A Comparison of Tacrolimus and Cyclosporine in Liver Transplantation: Effects on Renal Function and Cardiovascular Risk Status


*Corresponding author: Michael R. Lucey, mrl@medicine.wisc.edu


A retrospective chart review of 1065 consecutive liver allograft recipients in 11 centers from January 1997 to September 1998 was performed. Patients were followed for 3 years or until graft loss. Patients received either tacrolimus (n = 594), cyclosporine (n = 450) or no calcineurin inhibitor (n = 21). Model for end-stage liver disease (MELD) scores at time of transplant were similar between the two groups. During follow-up, more patients switched from cyclosporine to tacrolimus (26.7%) than from tacrolimus to cyclosporine (12.8%; p < 0.0001). Patient and graft survival were equivalent. Corticosteroid use was more common in cyclosporine-treated patients (p < 0.00001). Patients receiving tacrolimus experienced lower serum creatinine levels at months 3 through 36 (p < 0.0001). Systolic blood pressure was lower in patients receiving tacrolimus (p < 0.001) despite a reduced requirement for anti-hypertensive agents (p < 0.0001). In addition, tacrolimus was associated with lower total cholesterol and triglyceride levels for months 3 through 24 and 3 through 12, respectively (p < 0.01), despite a reduced requirement for anti-hyperlipidemic agents. The incidence of new-onset diabetes mellitus was similar in both groups. While both calcineurin inhibitors were associated with excellent patient and graft survival, renal function, blood pressure and serum lipid levels were significantly better with tacrolimus treatment.


Immunosuppression based on calcineurin inhibition with either cyclosporine or tacrolimus is standard of care in almost all primary liver allograft recipients in the United States (1). Renal dysfunction and cardiovascular disease are significant causes of morbidity and mortality in long-term liver allograft recipients (2,3). There is a 3-fold increase in relative risk of ischemic cardiac events in liver transplant recipients compared to age- and sex-matched populations, while the increase in relative risk for cardiac death in this population is 2.56 (3). Hyperlipidemia, hypertension and diabetes mellitus contribute to an increased risk for cardiovascular disease in liver allograft recipients (4–7).

Although both cyclosporine and tacrolimus appear very efficacious in promoting graft survival, differential effects of these two medications on markers of renal and cardiovascular health have been reported. Several studies suggest a greater prevalence of hypertension, renal failure and/or hyperlipidemia in liver allograft recipients receiving cyclosporine compared to those treated with tacrolimus (2,5–8). Conversely, new-onset diabetes mellitus has been reported to occur more commonly in liver recipients receiving tacrolimus rather than cyclosporine (9,10). Consequently, we undertook a multicenter study of renal and cardiovascular toxicity profiles, including new-onset diabetes mellitus, in a large number of liver allograft recipients in routine clinical practice.

Materials and Methods

We conducted a retrospective chart review of consecutive liver allograft recipients at 11 U.S. transplant centers. Pediatric and adult recipients of primary-deceased, split-deceased or living-donor livers from January 1997 to September 1998 were included. The exclusionary criteria were: recipients of multiple organs, accessory liver transplants defined as an allogeneic supplement to the native liver and patients lost to follow-up prior to achieving an endpoint (graft loss or death). Primary outcomes included kidney function and cardiovascular risk profile over 3 years following transplantation. Specifically, the cardiovascular risk parameters examined included the following: systolic and diastolic blood pressure, calculated mean arterial pressure, use of anti-hypertensive agents, total serum cholesterol, LDL cholesterol, triglycerides, HDL cholesterol and the use of anti-hyperlipidemic agents. Data were collected for the first 3 years after transplantation or until graft loss (death due to graft failure). Data were analyzed according to the initial calcineurin inhibitor selected: tacrolimus or cyclosporine.

