Randomized Controlled Trial of Tacrolimus Versus Microemulsified Cyclosporin (TMC) in Liver Transplantation: Poststudy Surveillance to 3 Years


*Corresponding author: John O'Grady, john.o'grady@kcl.ac.uk


The 1-year results of the tacrolimus versus microemulsified cyclosporin (TMC) study found a benefit with tacrolimus immunosuppression after primary liver transplants in adults with respect to freedom from graft loss and immunological failure. The integrity of the randomization process was preserved for a further 2 years for poststudy surveillance. The data after 3 years confirms the significant difference between tacrolimus and cyclosporin with tacrolimus less likely to meet the composite primary endpoint (log rank p = 0.01; relative risk 0.75; 95% CI 0.60–0.95; p = 0.016). However, freedom from death or retransplantation no longer achieves statistical significance (relative risk 0.79; 95% CI 0.62–1.02; p = 0.065). A total of 62.1% of patients randomized to tacrolimus were alive at 3 years with their original graft and still on their allocated study medication, as compared with only 41.6% in the cyclosporin limb (p < 0.001). No difference was detected between tacrolimus and cyclosporin in hepatitis-C-positive patients with the available data. The TMC study confirms after 3 years of follow-up the benefits of tacrolimus-based immunosuppression over cyclosporin using C0 monitoring.


The tacrolimus versus microemulsified cyclosporin (TMC) trial was designed to compare tacrolimus and microemulsified cyclosporin in a ‘head-to-head’ study of adults undergoing liver transplantation, with other aspects of care standardized in the two study groups (1). The primary endpoint was the combined frequency (whichever occurred first) of death, retransplantation or treatment failure for immunological reasons, analyzed by intention to treat. The latter component of the primary endpoint was included to reduce the distortion caused by successful rescue with the alternative calcineurin-inhibiting drug when intention-to-treat analysis was performed. The results at 1 year showed a statistically significant benefit with tacrolimus-based immunosuppression for the composite primary endpoint as well as for graft survival (1). In line with earlier studies, we decided to preserve the integrity of the randomization status and undertake a 2-year poststudy surveillance collecting salient data (2,3). We now report on the findings of the TMC study 3 years after randomization.

Materials and Methods

The study design and recruitment have been described in detail in the initial publication but some key characteristics are summarized below (1). The study population comprised adults undergoing their first liver transplant in any of the eight centers in the United Kingdom or Republic of Ireland. All etiologies of acute and chronic liver disease, including malignancy, were eligible. The primary endpoint was a composite of death, retransplantation and treatment failure for immunological reasons (defined as a change from protocol immunosuppression because of intractable rejection defined as not responding to two or more cycles of increased steroids or two or more cycles of any increased immunosuppressive therapy including antilymphocyte preparations or other investigational drugs established by liver biopsy). The main exclusion criteria were limited to multiorgan transplants, auxiliary grafts, incompatible donor blood group and patient refusal. Of particular note, patients with renal dysfunction or those requiring emergency transplantation were not excluded from the study. Patients were randomized to either tacrolimus or microemulsified cyclosporin. Concomitant medication in both arms included prednisolone and azathioprine. The target drug levels at the end of the first year were tacrolimus 5–15 ng/L and cyclosporin 150–300 ng/mL, but lower levels were tolerated in stable patients or those with evidence of drug toxicity. The protocol allowed weaning from prednisolone after 3 months. The switching of calcineurin inhibitors or the addition of mycophenolate mofetil or sirolimus was recorded.

The data pertinent to this report were collected as close as possible to 24 and 36 months after randomization. The data were not as extensive as during the first year of the study but included physical examination, immunosuppressive medication and standard laboratory parameters. Additional data were collected for any significant clinical event occurring at other timepoints during the surveillance period and were comprehensive for changes in study medication, rejection episodes and diagnoses of malignancy.

Data collection up to 3 years continued with resident research nurses. The trial co-ordination, data management and analysis were carried out at the Medical Statistics Unit of the London School of Hygiene and Tropical Medicine. Double entry and validation was performed on all reported primary and secondary outcomes. The composite primary outcome was defined as the time to death, retransplantation or immunological failure, whichever occurred first.

Local Research Ethics committee approval was granted for all the clinical centres.

Statistical analysis

Sample size calculation indicated that 596 patients (298 in each group) would be needed to detect a 10% difference in the composite primary endpoint at 1 year from 70–80% (tacrolimus) with 80% power and a two-tail alpha error of 5%. The primary analysis was by intention to treat, comparing the time to reach death, retransplantation or treatment failure for immunological reasons using Kaplan-Meier estimates and comparing differences using the log rank test. Other comparisons between treatment groups are presented as estimates of effect sizes (relative risks of difference of means) with 95% confidence intervals (CIs) and chi-squared, Fisher's exact, Mann-Whitney U or t-tests, where appropriate.


