Early Withdrawal of Calcineurin Inhibitors and Rescue Immunosuppression with Sirolimus-Based Therapy in Renal Transplant Recipients with Moderate to Severe Renal Dysfunction


* Correspondonding author: R. K. Wali, rwali@medicine.umaryland.edu


Mammalian Target-of-Rapamycin inhibitors (mTOR inhibitors) can be used to replace the calcineurin inhibitors (CNIs) to prevent progression in chronic kidney disease (CKD) following organ transplantation. Discontinuation of tacrolimus in 136 recipients of kidney transplants with progressive renal dysfunction significantly decreased the rate of loss of estimated glomerular filtration rate (eGFR, mL/min/1.73 m2) (pre-intervention vs. post-intervention slopes, −0.013 vs. −0.002, p < 0.0001). Discontinuation of tacrolimus was associated with a sustained and significant improvement in graft function (pre-eGFR vs. post-eGFR; 26.0 ± 1.1 vs. 47.4 ± 2.1, p < 0.0001) in 74% of patients. This intervention was ineffective if the mean and (median) values of creatinine (mg/dL) and eGFR were 3.8 ± 0.2 (3.4) and 18.4 ± 1.9 (22.4), respectively, at the time of conversion therapy. During the follow-up (range, 1.5–34.6, months), a total of 13 patients had their first acute rejection following the conversion therapy, an annual incidence of less than 10% and none of these episodes resulted in graft loss. The salutary effects of sirolimus therapy following discontinuation of tacrolimus in patients with moderate to severe graft dysfunction due to allograft nephropathy even in high-risk patients improves kidney function and prevents acute rejection.


Due to advances in organ preservation, surgical techniques and immunosuppressive agents, including calcineurin inhibitors (CNIs) such as cyclosporine and tacrolimus, the 1-year patient and graft survival following organ transplantation have reached a zenith (1–3). However, this progress has been marred by the growing incidence of chronic kidney disease (CKD) in the recipients of kidney allografts (4–6) and other solid organ transplants (7). One of the several factors implicated in the development and progression of renal dysfunction after organ transplantation could be dose- and time-dependent CNI nephrotoxicity. The role of CNI-related nephrotoxicity was first elucidated in the recipients of heart transplants (8,9). Since then a growing body of evidence has demonstrated that prolonged use of CNI has a negative impact on renal function, whether used in patients with normal kidney function (10), nonrenal solid organ transplants (7) or kidney transplants (11–13).

The progressive deterioration of renal function in the recipients of kidney transplants in the absence of histological features of acute rejection and other diseases is defined as allograft nephropathy (AN) and has also been inaccurately termed as chronic rejection (1,14). Due to the lack of treatment options, the onset of AN can often lead to an inexorable progression to graft failure (1,11,12). Consequently, the rate of graft loss due to AN has remained unchanged for the past decade.

Mammalian Target-of-Rapamycin Inhibitors (mTOR inhibitors) include sirolimus and everolimus. They both bind to the immunophilin, FK506-binding protein-12 (FKBP-12), without affecting the calcineurin activity (15,16). Combination therapy with sirolimus and CNI prevents acute allograft rejection (17,18). Such combination therapy however, also potentiates the different types of CNI-associated toxicity (19,20). Withdrawal of cyclosporine therapy in recipients of kidney transplants with stable function who were on a combination of sirolimus and cyclosporine therapy was associated with an improvement in renal function (21,22), albeit with an increased risk of acute rejection. Improvement in graft function was noted when cyclosporine was replaced with sirolimus either early (23,24) or several years (25,26) after transplantation.

The current study was based on the a priori hypothesis that discontinuation of tacrolimus- and replacement immunosuppression with sirolimus-based therapy would attenuate the renal parenchymal injury associated with long-term use of CNIs in recipients of kidney transplants with declining kidney function due to biopsy proven AN. Another objective of this study was to elucidate the efficacy and safety of combination therapy with sirolimus and mycophenolate mofetil: a novel approach to arrest the progression of allograft dysfunction.

Materials and Methods


Between November 12, 2000, and October 31, 2002, 159 recipients of kidney and kidney–pancreas transplants with progressive renal dysfunction and biopsy proven diagnosis of AN were treated with sirolimus-based therapy.


We examined: (1) slopes of renal function based on the estimated glomerular filtration rate (eGFR, mL/min/1.73 m2) by MDRD formula (27) as well as by serum creatinine (mg/dL); (2) acute rejection and AN scores based on Banff Criteria-1997 and (3) safety analysis including adverse events associated with sirolimus and mycophenolate mofetil combination therapy. The analysis of encrypted and annonymized database was performed with permission from the Institutional Committee on Human Research. Patients on maintenance therapy, consisting of a combination of tacrolimus, mycophenolate mofetil (MMF, CellCept) and/or corticosteroids, were included in the study. Corticosteroid therapy was discontinued in non-African American recipients within 4 weeks following transplantation unless complicated by histological diagnosis of acute rejection.

Allograft biopsies

Renal biopsies (n = 518) were performed before and after intervention and these were generally triggered by a persistent increase (≥20%) in serum creatinine from the baseline. Each patient had a biopsy performed prior to the conversion to sirolimus-based therapy. Similarly, all but two patients had biopsies performed in the post-conversion period. Fifty-five biopsy samples were deemed inadequate for this analysis. The remaining biopsy specimens while on tacrolimus (n = 341) and again following conversion therapy (n = 122) were analyzed cumulatively for each period to overcome the sampling differences and variable time intervals either before or after the conversion therapy. Biopsy specimens were graded for the type of acute rejection and the patterns of injury to the interstitium, tubules and arterioles as described by the Banff Criteria 97.

