- Top of page
- Materials and Methods
Calcineurin inhibitor (CNI) withdrawal has been used as a strategy to improve renal allograft function, however, it also carries risk of acute rejection. We conducted a systematic review of randomized trials that involved CNI withdrawal from a sirolimus-based immunosuppressive regimen. The search strategy yielded six trials (n = 1047 patients) reported in eight publications. CNI withdrawal from sirolimus-based therapy, was associated with an increased risk of acute rejection (risk difference, 6%; 95% CI 2–10%, p = 0.002) but a higher creatinine clearance (mean difference, 7.49 mL/min; 95% CI 5.08–9.89 mL/min, p < 0.00001) at 1 year compared to continued CNI and sirolimus therapy. Graft loss (relative risk, 0.87; 95% CI 0.46–1.64, p = 0.66) and death (relative risk, 0.88; CI 0.40–1.96, p = 0.76) were similar in both groups at 1 year. Hypertension was significantly reduced in the CNI withdrawal group (relative risk, 0.56; 95% CI 0.40–0.78, p = 0.0006). CNI withdrawal from sirolimus-based therapy is associated with an increased risk of acute rejection in the short term with a significant improvement in renal function and a reduction in hypertension. Longer follow-up is needed to determine if these changes will result in a significant improvement in patient and graft survival.
- Top of page
- Materials and Methods
Calcineurin inhibitors (CNIs), cyclosporine and tacrolimus, are potent medications that have become the cornerstone of immunosuppressive therapy following kidney transplantation. The early pivotal trials with cyclosporine showed a significant reduction in acute rejection and improvement in 1-year allograft survival (1,2). Subsequent studies with tacrolimus showed a significant reduction in acute rejection rates (3,4). Recent studies have shown that over 85% of kidney transplant recipients receive either cyclosporine or tacrolimus as part of their maintenance immunosuppressive regimen (5).
Although the era of CNI use brought about dramatic improvements in allograft survival, concern about acute and chronic nephrotoxicity quickly emerged (6,7). In order to overcome the long-term toxicity associated with immunosuppression, withdrawal strategies involving different medications has been attempted (8,9). Individual studies of CNI withdrawal have produced conflicting results (8). In a systematic review of 13 trials, cyclosporine withdrawal following renal transplantation was associated with a significant increase in acute rejection but no difference in allograft survival (8). However, in all of these trials, azathioprine and prednisone were used for immunosuppression after cyclosporine was withdrawn (8).
Sirolimus is a newer immunosuppressive agent whose mechanism of action differs from that of calcineurin inhibitors. Two large trials showed that combination of sirolimus with cyclosporine resulted in a significant reduction in acute rejection compared to either azathioprine or placebo (10,11). However, in both of these trials the sirolimus treated patients had a significantly lower creatinine clearance compared to the placebo or azathioprine groups. This finding suggested that sirolimus could potentiate the known nephrotoxicity of calcineurin inhibitors. A similar reduction in creatinine clearance was found when sirolimus was combined with tacrolimus (12). The finding of enhanced nephrotoxicity has prompted investigators to evaluate sirolimus in combination with low dose CNI as well as study CNI withdrawal in an effort to improve renal function (13). Immunosuppressive withdrawal strategies often carry a risk of acute rejection, which may or may not impact long-term graft survival (8). The potential risks and benefits of CNI withdrawal from a sirolimus-based regimen are not clearly known. Accordingly, the objective of this study was to systematically review all trials evaluating CNI withdrawal from a sirolimus-based immunosuppressive regimen in kidney transplant recipients. Endpoints to be studied include acute rejection, renal function, blood pressure, lipid parameters, patient and graft survival.
- Top of page
- Materials and Methods
The search strategy identified 162 articles from Medline, 272 from Embase, 61 from Cochrane Central, 2 from journal hand-search and 1 from conference proceedings hand-search (Figure 1). Of these, 12 were potentially eligible and reviewed as full articles (13,17–27). Four studies were excluded (13,17–19) for the following reasons: review article (n = 1), control group did not receive sirolimus (n = 1) and no outcome of interest (n = 2). Of the eight studies that met the inclusion criteria, three articles reported 1-, 2- and 3-year follow-up data of the same trial (Rapamune Maintenance Regimen Study) (22–24). The 1-year follow-up report (22) was used for the primary analysis and the 3-year report (24) was used in sensitivity analyses. Therefore, the primary analysis included six trials involving 1047 patients. There was 100% agreement among reviewers about the included studies.
