Sirolimus with Neoral Versus Mycophenolate Mofetil with Neoral is Associated with Decreased Renal Allograft Survival

Authors


*Corresponding author: Herwig-Ulf Meier-Kriesche, meierhu@medicine.ufl.edu

Abstract

To evaluate the association between a regimen of cyclosporine microemulsion (CsA) + sirolimus (Rapa) treatment versus CsA and mycophenolate mofetil (MMF) and renal allograft survival, we analyzed 23 016 primary recipients reported to the Scientific Registry of Transplant Recipients between January 1, 1998 and July 26, 2003.

Univariate Kaplan-Meier analysis and multivariate Cox proportional hazard models correcting for demographic and clinical covariates were used to estimate the relative risks for CsA+Rapa versus CsA+MMF-treated patients reaching study endpoints. Subgroup analyses were conducted for recipient ethnicity and donor type.

CsA+Rapa was associated with significantly lower graft survival (74.6% vs. 79.3% at 4 years, p = 0.002) and death-censored graft survival (83.7% vs. 87.2%, p = 0.003) compared to CsA+MMF. In multivariate analyses, CsA+Rapa was associated with a significantly increased risk for graft loss, death-censored graft loss and decline in renal function (HR = 1.22, p = 0.002; HR = 1.22, p = 0.018 and HR = 1.25, p < 0.001, respectively). Similar results were obtained in recipient ethnicity and donor type subgroups.

In summary, CsA+Rapa was associated with significantly worse graft survival and death-censored graft survival compared to CsA+MMF, and likely reflects full-dose CsA +Rapa. Outcomes regarding alternative strategies of Rapa utilization with reduced CsA, with alternative agents or with no calcineurin inhibitor cannot be extrapolated from these data.

Introduction

Many immunosuppression regimens are in use at the present time in renal transplantation, with triple therapy consisting of a calcineurin inhibitor (CNI; cyclosporine [CsA] or tacrolimus [tacro]), mycophenolate mofetil (MMF) and steroids being the most frequently used combinations in the United States. Because of advances in post-transplant clinical management, it has become difficult to discern differences in 1- and 3-year graft survival in prospective randomized trials of specific immunosuppression regimens (1).

Sirolimus (rapamycin [Rapa]) is an mTOR inhibitor that was approved by the Food and Drug Administration (FDA) in 1999. FDA registration trials showed that in combination with full-dose CsA and steroids, Rapa significantly lowered acute rejection when compared to azathioprine or placebo (2,3). However, impaired renal function was demonstrated in the Rapa group compared to control groups. Subsequently, studies were conducted in low-risk patients comparing full-dose CsA + Rapa to a strategy of CsA withdrawal and Rapa intensification 3 months post-transplantation (the Rapamune Maintenance Regimen [RMR] studies) to evaluate whether this regimen reduced the potential nephrotoxic insult (4,5). These studies demonstrated improved renal function in the CsA-withdrawal group, resulting in a revised label recommending the withdrawal of CsA in Rapa-based immunosuppression at 2–4 months post-transplantation in low-risk renal transplant recipients (6). Follow-up data from the RMR studies also demonstrated superior graft survival at 4 years for the CsA-withdrawal study arm compared to the CsA+Rapa control group (91.5% vs. 84.1%, p = 0.024). A statistically non-significant increase in acute rejection rates was observed during the CsA withdrawal in the study (7).

The data from the RMR studies indicate that low-risk patients experience better renal allograft survival on intensified Rapa monotherapy when compared to low-risk patients maintained on full-dose CsA + Rapa. However, these findings cannot be extrapolated to comparisons with immunosuppressive regimens that include alternative agents, such as MMF. As MMF-based regimens constitute the majority of de novo immunosuppressive regimens in renal transplant recipients in the United States (8), they probably serve as the most relevant comparator group in assessing other de novo regimens. Therefore, we retrospectively compared the outcomes for patients who underwent renal transplantation between 1998 and 2003 who were initially treated with CsA+Rapa to those initially treated with CsA+MMF, based on data supplied by the Scientific Registry of Transplant Recipients (SRTR).