There were 1065 cases identified that met inclusion criteria. Data were recorded at 3-month interval in the first year after transplant and then at 6-month interval through 36 months. Data entry was validated via a multiple iteration process. Gaps in data collection were identified for three of the participating centers. Data for the cases that were missing in the three centers provided as much information as possible, with the following remaining as outstanding and unverified: switch information for primary immunosuppressive therapy (8 patients), oral corticosteroid dose (2 patients), unconfirmed patient weights (2 patients), unconfirmed triglyceride levels (1 patient) and two records not entered due to loss of follow-up.

This study was conducted in accordance with the Institutional Review Boards at the individual transplant centers.


Electronic data capture and result generation utilized Synapse AXON technology. Data were analyzed using the SAS System (SAS Institute Inc., Cary, NC) for Windows Release 8.02. Kaplan-Meier estimates were calculated for patient and graft survival and retransplantation rates and compared using the Wilcoxon rank-sum test. Acute rejection data were compared using the log-rank test (11); switch of primary immunosuppressive therapy was compared with the chi square test. Causes of graft loss and death, pre-transplant diagnoses and percentage of patients receiving corticosteroids, anti-hypertensive and anti-hyperlipidemic agents and insulin were compared using Fisher's exact test. The Mann-Whitney U-test was used to compare serum creatinine, blood pressure and serum lipid measurements. Statistical tests used for comparison of patient demographics and characteristics are specified. Bonferroni correction was applied to repeated measures. Results are expressed as mean ± SD, except where indicated as median and range. p-values of less than 0.05 were considered statistically significant.


Patient disposition and demographics

A total of 1065 patient records were entered; data were complete for 1011, while verification was required for 54 patients. Patient characteristics were comparable between groups with a few exceptions—primary non-function and median cold ischemia time were both greater in the cyclosporine group (p = 0.03 and p < 0.0001, respectively), while positive T-cell cross-match was greater in the tacrolimus group (p = 0.01; Table 1). In addition, a significant difference in UNOS status was evident between groups (p = 0.04), with a higher proportion of Status 1 patients in the tacrolimus group (8.3%) compared to the cyclosporine group (5.6%; Table 1).

Table 1.  Baseline patient demographics and clinical characteristics
(n = 594)
(n = 450)

  1. CMV = cytomegalovirus; D = donor; R = recipient; HCV = hepatitis C virus; PBC = primary biliary cirrhosis; PSC = primary sclerosing cholangitis.

  2. *Chi-square probability; Mann-Whitney U-test; Fisher's exact test; §Cochran-Armitage trend test.

Gender–male (%)*
Median age in years (range)50 (0–78)48 (0–74)0.69
Race–African American (%) 4.9 6.50.26*
CMV: D(+) R(−) (%)18.321.60.20*
Median cold ischemia time in hours (range) 7.9 (1.4–6.8) 9.0 (1.87–0.00)<0.0001
Positive T-cell cross-match (%)14.7 5.60.01*
Type of transplant (%)
 Living donor 5.4 4.0 
 Split cadaveric 2.9 1.8 
Median MELD score (range)16.8 (6.4–46.7)16.2 (6.4–54.3)0.40
UNOS status (%)
 1 8.3 5.70.04§
 4 0.2 0.7 
Child-Turcotte-Pugh class (%):
 A: 5–6 3.2 7.10.42§
 B: 7–935.233.7 
 C: ≥1061.759.2 
Non-cholestatic cirrhosis55.961.30.09
HCV (%)35.742.40.03
Cholestatic cirrhosis (PBC or PSC) (%)15.814.20.50
Cryptogenic cirrhosis (%)10.1 9.60.83
Malignancy (%) 4.2 2.00.053
Other/unknown (%)21.916.7 

The proportion of patients presenting with single and multiple diagnoses was not different between the groups, with overall rates of approximately 83% and 17%, respectively (data not shown). Pre-transplant diagnoses were similar between the groups, with the exception of the prevalence of hepatitis C virus (HCV) infection, which was higher in the cyclosporine group (p = 0.03; Table 1).