Table 1 summarizes the baseline characteristics of the studies included in the analysis. Only one trial was performed in the United States (20) with the remainder from Canada, Europe and Australia (21,22,25–27). Five trials were published as full articles in peer-reviewed journals (20–22,25,26) while only one report was in abstract form (27). Three trials had a sample size of less than 100 patients (21,25,26). Five studies used cyclosporine (20–22,25,27) and only one used tacrolimus (26) as the CNI. Most trials involved low risk patients with a large proportion of Caucasians, very few diabetics and the majority undergoing primary transplantation (Table 1). Planned antibody induction, delayed graft function or highly sensitized patients were specifically excluded in some of the trials (Table 1). Four studies reported specific eligibility criteria to proceed with CNI withdrawal (20–22,26). Most trials required that the patient had adequate and stable renal function with no significant acute rejection in the preceding month (20–22,26). Calcineurin inhibitor withdrawal was completed in 75 to 93% of randomized patients (Table 1). The methodological quality scores were generally low with a median value of 2 (Table 1). None of the six trials was reported as double blind. Table 2 outlines the target drug levels for sirolimus, cyclosporine and tacrolimus as well as lists the other immunosuppression used in the six included trials.
Table 1. Baseline characteristics of studies included in analysis. Figures are given for withdrawal/control group as appropriatea
|Region||Europe, Australia, Canada||Europe, USA||United Kingdom, Ireland||Italy||United Kingdom||Spain|
|Primary transplant (%)||90/92||100/100||NR||NR||NR||89/93|
|Mean recipient age (years)||45/46||45/45||45/45||40/47||NR||42/47|
|Deceased donor (%)||88/88||100/100||85/91||100/100||NR||100/100|
|Mean donor age (years)||42/44||NR||NR||40/46||NR||41/43|
|Time of randomization post-transplantation||3 months||Within 48 h||3 months||3 months||3 months||Within 24 h|
|Criteria for calcineurin inhibitor withdrawal ||Cr < 400; absence of severe rejection in 4 weeks||Stable function; absence of rejection in 3 weeks||Cr < 400; absence of severe rejection||NR||NR||Stable function; absence of rejection in 3 weeks|
|Exclusion criteria||Planned antibody induction||DGF > 7 days, repeat transplant or live donor||Planned antibody induction||NR||NR||PRA > 50, repeat transplant if early first loss|
|Time of withdrawal post-transplantation (months)||3||2||3||3||3||3|
|Completed withdrawal (%)||93||78||81||NR||NR||75|
Table 2. Target drug levels and other immunosuppression of the studies included in the analysisa
|Study||Target sirolimus levels (ng/mL)||Target calcineurin inhibitor levels (ng/mL)||Other immunosuppression|
|Before withdrawal||After withdrawal||Before withdrawal||After withdrawal|
| Withdrawal||10–20||10–20||100–175 then 100–150||N/A||Induction: NR Both groups: prednisone|
| Control||Fixed dose (6 mg load then 2 mg/day)||Fixed dose (6 mg load then 2 mg/day)||200–400 then 200–300||150–250|| |
| Withdrawal||4–12||8–16||200–400 then 125–250||N/A||Induction: not used Both groups: prednisone|
| Control||4–12||8–16||200–400 then 125–250||50–100|| |
| Withdrawal||>5||20–30||200–400 then 150–300||N/A||Induction: Not used Both groups: prednisone|
| Control||>5||>5||200–400 then 150–300||75–200|| |
| Withdrawal||Fixed dose (15 mg load then 2 mg/day)||10–15||150–250||N/A||Induction: NR Both groups: prednisone|
| Control||Fixed dose (15 mg load then 2 mg/day)||Fixed dose (2 mg/day)||150–250||150–250|| |
| Withdrawal||4–12||8–16||125–250||N/A||Induction: NR Both groups: prednisone|
| Control||4–12||8–16||125–250||50–100|| |
| Withdrawal||8–16||12–20||3–8||N/A||Induction: Not used Both groups: prednisone|
| Control||4–8||4–8||8–12||5–10|| |
Three studies reported acute rejection occurring after CNI withdrawal (20–22). The studies were homogeneous (Q = 0.08, p = 0.96) and all showed a trend toward a higher incidence of acute rejection following CNI withdrawal. The pooled estimate showed that CNI withdrawal significantly increased the risk of acute rejection (relative risk 2.36; 95% CI, 1.34–4.15; p = 0.003). The absolute risk difference in acute rejection was 6.0% (95% CI, 2–10%; p = 0.002) (Figure 2). This implies that CNI withdrawal is associated with an incremental risk of acute rejection of 6%. A sensitivity analysis using the 3-year data instead of the 1-year data from the Rapamune Maintenance Regimen Study (24) had little impact on the relative or absolute risk of rejection given that very few rejections occurred after the first year (relative risk 2.04; 95% CI, 1.2–3.46; p = 0.009 and absolute risk difference 6%; 95% CI, 2–10%; p = 0.006).
