Cyclosporine (CsA) nephrotoxicity is enhanced by sirolimus (SRL). Tacrolimus is perceived to be less nephrotoxic than CsA, and therefore, CsA has been largely replaced by tacrolimus (TAC) when calcineurin inhibitors are used with SRL.
We analyzed 44 915 adult renal transplants in the Scientific Renal Transplant Registry (SRTR) from 2000 to 2004. Three thousand five hundred twenty-four (7.8%) patients received a baseline immunosuppressive regimen of TAC/SRL, with an inferior overall (log-rank p < 0.001) and death-censored graft survival (p < 0.001) as compared to TAC/MMF (N = 27 007). This effect was confirmed in multivariate Cox models; the adjusted hazard ratio (AHR) for overall graft loss with TAC/SRL was 1.47 (95% CI = 1.32, 1.63) and for CsA/SRL 1.38 (95% CI = 1.20, 1.59) relative to TAC/MMF. These effects were most apparent in high-risk transplants. Six-month acute rejection rates were low (11.5–12.6%) and not different between groups.
In summary, national data indicate that TAC/SRL as compared to TAC/MMF is associated with significantly worse renal allograft survival in all subgroups of patients and, in particular, higher-risk transplants. These results have to be interpreted in the context of the inherent limitations of any retrospective database analysis and evaluated in context with data from prospective clinical trials.
Early studies by Andoh et al. in animals did not show any effects of sirolimus (SRL) alone on glomerular function or on interstitial fibrosis; however, subtle morphologic and functional tubular changes could be observed (1). However, as SRL in combination with cyclosporine (CsA) was being investigated in kidney transplantation in the United States, Andoh et al. reported a rapid communication that rapamycin could potentiate CsA-induced interstitial scarring in their well-described rat model (2). This was shortly followed by the phase III studies that led to the approval of SRL in the United States for kidney transplantation, where impaired renal function in patients on SRL and CsA was reported as compared to the control group (3–5). This finding was felt to be due to SRL potentiating CsA nephrotoxicity and led to the conception of the Rapamune Maintenance Regimen (RMR) study, where CsA was withdrawn at 2 months post-transplantation and patients remaining on full dose SRL and CsA were compared (6,7). The subjects in the RMR study experienced renal function improvements when taken off CsA and in the longer follow-up, achieved better graft survival despite a statistically nonsignificant increase in acute rejection during CsA withdrawal (6,7). This finding in the RMR study led to the labeling change suggesting withdrawal of CsA from regimens containing Rapamycin at 3–4 months post-transplantation in low-risk subjects (8). Subsequently, a large population study confirmed the original concerns raised in the phase III trials of the potential for worse graft survival in patients treated with the combination of CsA and SRL (9).
Given the overall perception of reduced nephrotoxicity with tacrolimus (TAC) compared to CsA (10–12), and the preliminary observations by MacDonald et al. (13), TAC-SRL in combination with corticosteroids became increasingly accepted as a regimen that was both efficacious and afforded stable renal function. However, the best renal function with this combination was most evident in studies utilizing TAC in low doses (13–16). It is important to note that none of these studies compared renal function to a control group of TAC and MMF. Given the reciprocal nature of serum creatinine and GFR, it is possible that at the low values of serum creatinine seen with TAC, small increments in serum creatinine that might translate into important differences in GFR, could easily be overlooked. In fact, the only randomized prospective trial comparing SRL and MMF in combination with TAC, showed significantly worse renal function, and a trend toward inferior graft survival at 1 year in the patients treated with SRL and TAC versus MMF and TAC (17,18).
The latest released update of U.S. transplant data provided by the Scientific Renal Transplant Registry (SRTR), affords sufficient follow-up to retrospectively analyze results with the combination of TAC and SRL versus the current most prevalent combination of TAC and MMF, and put the results of accumulating prospective evidence in the context of the national usage of these drug regimens.