Materials and Methods

Data reported to the SRTR for patients between the ages of 18 and 80 years who underwent solitary renal transplantation in the United States between January 1, 1998 and July 26, 2003, were included in the analysis. Patients who received cyclosporine microemulsion-based immunosuppression regimens were divided into two treatment groups based on medications at hospital discharge, as recorded on UNOS follow-up forms: CsA+Rapa (n = 1999) and CsA+MMF (n = 21 017), and were analyzed in an intent-to-treat fashion. Patients who were recorded as being treated with both Rapa and MMF at discharge (n = 233) were excluded from the analysis. Subgroup analyses based on recipient ethnicity (African American vs. Caucasian/all others) and donor type (living vs. cadaveric) were also conducted. An expanded analysis, including all patients who received any CNI (cyclosporine microemulsion, cyclosporine or tacrolimus), was performed to investigate the association between Rapa versus MMF therapy and the primary endpoints. The primary outcomes were graft survival, death-censored graft survival and patient survival. Acute rejection, late acute rejection and renal function decline were secondary endpoints. The definition for an acute rejection episode was that the patient received treatment for acute rejection, as recorded on UNOS follow-up forms. Late acute rejection was defined as a rejection episode that occurred beyond 12 months after transplantation. Renal function decline was defined by a 20% decrease in calculated creatinine clearance from baseline (i.e. 6 months post-transplantation). Creatinine clearance (CrCl) values were estimated using the Cockroft-Gault methodology (9); recipients who had no baseline serum creatinine value were excluded from the analysis (50 CsA+Rapa; 364 CsA+MMF). Follow-up data were obtained at 6 and 12 months post-transplantation, and annually thereafter, as long as data were available. The analysis of late acute rejection (853 CsA+Rapa; 11902 CsA+MMF) and renal function decline (803 CsA+Rapa; 11538 CsA+MMF) included patients who had been treated continuously for a minimum of 1 year post-transplant, and were analyzed using a methodology reported previously (10). The mean estimated CrCl was calculated for treatment groups by transplant year. For this analysis, tacrolimus-based regimens were also analyzed (2712 Rapa+tacro; 24 503 MMF+tacro).

Univariate Kaplan-Meier analysis was used to estimate survival outcomes for each endpoint. P-values were calculated using non-parametric log-rank tests. Cox regression models were used to investigate the independent effects of treatment on graft survival, death-censored graft survival and renal function decline. Potential confounding variables were corrected for in multivariate analyses, including recipient age and ethnicity, donor age, donor type, degree of human leukocyte antigen (HLA) matching, cold ischemia time, donor and recipient cytomegalovirus (CMV) status and length of time on dialysis prior to transplantation. Induction therapy (T-cell depleting antibodies, IL-2 receptor blockers and recipients of multiple induction agents) was also investigated as a potential confounding factor. Missing race was considered as other race (12 [0.6%] CsA+Rapa; 165 [0.8%] CsA+MMF). Absent donor (7 [0.4%] CsA+Rapa; 61 [0.3%] CsA+MMF) ages were replaced with mean values from complete data. A separate category was created for missing cold ischemia time (499 [25.0%] CsA+Rapa; 4190 [19.9%] CsA+MMF) and HLA mismatches (24 [1.2%] CsA+Rapa; 350 [1.7%] CsA+MMF). Missing weight (98 [11.5%] CsA+Rapa; 2508 [21.1%] CsA+MMF) was replaced with a class mean based on age, race and gender. Demographic and baseline clinical characteristic data were summarized with descriptive statistics. A probability of type 1 error α= 0.05 was considered the threshold of statistical significance. All analyses were performed using SAS Release 8.2 (SAS Institute, Cary, NC) under Windows XP.

Results

Table 1 shows the baseline study population characteristics for the CsA+Rapa and CsA+MMF treatment groups.