Figure 2. Absolute risk difference of acute rejection after calcineurin inhibitor withdrawal compared to the control group receiving combined calcineurin inhibitor and sirolimus therapy. RD, risk difference; 95% CI, 95% confidence interval.
Download figure to PowerPoint
All six of the included trials reported the total rejection rate from the time of transplantation to the end of follow-up. The studies were homogeneous (Q = 3.4, p = 0.64) and most showed a nonsignificant trend toward more overall acute rejection in the CNI withdrawal group (Figure 3). Pooling of the trials showed that the overall risk of acute rejection (includes rejection occurring before and after CNI withdrawal) was increased in the withdrawal group (relative risk 1.59; 95% CI, 1.2–2.1%; p = 0.001). The CNI withdrawal group had an 8% (95% CI, 3–12%; p = 0.0006) absolute increase in total acute rejection since transplantation compared to the control group (Figure 3). A sensitivity analysis using the 3-year data from the Rapamune Maintenance Regimen Study (24) had no significant impact on the risk of total rejection following transplantation (relative risk 1.55; 95% CI, 1.18–2.04; p = 0.002). Similarly, the overall risk of rejection following transplantation did not significantly change if the analysis was repeated using the five studies (20–22,25,26) published in a peer-reviewed journal (relative risk 1.48; 95% CI, 1.09–2.02; p = 0.01). A sensitivity analysis involving only trials that used cyclosporine (i.e. the one tacrolimus trial (26) was excluded) did not significantly change the relative (relative risk 1.56; 95% CI, 1.17–2.07, p = 0.003) or absolute (absolute risk difference 7%; 95% CI, 3–12%, p = 0.002) risk of acute rejection. In the three studies (20–22) that reported acute rejection before and after withdrawal, there was no increased risk of rejection prior to CNI withdrawal (relative risk 1.0; 95% CI, 0.64–1.56; p = 1.0).
Figure 3. Absolute risk difference of acute rejection since transplantation for the calcineurin inhibitor withdrawal group compared to the control group receiving combined calcineurin inhibitor and sirolimus therapy. RD, risk difference; 95% CI, 95% confidence interval.
Download figure to PowerPoint
Graft and patient survival
All six of the included trials reported both patient death and allograft loss. There was no evidence of heterogeneity for the outcome of graft loss (Q = 2.42, p = 0.79) or patient death (Q = 0.57, p = 0.99). CNI withdrawal had no significant effect on the relative (relative risk 0.87; 95% CI, 0.46–1.64; p = 0.66) (Figure 4) or absolute risk of graft loss (absolute risk difference 0%; 95% CI, −3–2%; p = 0.32) at 1 year. The results did not change when the study with only 6-month follow-up (21) was included. When a sensitivity analysis using the 3-year data from the Rapamune Maintenance Regimen Study (24) was performed, there was a trend toward decreased graft loss in the CNI withdrawal group (relative risk 0.70; 95% CI, 0.45–1.10; p = 0.12). This trend was more pronounced when the trial reported in abstract form (27) was excluded from analysis (relative risk 0.69; 95% CI, 0.43–1.05; p = 0.08). CNI withdrawal had no effect on patient death (relative risk 0.88; 95% CI, 0.40–1.96; p = 0.76).
Figure 4. Relative risk of graft loss since transplantation for the calcineurin inhibitor withdrawal group compared to the control group receiving combined calcineurin inhibitor and sirolimus therapy. RR, relative risk; 95% CI, 95% confidence interval.
Download figure to PowerPoint
Creatinine clearance was reported by all studies; however, one study (25) only reported median and inter-quartile range and thus could not be included in the pooled analysis. We estimated the mean difference in creatinine clearance between the withdrawal and control group at the end of follow-up. There was no evidence of heterogeneity between the studies (Q = 2.58, p = 0.63). CNI withdrawal was associated with a significant increase in the creatinine clearance (weighted mean difference 7.49 mL/min; 95% CI, 5.08–9.89; p < 0.00001) (Figure 5). A sensitivity analysis involving only trials that used cyclosporine (i.e. the one tacrolimus trial (26) was excluded) did not rsignificantly change the effect on creatinine clearance (weighted mean difference 7.78 mL/min; 95% CI, 5.31–10.24; p < 0.00001). Similarly, serum creatinine, which was reported in four studies (20–22,27) was significantly lower in the CNI withdrawal group (weighted mean difference −0.19 mg/dL; 95% CI, −0.28 to −0.10; p < 0.0001) compared to control arm at the end of follow-up.