We analyzed 44 915 adult solitary renal transplant recipients in the United States available in the SRTR database from 2000 to 2004 on a discharge regimen of either TAC or CsA and either SRL or MMF. Patients treated with CsA derived only from the use of Neoral (Novartis Pharma) in their baseline regimen as indicated in the transplant database, other generic forms of CsA were not utilized for this analysis. Recipients with multiple indications of TAC and CsA or SRL and MMF were eliminated from the sample. We generated Kaplan-Meier plots for overall and death-censored graft survival and patient survival. Log-rank tests were used to test significance of comparative groups in Kaplan-Meier survival plots. We also constructed multivariate Cox proportional hazard models for these outcomes adjusted for potential confounding factors. Multivariate models were adjusted for donation type (living or deceased donor), recipient race, recipient age group, donor race, donor age group, pre-transplant dialysis time and recipient primary diagnosis. We generated a multivariate logistic model for the outcome of 6-month acute rejection for those recipients with a minimum of 6-month follow-up adjusted for the same covariates as the survival models with the addition of use of induction therapy. In addition, we examined survival models stratified by donor type, recipient race, the use of expanded criteria donations (ECD), and we analyzed subsets of data limited to primary transplants. We also examined graft survival rates between the regimens in centers with high TAC/SRL utilization (defined as >30% of recipients on this baseline regimen) and in centers with lower usage. This threshold of ‘high’ utilization was selected as approximately half the recipients on a TAC/SRL regimen were transplanted in these centers. Recipient age groups were categorized as 18–34, 35–54, 55–64 and 65+. Donor age groups were categorized as 0–6, 7–11, 12–17, 18–29, 30–39, 40–49, 50–59, 60–69 and 70+. Acute rejection rates were calculated for only those patients with follow-up time through the applicable period. Rates of discontinuation were assessed by examining the cohort of patients with the applicable follow-up time and with indications of immunosuppressive regimen in follow-up files. Discontinuation rates were calculated based on the percentage of recipients that were not indicated as receiving the same immunosuppressive regimen at follow-up as they were at baseline of all those patients on a particular regimen. The last possible follow-up time was through July 2004 as indicated in the transplant files. All analyses were conducted on SAS software version 9.1 (Cary, NC).
Demographic characteristics of the study cohort are displayed in Table 1. The overall cohort included 58% deceased-donor transplants, 24% African American recipients, 12% African American donors, 34% of recipients over the age of 54, 24% of donors over the age of 49, 13% retransplant recipients, 22% of recipients had a primary diagnosis of diabetes and 27% of recipients had a pre-transplant dialysis time greater than 36 months. The relative frequency of the regimens in the cohort within the study period (from 2000 to 2004) was 60.1% (TAC/MMF), 27.9% (CsA/MMF), 7.8% (TAC/SRL) and 4.2% (CsA/SRL). Utilization patterns from 1998 through 2004 are displayed in Table 2. As indicated in the table, the TAC/SRL regimen was utilized regularly after 1999 (between 6.2% and 10.1% of the cases in this cohort) and the TAC/MMF regimen has increased the utilization steadily from 1998 to 2004. Approximately, half (49.7%) of recipients on a baseline regimen of TAC/SRL were transplanted in centers with 30% or higher utilization rate of this regimen (representing 7.7% of all transplants from this cohort).
