A Randomized, Controlled Study to Assess the Conversion From Calcineurin-Inhibitors to Everolimus After Liver Transplantation—PROTECT


Prof Dr. med. Hans. Schlitt, hans.schlitt@klinik.uni-regensburg.de


Posttransplant immunosuppression with calcineurin inhibitors (CNIs) is associated with impaired renal function, while mTor inhibitors such as everolimus may provide a renal-sparing alternative. In this randomized 1-year study in patients with liver transplantation (LTx), we sought to assess the effects of everolimus on glomerular filtration rate (GFR) after conversion from CNIs compared to continued CNI treatment. Eligible study patients received basiliximab induction, CNI with/without corticosteroids for 4 weeks post-LTx, and were then randomized (if GFR > 50 mL/min) to continued CNIs (N = 102) or subsequent conversion to EVR (N = 101). Mean calculated GFR 11 months postrandomization (ITT population) revealed no significant difference between treatments using the Cockcroft-Gault formula (−2.9 mL/min in favor of EVR, 95%-CI: [−10.659; 4.814], p = 0.46), whereas use of the MDRD formula showed superiority for EVR (−7.8 mL/min, 95%-CI: [−14.366; −1.191], p = 0.021). Rates of mortality (EVR: 4.2% vs. CNI: 4.1%), biopsy-proven acute rejection (17.7% vs. 15.3%), and efficacy failure (20.8% vs. 20.4%) were similar. Infections, leukocytopenia, hyperlipidemia and treatment discontinuations occurred more frequently in the EVR group. No hepatic artery thrombosis and no excess of wound healing impairment were noted. Conversion from CNI-based to EVR-based immunosuppression proved to be a safe alternative post-LTx that deserves further investigation in terms of nephroprotection.


adverse event


analysis of covariance


biopsy-proven acute rejection


GFR calculated with the Cockcroft-Gault formula


confidence interval




cyclosporine A


calcineurin inhibitor(s)




giga per liter


hepatic artery thrombosis


hepatitis C virus


interstitial lung disease


intent-to-treat (population)


last observation carried forward


least square


liver transplantation


GFR calculated using the MDRD formula


Medical Dictionary for Regulatory Activities


Model of End-Stage Liver Disease




mycophenolate mofetil


mammalian target of rapamycin


per protocol (population)


randomized controlled study


standard deviation


standard error (of the mean)


system organ class (MedDRA)




treatment-emergent adverse event


Chronic progressive deterioration of renal function following solid organ transplantation is a major clinical concern in patients receiving a nonrenal graft. A population-based cohort study (1) estimated the 5-year risk of chronic renal failure posttransplant at 18.1% in patients after liver transplantation (LTx).

In addition to several other factors such as patient age, prevalent renal disorders, or metabolic diseases (1–6), immunosuppressive regimens employing calcineurin inhibitors (CNIs), for example cyclosporine A (CsA) or tacrolimus (TAC) are likely to promote posttransplant renal deterioration because of their nephrotoxic side effects also in LTx patients (7). This potential drawback of CNIs has promoted the search for alternative immunosuppressive regimens following LTx that are equally effective, but with no adverse impact on renal function (8).

Mammalian target of rapamycin (mTOR) inhibitors such as rapamycin (sirolimus) and everolimus might represent such an alternative. These drugs exhibit both immunosuppressive and antiproliferative properties with no relevant nephrotoxicity. Consequently, numerous studies were performed to investigate whether CNIs can be reduced or even eliminated following conversion to mTOR inhibitors while maintaining immunosuppressive efficacy. The majority of data are derived from patients after renal transplantation, whereas the experience with the use of mTOR inhibitors after LTx is rather limited (8). Initial reports in LTx patients raised the concern that the early use of sirolimus might be associated with increased rates of hepatic artery thrombosis (HAT), wound healing disturbances, and mortality (9). However, the results across all studies were conflicting, and a recently published review of 688 liver transplant patients receiving sirolimus also did not support these apprehensions (10).

In LTx recipients with renal dysfunction in later stages after transplantation the conversion of CNI to sirolimus did not show conclusive results. Although an improvement in GFR could be demonstrated in a few studies including small numbers of patients, a large randomized controlled study (RCT) involving more than 600 patients failed to demonstrate an improvement of renal function, while conversion to sirolimus was associated with an increased rate of rejection episodes (11). Another RCT in LTx patients revealed that late conversion to everolimus was safe, but again no significant improvement of renal function could be shown (12). For both studies the mean time after LTx was more than 3 years and it was concluded that future trials should target earlier conversion to mTOR based immunosuppression for improving renal function in these patients.

The aim of this study, therefore, was to evaluate the feasibility and nephroprotective effects of the mTOR inhibitor everolimus started 30 days after LTx within an early CNI-discontinuing regimen in transplant recipients with normal renal function or mild renal dysfunction.

Material and Methods

Study design, eligibility of patients and study treatments

We performed a multicenter, prospective, open-label, parallel-group RCT employing either a continuous CNI-based regimen (comparator arm) or a CNI-discontinuation regimen introducing everolimus (investigational arm) for immunosuppression in patients who underwent de novo LTx.

Primary study objective was the evaluation of renal function over 1-year following transplantation; primary efficacy endpoint was the GFR, estimated according to Cockcroft-Gault after 12 months posttransplant.

This study (NCT00378014) was conducted according to the ethical principles of the Declaration of Helsinki, ICH-GCP, and all applicable local regulations. The protocol (No. CRAD001HDE10) was approved by all competent Ethics Committees and regulatory authorities. Study candidates gave written informed consent prior to any study-specific measures.

Generally, de novo liver allograft transplant recipients aged between 18 and 70 years of either sex were eligible for screening. Among others, exclusion criteria precluded previous transplants, multiple solid organ transplants, severe systemic infections, significant renal dysfunction and treatment with any immunosuppressants within 2 months prior to transplantation. Basiliximab 20 mg was administered intravenously at Days 0 and 4 relative to LTx. CNI treatment with or without corticosteroids (CS) as well as infection prophylaxis was started after LTx according to local best center practice (no protocol directions). After 4 weeks on CNI treatment, patients were re-checked and eligible for randomization if they had been rejection-free for at least 2 weeks prior to randomization, showed sufficient hematology values (leukocytes >2.5 G/L, platelets >50 G/L, hemoglobin >8 g/dL), did not have severe hyperlipidemia in spite of lipid-lowering therapy (total cholesterol ≥9 mmol/L, triglycerides ≥8.5 mmol/L), and had sufficient renal function (cGFR >50 mL/min).

If patients failed to meet these criteria, randomization criteria could be reassessed until 8 weeks post-LTx; patients still not meeting the criteria were excluded from further study participation. Eligible patients were centrally randomized (computer-based) and allocated in a 1:1 ratio to the treatment arms (randomization = Baseline). The control arm continued the CNI-based regimen. In the investigational arm, everolimus was started at a dose of 1.5 mg bid and adjusted to achieve a target trough level of 5–12 ng/mL (8–12 ng/mL in patients on treatment with CsA), when the CNI was tapered by 70% of the initial CNI dose. After 8 weeks of CNI reduction, CNIs were completely stopped, provided the patient was rejection-free for at least 4 weeks. If CNI cessation could not occur at 4 months after LTx the patient was withdrawn from study drug treatment but received a final follow-up visit 11 months postbaseline. CS treatment was optional in either group (Figure 1).

Figure 1.

Schematic overview of CNI reduction/discontinuation during the course of the study. LTx, liver transplant. 8 weeks post-LTx at the latest. 16 weeks postbaseline at the latest. aStart with 2 doses of basiliximab (each 20 mg iv) at Days 0 and 4 related to time of LTx; afterwards treatment with any CNIs with or without systemic corticosteroids at the discretion of the investigator. bMode of CNI reduction in patients randomized to the everolimus group: Start of everolimus at Baseline with 1.5 mg bid orally until achievement of a whole blood target level of 5–12 ng/mL; subsequent CNI dose reduction by 70%. cComplete cessation of CNI treatment, if patient had been on reduced (−70%) CNI treatment for at least 2 months and did not experience a rejection episode within the 4 weeks before scheduled CNI discontinuation.

Statistical analyses

Primary analysis—determination of GFR:  Several methods were applied for GFR estimation. This report focuses on the results obtained with Cockcroft-Gault (13) and MDRD (14) formulas (CG-GFR and MDRD-GFR, respectively).

The primary study hypothesis postulated superiority of the everolimus regimen compared with the CNI continuation regimen at Month 11 postbaseline (i.e. 12 months posttransplant) in terms of a mean difference by at least 8 mL/min in CG-GFR. A sample size estimation assuming a standard deviation of 20 mL/min resulted in 100 patients per arm to detect the anticipated difference with a statistical power of 80% and a two-sided significance level of 5%. This confirmatory analysis was tested in the “Intent-to-Treat” (ITT) population using the “last-observation carried forward” (LOCF) approach with a covariance model (ANCOVA) with “treatment” and “center” as factors and “Baseline values” as covariate. Additional sensitivity analyses applied the same model in the per-protocol population and the other approaches for GFR estimation. No adjustment for multiple testing was made; all reported p values, apart from the confirmatory analysis, are thus of explorative nature.

Immunosuppressive efficacy:  To ensure sufficient drug exposure, drug level monitoring for everolimus and for CNIs was performed at the discretion of investigators throughout the study. The composite endpoint “efficacy failure” consisting of the single events “biopsy-proven acute rejection” (BPAR), graft loss, death, or loss to follow-up from any reason was used to compare the immunosuppressive efficacy of treatment regimens. Crude event rates were compared with Fisher's exact test, while time to events were analyzed using Kaplan–Meier estimates and log-rank tests.

Safety of treatment regimens, coding systems:  Safety/tolerability of treatments was evaluated by continuous monitoring of adverse events (AEs), vital signs, and safety laboratory variables.

AEs and concomitant diseases were coded using the “Medical Dictionary for Regulatory Activities” (MedDRA) terminology, drugs were classified using the WHO Drug-Dictionary. All analyses were performed using the software package SAS® (Version 8.2).


Disposition of patients

Patient enrollment started in August 2006, last patient's last regular study visit was on March 17, 2010. Altogether 375 patients entered the study and received treatment with basiliximab/CNIs. After the screening period, 203 patients (54.1%) were randomized at 16 specialized transplantation sites, while 172 patients (45.9%) failed randomization (Figure 2). Major reasons for nonrandomization were AEs (16.3%), withdrawal of consent (12.8%), cGFR ≤ 50 mL/min (11.6%) and graft loss (10.5%). A comparison of nonrandomized vs. randomized patients showed that the nonrandomized patients had poorer mean MELD scores (20.1 ± 8.2 vs. 16.9 ± 8.3), poorer mean CG-GFR values (74.0 ± 32.4 vs. 91.1 ± 29.6 mL/min), were more frequently on dialysis (30.8% vs. 9.9% of patients) and experienced more rejection episodes in the screening period (22.7% vs. 12.8%).

Figure 2.

Disposition of study patients. ITT = intent-to-treat (population); LTx = liver transplant; PP = per protocol (population); RX = randomized study treatment. *Basliliximab and/or CNI. The number of patients who “completed the study” does also include those patients who were prematurely withdrawn from study treatment but, nevertheless, showed up at Month 11 postbaseline for a follow-up visit (please note that the term “Baseline” refers to the time of randomization). The number of patients who prematurely discontinued randomized study drug treatment includes all patients who stopped study drugs prior to Month 11 postbaseline irrespective of whether they had a follow-up visit at Month 11 postbaseline or not.

Of the 203 patients, 101 were randomized to the everolimus arm and 102 to the CNI arm. 49.5% of patients in the everolimus group and 38.2% in the CNI group discontinued study drug permanently. AEs were the prevailing reason for premature treatment termination in either group (everolimus: 27.2%; CNI: 15.7%).

All 203 randomized patients were included in the safety population, while 9 patients were excluded from the ITT population (resulting in 194 patients; everolimus: 96, CNI: 98) because of missing postbaseline GFR data. The per-protocol (PP) population (i.e. those patients with no major protocol deviations) consisted of 106 patients (everolimus: 48, CNI: 58). Seven patients in the everolimus group (6.9%) had never received everolimus; they were included in the ITT analysis but not in the PP population.

Exposure to study drugs

The duration of CNI treatment before Baseline was 41.9 ± 11.3 days (range: 18–68 days); the mean daily dose of TAC (109 patients) was 9.5 ± 3.8 mg (range: 3.8–21.4 mg) and the mean daily dose of CsA (109 patients) was 429.1 ± 150.3 mg (range: 25.0–917.6 mg). No relevant group differences were noted.

After Baseline, patients in the everolimus group received reduced CNIs (TAC: 56 patients, CsA: 44 patients) for 89.7 ± 68.3 days (median: 72.0 days, range: 1.0–373.0 days) and everolimus for 213.5 ± 121.3 days (median: 298.0 days, range: 1.0–352.0 days) at a mean dose of 4.4 ± 1.7 mg/day (range: 1.4–9.9 mg/day). