This paper is based upon a Cochrane report: Haddad EM, McAlister VC, Renouf E, Malthaner R, Kjaer MS, Gluud LL. Cyclosporin versus tacrolimus for liver transplanted patients. In: Cochrane Database of Systematic Reviews, 2006. It was presented to the Joint International Congress of ILTS, ELITA and LICAGE at Milan, Italy in May 2006.
Conflict of interest
No funding was received for this study. Companies that make cyclosporin and tacrolimus and the principal investigators of each study were contacted in the course of the review. Vivian McAlister took part in clinical trials sponsored by each company and received, more than 10 years ago, laboratory grants-in-aid from each company.
A systematic review of randomized clinical trials (RCT) was undertaken to evaluate the beneficial and harmful effects of immunosuppression with cyclosporin versus tacrolimus for liver transplanted patients. MEDLINE, EMBASE, Cochrane Central and Hepato-Biliary Group Controlled Trials Registers were searched. Using fixed and random effects model, relative risk (RR), values <1 favoring tacrolimus, with 95% confidence intervals (CI) were calculated. Of 717 potentially relevant references, 16 RCTs were eligible for inclusion. Mortality and graft loss at 1 year were significantly reduced in tacrolimus-treated recipients (Death: RR 0.85, 95% CI 0.73–0.99; graft loss: RR 0.73, 95% CI 0.61–0.86). Tacrolimus reduced the number of recipients with acute rejection (RR 0.81, 95% CI 0.75–0.88) and steroid-resistant rejection (RR 0.54, 95% CI 0.47–0.74) in the first year. Lymphoproliferative disorder or dialysis rates were not different but more de novo diabetes (RR 1.38, 95% CI 1.01–1.86) occurred with tacrolimus. More patients stopped cyclosporin than tacrolimus (RR 0.57, 95% CI 0.49–0.66). Treating 100 recipients with tacrolimus instead of cyclosporin would avoid rejection and steroid-resistant rejection in nine and seven patients respectively, graft loss and death in five and two patients respectively, but four additional patients would develop diabetes after liver transplantation.
The introduction of cyclosporin into liver transplantation care, 20 years ago, coincided with a dramatic improvement in outcome (1). Ten years later, the first randomized clinical trials (RCT) of immunosuppression after liver transplantation compared a new drug, tacrolimus, to cyclosporin. Two large registration trials showed a reduction in the rate of acute rejection with tacrolimus but reductions in post-transplantation mortality and graft loss were not statistically significant (2,3). Subsequently both tacrolimus and cyclosporin were found to have a common mechanism of action (i.e. inhibition of calcineurin phosphatase) even though they bound different intracellular proteins. These intracellular proteins belong to the immunophilin family. Cyclosporin binds cyclophilin and tacrolimus binds FKBP12 (4).
Cyclosporin reformulation gave rise to further RCTs, which confirmed the advantage of tacrolimus with respect to the prevention of rejection. One of these trials suggested that the failure to confirm a survival advantage with tacrolimus was due to sample size but clinically substantial nonetheless (5). Even though cyclosporin and tacrolimus share a mode of action that is responsible for both the desired effect and the major unwanted effects, they have different pharmacokinetic profiles, different serum binding mechanisms and different minor side effects. Therefore, cyclosporin and tacrolimus could have different benefit and harm profiles. Systematic reviews of cyclosporin versus tacrolimus for kidney transplanted patients has been done, but we have not identified previous meta-analyses or systematic reviews for liver transplanted patients (6,7). We performed a systematic review of RCTs to evaluate the beneficial and harmful effects of immunosuppression with cyclosporin versus tacrolimus for liver transplanted patients.
We included randomized comparisons of tacrolimus versus cyclosporin in patients undergoing their first liver transplantation, irrespective of language or publication status. Collateral interventions were allowed if received by all intervention arms with the exception of azathioprine. The following outcome measures were evaluated 1 year after randomization: all-cause mortality (primary outcome measure), graft loss, acute rejection, steroid-resistant rejection, new-onset diabetes, new-onset dialysis-dependent renal failure, post-transplant lymphoproliferative disease (PTLD), other unspecified adverse events and withdrawals. In studies where 1-year follow-up was not available even after correspondence with the principal investigator, those outcomes that are available at the nearest time point to 1 year were included in the general and subgroup analyses.