During the recruitment period from May 1997 to April 1999, 606 patients were randomized (301 tacrolimus, 305 microemulsified cyclosporin). The characteristics of the randomized groups were similar, as they were for those eligible for follow-up to 36 months (250 patients on tacrolimus and 233 patients on cyclosporin) (Table 1). The outcome data for the primary endpoint are presented in Table 2 and Figure 1. By 3 years, patients immunosuppressed with tacrolimus were less likely to meet the composite primary endpoint (log rank p = 0.01; relative risk 0.75; 95%, CI [0.60 to 0.95] chi-squared p = 0.016). All patients meeting the primary endpoint between 12 and 36 months were deaths or retransplants as there were no further episodes of failure for immunological reasons. Unlike the 1-year data, the combined endpoint of death or retransplantation rate was not statistically different between the two groups with a relative risk of 0.79 (95% CI 0.62–1.02; p = 0.065). As previously, the result was similar for analyses stratified by center, elective versus emergency transplantation and by high (>85% of eligible patients) or low recruiting centres, as well as per protocol analysis.

Table 1.  Characteristics at randomization for patients alive at 12 months (number, [%] unless otherwise stated).
 Eligible for 3-yr follow-up
Tacrolimus n = 250 (%)Microemulsified cyclosporin n = 233 (%)
  1. 1Mixed etiology includes patients with two or more of hepatitis B (including delta), hepatitis C or alcoholic cirrhosis diagnoses.

Elective transplants220 (88)205 (82)
Primary diagnosis
 Chronic liver disease—cirrhosis130 (52)115 (49)
  Hepatitis B (delta)10 (4)7 (3)
  Hepatitis C27 (11)23 (10)
  Alcoholic51 (20)45 (19)
  Mixed etiology15 (1)6 (3)
  Cryptogenic11 (4)12 (5)
  Other26 (10)22 (9)
 Primary biliary cirrhosis29 (12)43 (18)
 Primary sclerosing cholangitis26 (10)19 (8)
 Malignancy26 (10)27 (12)
 Acute liver failure31 (12)27 (12)
 Miscellaneous8 (3)2
Males151 (60)131 (56)
Age (y) median {IQR}52 {42–58}50 {40–57}
Serum creatinine (μmol/L) med {IQR}87 {71–109}84 {71–103}
Gender match (donor–patient)
 Male–male98 (39)80 (34)
 Female–female50 (20)70 (30)
 Female–male52 (21)50 (21)
 Male–female49 (20)32 (14)
 Not applicable (not transplanted)01
Age of donor liver (years) median {IQR}39 {27–52}44 {31–52}
Cold ischemic time of liver (h) mean [sd]11.6 [3.6]11.4 [3.2]
Blood transfused during operation (units) med {IQR}6 {3–10}6 {3–10}
Table 2.  Primary outcomes from randomization to 36 months
 Tacrolimus (n = 301)Microemulsified cyclosporin (n = 305)
  1. 1A change from protocol immunosuppression because of intractable rejection is defined as not responding to two or more cycles of increased steroids or two or more cycles of any increased immunosuppressive therapy including antilymphocyte preparations or other investigational drugs established by liver biopsy.

  2. Composite outcome (death or retransplant or treatment failure). Relative risk (RR) 0.75 95% CI (0.60, 0.95) p = 0.016.

  3. Graft failure (death or retransplant) RR 0.79 95% CI (0.62, 1.02) p = 0.065.

Treatment failures for immunological reasons1612
Death or retransplantation80102
Death or retransplantation or treatment failure for immunological reasons184113
Figure 1.

Survival plot showing freedom from primary endpoint at 3 years (combined treatment failure, retransplantation or death). Log-rank test for equality of survivor functions: chi-squared (1 df) = 5.73, p = 0.01.

During the poststudy surveillance period there were 20 new deaths in the tacrolimus limb (8% of those at risk) and 8 new deaths in the cyclosporin limb (3.5% of those at risk), resulting in the overall number of deaths being 71 and 80, respectively. An apparent discrepancy between these and the original data is explained by the fact that 1 patient classified as lost to follow-up in the initial report is now known to have died within the first 12 months of the study. The causes of the new deaths are outlined in Table 3. The figures for retransplants during the same period were 3 in the tacrolimus limb and 6 in the cyclosporin limb resulting in the overall number of retransplants being 14 and 37, respectively. The indications for the new retransplants are also outlined in Table 3. Over the duration of the study, 38 patients developed malignant disease with 21 in the tacrolimus group and 17 in the cyclosporin group. Of these, 23 were de novo tumors (8 skin, 3 colon, 3 metastatic from unidentified primaries, 2 melanoma, 2 testis, 1 lung, 1 oropharynx, 1 kidney and 1 desmoid tumor), 7 were de novo lymphoma and 8 were recurrence of hepatocellular carcinoma. A total of 21 were diagnosed after the first year with the 11 in the tacrolimus group including 3 de novo lymphomas, 8 de novo other malignancies and no recurrent tumours; the comparable figures in the cyclosporin group were 1, 8 and 1. The laboratory data relating to liver and renal function showed higher bilirubin and alinine aminotransferase levels in the microemulsified cyclosporin group, but no other statistically significant differences between the 2 groups at 3 years of follow-up (Table 4).