Maintenance immunosuppression

Patients received either tacrolimus (n = 594), cyclosporine (n = 450) or no calcineurin inhibitor (CNI) (n = 21) as initial immunosuppressive therapy. The proportion of patients receiving tacrolimus or cyclosporine over the 3-year period is shown in Figure 1. Mean daily doses of calcineurin inhibitors and corresponding trough blood concentrations were appropriate, given 5–15 ng/mL for the first 3 months and 5–12 ng/mL thereafter for tacrolimus, and 250–400 ng/mL for the first 6 months and 100–250 ng/mL thereafter for cyclosporine (data not shown).

Figure 1.

Percentage of patients receiving tacrolimus, cyclosporine or no calcineurin inhibitor for maintenance immunosuppression following liver transplant.

Follow-up data on maintenance immunosuppression were not available beyond day 0 for 22 patients in the tacrolimus group and 20 in the cyclosporine group. Of patients switching from one primary immunosuppressant to the other (tacrolimus to cyclosporine or cyclosporine to tacrolimus), 70 patients (12.8%) switched from tacrolimus to cyclosporine, while 111 (26.7%) patients switched from cyclosporine to tacrolimus (p < 0.0001). Fifty-eight (10.6%) patients in the tacrolimus group and 81 (19.5%) patients in the cyclosporine group were switched during the first year following transplant.

Corticosteroid use

The use of corticosteroids was highly prevalent in both cohorts at 1 month following transplantation (88.6%). The percentage of patients receiving corticosteroids was not different between the tacrolimus and cyclosporine groups at this time point. The use of corticosteroids decreased over the 3-year follow-up period in both the tacrolimus and cyclosporine cohorts. However, at months 12, 24 and 36 following transplantation, significantly more patients in the cyclosporine group were receiving corticosteroids (p < 0.00001; Figure 2).

Figure 2.

Corticosteroid use overall and in patients receiving either tacrolimus or cyclosporine from months 1 to 36 following transplant. Significantly fewer patients treated with tacrolimus were on corticosteroids at 12, 24 and 36 months follow-up.

The median daily dose of corticosteroids was 18.75 mg (range: 0–150 mg) at 1 month and decreased to 2.5 mg (range: 0–30 mg) by 1 year (Figure 2). The median daily dose dropped to 0 mg for both 2 years (range: 0–40 mg) and 3 years (range: 0–60 mg) following transplantation. Patients on tacrolimus had significantly lower median daily oral doses of corticosteroids at both 1 month and 1 year after transplantation (p < 0.00001 and p < 0.01, respectively).

Patient/graft survival and retransplantation

No significant difference between treatment groups was observed for either patient or graft survival (Figure 3). Patient survival rates, censored for graft loss, at 3 years after transplantation were 88.9% in the tacrolimus group and 87.8% in the cyclosporine group (p = 0.62). The total number of deaths observed over the 3-year period was 134. Graft survival rates 3 years following transplantation were 90.7% in the tacrolimus group and 90.4% in the cyclosporine group (p = 0.90). A total of 106 grafts were lost over the 3-year period. The unadjusted graft survival rate including both death and graft loss was 79.0%.

Figure 3.

Kaplan-Meier estimates of 3-year patient (A) and graft survival (B) in the tacrolimus and cyclosporine treatment groups.

Treatment failures, defined as either death or graft loss, were not significantly different over the 3-year period when patients were stratified according to HCV status (data not shown). A failure rate of 20.3% was observed for HCV-negative patients and 26.2% for HCV-positive patients (p = 0.57). Treatment failure rates were not different between the treatment groups among HCV-negative patients, with 17.8% observed in the tacrolimus group and 19.3% in the cyclosporine group (p = 0.68). Similarly, no significant differences in treatment failure were observed between the groups among HCV-positive patients, with 25.0% for tacrolimus and 25.1% for cyclosporine (p = 0.88).

Primary causes of death were similar between treatment groups. The primary causes of death and overall frequency are depicted in Table 2. Similarly, no cause of graft loss was observed more frequently in either treatment group. The causes of graft loss and overall frequency are shown in Table 2.

Table 2.  Primary causes of death and graft loss*
n = 594 (%)
n = 450 (%)
  1. *No cause of death or graft loss was observed more frequently in either treatment group.

  2. CMV = cytomegalovirus.

Primary cause of death
 Infection18 (3.0)22 (4.9)
 Other20 (3.4)8 (1.8)
 Malignancy11 (1.9)9 (2.0)
 Other cardiovascular event6 (1.0)8 (1.8)
 Liver allograft failure4 (0.7)3 (0.7)
 Acute myocardial infarction5 (0.9)2 (0.4)
 Recurrence of underlying disease2 (0.3)3 (0.7)
Primary cause of graft loss
 Hepatic artery thrombosis16 (2.7)11 (2.4)
 Primary non-function16 (2.7)7 (1.6)
 Other9 (1.5)11 (2.4)
 Recurrence of underlying disease9 (1.5)8 (1.8)
 Infection (including CMV)6 (1.0)6 (1.3)
 Ductopenic/chronic rejection4 (0.7)3 (0.7)
 Acute rejection5 (0.3)2 (0.4)
 Non-compliance1 (0.2)1 (0.2)
 Surgical0 (0)0 (0)

The incidence of retransplantation over the 3-year period was low and comparable between the treatment groups. Seven percent of patients in the tacrolimus group and 6% in the cyclosporine group underwent retransplantation over the 36-month follow-up period (p = 0.49).

Acute rejection

The cumulative expected number of any acute rejection event was not significantly different between the treatment groups for the first year (0.621 for tacrolimus vs. 0.733 for cyclosporine; p = 0.06) or over the 3-year period following transplantation (0.763 for tacrolimus vs. 0.880 for cyclosporine; p = 0.07; Figure 4A). In contrast, tacrolimus was associated with a significant reduction in the incidence of acute rejection episodes requiring anti-lymphocyte antibody therapy. This reduction was apparent during both the first year (0.044 for tacrolimus vs. 0.080 for cyclosporine; p = 0.02) and over the 3-year period after transplant (0.051 for tacrolimus vs. 0.093 for cyclosporine; p = 0.01; Figure 4B).

Figure 4.

Kaplan-Meier estimates of any acute rejection episode (A) and acute rejection requiring anti-lymphocyte antibody treatment (B) by treatment group over 0–3 years.

Renal function

Mean baseline serum creatinine levels were equivalent between the tacrolimus and cyclosporine treatment groups (1.31 mg/dL and 1.20 mg/dL, respectively; p = 0.35). After transplantation, serum creatinine levels were significantly lower in patients receiving tacrolimus at all time points over the 3-year follow-up period (p < 0.0001 at all time points; Figure 5). In addition, the change in mean serum creatinine levels over months 3 through 36 post-transplant was significantly lower in patients treated with tacrolimus. The increase in mean serum creatinine from month 3 to 36 in patients receiving tacrolimus was 0.087 mg/dL as compared to 0.193 mg/dL for patients receiving cyclosporine (p = 0.003).

Figure 5.

Renal function months 3–36 following transplant in patients receiving tacrolimus or cyclosporine. Tacrolimus treatment was associated with significantly better mean serum creatinine levels at all time points following transplant (p < 0.0001).

Cardiovascular risk profile

Tacrolimus was associated with a markedly better cardiovascular risk profile. Patients in the tacrolimus cohort had lower mean systolic blood pressure measurements at months 3 through 36 post-transplant (p < 0.001 at all time points; Figure 6). Similarly, diastolic blood pressure measurements were significantly lower in patients receiving tacrolimus at months 3 through 24 after transplant (p < 0.01 at all time points; Figure 6). Concomitant with better blood pressure measurements seen in patients receiving tacrolimus, there was a decreased requirement for anti-hypertensive agents through the 3 years of follow-up (p < 0.0001 at all time points; Figure 7).