Table 1.  Demographic and clinical characteristics of patients undergoing solitary renal transplantation: CsA+Rapa and CsA+MMF treatment groups
 CsA+Rapa (n = 1999)CsA+MMF (n = 21 017)
Recipient characteristics
 Age (mean ± SD)45.45 ± 13.2047.30 ± 13.17
 Male gender (%)60.461.9
 Female gender (%)39.638.1
 African American (%)22.220.6
 Caucasian and other (%)77.879.4
 Delayed graft function (DGF) (%)17.917.5
 Immediate graft function (%)82.182.5
Induction treatment
 IL-2 receptor blocker (%)21.636.7
 T-cell depleting antibody (%)18.415.8
 Multiple induction agents (%)2.02.1
 No induction treatment (%)58.045.4
Donor characteristics
 Age (mean ± SD)37.12 ± 14.7037.78 ± 15.04
 Male gender (%)53.452.0
 Female gender (%)46.648.0
 Living donor (%)39.938.8
 Cadaveric donor (%)60.161.2

Graft survival

Kaplan-Meier analysis showed significantly lower graft survival in the CsA+Rapa-treated group compared to the CsA+MMF-treated group (p = 0.002; Figure 1). Graft survival at 3 and 4 years post-transplantation in the CsA+Rapa-treated group was 80.9% and 74.6%, respectively compared to 84.4% at 3 years and 79.3% at 4 years in the CsA+MMF-treated group. The Cox proportional hazard model (Table 2) demonstrated that CsA+Rapa therapy was associated with a 22% increased risk (HR = 1.22, p = 0.002) for graft loss compared to CsA+MMF therapy. Significant recipient risk factors for graft loss included African-American race (HR = 1.53, p < 0.001), increasing age (HR = 1.08 per decade, p < 0.001), extended time on dialysis preceding transplantation (HR = 1.03 per year, p < 0.001) and induction treatment with a T-cell depleting antibody (HR = 1.16, p = 0.002) or multiple agents (HR = 1.27, p = 0.032). Significant donor risk factors for graft loss included increasing donor age (HR = 2.60 for age >60 years, p < 0.001; HR = 1.86 for age 51–60 years, p < 0.001 and HR = 1.61 for age 41–50 years), and 6 HLA mismatches (HR = 1.34, p < 0.001). Having a living donor was associated with a reduced risk for graft loss (HR = 0.66, p < 0.001).

Figure 1.

Kaplan-Meier analysis of graft survival, CsA+Rapa versus CsA+MMF.

Table 2.  Multivariate analysis for graft loss (reference groups in parentheses)
Variables (reference group)Hazard ratio95% CIp-value
  1. aIncludes donor positive to recipient positive, donor positive to recipient negative and donor negative to recipient positive.

Treatment
 CsA+Rapa (CsA+MMF)1.221.07 1.380.002
Recipient factors
 Recipient age (per 10 years)1.081.05 1.11<0.001
 Time on dialysis (per year; zero time)1.031.02 1.04<0.001
 CMV neg donor to neg recipient (all other CMV statusa)0.950.86 1.040.231
Recipient race
 African American (Caucasian/all other)1.531.41 1.65<0.001
Donor factors
 Living donor (cadaveric donor)0.660.59 0.75<0.001
 Donor age 0–10 (donor age 11–20) years1.291.05 1.590.017
 Donor age 21–30 (donor age 11–20) years1.191.03 1.380.022
 Donor age 31–40 (donor age 11–20) years1.371.19 1.58<0.001
 Donor age 41–50 (donor age 11–20) years1.611.41 1.84<0.001
 Donor age 51–60 (donor age 11–20) years1.861.62 2.14<0.001
 Donor age >60 (donor age 11–20) years2.602.22 3.05<0.001
HLA mismatch
 1 HLA mismatch (exact)1.200.98 1.460.072
 2 HLA mismatch (exact)1.211.04 1.410.012
 3 HLA mismatch (exact)1.261.10 1.44<0.001
 4 HLA mismatch (exact)1.361.19 1.56<0.001
 5 HLA mismatch (exact)1.451.26 1.66<0.001
 6 HLA mismatch (exact)1.341.13 1.57<0.001
Cold ischemia time
 Cold ischemia time 12 to <18 h (<12 h)1.000.89 1.140.956
 Cold ischemia time 18 to <24 h (<12 h)1.060.94 1.200.355
 Cold ischemia time 24 to <30 h (<12 h)1.120.98 1.280.102
 Cold ischemia time ≥30 h (<12 h)1.181.02 1.380.030
Induction treatment
 T-cell depleting antibody (no induction)1.161.06 1.280.002
 IL-2 receptor blocker (no induction)0.980.90 1.070.670
 Multiple induction agents (no induction)1.271.02 1.580.032