Intervention: discontinuation of tacrolimus- and initiation of sirolimus-based therapy in combination with myocphenolate mofetil (MMF) and/or steroids

Patients were asked to stop the tacrolimus after taking the second dose of sirolimus (day 2), an overlap of 24 h with the combination of sirolimus and tacrolimus therapy before discontinuing the tacrolimus therapy.

First phase (learning phase):  From November 2000 to September 2001, sirolimus therapy was initiated with a loading dose of 10 mg/day for the first 3 days, followed by 5 mg/day until the trough levels became available (3–5 days). Doses of MMF and/or prednisone therapy were maintained at baseline. However, during this period we observed frequent hematological, gastrointestinal and liver function abnormalities that necessitated the discontinuation of sirolimus before completion of 90 days therapy in [14/64 (22%)] (Figure 1). These patients were excluded from this modified intent-to-treat analysis.

Figure 1.

Numbers of patients with biopsy confirmed diagnosis of chronic allograft nephropathy and worsening renal function in the absence of histological features of acute rejection. The recipients of combination of pancreas and kidney transplants, and also those who were not on mycophenolate mofetil therapy at the time of diagnosis of allograft nephropathy were excluded. Since this is a modified intent-to-treat analysis, only patients who completed more than 90 days of combination therapy (sirolimus and mycophenolate mofetil) were included in this analysis.

Second phase and the current practice:  After September 2001 in-house sirolimus assays became available. Subsequently, the loading dose of sirolimus was decreased to 5 mg/day for 2 days, with further dose adjustments to maintain whole-blood concentration controlled 24-h goal trough level of 12–20 ng/mL. This goal was once again modified in June 2002, to maintain the trough level of 10–12 or 8–10 ng/mL during or beyond the first-12 month's post-transplant, respectively.

Concomitant therapies at the time of initiation of sirolimus-based therapy:  At the time of initiation of sirolimus therapy, patients were also treated with following therapies. Prophylaxis against Pneumocystis jiroveci-carinii infection for 6 months.Lipid-lowering therapy: Statins (atorvastatin 20 mg daily) were initiated in those who were not receiving such therapy at baseline. In addition, a reduced dose of gemfibrozil (600 mg daily) was added if the fasting triglyceride levels exceeded 250 mg/dL on two consecutive occasions.

MMF dose modification: After September 2001, MMF dose was reduced by 25% in those who were taking 1.5 g or more per day at the time of initiation of sirolimus-based therapy.

Patient follow-up

Following initiation of sirolimus-based therapy, detailed laboratory evaluations were performed once a week during the first 4 weeks, every 2 weeks in the second month and once per month thereafter. Albuminuria was determined by a simultaneous measurement of urine albumin and creatinine ratio (UACR) in the early morning spot urine collected during the office visit at the time of initiation of sirolimus therapy. UACR was repeated again at the time of completion of 12-months of follow-up.

Definition of hematological and metabolic abnormalities

Hematologic and metabolic abnormalities were categorized on the a priori definitions based on the literature evidence and standard clinical practice as described in the footnote of Table 6.

Table 6.  Changes in laboratory values (hematology and lipid profiles) from baseline while on tacrolimus therapy and during the follow-up while on sirolimus-based therapy
VariablesBaseline N = 1363 Months N = 134p-Value16 Months N = 125p-Value212 Months N = 111p-Value318 Months N = 80p-Value424 Months N = 57p-Value5
  1. Linear regression analysis was performed for mean changes in WBC, platelet count, hematocrit, cholesterol, LDL, HDL and triglycerides at different time intervals during the follow-up period. Duration of exposure to sirolimus was a significant predictor for a change in the WBC count at month six from the baseline (t = 2.52; p = 0.01), and also a significant predictor of change in the triglycerides at month 3 (t = -3.04; p = 0.003) and month 6 (t =–-2.05; p = 0.04) from their baseline levels.

  2. Hematologic and metabolic abnormalities were categorized on the a priori basis, based on the literature evidence and standard clinical practice; leucopenia: total white cell count was <4000/mm3 and neutropenia as absolute neutrophil count (ANC) < 500 cells/mm3; thrombocytopenia: platelet count <150.000/mm3; anemia: hematocrit (HCT) <37%. Definition of hyperlipidemia: total cholesterol >201 mg/dL, LDL-cholesterol >110mg/dL, HDL-cholesterol <50 mg/dL, or triglyceride level >200 mg/dL. A separate analysis was performed to measure the number of patients within these categories at different time points during the follow-up period.

  3. 1Leucopenia decreased gradually after the first 6 months of sirolimus therapy, 20%, 11% 10% at month 3, 6 and 12, respectively, and persisted in 3% of patients at months 18 and 24. 12/136 (9%) developed neutropenia and responded to standard therapy without requiring changes in either sirolimus or MMF dosage.

  4. 2Thrombocytopenia was present in 26% at baseline, and improved despite the combination of sirolimus and CellCept therapy and the persisted in 18%, 14%, 12% and 10% at months-3, 6, 12 & 18 and 24, respectively. It was tolerated without complications.

  5. 3Anemia was present in 73%, of patients at the time of initiation of the sirolimus therapy and 23% were on erythropoeitin therapy. Erythropoietin use increased to 98% at month-24.

  6. 4Lipidemia and its management: Atorvastatin 20 mg/day was started concomitantly with sirolimus therapy (67%) in those who were not taking any lipid lowering therapy at the time of conversion therapy.