Figure 6.

Comparison of the mean systolic (A) and diastolic (B) blood pressure measurements in patients receiving tacrolimus and cyclosporine.

Figure 7.

Comparison of the proportion of patients in the tacrolimus and cyclosporine treatment groups requiring anti-hypertensive agents, months 3 through 36 following transplant.

One hundred and sixty-nine patients were diagnosed as having diabetes pre-transplant (on insulin or oral medications at the time of transplant), while 14 patients did not have a pre-transplant diabetes status reported. The remaining 882 patients (tacrolimus: 488, cyclosporine: 378) were considered at risk for the development of new-onset diabetes mellitus. New-onset requirement for anti-diabetic medications (oral agents and/or insulin) following transplantation is shown in Table 3. The incidence of new-onset, insulin-requiring diabetes mellitus was comparable between treatment groups through the 3 years of follow-up (Table 3).

Table 3.  Incidence of new-onset requirement for anti-diabetic medications by treatment group
Time after transplantTacrolimus (n = 488)Cyclosporine (n = 378)p-values
Oral agent (%)Insulin (%)Oral agent (%)Insulin (%)p*p
  1. *p-value for use of insulin in tacrolimus versus cyclosporine group.

  2. p-value for combined use of insulin and an oral agent in tacrolimus versus cyclosporine group.

  3. For each interval, patients who had a reversal in status (reported to be on an anti-diabetic agent and later reported as no longer on an anti-diabetic agent within the same interval) were excluded.

1–12 months2.6619.232.9614.070.100.15
>12–24 months4.9114.424.20
>24–36 months5.4113.694.22 8.860.080.06

Mean total cholesterol levels were lower in patients receiving tacrolimus at months 3 through 24 post-transplant compared to patients receiving cyclosporine (p < 0.01 at all time points; Figure 8). Mean serum triglycerides were also significantly lower in patients receiving tacrolimus for months 3 through 12 after transplant. Triglyceride measurements at 3 months were 169.64 mg/dL for tacrolimus- versus 231.03 mg/dL for cyclosporine-treated patients, at 6 months were 167.92 mg/dL for tacrolimus- versus 204.28 mg/dL for cyclosporine-treated patients and at 12 months were 169.57 mg/dL for tacrolimus- versus 216.84 mg/dL for cyclosporine-treated patients (p < 0.01 at all time points). The better serum lipid profile was observed in the tacrolimus cohort in spite of a reduced requirement for anti-hyperlipidemic agents (p < 0.0001 at 2 and 3 years; Figure 8).

Figure 8.

Mean serum cholesterol levels in the tacrolimus and cyclosporine treatment groups and the proportion of patients requiring anti-hyperlipidemic agents, months 3 through 36 following transplant.


The present study, which demonstrated excellent patient and graft survival for up to 3 years among liver transplant recipients treated with tacrolimus- and cyclosporine-based immunosuppressive protocols in routine use in 1997–1998, complements similar observations in prior multicenter, prospectively randomized clinical trials (12–14). Analyses by treatment group, in the present study, showed that freedom from retransplantation was comparable at 3 years for patients receiving tacrolimus (93%) or cyclosporine (94%). The rates of patient and graft survival observed in the present study were approximately 10% higher, at 88% and 90%, respectively, than those associated with tacrolimus and cyclosporine in national and international databases and studies (13–17). It is possible that the higher rates of patient and graft survival observed herein may reflect our reliance on liver transplant centers performing more than 60 transplants per annum in this retrospective analysis, although the impact of center size on outcome after liver transplantation remains controversial (18,19).

We observed no difference in treatment failure, defined as death or graft loss, over the 3-year period according to HCV status. In the same way, no differences were seen between the treatment groups for HCV-positive or -negative patients. This is in contrast to the North American Phase 3 registration study comparing tacrolimus and cyclosporine, in which patient survival at 5 years was found to be significantly higher among HCV-positive patients treated with tacrolimus compared with cyclosporine (14).

Previous studies have documented a lower incidence of acute rejection in liver transplant patients receiving tacrolimus at both 1 and 3 years following transplantation (9,12,13). Conversion from cyclosporine to tacrolimus has also been efficacious in controlling acute and chronic rejection (20,21). Furthermore, a recent prospective, randomized clinical trial in primary liver allograft recipients, wherein a composite endpoint consisting of death, retransplantation or treatment failure for immunological reasons was used, demonstrated a treatment benefit of tacrolimus over cyclosporine at 1 year (9). In the present study, the incidence of acute rejection was found to be similar among patients in the tacrolimus and cyclosporine cohorts for the 3-year follow-up period. However, significantly fewer tacrolimus-treated patients had acute rejection requiring anti-lymphocyte antibody therapy.

Cardiovascular disease and renal failure are important causes of mortality and morbidity following liver transplantation (2,3,6,22). For example, in a single-center retrospective study of 110 liver allograft recipients at a median of 3.9 years following transplant, there were 25 ischemic cardiovascular events compared to an expected number of 8.15 in the matched population (relative risk: 3.07; 95% CI: 1.98–4.53), whereas the number of deaths from cardiovascular disease among the allograft recipients was 18 compared with an expected number of 7.03 (relative risk: 2.56; 95% CI: 1.52–4.05) (3). In a study of 69 321 American recipients of non-renal organs between 1990 and 2000, the 5-year incidence of chronic renal failure in recipients of orthotopic liver transplantation was 18.1%, and the use of cyclosporine compared to tacrolimus was associated with an increased risk of chronic renal failure (relative risk: 1.25; p < 0.001) (2).

Our primary objectives were to study renal and cardiovascular function in liver transplant recipients managed according to immunosuppressive protocols in common use in 1997 and 1998. Our data show that, despite comparable baseline serum creatinine levels in the two treatment groups, patients receiving cyclosporine demonstrated significantly higher levels of serum creatinine and a greater change in serum creatinine (delta serum creatinine) compared with the tacrolimus cohort at all time points following transplantation. Similar results have been reported by two independent groups (6,23).

In our study, patients in the cyclosporine cohort demonstrated significantly higher systolic and diastolic blood pressure measurements compared to patients in the tacrolimus group. Coincident with the elevations in blood pressure, more anti-hypertensive agents were required by patients in the cyclosporine group through all 3 years of follow-up. The prevalence of hypertension in cyclosporine-treated patients is consistent with evidence in the literature both from the United Network for Organ Sharing (UNOS) database as cited above and from single-center studies (5,6,23).

In the present study, the incidence of new-onset diabetes mellitus was comparable between the cyclosporine and tacrolimus treatment groups at all time points following transplantation. In contrast, two recent prospective clinical trials showed higher rates of new-onset diabetes mellitus among liver transplant recipients receiving tacrolimus therapy (9,10). Still other studies have found that the choice of calcineurin inhibitor does not affect the incidence of new onset diabetes in the first 3 years following liver transplantation (6,24).

In this study, patients receiving tacrolimus had markedly lower total cholesterol levels over the 3-year follow-up period. The accompanying requirement for anti-hyperlipidemic agents was significantly lower in the tacrolimus cohort at 2 and 3 years. These results corroborate previous studies that have documented better lipid profiles associated with tacrolimus compared with cyclosporine (5,7) and those in which liver transplant recipients underwent conversion from cyclosporine to tacrolimus (25,26).