Death-censored graft survival

Kaplan-Meier analysis for death-censored graft survival (Figure 2) showed significantly lower graft survival in the CsA+Rapa-treated group compared to the CsA+MMF-treated group (p = 0.003). Death-censored graft survival was 87.6% at 3 years and 83.7% at 4 years in the CsA+Rapa-treated group, compared to 90.7% at 3 years and 87.2% at 4 years in the CsA+MMF-treated group. The Cox proportional hazard model (Table 3) demonstrated that CsA+Rapa therapy was associated with a 22% increased risk (HR = 1.22, p = 0.018) for death-censored graft loss compared to CsA+MMF therapy. Significant recipient risk factors for death-censored graft loss included African-American race (HR = 1.72, p < 0.001), extended time on dialysis preceding transplantation (HR = 1.03 per year, p < 0.001) and treatment with a T-cell depleting antibody (HR = 1.25, p < 0.001). Significant donor risk factors for death-censored graft loss included increasing donor age (HR = 3.49 for age >60 years, p<0.001; HR = 2.12 for age 51–60 years, p < 0.001 and HR = 1.65 for age 41–50 years), and 6 HLA mismatches (HR = 1.69, p < 0.001). Having a living donor was associated with a reduced risk for death-censored graft loss (HR = 0.65, p < 0.001), while increasing cold ischemia time (CIT) was associated with an increased risk for death-censored graft loss (HR = 1.29 for CIT ≥30 h, p = 0.014; HR = 1.29 for CIT between 24 and <30 h, p = 0.004).

Figure 2.

Kaplan-Meier analysis of death-censored graft survival, CsA+Rapa versus CsA+MMF.

Table 3.  Multivariate analysis for death-censored graft loss
Variables (reference group)Hazard ratio95% CIp-value
  1. aIncludes donor positive to recipient positive, donor positive to recipient negative and donor negative to recipient positive.

Treatment
 CsA+Rapa (CsA+MMF)1.221.031.430.018
Recipient factors
 Recipient age (per 10 years)0.810.780.84<0.001
 Time on dialysis (per year; zero time)1.031.011.04<0.001
 CMV neg donor to neg recipient (all other CMV statusa)0.900.801.020.101
Recipient race
 African American (Caucasian/all other)1.721.561.90<0.001
Donor factors
 Living donor (cadaveric donor)0.650.560.76<0.001
 Donor age 0–10 (donor age 11–20) years1.381.061.800.016
 Donor age 21–30 (donor age 11–20) years1.000.821.230.970
 Donor age 31–40 (donor age 11–20) years1.381.151.66<0.001
 Donor age 41–50 (donor age 11–20) years1.651.391.97<0.001
 Donor age 51–60 (donor age 11–20) years2.121.772.53<0.001
 Donor age >60 (donor age 11–20) years3.492.854.28<0.001
HLA mismatch
 1 HLA mismatch (exact)1.230.931.630.147
 2 HLA mismatch (exact)1.531.251.88<0.001
 3 HLA mismatch (exact)1.511.251.82<0.001
 4 HLA mismatch (exact)1.631.351.97<0.001
 5 HLA mismatch (exact)1.771.462.14<0.001
 6 HLA mismatch (exact)1.691.352.10<0.001
Cold ischemia time
 Cold ischemia time 12 to <18 h (<12 h)1.060.901.250.490
 Cold ischemia time 18 to <24 h (<12 h)1.140.971.350.115
 Cold ischemia time 24 to <30 h (<12 h)1.291.081.540.004
 Cold ischemia time ≥30 h (<12 h)1.291.051.570.014
Induction treatment
 T-cell depleting antibody (no induction)1.251.111.41<0.001
 IL-2 receptor blocker (no induction)0.990.881.100.804
 Multiple induction agents (no induction)1.200.891.610.229