  7. (a) Hypercholesterolemia was present in 59%, and 33% of these patients were on lipid lowering therapy at the time of initiation of the sirolimus therapy. Hypercholesterolemia increased to 74% at month-3, 86% at month-6, followed by a gradual decrease over time. TC levels remained above the goal in 47% at month-24 of follow-up.

  8. (b) Low-density lipoprotein (LDL >110 mg/dL) was present in 51% at baseline line and these levels persisted in 45% at month-3, and improved with statin therapy during the follow-up.

  9. (c) High-density lipoprotein (HDL) (defined by HDL <50 mg/dL): was present in almost all patients and these levels persisted during the follow-up period.

  10. (d) Hypertriglyceredemia was present in 65% of patients at baseline and increased to 94% at month-3, and 95% at month-6. After first 6 months, TG levels continued to decrease slowly during the follow-up period.

  11. Comparisons between baseline and discrete time points were done by paired sample t-test. Baseline vs. 3 months1, 2 months2, 12 months3, 18 months4 and 24 months5.

  12. *p-Value < 0.05 and **p-value < 0.01.

WBC (mean ± SEM)   7.2 ± 0.47   6.6 ± 0.260.55   6.8 ± 0.220.86   6.8 ± 0.280.79   7.3 ± 0.470.49   8.2 ± 0.900.25
Platelet count (mean ± SEM)211.49 ± 7.67216.43 ± 7.840.42220.78 ± 7.390.57233.26 ± 9.170.33 243.07 ± 15.200.15 252.70 ± 25.170.12
Hematocrit (mean ± SEM)    32 ± 0.50    34 ± 1.620.09    35 ± 0.560.00004**    35 ± 0.630.005**    35 ± 0.800.05*    33 ± 1.540.46
Total cholesterol (mean ± SEM)220.5 ± 5.6239.6 ± 5.30.0005*234.2 ± 5.30.01*218.1 ± 2.80.66202.7 ± 2.40.03*203.3 ± 2.70.02*
LDL108 .05 ± 1.84    105 ± 1.200.03* 94.32 ± 1.08<0.0001** 87.92 ± 1.16<0.0001** 89.33 ± 1.53<0.0001** 82.57 ± 1.59<0.0001**
HDL  27.6 ± 0.43  29.1 ± 0.530.0007*  31.3 ± 0.50<0.0001**  32.8 ± 0.43<0.0001**  32.1 ± 0.83<0.0001**  34.0 ± 0.58<0.0001**
Triglycerides 263.5 ± 5.99 275.2 ± 5.940.001** 274.1 ± 5.340.17 216.7 ± 3.72<0.0001** 200.5 ± 4.57<0.0001** 202.6 ± 3.94<0.0001**

Statistical analysis

Estimated glomerular filtration rate was calculated by MDRD formula (27) for each creatinine value obtained in the post-transplant period except for those obtained immediately after surgery (the first 4 weeks and those with the diagnosis of delayed graft function). Individual regression models were fit regressing MDRD to time, based on 5338 and 5920 eGFR values before and after conversion therapy, respectively.

The chi-square test and one-way analysis of variance (ANOVA) were used for discrete and continuous variables, respectively. Group means of a continuous outcome at each time point were compared by the paired t-test or Wilcoxon-Signed Rank test, as necessary. Proportions of a categorical outcome were compared by Chi-square test or Fisher's test, as necessary.

Multiple regression analysis was performed to determine those variables that were significant predictors of outcome that included a change in the mean serum creatinine (mg/dL) and eGFR (mL/min/1.73 m2) from baseline compared to different time intervals at 3, 6, 12, 18 and 24 months, with values set at zero for those with either graft loss or death. Similar analysis was also performed for a change in mean WBC, platelet count, hematocrit, cholesterol, low density and high-density lipoproteins and triglyceride levels at different time intervals. This adjusted analysis accounts for the potential confounders, including age, race, sex, body mass index, diabetes mellitus, baseline urine albumin to creatinine ratio, acute rejection episodes, time to conversion, graft type and tacrolimus level at the time of conversion.

Logistic regression analysis was performed to predict categorical outcomes, including graft loss, patient death and acute rejection in the post-conversion period and the independent variables included all those that were used in the multivariable regression analysis as above.

Effect size (ES) was measured to determine the magnitude of a treatment effect. Effect size was computed as Cohen's d where a positive effect size (responders) represented improvement and a negative effect size represented worsening or no change (nonresponders). Using the equation: {d = SCrb− SCr3/ spooled1}, d = is the effect size, SCrb and SCr3 are mean serum creatinine (mg/dL) at baseline and at month 3, respectively, and spooled is the pooled standard deviation. These two groups were compared using paired t-test for continuous variables and Fisher's test for the categorical variables. In addition, the mean eGFR at baseline and those that completed 12-months of follow-up was calculated with and without (values set at zero) for patients with graft loss and death.

We used SPSS statistical software (SPSS version 13.0; SPSS Chicago, IL) and SAS (SAS version 9.1, Cary, NC) for these analyses. All p-values are two-sided, and a value of 0.05 or less was considered significant. Unless specified otherwise, continuous variables are expressed as mean ± SEM.


Patient characteristics

A total of 159 patients with biopsy-confirmed diagnosis of AN and without acute rejection for 3 months prior to conversion therapy were enrolled. Patients who completed more than 90 days of sirolimus-based therapy (n =136) were included in this analysis (modified intent-to-treat analysis). The description and outcomes of the other 23 patients are provided in Figure 1.

Table 1 describes the baseline demographics of these recipients and their corresponding donors. Fifty-nine percent (80/136) patients were of black ethnicity and 14% were retransplants. The majority (66%) were recipients of deceased donor kidneys as compared to living donor kidneys (p = 0.006). Seventeen percent had their first acute rejection while on tacrolimus therapy, with a cumulative prevalence of a total of 62 episodes while on tacrolimus therapy.

Table 1.  Baseline demographics of recipients and their corresponding donor characteristics
DescriptionAll patients, n = 136
  1. 1DDK vs. living donors (p = 0.006).

  2. 2Deceased donor kidney (DDK) recipients only.

  3. 3A cumulative frequency of 62 episodes of acute cellular rejections of different grades was diagnosed while the patients were on tacrolimus based therapy. 41/62 (66%) of these cases were grade IA or less and were treated with three bolus doses of methylprednisone. Five of these episodes were steroid-resistant and, as in patients with grade IB or higher, were treated with monoclonal antibody (Thymoglobulin or OKT3) followed by corticosteroid taper for 4 weeks. These episodes could have happened more than once in the same patient.

Age (years ± SE)50.5 ± 1.2
Race, n (%)
 Caucasians52 (38)
 Blacks80 (59)
 Asians4 (3)
Gender, n (%)
 Males74 (54)
 Females62 (46)
 Diabetes mellitus, n (%)51 (36)
 Body mass index, mean ± SE28.7 ± 0.54
Graft type1, n (%)
 Deceased donors (DDK)89 (66)
 Living donors46 (34)
 Re-transplantation, n (%)19 (14)
Donor race, n (%)
 Whites85 (63)
 Blacks39 (29)
 Others11 (8)
Donor gender, n (%)
 Males77 (57)
HLA mismatch (MM), n (%)
 A1-MM106 (78)
 A2-MM118 (88)
 DR1-MM120 (90)
 DR2-MM126 (93)
 Cold Ischemia time2 (h)21.5 ± 1.2
 First acute rejection,3 n (%)23 (17)

Renal function as defined by serum creatinine and eGFR

Based on the slopes of renal function assessed by eGFR while on tacrolimus- (pre-) and after discontinuation of the tacrolimus-based therapy (post-), a significant decrease in the decay of renal function was observed [pre-slopes vs. post-slopes (mL/min), −0.013 vs. −0.002, p < 0.0001]. At the time of discontinuation of tacrolimus therapy (considered baseline value), moderate to severe graft dysfunction (mean serum creatinine = 3.7 ± 0.12 mg/dL and mean eGFR = 26.0 ± 1.1 mL/min/1.73 m2) was present. A significant deterioration in graft function in the absence of acute rejection was observed preceding the withdrawal of tacrolimus therapy, [–3 months vs. baseline, creatinine (p = 0.0009) and eGFR (p < 0.0001), Table 2]. Following conversion to sirolimus-based therapy, the improvement in graft function measured by either serum creatinine or eGFR was maintained at all time points during the median follow-up period of 12.8 months (mean ± SE: 14.4 ± 0.6, range: 1.5–34.6 months) (Figures 2A and B, respectively). Inter-group comparisons of graft function by serum creatinine and eGFR at baseline (the time of conversion to cessation of tacrolimus and initiation of sirolimus threpay) compared to 3 months prior to conversion and during the follow-up period are described in detail in Table 2.

Table 2.  Changes in serum creatinine (mg/dL) and eGFR (mL/min/1.73 m2) 3 months prior to conversion and during the follow-up period after the conversion therapy
Variables−3 Months N = 1361−1 Month N = 136Baseline N = 13623 Months N = 1346 Months N = 12512 Months N = 11118 Months N = 8024 Months N = 57
  1. The eGFR mean ± SEM (median) at baseline (n = 111) and at 12 months (n = 111 without censoring, values set at zero for graft loss and death) was 28.8 ± 1.1(26.4) and 39.2 ± 1.8 (41.4), respectively. eGFR mean ± SEM (median) at 12 months (n = 111) with censoring (values missing for graft loss and death) was 47.1 ± 2.2 (50.9).

  2. Inter-group comparisons were made by Wilcoxon-Signed Rank test or paired sample t-test, as necessary.

  3. 1−3 months and −1 month values obtained 3 months and 1 month before the conversion therapy and while the patients were on tacrolimus-based therapy.

  4. 2Baseline is the time of initiation of sirolimus therapy and tacrolimus was discontinued 24 h later.

  5. 395% CI: 95% confidence intervals for the mean difference.

  6. *p-Value < 0.05 and **p-value <0.01

Serum creatinine (mean ± SEM)3.4 ± 0.113.7 ± 0.123.7 ± 0.113.5 ± 0.133.2 ± 0.143.1 ± 0.152.7 ± 0.122.6 ± 0.12
p-Value −3 months vs. baseline3 months vs. baseline6 months vs. baseline12 months vs. baseline18 months vs. baseline24 months vs. baseline
95% CI3 −0.49; −0.13−0.46; −0.06−0.62; −0.14−0.68; 0.40−0.96; −0.40−1.05; −0.45
eGFR (mean ± SEM)29.7 ± 1.326.72 ± 1.126.0 ± 1.132 ± 1.436 ± 1.639.2 ± 1.844.6 ± 1.947 ± 2.1
p-Value −3 months vs. baseline3 months vs. baseline6 months vs. baseline12 months vs. baseline18 months vs. baseline24 months vs. baseline
95% CI3 2.42; 5.094.38; 7.476.85; 10.937.12; 11.9110.68; 16.9912.31; 19.99
Figure 2.

(A) Serum creatinine (mg/dL), (B) eGFR (mL/min/1.73m2). Each value is mean ± SEM. Pre-mon3 and pre-mon1: (3 and 1 months before the discontinuation of tacrolimus) At con: at the time of initiation of sirolimus therapy. Post-mon 3, 6, 12, 18 and 24: while on sirolimus therapy at months 3, 6, 12, 18 and 24, respectively.

Using multiple regression adjusted analysis, the tacrolimus level and the dose at baseline were significant determinants of improvement in graft function during the first 6 months after conversion therapy. In addition, time to conversion to sriolimus-based therapy after transplantation was a significant predictor of an improvement in graft function from baseline at all time points during the follow-up period. The detailed description of this analysis is attached as Appendix A.

Subgroup analysis (Table 3)

Table 3.  Subgroup analysis of patients based on the effect size
Group 1 responders N = 101Group 2 non-responders N = 33
  1. Based on the Effect size: Group 1, ‘responders’ and Group 2, ‘non-responders’ based on the positive or negative effect size, respectively. Paired sample t-test was performed to compare the effect sizes for the continuous variables across the two groups and Fisher's test was performed for the categorical variables.

  2. 1 Two patients died during the first 3 months with functioning graft (motor vehicle accident and following gall bladder surgery).

  3. 2, 3Cumulative frequency of episodes of acute rejection while on tacrolimus and following the conversion therapy, respectively.

Age, mean ± SD (years)51 ± 14 49 ±12T = 0.930.36
Gender, n (%) males52 (51)20 (61%)Fisher's test = 0.830.87
Race, n (%) Whites38 (39%)13 (39%)Fisher's test = 0.00050.98
Body mass index, (Mean ± SD)28 ± 6 29 ± 6T = 0.270.78
Pre-conversion acute rejection episodes, n (%)240 (40%)22 (67%)Fisher's test = 5.740.03
Post-conversion acute rejection episodes, n (%)321 (21)8 (24)Fisher's test = 0.880.65
Time to conversion, mean ± SD (months)17 ± 12 34 ± 22T= 3.580.001
Tacrolimus level, mean ± SD (ng/mL)8 ± 3 7 ± 2T=−2.250.03
eGFR at baseline, mean ± SD (mL/min/1.73 m2)28 ± 1319 ± 8−4.04.0001
Urine albumin creatinine ratio at baseline, mean ± SD (ug/mg)93±49113 ± 661.660.10
Graft loss, n (%)21 (21%)24 (73%)Fisher's test = 29.85<0.0001
Death7 (7%)6 (18%)Fisher's test = 3.180.08

Effect size (ES) was measured in 134 patients to determine the magnitude of treatment effect. Seventy-six percent (101/136) had a positive effect size following discontinuation of tacrolimus therapy. Another 24% (33/136) with a negative effect size had a baseline mean and (median) values of creatinine and eGFR of 3.8 ± 0.2 (3.4) and 18.4 ± 1.9 (22.4), respectively. Pre-conversion lower eGFR levels, frequent episodes of acute rejection, tacrolimus levels and increased time interval from transplant to conversion therapy were all significant determinants in non-responders as described in Table 3.

Histological findings (allograft nephropathy scores as per the Banff 97 Criteria) (Table 4)

Table 4.  Histo-pathological analysis of allograft biopsies for grades of allograft nephropathy (AN) and acute rejection based on Banff Criteria 97 while patients were on tacrolimus (pre-) and following conversion therapy
  1. All but two patients had a biopsy in the post-conversion period. A total of 142 biopsies were performed following the conversion therapy. Twenty of these samples were deemed inadequate for CAN grading. Hence post-conversion data were analyzed for 122 biopsy specimens and matched with corresponding 122 patients for comparisons of CAN scoring.

  2. A total cumulative frequency of 29 episodes of acute rejection developed during the follow-up period. The majority of these episodes 24/29 (83%) were Grade IA or less and 5 episodes were grade IB or higher.

  3. 1Interstitial fibrosis/tubular atrophy, tubular degeneration and arteriolar hyalinosis were graded based on the scores of 0 to 3, with higher scores indicating severe abnormalities.

  4. 2Tubular degeneration was defined by the presence of one or more components of: (a) tubular epithelial cell blebbing, (b) spotty tubular drop outs and (c) tubular micro-calcifications.

  5. 3IA or less (n = 8), IB (n = 3) and IIA (n = 2).

  6. *p Inter-group comparisons were performed by Wilcoxon-Signed Rank test.

Number of biopsies122122 
Interstitial fibrosis/Tubular atrophy1 (mean ± SE)1.01 ± 0.061.99 ± 1.110.064
Tubular degeneration1,2 (mean ± SE)0.95 ± 0.700.75 ± 0.850.013
Arteriolar hyalinosis1 (mean ± SE)0.32 ± 0.040.22 ± 0.030.051
First acute cellular rejection,3 n (%)23/136 (17%)13/136 (9.6%)0.05 

All but two patients had biopsies following conversion therapy. A total of 142 biopsies were performed and 122 were deemed adequate for grading of CAN. Analysis of biopsy specimens while on tacrolimus (n = 341) and following conversion therapy (n = 122) demonstrated a significant improvement in the degree of tubular degeneration and arteriolar hyalinosis (Table 4). On the contrary, interstitial fibrosis and tubular atrophy continued to progress. The major limitation of biopsy analysis is that these biopsies were obtained at different time intervals.