Table 1. Transplant characteristics by immunosuppressive regimen
As displayed in Figure 1, overall graft survival was significantly different between the four regimen groups overall (p < 0.001). The individual comparison of TAC/MMF and TAC/SRL was also significant (p < 0.001). The 3-year overall graft survival rates were 85.9% (TAC/MMF), 85.3% (CsA/MMF), 82.2% (CsA/SRL) and 80.3% (TAC/SRL). Three-year survival rates for the overall cohort and stratified by donor type, recipient ethnicity, and for ECD transplants that were on TAC/MMF or TAC/SRL at baseline are displayed in Table 3. The difference is survival between the TAC/MMF and TAC/SRL regimen was most evident among the highest-risk donor transplants, as evidence by the results in expanded donor-criteria transplants (Figure 2). Recipients at centers with high TAC/SRL utilization (>30%) had significantly different overall graft survival (p = 0.012) between TAC/MMF (3-year graft survival = 83.8%) and TAC/SRL (3-year graft survival = 80.8%). Similarly, centers with low utilization of the TAC/SRL regimen (≤30%) had significantly different overall graft survival (p < 0.001) between TAC/MMF (3-year survival = 86.0%) and TAC/SRL (3-year survival = 79.9%). The TAC/SRL regimen was also associated with a significant hazard for overall graft loss in the multivariate Cox proportional hazard model (adjusted hazard ratio (AHR = 1.47, 95% CI = 1.32, 1.64) relative to the TAC/MMF regimen. In addition, the CsA/MMF (AHR = 1.15, 95% CI = 1.07, 1.23) and the CsA/SRL (AHR = 1.39, 95% CI = 1.21, 1.60) were significantly associated with greater hazards for graft loss relative to the TAC/MMF regimen. Limiting the patient population to only primary transplants adjusted for the same covariates, indicated similar results, with TAC/MMF as reference, recipients on TAC/SRL (AHR = 1.48, 95% CI = 1.31, 1.66), CsA/SRL (AHR = 1.34, 95% CI = 1.15, 1.57), and CsA/MMF (AHR = 1.13, 95% CI = 1.05, 1.22) also demonstrated a significant hazard for overall graft loss.
Table 3. Overall and death-censored graft survival in TAC/MMF and TAC/SRL regimens
^p-value for log-rank test comparing TAC/SRL and TAC/MMF survival.
Death-censored graft survival
Death-censored graft survival was significantly different among the four regimen groups (p < 0.001). The 3-year death-censored graft survival rates were 92.0% (TAC/MMF), 92.3% (CsA/MMF), 90.1% (CsA/SRL) and 87.4% (TAC/SRL). Recipients with a TAC/SRL baseline regimen had significantly inferior death-censored graft survival (p < 0.001) for deceased-donor transplant recipients relative to TAC/MMF as displayed in Figure 3. Three-year death-censored survival rates for all transplant recipients and stratified by donor type, recipient ethnicity, and for ECD transplants that were on TAC/MMF or TAC/SRL at baseline are displayed in Table 3. The cohort of patients on a baseline regimen of TAC/MMF had significantly superior survival overall and among each of the subgroups that we examined. The adjusted model for death-censored graft survival also indicated that TAC/SRL (AHR = 1.54, 95% CI = 1.34, 1.77) and CsA/SRL (AHR = 1.28, 95% CI = 1.06, 1.56) were significantly associated with inferior survival with TAC/MMF as reference. CsA/MMF was not significantly associated with worse death-censored graft survival in the multivariate model (AHR = 1.09, 95% CI = 0.99, 1.20).
Patient survival was significantly different among the four regimen groups (p < 0.001). The 3-year patient survival rates were 92.2% (TAC/MMF), 91.0% (CsA/MMF), 90.0% (CsA/SRL) and 89.9% (TAC/SRL). Three-year patient survival rates for TAC/MMF and TAC/SRL, respectively, were 90.3% and 87.2% (p < 0.001) in deceased-donor transplant recipients, 91.5% and 88.3% for all African American recipients (p = 0.004) and 84.6% and 84.2% for ECDs (p = 0.45). The adjusted model for patient survival indicated that TAC/SRL (AHR = 1.41, 95% CI = 1.21, 1.63), CsA/SRL (AHR = 1.49, 95% CI = 1.24, 1.79) and CsA/MMF (AHR = 1.22, 95% CI = 1.11, 1.34) were significantly associated with inferior patient survival rate relative to TAC/MMF.