The Cochrane Hepato-Biliary Group Controlled Trials Register (CHBG CTR), The Cochrane Controlled Trials Register on The Cochrane Library (CENTRAL), MEDLINE and EMBASE were searched using terms and strategies developed with the help of the Trials Search Coordinator of the Cochrane Hepato-Biliary Group (8). We also scanned bibliographies in relevant articles and conference proceedings. We wrote to authors of included trials and pharmaceutical companies that are involved in the production of tacrolimus or cyclosporin.
Data extraction was performed by two reviewers (VM, EH) using standardized forms and reports, which were blinded as to authors, location and journal. Methodological quality was assessed on the basis of randomization and follow-up (9). Analyses were performed with the Cochrane Collaboration review manager programme (RevMan 4.2) (8). The number of events and number of patients in all intervention arms were used to calculate relative risks (RR) and risk differences with 95% confidence intervals (CI). The following subgroup analyses were carried out: pediatric recipients; patients infected with hepatitis C virus (HCV) at the time of transplantation; studies using oil-based cyclosporin; studies where cyclosporin was combined with azathioprine; studies where tacrolimus and cyclosporine are combined with mycophenolate mofetil (MMF) or sirolimus; trials without follow-up data for 1 year.
Searches performed on August 30, 2005 resulted in 717 hits, which yielded 114 reports when duplicates were removed. After initial review, 20 potential RCTs were identified. Four were excluded on further examination because they were either a review of other studies (10), a sub-analysis of another study (11), or designed for other purposes, usually regarding perioperative care, without any of the outcomes being studied in this systematic review (12,13). Data at 1 year after liver transplantation were available in all the remaining 16 studies except for two in which data were only available at 3 months (14) and at 6 months (15). After contacting principal investigators and sponsoring pharmaceutical companies, supplementary information was supplied regarding seven reports (11,16–21).
The 16 included studies had 3813 participants of whom 1899 were randomized to tacrolimus and 1914 to cyclosporin (2,3,5,14–26). Seven of the studies were conducted at single center sites (15–17,20,21,25,26) while the remaining 9 studies were multi-centered. In all, the studies involved 59 liver transplantation centers in 18 countries. Most of the RCTs restricted enrolment to adults but one included children (3) and one was restricted to children (22). HCV cirrhosis was included as an indication for transplantation in all studies but only one identified the outcome in patients with HCV in a post hoc analysis (18). Two RCTs confined entry to patients with HCV (23,26). The three earliest studies with 1157 participants compared tacrolimus with the original oil-based formulation of cyclosporin (Sandimmune®) (2,3,17) whereas the other 13 studies with 1656 participants compared tacrolimus with the microemulsion formulation of cyclosporin (Neoral®). All of the studies used trough-level monitoring to guide cyclosporin and tacrolimus dosing except one study (18) which used the 2-h post-dose level to guide the dose of cyclosporin. Table 1 gives details of the RCTs including concomitant immunosuppression. In four studies, azathioprine was given only to cyclosporin-treated patients according to local best practice (2,3,14,22).
Table 1. Meta-analysis of tacrolimus versus cyclosporine in liver transplant recipient: description of included studies
1st author (name, year)
Study period in months
Pred = prednisone; Aza = azathioprine; MMF = mycophelonolate mofetil; HCV = hepatitis C virus.
The primary outcome favored tacrolimus (Figure 1). Mortality at 1 year was reduced by 15% in the tacrolimus patients (RR 0.85, 95% CI 0.73–0.99). Graft survival was reported in 15 studies favoring tacrolimus with 22% less grafts lost (RR 0.78, 95% CI 0.68–0.79) (Figure 2). Rejection and steroid-resistant rejection were reduced by 18% and 43%, respectively in the tacrolimus-treated recipients (RR 0.82, 95% CI 0.76–0.88; RR 0.57, 95% CI 0.46–0.71) (Figure 3). These results are from intent-to-treat analyses. Substantially more patients discontinued cyclosporin than tacrolimus (RR 0.57, 95% CI 0.49–0.66) (Figure 4).
