Potential conflicts of interest: C.J.E.W. has received research grants from Wyeth, Astellas, Novartis, Roche, and Organ Recovery Systems, and has received sponsorship to attend scientific meetings from Astellas, Novartis, Roche and Wyeth. G.J.A. has received financial support from Wyeth and Astellas (Fujisawa) for clinical research fellows. A.E.G. has received sponsorship from Roche. M.E.D.A. has previously received sponsorship from Roche to attend scientific meetings. P.G. has received sponsorship to attend scientific meetings from Astellas, Novartis, Roche, and Wyeth. J.C.S. has received sponsorship from Wyeth to attend scientific meetings. C.R.P. has no potential conflict of interest. J.A.B. has not received any personal sponsorship or consultancy fees but his department has received funding to support clinical research from Astellas, Novartis, Roche, and Wyeth. Wyeth Laboratories supplied the sirolimus tablets and sirolimus assay service; they played no role in data collection, analysis, or interpretation, nor in the writing of this report or the decision to submit for publication.
Calcineurin inhibitors (CNIs) have been the mainstay of immunosuppression in liver transplantation since their introduction over 2 decades ago.1 Although the use of cyclosporin, and more recently tacrolimus, has markedly improved liver transplant outcome both agents are potentially nephrotoxic. In one UK study 4% of patients surviving 12 months posttransplantation had severe chronic renal failure (defined as a creatinine level >250 μmol/L, 2.8 mg/dL).2 In a registry analysis of 36,849 liver transplant recipients in the United States, the incidence of chronic renal failure (the Modification of Diet in Renal Disease Study Group calculated glomerular filtration rate [GFR] ≤29 mL/minute/1.73 m2) was 18.1% at 5 yr.3
Not all of the renal impairment in liver recipients is due to CNI therapy, and many patients have preexisting abnormalities on renal biopsy.4 Nevertheless it is clear that compounding pretransplantation renal insufficiency with CNI toxicity is not conducive to long-term renal function.
To avoid long-term CNI nephrotoxicity, a number of strategies have been employed, including replacing CNI with mycophenolate5 or simply withdrawing CNI and continuing maintenance on prednisolone and azathioprine. A potential concern with CNI withdrawal is the risk of underimmunosuppression and consequent rejection, and in some reports this has led to graft loss.6
Sirolimus is a potent and nonnephrotoxic immunosuppressant currently licensed for use in renal transplantation. In this setting it has been shown to be associated with improved renal function when used without a CNI.7–9 However, when combined with cyclosporin, and by inference tacrolimus, nephrotoxicity is enhanced.10, 11 We have recently shown that replacement of CNI with sirolimus in kidney transplant recipients is associated with a sustained improvement in renal function.12 Consequently conversion from CNI to sirolimus might also be an attractive option for patients with renal impairment following liver transplantation.
We report here the results of an open-label single-center investigator-led study in which liver transplant recipients receiving CNI-based immunosuppression who had impaired renal function were randomized to switch to sirolimus-based immunosuppression or to remain on CNI-based therapy. Although the power of the study was limited, our findings suggest that switching to sirolimus results in a modest improvement in renal function as measured by radioisotope-labeled ethylenediaminetetraacetic acid-GFR.
The study was an investigator-led, open-label, single-center, randomized controlled trial designed to determine whether in liver transplant recipients with impaired renal function conversion from CNI-based to sirolimus-based maintenance immunosuppressive therapy results in improved kidney function.
The primary outcome measure was the change in GFR from baseline (time of randomization) to 12 months. Secondary end-points were graft and patient survival, dialysis requirement, therapy for hypertension, number of acute rejection episodes, incidence of hypercholesterolemia and hypertriglyceridemia, changes in serum creatinine and uric acid concentrations, systolic and diastolic blood pressure, and quality of life (using the Short Form 36 Quality of Life instrument).
Adult recipients of liver transplantations performed at our own center at least 6 months previously, and who were receiving CNI-based maintenance treatment, were invited to participate. The 6-month time point was chosen to minimize the risk of early acute rejection associated with changing immunosuppression. The principal inclusion criterion was suboptimal renal function, defined as an estimated GFR (Modification of Diet in Renal Disease Study formula13) of less than 65 mL/minute. Patients with known allergies to macrolide antibiotics were excluded, as were patients who experienced an episode of acute rejection episode within the preceding 3 months. Other exclusion criteria included patients already on dialysis, abnormal liver function tests (values greater than twice the upper limit of the normal range), subnormal leukocyte or platelet counts, uncontrolled hypertension and hyperuricemia. Entry of female patients was dependent on a negative pregnancy test and, where relevant, females were asked to use an effective method of contraception for the duration of the study. Patients with hepatitis C were participating in another trial of sirolimus and were not eligible.
Consenting patients were randomized to either continue CNI (tacrolimus or cyclosporin) therapy or to undergo an abrupt switch to sirolimus-based immunosuppression. Concomitant immunosuppression, such as prednisolone, azathioprine, or mycophenolate was continued unchanged. Patients randomized to switch to sirolimus received their last dose of CNI the evening before conversion, and started 2 mg per day sirolimus on the following day; there was no loading dose. The sirolimus dose was then adjusted according to 24-hour whole blood trough concentration on days 4, 7, and 14 following conversion. The target range for whole blood sirolimus concentrations was between 5 and 15 ng/mL (measured by high-performance liquid chromatography).
GFR was measured by 6-point 51Cr-ethylenediaminetetraacetic acid clearance on the day of randomization (baseline), and at 3 and 12 months. In addition, resting blood pressure measurement and full hematological and biochemical screening were undertaken at each of the 3 time points. The Short Form 36 Quality of Life instrument was administered at baseline and at 12 months.
Although the principal study end-point was change in GFR (delta GFR), in the absence of such data on which to make a power calculation when designing the study, change in serum creatinine was chosen as a surrogate. The serum creatinine trends of 125 adult patients receiving their first liver transplant were reviewed and the change in creatinine over a 12-month period was measured. Between 6 and 18 months posttransplantation 57% of patients experienced a rise in serum creatinine, while between 12 and 24 months 55% displayed a rise. We assumed that sirolimus was not nephrotoxic, and that the rise in creatinine was due to CNIs such that patients switching to sirolimus would not experience a rise in creatinine. To detect a significant difference in the proportion of patients with a rise in creatinine in each group at the 5% level with 90% power would require 75 patients to be randomized. We acknowledged that such an approach was less precise than calculations based on GFR itself, but reasoned that the error was likely to overestimate the number of subjects required to show a significant change in GFR. Nevertheless the lack of good data necessitated a planned interim analysis to validate our assumptions. When the single planned interim analysis suggested the magnitude of the beneficial effect on GFR was greater than expected, recruitment to the study was stopped.
An a priori randomization sequence was determined by random numbers generated by a Microsoft Excel Software program (Microsoft Corporation, Redmond, WA), such that numbers between 0 and 4 were allocated to switch to sirolimus, while numbers between 5 and 9 inclusive remained on their CNI. The allocations were then placed in sequentially numbered sealed envelopes by a member of the research team, but concealed from staff involved in the enrolment of the participants. The patients themselves opened the envelopes at the time of randomization. Neither patients nor clinicians were blinded as to therapy. The study had received approval by the Local Research Ethics Committee and all patients gave informed consent prior to participation. The study was conducted according to the guidelines set out in the Declaration of Helsinki.
Analysis was by intention to treat. Mean values for each group were compared by Student's t-tests for independent samples; 2-sided significance tests were used throughout. Fisher's exact test was used to compare the proportion of adverse events in each group.
Entry and participant flow through the study are shown in Figure 1. A total of 27 subjects provided the data for this study, with 13 randomized to switch to sirolimus and 14 to continue on their CNI-based immunosuppression. Three further subjects who had been randomized were excluded; 1 had renal function outside inclusion criteria on the initial ethylenediaminetetraacetic acid-GFR and 2 refused to remain on CNIs when allocated to that arm.
In addition, 58 patients had refused to participate, largely because it would necessitate them travelling long distances for additional visits to the regional transplant center. A further 63 patients were excluded for reasons other than those specified in the inclusion and exclusion criteria, usually related to comorbidity or an intention to convert to sirolimus outside the study.
The 2 study groups were comparable for age, gender, and baseline immunosuppression. The time from transplantation to randomization was very variable, ranging from 11 months to 12 yr, with a median of 3 yr in the sirolimus and 5 yr in the CNI group (Table 1). The 51Cr-ethylenediaminetetraacetic acid-GFR measured at randomization (baseline) was similar in the sirolimus and CNI groups (49.8 and 47.2 mL/minute/1.73 m2, respectively). There was no difference in any biochemical or hematological parameter measured at baseline (data not shown). In concordance with the CONSORT guidelines for reporting of clinical trials no statistical comparison of the baseline data is presented.
Table 1. Baseline Characteristics of Each Group
Sirolimus (n = 13)
CNI (n = 14)
Abbreviation: IQR, interquartile range.
Years posttransplantation; median (IQR)
Age (yr); median (IQR)
Concomitant immunosuppression at baseline
Creatinine (μmol/L); median (IQR)
GFR (mL/minute/1.73 m2); median (IQR)
Underlying liver disease at transplant
Primary biliary cirrhosis
Primary sclerosing cholangitis
Fulminant liver failure
Three patients, all randomized to remain on their CNI, were eliminated at baseline. One had a measured GFR of over 90 mL/minute/1.73 m2, in spite of a low GFR by the Modification of Diet in Renal Disease Study formula. A total of 2 requested conversion to sirolimus immediately and did not wish to remain on their CNI. Of the remainder, 2 sirolimus patients withdrew from the study at 3 and 6 months due to fatigue and lethargy, while 1 patient who remained on their CNI died from complications of pulmonary hypertension 11 months following randomization. No patients on sirolimus died, and no patient required dialysis during the follow-up.
There was a modest but highly significant difference in delta GFR (change in GFR from baseline, the primary endpoint) of 7.7 mL/minute/1.73 m2 at 3 months (P = 0.001; 95% confidence interval, 3.5-11.9) and 6.1 mL/minute/1.73 m2 at 12 months (P = 0.024; 95% confidence interval, 0.9-11.4). This was characterized by a mean improvement in GFR of 6.7 mL/minute/1.73 m2 in patients switching to sirolimus, compared to 0.6 mL/minute/1.73 m2 in the patients remaining on CNIs (Fig. 2a and b). Most of the improvement in delta GFR was observed by the 3-month time point, with little change thereafter (Figs. 2b and 3). In spite of the clear difference in delta GFR there was no significant difference in delta serum creatinine between groups. The difference in absolute GFR between the study groups was statistically significant at 3 months (P = 0.02), but not at 12 months (P = 0.07).
There was no significant difference in mean changes in 24-hour urinary protein excretion between groups (Table 2), with a mean change of 0.30 gm at both 3 and 12 months in sirolimus-treated patients and 0.01 and 0.29 gm in CNI-treated patients at the same time points (P = 0.21 and 0.99 for 3 and 12 month differences, respectively).
Table 2. Changes in Renal Function and Biochemistry at 12 Months Postconversion
Change from baseline
Mean (95% confidence interval) difference between groups
The majority of patients in both study arms were on tacrolimus monotherapy at the time of randomization. Those converted to sirolimus continued on monotherapy (10 of 14 patients). Following conversion to sirolimus the median daily dose of sirolimus at 3 months was 2 mg (range 0.5-4.0 mg), and at 12 months 2 mg (range 0.5-3.0 mg). Median whole blood sirolimus concentrations were 6.3 ng/mL (range 3-10.9 ng/mL) at 3 months, and 7.3 ng/mL (range 3.8-9.9 ng/mL) at 12 months. Patients remaining on tacrolimus had median 12-hour trough concentrations of 7.7 and 7.0 ng/mL at 3 and 12 months, respectively; only 2 were on cyclosporin, with concentrations 86 and 107 ng/mL at 3 months, and 147 ng/mL at 12 months (the second patient died).
Two patients experienced acute rejection episodes following conversion to sirolimus, while there were no rejection episodes in patients remaining on CNIs. One patient with primary sclerosing cirrhosis suffered a rejection episode 2 months after conversion. Her sirolimus trough concentration at the time was 2.8 ng/mL, and she was also on prednisolone and azathioprine. She responded to pulsed steroid treatment. The second patient stopped taking sirolimus at 3 months, misunderstanding that the statin that had been started was in place of sirolimus and not in addition to it. She also responded to pulsed steroid therapy.