Acute allograft rejection (pre- and post-conversion) (Table 4)

A total of 13 patients developed their first acute rejection following conversion therapy, an annual incidence of approximately less than 10%. A total cumulative frequency of 29 episodes of biopsy confirmed acute allograft rejections were diagnosed during the mean follow-up of 14.6 ± 0.6 months (range: 1.5–34.6 months). The majority of these episodes of acute rejection 18/29 (62%) were classified as borderline or suspicious. None of these episodes resulted in graft loss. Based on the logistic regression analysis, blacks compared to Caucasians (odds ratio (OR), 4.03; 95% CI (confidence interval), 0.95–17.1) and patients with lower sirolimus levels (OR, 1.1; 95% CI, 1.0–1.3) during the first 3 months had an increased risk for acute rejection following conversion therapy.

Urine albumin creatinine ratio (UACR) (urine μg albumin/mg of creatinine)

There was a significant increase in UACR at 1 year (n = 111) following conversion therapy as compared to the baseline (105.8 ± 5.4 vs. 97.6 ± 4.6, p = 0.0004). This difference was not significant after adjusting for the potential confounders, including diabetes and other demographic characteristics (data not shown). However, 66% of our patients were taking different types of angiotensin converting enzyme inhibitors and/or angiotensin receptor blockers at baseline, and they continued this therapy during the follow-up period.

Graft and patient survival (Table 5)

Table 5.  Graft and patient outcomes during the follow-up period while on sirolimus-based therapy
VariablesGraft lossPatient deathFollow-up months1
Odds ratio95% CIp-ValueOdds ratio95% CIp-ValueParameter estimatep-Value
  1. Logistic regression analysis was performed after adjusting for the potential confounders for the categorical outcomes; graft loss (yes/no), patient death (yes/no). Adjusted variables included race (Caucasian) as a referent, recipient age, graft type (living donor as referent), eGFR at baseline (mL/min/ 1.73 m2), acute rejection (yes or no) before sirolimus therapy and time to conversion (months).

  2. Multiple regressions analysis provided parameter estimates or the regression coefficients that explain the variables significance in predicting the continuous outcome for follow-up (months).

  3. 1Follow-up time following conversion therapy: median:12.8 months, range: 1.5–34.6 months.

  4. 2These episodes could have happened more than once in the same patient.

  5. *p-Value < 0.05 and **p-value < 0.01.

Race (Caucasian, referent)2.570.72;; 11.820.291.530.21
Recipient age1.010.97; 1.050.631.111.03; 1.190.006**−0.0070.87
Graft type (living donor, referent)0.910.22; 3.710.891.860.37; 9.390.45−0.940.46
eGFR at baseline (mL/min/1.73 m2)0.730.63; 0.84<0.0001**0.850.75; 0.950.005**0.050.33
Acute rejection2 (yes, referent)0.990.29; 3.370.991.980.44; 8.940.37−0.400.75
Time to conversion therpay (months)1.051.02; 1.080.003**1.031.01; 1.060.02*−0.080.003**

A total of 47 (35%) patients lost their graft and 14/136 (10%) patients died during the follow-up period. Based on the regression analysis (Table 5) baseline eGFR (OR, 0.73; 95% CI, 0.63–0.84; p < 0.0001), time to conversion (OR, 1.05; 95% CI, 1.02–1.08; p = 0.003) and age (OR, 1.11; 95% CI, 1.03–1.19; p = 0.007) were the significant determinants of graft loss.

Changes in laboratory values (hematology and lipid profile) in the post-intervention period (Table 6)

During the first 6 months after initiation of this protocol (learning phase), we observed a significant decrease in the white cell and platelet count and new onset gastro-intestinal symptoms. These events necessitated discontinuation of sirolimus therapy in 14 patients (as described in Figure 1). After modifications in the loading dose of sirolimus and concomitant dose reduction in MMF, these side effects developed less frequently and did not require discontinuation of sirolimus therapy.

Anemia:  Following conversion therapy, nearly 76% of patients required erythropoietin therapy at month-3. However, anemia persisted in more than one-third of patients during the second year of follow-up despite erythropoietin therapy.

Hyperlipidemia:  Total cholesterol and LDL-cholesterol levels peaked at 6 months following conversion therapy despite statin therapy (Table 6). These levels decreased gradually and remained above the goal in less than 50% of patients at the last follow-up visit. Fifty-four percent and another 17% of these patients were on atorvastatin dose of 40 and 80 mg/day, respectively, and others remained on the baseline dose of statins (<40 mg/day). Three patients developed muscle pains and joint stiffness without an increase in muscle enzymes. Atorvastatin in these three patients was replaced with ezetimibe, and it was tolerated without side effects. Suboptimal HDL levels were present in almost all patients at baseline and continued to remain low despite statin therapy. Hypertriglyceridemia (HTG) was present in 65% of this cohort at baseline and in 94% of patients at 6 months following conversion therapy. HTG responded to standard statin therapy except in three patients. These three patients required low dose gemfibrozil therapy as their triglyceride levels increased to more than 500 mg/dL and persisted for two consecutive months. These levels returned to <200 mg/dL and gemfibrozil was discontinued at this stage. The combination of gemfibrozil and statin was tolerated without side effects.

Adverse events following conversion therapy (Table 7)

Table 7.  Adverse events (other than the changes in hematology and lipid profiles) while on sirolimus-based therapy
  1. 1Asymptomatic increases in transaminases resolved in two patients after stopping the poiglitazone therapy. In another patient, transaminases became normal after decreasing the sirolimus dose, and fluctuated in one patient without any untoward effects.

  2. 2Mouth ulcers resolved with a decrease in the dose of mycophenolate mofetil (MMF). One patient did not tolerate more than 500 mg of MMF per day.