Figure 5. Weighted mean difference in creatinine clearance since transplantation for the calcineurin inhibitor withdrawal group compared to the control group receiving combined calcineurin inhibitor and sirolimus therapy. WMD, weighted mean difference; 95% CI, 95% confidence interval.
Download figure to PowerPoint
Blood pressure, lipids and post-transplant diabetes mellitus
Three studies reported on the proportion of patients with hypertension (21,22,25) and there was no evidence of heterogeneity between the studies (Q = 2.92, p = 0.23). CNI withdrawal was associated with a significant reduction in the risk of hypertension (relative risk 0.56; 95% CI, 0.40–0.78; p = 0.0006). Mean systolic blood pressure was reported in three studies (20,22,26) and mean diastolic blood pressure was reported in two studies (22,26). There was no evidence of study heterogeneity for either systolic blood pressure (Q = 0.01, p = 1.0) or diastolic blood pressure (Q = 0, p = 1.0). CNI withdrawal was associated with a significant reduction in both systolic (weighted mean difference −7.02 mmHg; 95% CI, −10.25 to −3.79; p < 0.0001) and diastolic blood pressure (weighted mean difference −3.2 mmHg; 95% CI, −5.29 to −1.11%; p = 0.003).
Four studies reported total cholesterol and triglycerides (20–22,25). There was significant heterogeneity for the outcome of total cholesterol (Q = 63.1, p < 0.001) as well as for triglycerides (Q = 16.9, p < 0.001). Pooling of the studies was therefore conducted using a random effects model. CNI withdrawal was not associated with any significant effect on total cholesterol (weighted mean difference 0.53 mmol/L; 95% CI, −0.26 to 1.31; p = 0.19) or triglycerides (weighted mean difference 0.17 mmol/L; 95% CI, −0.33 to 0.68; p = 0.5). However, there was a trend toward more statin use in the CNI withdrawal group (relative risk 1.09; 95% CI, 0.97–1.21; p = 0.14) in the two studies that reported medication usage (20,22). The use of fibrates was not significantly different between the two groups (relative risk 0.95; 95% CI, 0.68–1.31; p = 0.73).
Four studies reported incidence of post-transplant diabetes mellitus (20,22,25,26). The studies were homogeneous (Q = 1.85, p = 0.60). There was no significant effect of CNI withdrawal on the risk of post-transplant diabetes mellitus (relative risk 0.93; 95% CI, 0.49–1.78; p = 0.83).
- Top of page
- Materials and Methods
This systematic review has demonstrated that compared to the continued use of a CNI and sirolimus, CNI withdrawal early post-transplantation is associated with a significant increase in the risk of acute rejection. However, despite the increase in rejection the strategy of CNI withdrawal was associated with a significant improvement in creatinine clearance, less hypertension and no unfavorable effect on patient or graft survival.
The incremental rejection risk from CNI withdrawal found in this study was slightly lower than a previous meta-analysis on cyclosporine withdrawal (8). Kasiske et al. found that patients undergoing cyclosporine withdrawal had an 11% increase in acute rejection compared to the control patients (8). They also found that the studies were heterogeneous and smaller studies tended to a report of higher risk of rejection after cyclosporine withdrawal (8). In our analysis, the studies were clinically homogeneous with respect to acute rejection and a relationship with sample size was not obvious. The higher rejection risk in the Kasiske meta-analysis may also reflect the era of the included trials and the other immunosuppression (azathioprine) used. Nonetheless, we still found that CNI withdrawal was associated with an increased risk of rejection even when a more potent immunosuppressant, such as sirolimus was used instead of azathioprine. In addition, most of the studies in this analysis included relatively low-risk patients. Had more high-risk patients been included, the incremental risk of rejection might have been even greater. This suggests that patients undergoing CNI withdrawal require close monitoring for rejection and a degree of caution is required before generalizing this strategy to higher risk populations.
Despite the increased risk of rejection, we found that CNI withdrawal had no deleterious effect on graft survival. Our sensitivity analysis using 3-year follow-up data even suggested a trend toward improved graft survival in the withdrawal arm. This finding was similar to previous trials of CNI withdrawal. Studies with short duration of follow-up tended to show that cyclosporine withdrawal was associated with an increased risk of allograft failure, whereas studies with more than 48-months follow-up tended to show a beneficial effect on graft outcome (8). It is not clear why the increased rate of acute rejection does not seem to negatively effect graft survival in the setting of cyclosporine withdrawal. It may be that the rejection episodes are mild and easily treated in the setting of controlled immunosuppression withdrawal. In the five studies that reported rejection grade in this analysis, only one Banff III acute rejection was reported and the majority of rejection events were mild (21,22,25–27).