There were no differences in the changes in hematological variables from baseline between groups. In particular, there was no significant difference in the changes in hemoglobin, leukocyte, or platelet count following conversion to sirolimus. However, it should be remembered that the study was not powered to look at changes in cell counts.
The significant changes in biochemistry parameters are illustrated in Table 2. There were significant changes in serum uric acid (−0.07 mmol/L; P = 0.019) and potassium (−0.5 mmol/L; P = 0.029) in the sirolimus-treated group at 3 months, but these were not seen at 12 months. There was a nonsignificant change in serum urea (P = 0.065 and 0.063 at 3 and 12 months, respectively). There was no significant difference in the changes in bilirubin, alkaline phosphatase, and alanine transaminase between groups.
There was a significant change in serum cholesterol at 3 months (mean difference 1.5 mmol/L; P = 0.002) in patients on sirolimus, which was sustained but no longer significant at 12 months. The change in total cholesterol was largely due to changes in the low-density lipoprotein fraction (0.8 and 0.7 mmol/L at 3 and 12 months, P <0.001 and P = 0.004, respectively), with no change in the high-density lipoproteins or triglycerides. In addition, 5 of 14 patients on sirolimus required introduction of statins to control their lipids, while none of the CNI group required new statin therapy.
There was no significant difference in the mean change in systolic or diastolic blood pressure, neither was there a difference in treatment for hypertension, with 3 of the sirolimus patients requiring reduction in antihypertensive therapy while 2 patients on sirolimus and 3 on CNIs required additional antihypertensive therapy.
Quality of Life
The Short Form 36 Quality of Life instrument was administered at baseline and 12 months later. There was no difference in any measured dimension between these time points in either sirolimus or control groups, although it should be remembered that the study was not powered to show differences in quality of life.
Adverse events are summarized in Table 3. There was a high incidence of rashes (9/13, 69%; P = 0.006) following conversion to sirolimus, most of which resolved on dose reduction. Likewise one-third of patients converted to sirolimus experienced aphthous-type mouth ulcers that resolved over the course of the first 2 weeks with dose adjustment to the lower end of the target range, although 3 patients developed coincidental herpes simplex. Two other sirolimus side effects are worthy of note. Three patients who switched to sirolimus developed leg edema and 1 developed tongue and facial edema, all of which were successfully managed by sirolimus dose reduction. Bone pain, said to be common in patients switching to sirolimus, was seen in a small number of patents in each arm (3 on sirolimus, 2 on CNI). The low incidence might reflect the lower target sirolimus levels used in this study.
Table 3. Principal Side Effects Following Randomization
The results of this single-center randomized controlled study demonstrate that, in recipients with impaired renal function 6 or more months following liver transplantation, conversion to sirolimus results in a modest but significant improvement in kidney function, as measured by the change in 51Cr-ethylenediaminetetraacetic acid GFR (delta GFR). The improvement in delta GFR after switching to sirolimus was evident by 3 months after conversion and maintained at 12 months. However, it is important to emphasize that while the absolute GFR was significantly different between the 2 study groups at 3 months, it just failed to reached significance at 1 yr. Patients who remained on CNI-based immunosuppression experienced no change in renal function.
During the design of this trial, initial statistical assessment suggested that twice the number of patients should be recruited, but because this prediction was based on changes in serum creatinine not GFR (the primary end-point), a planned interim analysis was built into the study design. The interim analysis confirmed that there was a highly significant difference in GFR between the groups at 3 months. Consequently further recruitment was suspended to allow further follow-up and when the difference in GFR was shown to persist to the 12-month time-point, the study was stopped. It is acknowledged that the reduced recruitment might compromise the interpretation of secondary end-points, and hence represents a potential weakness of the study. However, even a study of 75 patients would not be powered sufficiently to allow full evaluation of all secondary end-points, and the findings from our study strongly support the conduct of a large-scale randomized trial of conversion to sirolimus in liver transplantation.