  3. 3Skin rash was an accentuation of the ancneiform rash.

  4. 4Hospitalizations for pneumonias: None of these patients required ventilator support and did not necessitate discontinuation of sirolimus therapy.

  5. 5Malignancies: Non-melanotic skin cancers were multi-focal, without deep tissue invasion and were recurrent. Non-small cell lung cancer was treated with lobectomy with remission. Patient with PTLD remains in remission after chemotherapy.

Abnormal liver transaminases (ALT, AST)1, n (%)4 /136 (3)
Mouth ulcers 2, n (%)2/136 (1.5)
Skin rash3, n (%)1/136 (0.7)
Pneumonia4, n (%)6 (5)
Malignancy5, n (%)4 (3)
Non-melanotic skin cancers
Non-small cell lung cancer1
Lymphomas (PTLD)1

Asymptomatic increases in aminotransferases (ALT and AST) was observed in 4/136 (<2.9%) of patients after the learning phase. These levels returned to normal (n = 2). Three patients (2.2%) developed mucocutaneous symptoms. These responded to adjustments in the dose and target goal levels of sirolimus.

Hospitalizations for pneumonia were observed in 6/136 (4.4%) of patients and none of them required ventilator support. The majority of these (n = 4) were community-acquired pneumonias. One patient each developed respiratory syncitial viral (RSV) and cytomegalovirus (CMV) pneumonia. None of these patients required cessation of sirolimus therapy.

Four patients (2.9%) developed different types of malignancies while on sirolimus-based therapy. Two of these were non-invasive, non-melanoma skin cancers treated with local therapy, with recurrence in both patients. One patient developed non-small cell lung cancer and another patient developed Epstein-Barr virus related post-transplant lymphoproliferative disorder (PTLD). Both patients remained disease-free at the time of the last follow-up visit.


Despite remarkable improvements in 1-year patient and graft survival in recipients of kidney transplants (28) the progressive loss of graft function due to AN is the most commonly encountered clinical problem.

Allograft nephropathy encompasses a spectrum of cumulative damage to the allograft from time-dependent immunologic and non-immunologic factors (6). Among the non-immunologic factors, CNI nephrotoxicity is perhaps the most common factor leading to long-term graft damage and progression to graft failure (29–32). This is further supported by the evidence that long-term use of CNI-based therapy leads to deterioration in kidney function, even in recipients of non-renal organ transplants (33).

Withdrawal of CNI therapy in patients with AN has been demonstrated to increase the risk of acute allograft rejection (34,35). However, there are limited therapeutic options to prevent the ongoing CNI toxicity, particularly in patients who are on mycophenolate mofetil-based therapy at the time of onset of AN. Other options in the past included reducing the dose or withdrawal of CNI (35–38) or replacing azathioprine with mycophenolate mofetil (39–41). Following the availability of mTOR inhibitors, it became possible to replace CNI with mTOR inhibitors without increasing the risk of acute graft rejection (23,24,26).

We demonstrated that discontinuation of tacrolimus within 24-h after initiation of sirolomus therapy in patients with progressive graft dysfunction not only improves the graft function but also prevents the acute rejection. This observation is in agreement with other smaller studies (23,25,26). However, our patient population had an advanced graft dysfunction and included high-risk patients at the time of conversion therapy as compared to previously reported series (42).

How can we explain the improvement in renal function?

Exposure to CNIs (either cyclosporine or tacrolimus) results in abnormal renal hemodynamics as well as cumulative and incremental renal histological changes over time even in healthy subjects with normal kidney function (10). We postulate that renal hemodynamic effects of CNI are more pronounced in subjects who have impaired renal function at the time of initiation of CNI therapy.

The cessation of CNI therapy almost invariably leads to pre- and intra-glomerular hemodynamic changes. This is due to abolition of the afferent arteriolar and intrarenal vasoconstriction (43), in combination with a decrease in endothelin levels (44), resulting in an increase in glomerular filtration rate. These hemodynamic effects may be reversible before the onset of arteriosclerosis of the graft blood vessels if CNI is discontinued early after transplantation (45). In addition, these hemodynamic effects after withdrawal of CNI could abrogate the potential ischemic effects in the remaining viable nephrons.

Hence, it is possible that cessation of CNI in patients with compromised renal function could result in the recovery of tubular elements that are still viable but remain ischemic under the influence of CNI therapy. The recovery from ischemic effects could explain the significant improvement in the tubular degenerative changes as well as arteriolar hyalinosis after discontinuing the tacrolimus-based therapy. This could indirectly support the hypothesis that discontinuation of tacrolimus restores the function of those viable nephrons that remain ischemic in the presence of tacrolimus therapy.

Combined use of sirolimus and mycophenolate mofetil (MMF) can potentiate the hematopoietic and gastrointestinal side effects of individual drugs. Modification of MMF dose, particularly in those who are on more than 1.5 g of MMF per day, along with judicious adjustments in the maintenance trough levels of sirolimus, can to a greater extent prevent these side effects. Similarly, the use of sirolimus has been associated with variable effects on different lipid fractions (46). Our analysis of lipid fractions demonstrated an increase in total cholesterol, LDL-cholesterol and triglyceride levels, with a decrease in HDL-cholesterol along with initiation of sirolimus therapy. During the short period of follow-up in this study, there was no increase in cardiovascular morbidity and mortality rates. The long-term cardiovascular effects of sirolimus therapy cannot be ascertained from this study, however.