Acute rejection rates at 6 and 12 months for deceased-donor and living transplant recipients by regimen are displayed in Table 4. The overall 6-month rejection rates for this cohort were 12.9% (deceased-donor transplants) and 11.2% (living transplants). The overall 12-month rejection rates were 15.2% (deceased-donor transplants) and 13.1% (living transplants). There was no significant association between regimen group and rates of acute rejection at 6 or 12 months. The 6-month rejection rates by regimen for African American recipients were 13.3% (TAC/SRL), 12.3% (CsA/SRL), 14.5% (TAC/MMF) and 14.0% (CsA/MMF). The 6-month rejection rates by regimen for recipients of ECDs were 14.4% (TAC/SRL), 14.3% (CsA/SRL), 14.6% (TAC/MMF) and 17.1% (CsA/MMF). The multivariate logistic model also confirmed that there was no significant association of immunosuppressive regimen with 6-month acute rejection was present, TAC/SRL (adjusted odds ratio (AOR = 0.93, 95% CI 0.83, 1.05), CsA/MMF (AOR = 1.07, 95% CI = 0.99, 1.14) and CsA/SRL (AOR = 0.98, 95% CI 0.83, 1.15) with TAC/MMF as the reference group.
Table 4. Acute rejection rates by baseline immunosuppressive regimen
Chi-square p-value comparing rates among four regimens
Regimen discontinuation rates at 6-, 12- and 24-months post-transplant for the four regimen groups are displayed in Table 5. Discontinuation rates for deceased-donor transplants at 6 months were 17.9% (TAC/MMF), 28.2% (CsA/MMF), 30.4% (CsA/SRL) and 32.3% (TAC/SRL). Discontinuation rates for African American recipients at 6 months were 19.0% (TAC/MMF), 30.6% (CsA/MMF), 32.5% (CsA/SRL) and 33.0% (TAC/SRL). Discontinuation rates for ECD transplants at 6 months were 24.4% (TAC/MMF), 33.4% (CsA/MMF), 45.3% (CsA/SRL) and 35.8% (TAC/SRL).
Table 5. Rates of discontinuation of baseline immunosuppressive regimens*
*Based on patients with minimum follow-up to the applicable period and indications of immunosuppressive regimens at that follow-up period. Rates were calculated based on the percentage of recipients with a given regimen at baseline that were not on the same regimen at the applicable follow-up period.
Our present analysis of retrospective data of U.S. renal transplants provided by the SRTR indicates that the combination of SRL and TAC in kidney transplantation is associated with significantly worse graft survival when compared to TAC in combination with MMF. These results seem most clinically relevant in high-risk transplants, or transplants with a relatively shorter life expectancy. In fact in the deceased-donor and even more so in expanded criteria donor grafts, we observed significant clinically relevant differences in both short- and long-term follow-up. The interpretation of these results has to be done cautiously, considering the inherent limitations of retrospective database analyses, and must include evidence accumulated from prospective studies. Both prospective randomized studies and retrospective data have their own inherent biases and potential for misinterpretations. Prospective clinical studies are clearly the gold standard in the evaluation of any therapy, whereas retrospective studies can provide only ancillary evidence to either confirm or stimulate prospective studies. As prospective clinical trials are conducted in controlled circumstances and in selected patient populations, retrospective database studies can supplement these with data from global clinical practice and in nonselected populations. In no instance though should retrospective data be used as the only piece of evidence when trying to evaluate the efficacy of medication regimens.
Both epidemiologic studies and clinical investigations show that the TAC affords better renal allograft function than CsA, with excellent immunologic protection (10,11). Observations such as these and clinical experience have doubtless contributed to the wide acceptance of TAC/MMF as the standard regimen in most transplant centers (19).
The perception of the lower nephrotoxicity of TAC along with the renal function concerns about the CsA/SRL combination, also led to increasing use of TAC in combination with SRL (5,13–15).