To the best of our knowledge this is the first randomized controlled trial of switching from CNI-based to sirolimus-based immunosuppression after liver transplantation. The findings are in accord with other nonrandomized studies. In 1 prospective study, conversion of liver transplant recipients to sirolimus was associated with an improvement in estimated GFR of 9 mL/minute,14 which is similar to the improvement observed in the present study. Similar results have been reported from retrospective studies.15, 16 Other studies have suggested that using sirolimus to facilitate minimization of tacrolimus, rather than replacement, might also result in improved renal function,17, 18 although experience in renal transplantation suggests that the combination of CNI with sirolimus is not desirable since it might actually enhance nephrotoxicity.
In this study we chose to perform an abrupt switch to sirolimus, rather than to employ an overlap period with the CNI. No patient experienced acute rejection as an immediate consequence, although 1 suffered a rejection episode 2 months later in association with low levels of sirolimus, while a second patient suffered an acute rejection episode after mistakenly stopping sirolimus altogether. The abrupt switch has the advantage of being simple to institute, and it avoids the potential risks of overimmunosuppression and enhanced nephrotoxicity when both CNI and sirolimus are taken together during an overlap period. No loading dose of sirolimus was used in this study because our previous experience suggested that this was associated with increased side effects. Avoidance of a loading dose demands early assay of therapeutic drug levels to ensure adequate immunosuppression and to avoid drug toxicity. In this study sirolimus levels were measured at 4, 7, and 11 days following conversion.
Conversion to sirolimus was associated with development of both skin rashes and mouth ulcers, as well as biochemical changes including a rise in alanine transaminase, raised total and low-density lipoprotein cholesterol, and increased requirement for statin therapy. There was also a higher incidence of infection, particularly of herpes stomatitis. Whether this was an effect of the sirolimus per se, or opportunist infection following sirolimus induced aphthous ulceration is not clear. In general, the side effects of sirolimus, in particular rashes and mouth ulcers, responded to dose reduction.
In renal transplantation conversion to sirolimus has been reported to cause proteinuria19 but in the present study there was no significant change in urinary protein loss, and no patient developed nephrotic range proteinuria during the 1 yr follow-up.
One area that remains contentious is the optimal timing of conversion to sirolimus. We converted patients after a minimum of 6 months, and a median of 3 yr posttransplantation. Conversion before 6 months would seem to be appropriate for patients with evidence of renal impairment in the early posttransplantation period. In cases of early renal impairment, it is important to convert patients with CNI nephrotoxicity rather than acute tubular necrosis, since following renal transplantation sirolimus appears to delay recovery.20 In addition, wound healing can be a problem in patients on sirolimus, so where possible wounds should be well healed before conversion. The reports from 2 randomized studies of an increased incidence of hepatic artery thrombosis in patients treated with sirolimus from the time of transplant also argues against very early introduction of this agent,21, 22 although other reports from nonrandomized studies refute these observations.23, 24 The target concentrations used in our study for patients at least 6 months posttransplantation are lower than have been used for patients immediately after transplantation, unless sirolimus immunosuppression is combined with other agents such as azathioprine and steroids.
Although this study has demonstrated a clear advantage of switching to sirolimus, it is important to acknowledge the limitations of the study. Patients with hepatitis C infection were excluded from the study, although they now form a large proportion of patients requiring liver transplantation. The study is relatively small and based on a single center, so it will be important to confirm these findings in a larger multicenter randomized trial. Finally it is important to recognize that the benefit in renal function cannot necessarily be extrapolated to patients whose renal function is outside the inclusion criteria of the present study, and who are at such varying times posttransplantation (and hence with different durations of CNI exposure) as seen in this study, all things that can be more completely evaluated in a large multicenter trial.
In conclusion, the results of this randomized trial suggest that converting liver transplant recipients with impaired renal function from CNI-based to sirolimus-based immunosuppression may result in a modest improvement in delta-GFR.
We thank Kay Elston, the Nuclear Medicine Department at Addenbrooke's Hospital, and Addenbrooke's Centre for Clinical Investigation for their help during the conduct of the study. In addition, we thank Mike Allen and George Stanley, both of Wyeth, for their help in setting up the study, and their support during its course.
We acknowledge support from the Cambridge University Hospitals NHS Foundation Trust NIHR Biomedical Research Centre.