Table 3.  Primary cause of death and reason for retransplantation between 12 and 36 months (numbers)
Primary cause of deathTacrolimus n = 20Microemulsified cyclosporin n = 8
Sepsis/multiorgan failure32
De novo malignancy42
Recurrent malignancy20
Recurrent hepatitis B11
Recurrent hepatitis C21
Sudden death10
Primary reason for retransplantationn = 3n = 6
Hepatic artery thrombosis23
Chronic rejection01
Recurrent viral disease11
Recurrent PBC01
Table 4.  Laboratory parameters 36 months after randomization (medians and interquartile ranges)
 Tacrolimus n = 230Microemulsified cyclosporin n = 225Mann-Whitney U test
Serum bilirubin (μmol/L)11 (7–14) (n = 215)13 (9–18) (n = 207)p < 0.001
Aspartate aminotransferase (IU/L)24 (19–38) (n = 123)27 (22–40) (n = 120)p = 0.097
Alinine aminotransferase (IU/L)26 (17–38) (n = 159)39 (18–64) (n = 144)p = 0.002
Alkaline phosphatase (IU/L)120 (77–193) (n = 214)124 (85–225) (n = 206)p = 0.345
Gamma glutamyltransferase (IU/L)57 (24–105) (n = 86)54 (21–189) (n = 97)p = 0.296
Serum creatinine (μmol/L)109 (94–123) (n = 215)111 (94–130) (n = 207)p = 0.448

Significantly more patients remained on tacrolimus (n = 199) than cyclosporin (n = 133) (p < 0.001). A total of 187 (62.1%) patients randomized to tacrolimus were alive at 3 years with their original graft and still on their allocated study medication, as compared with 127 (41.6%) in the cyclosporin limb (p < 0.001). Only a minority of patients in both groups (tacrolimus 44, cyclosporin 47) remained on corticosteroids at 3 years. Numerically fewer patients in the tacrolimus than the cyclosporin limb remained on azathioprine (88 vs. 102), but the difference was not statistically significant. Mycophenolate mofetil had been introduced in 21 patients on tacrolimus and 24 patients on cyclosporin, while sirolimus was introduced in only 3 patients, all of whom were on cyclosporin.

Six patients were switched from tacrolimus to cyclosporin during the poststudy surveillance period; 3 for perceived toxicity (myopathy, depression, restrictive cardiomyopathy), 2 for rejection not meeting the criteria for inclusion in the composite endpoint and 1 because of difficulty getting supplies of tacrolimus in the country of residence. During the same period, 17 patients were switched from cyclosporin to tacrolimus. Perceived toxicity was the reason in 9 patients—gingival hyperplasia 3, renal dysfunction 3, seizures 1, hypertension 1 and 1 was not specified. Rejection or graft dysfunction not meeting the criteria for inclusion in the composite endpoint was the reason in 5 patients. One patient was switched because of noncompliance with cyclosporin therapy and the reason was not specified in one patient. The final patient was switched to tacrolimus prior to the initiation of interferon-based therapy for recurrent hepatitis C infection.

Among the hepatitis C virus-related cirrhosis group, 46 were randomized to the tacrolimus group and 58 in the microemulsified cyclosporin group. Death, retransplantation or treatment failure occurred in 11 tacrolimus (24%) versus 20 cyclosporin patients (34.5%). Retransplantation occurred in 4 patients on tacrolimus (9%) and 10 patients in the cyclosporin group (17%). There were 9 deaths at 3 years in the tacrolimus group (20%) versus 15 (26%) in the cyclosporin group. However, death was attributed directly to hepatitis C recurrence in only 2 patients in each group and only 1 patient in each group underwent retransplantation for recurrent hepatitis C. These events bring the overall number of deaths or grafts lost postrandomization to recurrent hepatitis C to 5. Freedom from the study endpoint was numerically higher in hepatitis C patients than in their hepatitis-C-negative counterparts, but these differences did not reach statistical significance (tacrolimus 76% vs. 71%, cyclosporin 66% vs. 62%; p > 0.05).