Recently, Light and Lawsin (20) described two cases where the TAC/SRL combination was associated with acute renal dysfunction in two otherwise low-risk live-donor recipients. This report suggested the potential for synergistic nephrotoxicity with the TAC/SRL combination may obtain in just the same manner as in the CsA/SRL combination.
In animal models of kidney transplantation, SRL clearly compounds injurious effects on the kidney by impairing the ability of the kidney to recover from ischemia reperfusion injury (21), and in experimental ischemic acute renal failure SRL impairs recovery from tubular injury. Potential mechanisms include impaired recovery of the tubular epithelium from ischemic injury due to blockade of the cell cycle, a pathway felt to be of importance in tubular regeneration (22). In a different study, in both the uninephrectomy and transplantation models, SRL used alone was associated with tubular injury and interstitial fibrosis to the same extent as TAC. This finding is of interest in that SRL is largely perceived as an agent with minimal intrinsic nephrotoxicity. In particular, this finding was most apparent in the transplantation model, which is inextricably tied to ischemia-reperfusion injury (23).
The tubular injury associated with the calcineurin inhibitors use is well known (24,25). As SRL could potentially impair the ability of the kidney to recover from calcineurin inhibitor induced tubular injuries as well, this effect of SRL should apply to both CsA and TAC.
In fact, in a retrospective study of patients treated with TAC or CsA in combination with SRL, who had signs of deteriorating renal function, withdrawal of SRL and substitution with MMF led to improvement in graft function and blood pressure (26), suggesting a possible synergistic nephrotoxicity between SRL and either CsA or TAC. In this study, the fraction of patients responding with improvements in renal function after SRL withdrawal were equivalent between CsA and TAC, with each group demonstrating a significant improvement in GFR after SRL withdrawal and substitution with MMF.
The only randomized prospective study reported by Mendez et al., comparing SRL and MMF in combination with TAC, provides currently 1-year follow-up data (18). In this multicenter study enrolling 361 renal transplants, patients were randomized 1:1 to receive corticosteroids and either TAC + SRL (N = 185) or TAC + MMF (N = 176). Renal function was significantly worse in the patients randomized to TAC + SRL, with 20% versus 11.2% of patients' serum creatinine exceeding 2.0 mg/dL at 1 year after transplantation. There was also a trend toward worse graft survival in the patients randomized to TAC + SRL as compared to TAC + MMF (91% vs. 94%, p = 0.22) (18). Our retrospective analysis produces almost identical 1-year graft survival rates when compared to that in the Mendez prospective study (TAC + SRL, 91.8% vs. TAC + MMF, 94.2%) but yielded a statistically significant difference because of the larger study sample (p < 0.01). In our retrospective analysis, we documented a progressive separation of the survival curves over time, with a clinically significant difference at 3 years after transplantation (TAC + SRL, 80.3% vs. TAC + MMF, 85.9%, p < 0.001).This difference in graft survival was similar in magnitude to the previously shown difference between CsA + SRL and CsA + MMF (9), in both the univariate analysis (Figure 1) and the multivariate analysis (Table 5).
It is important to note that these differences in graft survival develop gradually over time as opposed to limited to a very early effect, which in a retrospective study could denote the possibility of a selection bias. In addition, the progressive continued divergence of the survival curves might support the possibility of a nephrotoxic mechanism as the principal basis underlying the observed difference in graft survival. Also, our data indicate that the effect is significantly stronger in more vulnerable grafts with pre-existing damage such as deceased donor grafts and even more so in ECD kidneys. High-risk grafts such as ECD kidneys likely represent a group of kidneys with less functional nephron mass, less regenerative ability and decreased functional reserve. Thus, these kidneys may be particularly vulnerable to nephrotoxic insults. While some have speculated that potentiation of CNI nephrotoxicity may be less with TAC than with CsA, this analysis demonstrates that at least in terms of graft survival this deleterious association is of equal magnitude and would strongly suggest that the potentiation of nephrotoxicity may also be of equal magnitude.