The use of sirolimus has been associated with variable effects on proteinuria in recipients of kidney transplants with varying degrees of graft dysfunction (25,47). This may be a cause for concern. However, it is not clear if this increase in proteinuria is due to the progression of glomerular disease often present in patients with AN (48) or due to hemodynamic and paracrine effects that occur following cessation of CNIs (43,44). The increase in albuminuria in our series was not significant during the median follow-up of more than 12 months.

Long-term use of immunosuppression therapy increases the risk for infections and malignancies. The rates of non-systemic malignancy and PTLD were consistent with the previously reported UNOS registry data (49,50).


Although randomized controlled clinical trials eliminate the selection bias and treatment indication bias (51,52), there is always the possibility of a different type of a selection bias due to strict inclusion and exclusion criteria. Due to the regulatory oversight and the necessity to ensure smooth approval in phase 3 trials, perhaps phase 4 clinical studies offer the opportunity that the efficacy and the safety of intervention in high-risk populations can be evaluated (53). In a study population such as ours with a significant magnitude of graft dysfunction, neither cohort nor randomized studies can assure the equal distribution of unmeasured confounders. Findings of our study should be viewed in light of limitations of a cohort study and the associated selection bias.

Clinical implications

Notwithstanding these biases, observations gained from this study have several important clinical implications. First, discontinuation of tacrolimus in patients with suboptimal graft function results in an immediate and sustained improvement in renal function, although one-third of our patients developed graft failure within a median of 12.8 (range 1.5–34.6) months of follow-up. However, preservation of graft function is particularly important for those patients who are at increased risk for CNI-related nephrotoxicity (patients with underlying renal dysfunction due to suboptimal quality of the graft, acute rejection or both). The number of recipients with suboptimal graft function will likely continue to increase in the future, due to the increased use of extended criteria donor (ECD) kidneys. Second, sirolimus-based therapy can be safely used even in high-risk patients (African American, re-transplantation, previous acute allograft rejection). Third, the negative impact of tacrolimus on graft survival is time-dependent, CNI-related changes can progress over time and may ultimately be irreversible. It is possible that a substantial reduction in the tacrolimus dose and level could have similar benefits in such patients. Finally, the adverse events associated with the use of sirolimus and mycophenolate mofetil therapy are avoidable with the judicious use of the loading dose of sirolimus, time to time adjustments in the dose and trough levels of sirolimus, concomitant modification of the MMF dose and supplemental preventive strategies for pneumonia and hyperlipidemia. Furthermore, patients with severely advanced CKD (eGFR <18mL/min/1.73 m2) following transplantation apparently did not benefit from discontinuation of tacrolimus-based therapy.

In conclusion, discontinuation of CNI-based immunosuppression in patients with graft dysfunction due to AN could lead to preservation of renal function. The combination of sirolimus and mycophenolate mofetil with or without concomitant use of corticosteroids results in improvement in graft function with a low incidence of acute rejection and tolerable toxicity. Longer follow-up is needed to determine if such an approach ultimately improves the long-term graft survival. Given the high prevalence of CKD in the recipients of renal and other solid organ transplants, there is an urgent need for prospective studies to establish the benefits of non-CNI-based therapy compared to different dosing regimens of CNI-based therapy.


Table Appendix A:.  Multiple regressions analysis for changes in eGFR (eGFR mL/min/1.73 m2) and serum creatinine (mg/dL) 3 months prior to conversion and during the follow-up after conversion therapy
Independent variablesDependent variable (changes in eGFR at different time periods)
Baseline-(−3 months)3 months-baseline6 months-baseline12 months-baseline18 months-baseline24 months-baseline
  1. Multiple regressions analysis for changes in eGFR (eGFR mL/min/1.73 m2) and serum creatinine (mg/dL) 3 months prior to conversion therapy and during the follow-up period after discontinuation of tacrolimus therapy.

  2. Variables in the model included age (years), gender (males), race (Caucasian), body mass index, acute rejection episodes (yes or no), time to conversion (months), graft type (living donor), dose and level of tacrolimus at baseline.

  3. 1PE: point estimates are the values for regression coefficients.

  4. 2p-Value: *p-value < 0.05 and ** p-value < 0.01.

Age (years)0.020.610.
Acute rejection episodes−1.080.38−1.140.47−2.880.
Time to conversion (months)0.030.19−0.090.004**−0.160.0003**−0.060.46−0.230.003**−0.240.04*
Graft type−0.470.71−0.380.811.520.454.110.30−1.720.614.340.33
Tacrolimus dose0.−0.010.970.300.51−0.390.27−0.050.91
Tacrolimus level−0.140.530.870.003**0.970.01*0.030.960.910.121.290.08
 Dependent variable (changes in serum creatinine at different time periods)
Age (years)0.0030.67−0.0030.72−0.020.15−0.0070.59−0.0080.61−0.0030.88
Body mass index0.0060.700.−0.040.19−0.010.72
Acute rejection episodes0.050.810.660.01*0.630.04*0.860.02**0.480.210.520.29
Time to conversion (months)−0.0080.03*0.0020.670.0070.260.0070.390.0040.59−0.0020.85
Graft type0.350.13−0.410.12−0.400.20−0.170.64−0.030.93−0.400.38
Tacrolimus dose−0.020.36−0.010.960.020.630.040.310.
Tacrolimus level0.0050.89−0.050.28−0.070.23−0.080.21−0.090.15−0.140.07


We are indebted to William L. Henrich, M.D. and Frank Calia, M.D., for their continuous support and mentorship as former and current chairmen Department of Medicine, respectively. In addition, our gratitude to all the participating patients and our special thanks to Ms. Geetha Stachowiak for her skillful administrative help.