The use of corticosteroids was significantly greater in patients in the cyclosporine cohort compared to the tacrolimus cohort at all times after the first post-operative month. In part, this reflects clinical practice in the era under observation and has been noted by other authors (6). Given that corticosteroids may lead to hyperglycemia, hypertension and hyperlipidemia, it is possible that their greater use in the cyclosporine cohort may have ameliorated the diabetogenic impact of tacrolimus, while exacerbating the hypertensive and hyperlipidemic effects of cyclosporine. The present study does not support this contention regarding blood pressure in that the absolute levels of systolic pressure are maintained over the 36 months rather than declining with the gradual reduction in corticosteroid use. Similarly, when we consider the lipid profiles, while taking into account the greater use of anti-lipidemic agents as time goes on, we note that there is greater use of anti-lipidemics in the cyclosporine cohort at 36 months when the use of corticosteroids is least. This observation is in accord with the results of Aguirrezabalaga and colleagues who found that blood lipid levels were significantly greater in liver transplant recipients treated with cyclosporine compared to those receiving tacrolimus at times when corticosteroid use was not significantly different (27). Similarly, Trotter et al. found that using a immunosuppressive protocol based on sirolimus, a calcineurin inhibitor and no or at most 3 days of prednisone, hypercholesterolemia was significantly more common in the cyclosporine-treated patients (8). Finally, we note that the rate of diabetes in either cohort did not change in the last 24 months of observation, despite a reduction by half in the use of corticosteroids in the study cohort. Consequently, we infer that the choice of calcineurin inhibitor may have a substantial impact on hypertension and lipid profile.

This study was undertaken to examine renal and cardiovascular toxicity profiles in a large number of liver allograft recipients to assess not only the effects of calcineurin inhibitors but also the longitudinal practice patterns on patient outcome. As such, a univariate analysis was used to compare data in the cyclosporine and tacrolimus cohorts. Use of a multivariate analysis would have allowed for consideration of the effect on specific outcomes of pre-transplant factors other than selection of calcineurin inhibitor. Thus, multivariate analysis of peri-transplant variables to look for associations with specified outcome variables would be interesting in its own right. At the same time, we believe that use of a multivariate analysis for this study would have sacrificed the insights related to the comparison of the effect of the two calcineurin inhibitors gained from serial observation of the data set at intervals across 36 months.

There are limitations to our study. All retrospective studies are open to selection bias. We believe that our cohort size, reiterative data gathering and verification procedures limited such bias. First, the use of a recent cohort limits the bias due to changing patterns of practice. Second, the cohorts were constructed based on the initial choice of immunosuppressant. However, patients in the cyclosporine cohort were switched to tacrolimus significantly more frequently over the 3-year follow-up period. Thus, our data represent an underestimation of the advantage of tacrolimus over cyclosporine, further dampening any potential selection bias. Third, our study, which was based on standard clinical practice at participating centers in 1997 and 1998, demonstrated a significantly greater use of corticosteroids in the cyclosporine cohort compared with the tacrolimus cohort. As discussed above, the use of corticosteroids may have had an impact on the development of diabetes and influenced the prevalence of hypertension and hyperlipidemia in the patients receiving cyclosporine. Finally, it is worth noting that while many statistically significant differences in renal and cardiovascular clinical markers were observed between tacrolimus-treated and cyclosporine-treated cohorts, we have not shown that these markers were associated with significant mortality or graft loss in this study group during the period of follow-up.

In summary, we found that both tacrolimus- and cyclosporine-based immunosuppressive regimens in routine clinical use provided excellent patient and graft survival through 3 years following liver transplantation. Tacrolimus was associated with better renal function, lipid profile and blood pressure. The incidence of new-onset diabetes mellitus was similar in the tacrolimus and cyclosporine cohorts. The advantages of tacrolimus therapy on renal and cardiovascular risk profiles suggest possible benefits for the health of liver allograft recipients. However, larger, prospective studies with longer follow-up are warranted to evaluate the consequences of renal and cardiovascular risk changes that accompany long-term immunosuppressive agent administration.


The authors are grateful to the surgeons, study coordinators and hepatologists at the participating transplant centers. The authors thank The Synapse Group for assistance with collation of the data, Dr. Jamie Myles for statistical consultation and Drs. Susan Elliott and Carolynn Pietrangeli for preparation of the manuscript.

This study was sponsored by an unrestricted educational grant from Fujisawa Healthcare, Inc.