The TMC poststudy surveillance confirms the benefit with respect to freedom from the primary endpoint in patients who are immunosuppressed with tacrolimus up to 3 years after primary liver transplantation in adults. The endpoints were set as ones considered to be most relevant in assessing immunosuppression, a design feature that in editorial comment was described as a new standard in the field of immunosuppression trials in liver transplantation (4). As might be expected, immunological failure was not encountered beyond 1 year and all events contributing to the study endpoint during poststudy surveillance were deaths or retransplants. There was a degree of convergence of the graft survival curves such that a significant difference detected at 12 months was just lost at 36 months. Analysis of the causes of death or retransplantation did not identify a specific pattern accounting for this change.

The number of patients who are still alive with their primary graft and on their original immunosuppression is a good overall clinical marker of performance and, in this respect, tacrolimus was again superior to cyclosporin with 62.1% of patients meeting these criteria as compared with only 41.6% of patients allocated to cyclosporin therapy. The equivalent figures at 12 months were 76% for tacrolimus and 56% for cyclosporin. This endpoint could potentially be subject to bias, particularly in the context of the favorable findings with tacrolimus after 12 months. However, an analysis of the reasons for switching calcineurin inhibitors, which is one assessable component of potential bias, did not identify an obvious behavioral pattern supporting this possibility.

Calcineurin inhibitors remain pivotal to primary maintenance immunosuppressive therapy but there have been a number of changes in practice since the design of this study. One significant change was the adoption of C2 monitoring for cyclosporin and this was compared with tacrolimus in a study of 495 patients followed for 6 months after liver transplantation in the LIS2T study (5). The key performance indicators for the two drugs were comparable. However, the trials are difficult to compare meaningfully as the primary endpoint in the LIS2T study was acute cellular rejection within the first 3 months after transplantation and this is of considerably less clinical significance than the endpoint in the TMC study. The durations of follow-up reported to date are also obviously different. In addition, a critical difference was the exclusion of patients with renal dysfunction and patients with acute liver failure with hemodynamic instability from the LIS2T study which potentially biased the study outcome toward neutrality. Another change since the design of the TMC study has been that azathioprine is now rarely used in primary immunosuppression in liver transplantation and would not be included in a modern trial design. The issue of whether or not it should be substituted with a newer agent, such as mycophenolate mofetil or sirolimus, cannot be answered by this trial although it is of interest that by 3 years after transplantation the perceived need for either of these drugs was less than 10%.

There is currently a great level of interest in possible differences between tacrolimus and cyclosporin with respect to hepatitis C recurrence. To date, most commentators have considered the effect of the individual calcineurin inhibitors to be neutral with regard to the impact on recurrence of hepatitis C (6,7). However, recent in vitro data suggest that cyclosporin may have some intrinsic antiviral properties that could be beneficial in patients with hepatitis C infection (8). The study design used in the TMC study ought to give an opportunity to determine if this theoretical benefit translates to clinical practice. However, the data to date are not particularly helpful in this regard as there were only five grafts lost to hepatitis C recurrence during the first 3 years. This incidence of graft loss may be less than might be expected and, if so, a possible contributing factor may be the fact that antilymphocyte preparations were not used in this study. In addition, routine histological evaluation of the severity of hepatitis C recurrence was not incorporated in the study design but this is now being addressed in a separate extension to the study.

The data collected during the poststudy surveillance was not as comprehensive as during the first 12 months of the study. In particular, detailed data on rejection episodes, drug levels and toxicity that did not lead to a change in the immunosuppressive regimen were not collected. Nevertheless, the differences found between tacrolimus and cyclosporin in this extension of the TMC trial appear robust in both statistical and clinical terms and confirm the superiority of tacrolimus-based immunosuppression in adult recipients of primary liver grafts.


Members of the UK and the Republic of Ireland Liver Transplant Study Group

Alex Gimson, Neville Jamieson (Addenbrooke's Hospital, Cambridge)

Mark Hudson, Michael Thick (Freeman Hospital, Newcastle)

John O'Grady, Nigel Heaton (King's College Hospital, London)

David Mayer, James Neuberger (Queen Elizabeth Hospital, Birmingham)

Alistair Macgilchrist, John Forsythe (Royal Infirmary, Edinburgh)

Andrew K Burroughs, Keith Rolles (Royal Free Hospital, London)

Mervyn Davies, Stephen Pollard (St. James's Hospital, Leeds)

Aiden McCormick, Oscar Traynor (St. Vincent's Hospital, Dublin)

Diana Elbourne, Pollyanna Hardy, Ann Truesdale (Data co-ordinating Centre; Medical Statistics Unit, London School of Hygiene and Tropical Medicine)


We would like to thank all the patients who participated in the follow-up study; the research nurses from all the clinical centres; and Andy King, Steven Robertson and Keith Tomlin at the LSHTM. Both Astellas (formerly Fujisawa) and Novartis extended financial support in equal amounts for the poststudy surveillance. Full disclosures of conflict of interest have been published elsewhere (1).