Another parallel between our retrospective U.S. population study and the Mendez study was the similar early acute rejection rates between the regimens. In the Mendez study, that had been powered for the primary endpoint of acute rejection at 6 months, 11.4% versus 13% of patients had acute rejection in the SRL + TAC versus MMF + TAC at 6 months after transplantation (p = 0.64) (18). In our study we found very similarly low rates and no difference of acute rejection rates between SRL + TAC versus MMF + TAC in both the univariate and multivariate analysis. Interestingly these data confirm the hypothesis that acute rejection is not a good surrogate endpoint for long-term graft survival (27).
As could be expected according to the worse outcomes, drug regimen discontinuation rates, as measured by the change or substitution in either drug from baseline, were substantially higher in the SRL + calcineurin inhibitor combination regimens, and lowest in the TAC + MMF group. This was another finding of the prospective Mendez study, where patients on SRL + TAC experienced a significantly higher drug regimen discontinuation rate compared with patients on MMF + TAC (18). Because of the high discontinuation rate and the unavailability of data about the reasons for a switch, intent-to-treat analysis, based on the drug regimen administered at discharge from the hospital, is the only viable option to assess outcomes from this type of retrospective data.
There are several potential biases in this type of retrospective registry analysis. Higher-risk patients might have been selected for the regimen combination that was associated with worse graft survival rates. However, when this potential bias was investigated in the context of immunological risk factors for graft loss available from the database, this did not seem to be the case. In addition, acute rejection rates within the first 6 and 12 months were remarkably similar between the regimen groups and in particular between the TAC/SRL and the TAC/MMF regimens. It is reassuring also that similar results were obtained in all sub-group analyses (Table 3), reaffirming that demographic variables important for graft loss were not the sole drivers of a potential selection bias. Moreover, when correcting in the multivariate analysis for differences in baseline variables of potential impact on graft survival, the differences observed in the univariate analysis were confirmed.
Even though some inferences and hypotheses can be generated about the possible mechanisms of the increased graft loss in the SRL + calcineurin inhibitor groups, our present analysis does not allow us to come to any conclusions in this regard.
In addition, no conclusions about alternative dosing strategies involving these agents with later introduction of Rapamycin during the post-transplant course can be drawn from this data. Importantly, the database used does not provide any dosing or drug concentration data and, for that reason, any associations presented in this paper, can be extrapolated only to the pattern of use of these drug combinations during the timeframe analyzed. The results displayed in this paper can, therefore, be interpreted only as an expression of the clinical usage of these drug combinations during this time period.
On the other hand, the 1-year results presented in our study, are mirrored closely by the data of Mendez randomized prospective study. When using retrospective data, as in our analysis, it is important that findings have some pathophysiological or clinical correlate, provide another assurance that the data do not represent a selection bias or artifact. The final verdict about the validity of our conclusion will be known only when the 3-year results from the Mendez study will be available.
In summary, since the growing concerns about the CsA + SRL combination, there has been a significant increase in the use of TAC with SRL. National data indicate that TAC + SRL as compared to TAC + MMF is associated with significantly worse graft survival rates in all subgroups of patients, but particularly in high-risk transplants, which might show earlier susceptibility to nephrotoxic insults. The magnitude of this effect is similar to the previously described deleterious effect of CsA + SRL. These results have to be interpreted in the context of the inherent limitations of any retrospective database analysis and evaluated in context with the data from prospective clinical trials.
The data reported here have been supplied by the University Renal Research and Education Association (URREA) as the contractor for the Scientific Registry of Transplant Recipients (SRTR). The interpretation and reporting of these data are the responsibility of the author(s) and in no way should be seen as an official policy of or interpretation by the SRTR or the U.S. Government. IRB approval or exemption determination is the responsibility of the authors as well. We would like to express our appreciation to Suzanne C. Johnson who has helped with the editing and review of the paper.