Acute rejection

Acute rejection rates were higher in the CsA+Rapa group compared to the CsA+MMF group (p = 0.002). Of the recipients treated with CsA+Rapa 20.5% experienced rejection in the first year, compared to 17.4% of recipients treated with CsA+MMF. The Cox proportional hazard model demonstrated that CsA+Rapa therapy was associated with a 10% increased risk (HR = 1.10) for acute rejection compared to CsA+MMF therapy, but this result did not reach statistical significance (p = 0.079).

Acute rejection rates for the first year post-transplantation by transplant year and treatment group are presented in Table 4. The acute rejection rates decreased in later transplant years in both study groups. The acute rejection rate in the CsA+Rapa group was significantly higher than that of the CsA+MMF group in all years studied, with diagnosis rates of 27.1%, 26.2% and 20.8% in 1998, 1999 and 2000, respectively, compared to 20.0%, 17.7% and 16.2% in the CsA+MMF group (p = 0.001, 0.002 and 0.007, respectively). There were no differences in acute rejection rates in patients who underwent transplantation in 2001 (p = 0.951).

Table 4.  Acute rejection rates (by Kaplan-Meier estimate) in the first year post-transplantation and Cox regressiona analysis findings by transplant year and treatment group
 Acute rejection (%) by transplant year
1998199920002001
  1. aBased on data from discharge, 6-month and 12-month follow-up.

  2. bLog-rank test.

Kaplan-Meier analysis
CsA+Rapa (%)27.126.220.816.8
CsA+MMF (%)20.017.716.216.6
p-valueb0.00110.00240.00720.9513
Cox regression analysis
HR CsA+Rapa (CsA+MMF)1.441.531.180.82
95% CI1.15–1.801.14–2.040.97–1.440.63–1.06

Kaplan-Meier analysis showed significantly higher late acute rejection rates in the CsA+Rapa-treated group compared to the CsA+MMF-treated group (p = 0.006). Late acute rejection at 4 years, post-transplantation in the CsA+Rapa-treated group was 9.2% compared to 7.2% in the CsA+MMF-treated group. The Cox proportional hazard model demonstrated that CsA+Rapa therapy was associated with a 28% increased risk (HR = 1.28, p = 0.053) for late acute rejection compared to CsA+MMF therapy, but this result did not reach statistical significance.

Renal function

The mean estimated CrCl at 1-year post-transplantation by transplant year and treatment group is presented in Figure 3. CrCl at 1-year post-transplant improved in later transplant years, in both study groups. The mean CrCl in the CsA+Rapa group was lower than that of the CsA+MMF group in all years studied, ranging from 59.89 ± 23.46 cc/min to 67.69 ± 23.39 for CsA+Rapa versus 69.59 ± 26.57 cc/min to 70.35 ± 27.43 for CsA+MMF (p < 0.0001 in 1998, p = 0.328 in 2002).

Figure 3.

Mean creatinine clearance at 1 year after transplantation for Rapa- and MMF-based treatment regimens, by transplant year.

Kaplan-Meier analysis of renal function, as measured by a 20% decrease from the baseline in estimated CrCl showed a significantly higher rate of renal function decline in the CsA+Rapa-treated group compared to the CsA+MMF-treated group (p = 0.001). The rate of renal function decline in the CsA+Rapa-treated group was 39.4% at 3 years and 47.0% at 4 years, compared to 33.2% at 3 years and 38.6% at 4 years in the CsA+MMF-treated group. A higher rate of renal function decline for the CsA+Rapa group was confirmed in a multivariate analysis (HR = 1.25, p < 0.001; data not shown).

