 1996-1999. The number of biopsies in the tacrolimus group was slightly but not significantly less than in the cyclosporine group; this difference would reduce, if anything, the risks of detecting recurrence in those receiving tacrolimus. While it could be argued that the different lengths of follow-up for the two groups could confound comparisons of recurrence, when the 3 year follow-up is assessed, the increased rate of recurrence in the tacrolimus group persists (P = 0.001).
Several studies have looked at rates and risk factors for recurrence of PBC. However, in many series, the number of patients studied is small, confidence intervals not given and many studies have not been based on protocol biopsies. Since histological features of recurrence may be present in the presence of normal liver tests (11,12,13), liver biopsies taken only when liver tests are abnormal may underestimate the extent of the problem. In addition, it must be stressed that the number of patients evaluated for recurrence after 5 years is relatively small, so comparison of rates of recurrence after this time with other reports, should be done with considerable caution.
The role of immunosuppression and recurrence of PBC is conflicting. Withdrawal of immunosuppression may unmask PBC.14, 15 Corticosteroids have been postulated to protect the graft from recurrent PBC since lower rates of recurrence may be seen in those on regimes with corticosteroids.3 This suggestion was based on lower rates of recurrence in patients in series where corticosteroids were maintained. Because most of our patients were not taking corticosteroids after 3 months, this study cannot answer whether steroids delay disease recurrence. With respect to the CNI used, several studies have failed to show a difference between the 2 agents, however, numbers have been small3: thus, Khettry et al.10 found that 8 of 43 patients had histological evidence of recurrence, and there was no effect of CNI. Similar findings were reported by Levitsky et al.7 who found no difference in the rate of recurrence related to CNI use where recurrent disease was seen in 7 of 43 patients. Sanchez et al.8 reported the Baylor experience of recurrent PBC in 169 patients, of whom 13 were excluded from analysis. Recurrence developed in 8.4% of recipients taking cyclosporine, azathioprine and steroids, compared with 12.2% of those receiving cyclosporine and steroids alone and 16.7% of patients taking tacrolimus and steroids (with or without mycophenolate). These differences were not statistically significant. The largest study from Berlin examined recurrence rates in 91 patients grafted for PBC between 1989 and 2000.9 Protocol biopsies were carried out: the recurrence rates were 17% for cyclosporine treated patients and 18% for tacrolimus-treated patients. In contrast, an increased rate of recurrence in those taking tacrolimus was suggested by our own unit6 based on an analysis of liver histology in a prospective study comparing cyclosporine and tacrolimus but numbers were small (7 of 16 patients on tacrolimus and 1 of 11 on cyclosporine had evidence of recurrence).
Other factors that might be implicated in the rates of recurrent PBC include use of corticosteroids, HLA match or mismatch. In this study, we could not analyze the effect of corticosteroids as there were too few patients maintained on long-term steroids and we did not have sufficient HLA matches known to analyze the effect.
The reasons for the different effects of the 2 CNIs is not clear. Although cyclosporine and tacrolimus are both CNIs, there are some significant differences which may shed light on the pathogenesis of PBC. Cyclosporine has been shown to have some beneficial effect in the treatment of patients with PBC,5, 17 with improved liver tests and a lower liver related mortality; clinical trials assessing a possible beneficial therapeutic effect of tacrolimus have not, to the authors' knowledge, been reported. Furthermore, development of inflammatory bowel disease after liver transplantation is affected by the choice of immunosuppression. Inflammatory bowel disease-free survival is greater in patients not receiving tacrolimus and in patients receiving azathioprine.18 The mechanisms for this difference between the two CNIs are not clear. Cyclosporine, but not tacrolimus, stimulates transforming growth factor (TGF)-B1 production by biliary epithelial cells.19 The 2 drugs also have differential effects on retroviral replication,20 and retroviruses have been suggested to be implicated in the pathogenesis of PBC.21In vitro, calcineurin inhibits tumor necrosis factor-induced apoptosis; cyclosporine, but not tacrolimus, inhibits mitochondrial permeability transition and cytochrome c release22, 23 and so may reduce apoptosis.24
Although recurrence of PBC in the allograft does not have a major impact on graft survival, some grafts have been lost because of recurrent disease.1 In view of the long natural history of recurrent disease, longer term studies are needed to determine the consequences of the reduced time to recurrence seen in tacrolimus-treated patients.