No differences were seen in the rates of chronic renal failure requiring dialysis or of lymphoproliferative disorder after liver transplantation (RR 1.55, 95% CI 0.64–3.78; outcome 09.01: RR 1.01, 95% CI 0.36–2.86, respectively). Differences in the serum creatinine at 1 year favoring tacrolimus were not statistically significant but data were available from only two studies with a total of 672 patients (–6.62 μmol/L, 95% CI –22.86 to 9.63). However the rate of new-onset diabetes was increased by 27% in the tacrolimus-treated patients (RR 1.27, 95% CI 1.12–1.44) (Figure 5). Insufficient data were reported regarding other adverse events for systematic analysis. These results can be examined in detail in the Cochrane report (8).
The number of deaths was 254 in the tacrolimus group (1899 patients) and 302 in the cyclosporine group (1914 patients). The actual number of patients and events are presented in Table 2 with the absolute risk differences and 95% CI for each outcome. Treating 100 recipients with tacrolimus instead of cyclosporin would avoid rejection and steroid-resistant rejection in nine and seven patients respectively, graft loss and death in five and two patients respectively, but four additional patients would develop diabetes after liver transplantation.
Table 2. Meta-analysis of tacrolimus versus cyclosporin immunosuppression after liver transplantation: absolute risk difference
Tacrolimus n/N (%)
Cyclosporin n/N (%)
Risk difference (95%CI)
Relative risk <1 favors tacrolimus; Risk difference <0 favors tacrolimus; 95%CI = 95% confidence interval; n/N = number with adverse event/number at risk; PTLD = post-transplant lymphoproliferative disorder.
−2% (−5, 0)
−5% (−8, −2)
−9% (−12, −6)
−7% (−9, −4)
0% (−0.01, 0.01)
Dialysis (de novo)
0.01% (−0.01, 0.03)
Diabetes (de novo)
4% (2, 7)
−11% (−13, −8)
Regression asymmetry tests showed no significant evidence of publication bias or other biases (p = 0.328). In meta-regression analyses, the treatment effect was not significantly associated with the allocation sequence generation (regression coefficient −0.022, 95% CI −0.57 to 0.52) or allocation concealment (regression coefficient −0.17, 95% CI 0.48–0.15). Identical results for the primary outcome were found whether fixed or random effect meta-analysis is used because the studies lack heterogeneity (I2= 0%) (31). Random effect meta-analyses did not change the other outcomes (8).
The formulation of cyclosporin, oil-based or microemulsion, did not affect the primary or other outcomes. No statistical heterogeneity was found between studies in each subgroup (Figure 1) (31). The mean reduction in mortality risk with tacrolimus is 20% and 13% in the oil-based and microemulsion cyclosporin subgroups, respectively, but the 95% confidence intervals cross unity so that the aggregate of almost all 3813 patients is required for statistical significance with respect to the primary outcome (Figure 1). Graft loss was reduced in the tacrolimus group by 20% (p = 0.04) and by 27% (p = 0.00008) compared to oil-based cyclosporin and microemulsion cyclosporin, respectively (Figure 2). Acute rejection was reduced by 13% (p = 0.02) and by 18% (p < 0.0001) and steroid-resistant rejection was reduced by 52% (p < 0.01) and by 30% (p = 0.02) compared to the oil-based and microemulsion formulations of cyclosporin, respectively (Figure 3). Proportionately more patients discontinued microemulsion cyclosporin than oil-based cyclosporin (Figure 4). Results from the single study that used the 2-h post-trough level to administer cyclosporin did not vary from those expected by analysis of the others (18,30). The outcomes were not altered by inclusion of the following subgroups: (i) studies with children, (ii) studies not reporting 12 month data, (iii) studies using MMF concomitantly and (vi) studies where azathioprine was given to some cyclosporin patients according to the best local practice (8).
HCV was the commonest diagnosis in studies after 1994, prior to which time these patients were diagnosed with non-A, non-B hepatitis. Subgroup analyses of studies confined to patients with HCV (23,26) and those which included HCV among other diagnoses did not show an effect related to HCV (Table 3). Reporting of actual doses and levels of drug used was too sparse to permit more detailed analyses.
Table 3. Mortality in a meta-analysis of tacrolimus versus cyclosporin immunosuppression after liver transplantation: subgroup analysis for studies confined to patients with hepatitis C virus
Tacrolimus n/N (%)
Cyclosporin n/N (%)
Relative risk (95%CI)
Risk difference (95%CI)
Relative risk <1 favors tacrolimus; Risk difference <0 favors tacrolimus; 95%CI = 95% confidence interval; n/N = number with adverse event/number at risk; HCV = hepatitis C virus.