Recipient ethnicity subgroup analysis

A subgroup analysis was performed to investigate the effects of CsA+Rapa therapy on graft survival; death-censored graft survival and renal function decline in patients of African-American (443 CsA+Rapa; 4338 CsA+MMF) and Caucasian/other race (1556 CsA+Rapa; 16 679 CsA+MMF). For African-American recipients, the Kaplan-Meier analyses showed significantly lower graft survival in the CsA+Rapa group (p = 0.029) compared to the CsA+MMF group (64.1% vs. 69.9%, respectively, at 4 years). There was no statistical difference between the two treatment groups for death-censored graft survival (p = 0.237). Estimated rates of renal function decline were higher in the CsA+Rapa group (47.4% at 4 years compared to 55.8% in the CsA+MMF group), but this difference was not statistically significant (p = 0.136). Increased graft loss, death-censored graft loss and renal function decline were observed for CsA+Rapa versus CsA+MMF in the Caucasian/other subgroup (results not shown).

A Cox proportional hazard model used to assess the combined risks of recipient ethnicity and treatment on study endpoints confirmed that African-American patients treated with CsA+Rapa had the greatest increased risk for graft loss, death-censored graft loss and death compared to all other recipient ethnicity and treatment combinations (Table 5).

Table 5.  Summary of multivariate analysesa for graft loss, death-censored graft loss and patient death, by recipient ethnicity
OutcomeCovariateHazard ratio95% CI of hazard ratiop-value
  1. aIn this multivariate analysis, combined drug and recipient ethnicity terms were included. The reference group was MMF+CSA Caucasian/all other.

Graft lossCsA+Rapa African American1.931.572.38<0.001
 CsA+MMF African American1.521.401.65<0.001
 CsA+Rapa Caucasian/all other1.191.021.390.031
Death-censored graft lossCsA+Rapa African American1.981.522.59<0.001
 CsA+MMF African American1.741.561.93<0.001
 CsA+Rapa Caucasian/all other1.261.031.540.025
Patient deathCsA+Rapa African American1.931.452.55<0.001
 CsA+MMF African American1.221.091.37<0.001
 CsA+Rapa Caucasian/all other1.110.901.370.313

Donor type subgroup analysis

A subgroup analysis was performed to investigate the effects of CsA+Rapa therapy based on donor type (living [798 CsA+Rapa; 8156 CsA+MMF] vs. cadaveric [1201 CsA+Rapa; 12 861 CsA+MMF]) for the endpoints of graft survival, death-censored graft survival and renal function decline. For recipients of living donor organs, the Kaplan-Meier analyses showed lower graft survival in the CsA+Rapa group compared to the CsA+MMF group, but this result did not reach statistical significance (p = 0.071). Graft survival at 4 years was 82.6% in the CsA+Rapa group compared to 85.0% in the CsA+MMF group. There were small, but not statistically significant, differences in death-censored graft survival (p = 0.121) (89.5% in the CsA+Rapa group vs. 90.8% in the CsA+MMF group at 4 years, respectively), and in renal function decline (p = 0.510) (45.1% at 4 years compared to 37.0% in the CSA+MMF group). CsA+Rapa living donor recipients showed an increased risk for graft loss (HR = 1.25, CI 0.96–1.61, p = 0.091) and death-censored graft loss (HR = 1.28, CI 0.91–1.79, p = 0.156), but these results did not reach statistical significance. There was a small, statistically non-significant difference in renal function decline between the two treatment groups (HR = 1.09, CI 0.86–1.37, p = 0.480).

CsA+Rapa cadaveric organ recipients had increased risks for graft loss (HR = 1.23, CI 1.06–1.42, p = 0.005), death-censored graft loss (HR = 1.22, CI 1.02–1.47, p = 0.034) and renal function decline (HR = 1.32, CI 1.13–1.53, p < 0.001) compared to CsA+MMF cadaveric organ recipients.