We found that CNI withdrawal was associated with a significant increase in the creatinine clearance. This improvement in renal function occurred despite the increase in rejection noted above. This suggests that reducing the nephrotoxicity of CNIs may be more advantageous to overall function than a rejection episode in the setting of medication withdrawal. Experimental data have shown that a pharmacokinetic interaction with sirolimus leads to increased blood and renal tissue levels of cyclosporine, thereby increasing CNI toxicity (28). Meier–Kriesche et al. recently showed that the combination of cyclosporine and sirolimus was associated with worse 4-year graft survival (74.6% vs. 79.3%) compared to the combination of cyclosporine and mycophenolate mofetil (29). Thus, the improvement in renal function found in our study is consistent with the known nephrotoxicity of the sirolimus–CNI combination. Histological data from two studies included in this analysis appear to support the functional improvement we have shown (19,25). In both of these studies, the incidence of new cases of chronic allograft nephropathy was significantly lower in the CNI withdrawal group (19,25). In addition, one study showed that the vascular lesions were significantly more severe in the group that remained on the CNI (25). Although long-term data are currently lacking, it appears that early functional and structural changes are seen with CNI withdrawal from a sirolimus-based regimen that may prove beneficial to long-term allograft survival.
Recent studies have found the 1-year serum creatinine (30–33) and creatinine clearance (34,35) to be strong predictors of long-term graft survival. Hariharan et al. showed that a 1.0 mg/dL increase in the serum creatinine at 1 year was associated with a 63% increased risk of graft failure (30). Using this data, the difference in serum creatinine of 0.19 mg/dL we found, would result in a 12% reduction in graft failure with CNI withdrawal. Following renal transplantation, the average decline in creatinine clearance is approximately 1.5 mL/min/year (36,37). Therefore, the increase in creatinine clearance of 7.49 mL/min found in this analysis is clinically important. Such an increase in function would be expected to improve graft survival by several years (36,37).
The effect of CNI withdrawal on cardiac risk profile is important, given that cardiovascular disease is the most common cause of death following renal transplantation (38,39). We have found that the effect of CNI withdrawal on total cholesterol and triglycerides was heterogeneous across studies. One study (21) showed a trend toward lower cholesterol and triglyceride levels in the withdrawal arm, while the other studies showed a trend toward increased lipid levels (20,22,25). Different target levels for sirolimus may help explain some of the variability in the lipid levels. In addition, the use of lipid lowering medications was reported in only two studies. Differences in the use of statins and fibrates between the withdrawal arm and the control arm could also explain some of the heterogeneity. In contrast, the effect of CNI withdrawal on blood pressure was uniform across the studies in this analysis. The lower systolic and diastolic blood pressures were both statistically significant and clinically meaningful. We found that the systolic blood pressure was 7.02 mmHg lower in the CNI withdrawal group compared to the control patients. Mange et al. showed that for every 10 mmHg increase in the systolic blood pressure the rate of renal allograft survival fell by 15% (40). Although unproven in this situation, the effect of CNI withdrawal on blood pressure may prove beneficial in the long term.
Limitations to this study should be noted. First, there were only six trials that could be included in the analysis. The number of available trials is a potential limitation for any meta-analysis; however, the clinical outcomes were homogeneous, which gave confidence to the pooled results. Second, none of the studies was blinded and therefore diagnostic suspicion bias could have occurred contributing to the increased risk of rejection seen in the CNI withdrawal group. This bias could have occurred if investigators felt that the withdrawal arm should be monitored more closely and undergo biopsy for smaller changes in renal function. Third, only one study examined the withdrawal of tacrolimus (26). Although the change in renal function and risk of acute rejection in this study was in the same direction as the cyclosporine trials, no firm conclusion can be made about the effect of tacrolimus withdrawal from a sirolimus-based therapy. Finally, this analysis generally involved trials with low risk patients who remained stable in the month before withdrawal. This limits the generalizability of our findings to other potentially higher risk populations.
In conclusion, CNI withdrawal from a sirolimus-based therapy is associated with an increased risk of acute rejection with a significant improvement in renal function, reduction in hypertension and no harmful effect on short-term graft survival. Based on this analysis, however, no firm conclusion can be drawn on long-term outcomes. Longer follow-up of the studies in this analysis is needed to determine if this strategy, which has important short-term risks and benefits, will result in a significant improvement in long-term patient and graft survival.