0.84 (0.36, 1.04)
−4% (−19, 1)
HCV and other diagnoses
0.85 (0.73, 0.99)
−2% (−5, 0)
0.85 (0.73, 0.99)
−2% (−5, 0)
Cyclosporin was introduced without the benefit of RCTs into care of the recipient after liver transplantation. A decade later, tacrolimus was developed, but considerable hesitancy remained regarding the robustness of liver transplantation to allow for randomization. This concern is apparent in descriptions of the earliest RCT in this review (17,27) and in a discussion of the two registration RCTs of tacrolimus (32). Since then a further 13 RCTs of tacrolimus and cyclosporin in liver transplantation have been performed which suggests an increasing level of comfort with the procedure. The current systematic review shows why this has occurred. Outcomes after liver transplantation are very good. Overall patient and graft survival rates are 85% and 80%, respectively.
Compared to cyclosporin, tacrolimus significantly reduced the risks after liver transplantation of death, graft loss, acute rejection and steroid-resistant rejection. Tacrolimus increased the risk of new-onset diabetes. More patients discontinued cyclosporin than tacrolimus. Tacrolimus's superiority to cyclosporin after liver transplantation has to be considered in the context of the excellent overall results using either drug. Calculating the risk difference of each treatment helps us understand its impact. Treating 100 liver recipients with tacrolimus instead of cyclosporin would result in two less deaths, five less graft losses, nine less patients with rejection and seven less with steroid-resistant rejection but four more patients would develop diabetes. Table 1 should be consulted to see the 95% confidence interval for these differences. For example, the patient and graft survival benefit with tacrolimus could be as high a 5 and 8 per hundred, respectively, but not below 0 and 3, respectively.
The involvement of the majority of transplant centers throughout the world in the RCTs reviewed combined with the lack of heterogeneity of the studies support the veracity of this review. Subgroup analyses show that the results are robust in a wide variety of situations. Cyclosporin reformulation as a microemulsion did not alter the outcomes. RCT results did not vary with era but instead clustered around the mean as would be expected according to sample size. Very similar results with respect to graft survival, rejection and diabetes outcomes were achieved by a recent Cochrane review of tacrolimus and cyclosporin in kidney transplantation (7).
Both cyclosporin and tacrolimus are immunosuppressive because they inhibit calcineurin phosphatase in lymphocytes. Inhibition of the same pathway in the beta-cells of the pancreas reduces insulin production. The superior effect of tacrolimus in the prevention of rejection was accompanied by an increase in the rate of diabetes in this review and in the kidney transplantation meta-analysis (7). The difference between cyclosporin and tacrolimus may, therefore, be related to the potency of calcineurin phosphatase inhibition. Insufficient data regarding exposure to cyclosporin or tacrolimus was reported in the RCTs to determine if the outcomes would merge with particular dosing protocols. The most recent RCT was built on the premise that superior outcomes would be seen with cyclosporin if dosing was guided by 2-h post-dose drug levels rather than the traditional trough level used in the other studies. Our analysis did not show the results of this trial to fall outside the expected range (18).
The higher rate of cyclosporin discontinuation seen consistently throughout the review suggests that clinicians find it more difficult to achieve the balance between efficacy and unwanted effects with that medication than with tacrolimus. Differences in pharmacokinetic profiles or in secondary side effects between cyclosporin and tacrolimus may account for this difficulty. However, switching from cyclosporin to tacrolimus does not appear to ameliorate the disadvantage of initial prescription with cyclosporin. The higher incidence of diabetes in tacrolimus-treated patients may have a negative impact on outcomes beyond the range of this review and it may reduce or negate the benefit of tacrolimus in subsequent years. Further progress in the care of liver transplant recipients requires greater disassociation of the immunosuppressive effect of calcineurin phosphatase inhibition from its unwanted effects.
We thank Sarah Klingenberg, Dimitrinka Nikolova and Dr. Christian Gluud of the Cochrane Hepato-Biliary Group for their advice and support.
This article was carried out using the recommendations of The Cochrane Collaboration and The Cochrane Hepato-Biliary Group. This review will be published in The Cochrane Database of Systematic Reviews. Cochrane Reviews are regularly updated as new evidence emerges and in response to comments and criticisms. The Cochrane Database should be consulted for the most recent version of this review.