CNI (CsA or tacrolimus)+Rapa versus CNI+MMF analysis

Kaplan-Meier analyses showed significantly lower graft survival in the CNI+Rapa group (p < 0.001) compared to the CNI+MMF group (71.0% vs. 77.9%, respectively, at 4 years), death-censored graft survival (p < 0.001) (80.5% vs. 85.9%, respectively, at 4 years) and renal function decline (p < 0.001) (55.3% vs. 62.8%, respectively, at 4 years). Results were confirmed in the multivariate analysis and are presented in Table 6.

Table 6.  Summary of analyses for graft loss, death-censored graft loss and renal function decline, CNI(cyclosporine or tacrolimus)+Rapa (n = 5266) versus CNI+MMF (n = 50 015)
Outcome% Event-free survival (CNI+Rapa vs. CNI+MMF)p-valueHazard ratio95% CI of hazard ratiop-value
Graft loss71.0 vs. 77.9<0.0011.321.21, 1.43<0.001
Death-censored graft loss80.5 vs. 85.9<0.0011.331.20, 1.48<0.001
Renal function decline55.3 vs. 62.8<0.0011.281.17, 1.40<0.001
Late acute rejection92.0 vs. 92.90.0051.201.00, 1.430.052

Discussion

The results of this study demonstrate that CSA+Rapa between 1998 and 2003, was associated with decreased graft survival (p = 0.002; 74.6% vs. 79.3% 4 years post-transplantation) when compared to CsA+MMF use during the same time period. Death-censored graft survival was also significantly lower in the CsA+Rapa group (p = 0.003; 83.7% vs. 87.2% 4 years post-transplantation, respectively). For both graft survival and death-censored graft survival endpoints, the difference between the two treatment groups increased over time. Since primary data about the reason for graft loss are not available from the database, we cannot determine the mechanism of the increased graft loss from this study. Poor renal function has been shown to be associated with poor outcomes in numerous studies (11–13). It has been shown that CsA+Rapa have a synergistic effect on renal function deterioration in animal models of chronic nephrotoxicity (14). In addition, registration trials showed that Rapa in combination with CsA resulted in impaired renal function (1,2) and subsequent studies resulted in the FDA recommendation to withdraw CsA from Rapa at 3 months (3,4). Renal function in the CsA+Rapa group was inferior, with 47% of patients experiencing renal function decline 4 years after transplantation, compared to 38.6% in the CsA+MMF group, a finding consistent with lower graft survival in the CsA+Rapa group. These results are possibly caused by the clinical practice of using full-dose CsA with Rapa, which was the standard of care early after Rapa's approval. Insufficient data and follow-up time have accmulated in the more recent era to evaluate whether CsA sparing or withdrawal in patients treated with Rapa might be associated with more favorable results. The improvement in renal function over time for Rapa in combination with CNIs provides evidence that there has been a positive evolution in the clinical usage of Rapa. With this positive learning curve, it is possible that graft survival differences noted in this analysis may not be evident in future analyses.

Acute rejection rates were similar between the two groups, with slightly lower rates in the CsA+MMF group. This difference was statistically significant, but probably represents little noticeable clinical difference between the two regimens. When analyzed by the year of transplantation, the difference between the two treatment groups disappeared in the more recent years. It is possible that more renal biopsies were performed in the CsA+Rapa group driven by the inferior renal function, and consequently, more episodes of acute rejection were diagnosed. This may have resulted in higher rates of acute rejection diagnosis reported, specifically in the CsA+Rapa group in the early years of this study (i.e. 1998–2000) when renal function was at its poorest in the group. It is well known that subclinical rejection is frequently found on biopsy (15), and that more biopsies may lead to increased diagnoses of acute rejection. This hypothesis is supported by the fact that acute rejection rates were not different in the more recent transplant years when renal function was more similar in the two comparison groups. The overall rates of acute rejection (20.5% in the CsA+Rapa group and 17.4% in the CsA+MMF group) were similar to rates described in prospective trials. In the registration trials, biopsy-proven rejection within the first 6 months post-transplantation for patients on Rapa was between 11.3% and 24.7% (1,2). Similarly, for patients on MMF, biopsy-proven acute rejection ranged between 13.8% and 19.8% (16–18).

In a subgroup analysis conducted to assess the impact of recipient ethnicity and treatment group on outcomes, the greatest risk for graft loss occurred in the African-American subgroup who received CsA+Rapa. This is not surprising since African-American recipients are known to be at increased risk for acute rejection (19), and for that reason, may not have tolerated well the approaches of CsA sparing with Rapa, or may not have been candidates for CsA sparing at all (20).

Within the subgroup who received organs from cadaveric donors, treatment with CsA+Rapa was associated with a statistically significant increased risk for graft loss, death-censored graft loss and renal function decline. For recipients of living-donor organs, trends of similar magnitude were observed, but the results in this subgroup did not reach statistical significance. When creating subgroups of different size, statistical power might be the reason for finding statistically significant numbers in one but not the other group. If this difference was not due to statistical power, the discrepancy in achieving statistical significance in cadaveric, but not living donor recipients, might be explained by the tendency to use higher doses of immunosuppressives in recipients of cadaveric organs, and thus increasing the nephrotoxic exposure in the cadaveric group. Second, the higher quality living donor kidneys may have shown less impact of the potentially increased nephrotoxicity.

This study is a retrospective analysis utilizing data reported to the SRTR. While registry analyses can elucidate important differences in long-term outcomes, results must be interpreted with caution. The most important caveat for this analysis is that no claim can be made about the dose or duration of exposure of the treatment regimens compared in this study. The SRTR database contains no information about drug dosing, and therefore, we cannot determine the dose or exact duration of Rapa, MMF or CsA exposure. Therefore, from this study, we cannot determine whether early or continued exposure contributes to the increase in graft loss. In addition, sufficient follow-up data have not yet accmulated for tacrolimus-based regimens, and no extrapolations of these data may be made to Rapa or MMF in conjunction with tacrolimus. The improvement in renal function over time for Rapa in combination with both CsA and tacrolimus suggests that improved outcomes might be obtained with CNI-sparing strategies, but the long-term safety of such approaches remains uncertain.

Further, potential limitations should be considered. In an attempt to address the possibility that many of the CsA+Rapa recipients were treated in a single center, we extended the analysis to include all CNIs. Because the results from this expanded analysis were consistent with the analysis limited to CsA+Rapa, our study suggests that the outcomes are likely not limited to a high-risk population in a single center. Another potential limitation is that reporting to any registry may be inaccurate or incomplete. The accuracy of post-transplant survival data submitted to the OPTN (UNOS) by individual transplant centers has been assessed. By using publicly available mortality data from the Social Security Death Master File (SSDMF), SRTR analyses show that for the vast majority of centers, this self-reported survival data is accurate (21). In addition, the SRTR utilizes a measure comparing expected follow-up and reported follow-up for a given endpoint, and publishes reports that demonstrate that the data are largely complete (22). Finally, if under- or over-reporting is occurring, it is reasonable to expect that it occurs equally between the two comparator drug groups, and does not affect the conclusions of this study.

In summary, our study demonstrates that treatment with the immunosuppressive regimen of CsA+Rapa in the United States between 1998 and 2003 was associated with an increased risk for both graft loss and renal function decline, when compared to one of the most frequently used immunosuppression regimens, CsA+MMF. The relative efficacy of Rapa utilized in a CNI-free regimen or a regimen in which CNI doses are reduced, compared to CNI and MMF, awaits longer follow-up and well-designed prospective trials.

Acknowledgment

The data reported here have been supplied by the United Network for Organ Sharing and University Renal Research and Education Association under contract with the Department of Health and Human Services. The interpretation and reporting of these data are the responsibility of the authors and do not represent an official policy or interpretation of the United States Government or any of its representatives. The analyses reported here have been conducted in ProSanos Corporation facilities, and were funded by Roche Laboratories. This work was presented in part at the 2004 American Transplant Congress, Boston, MA.

Ancillary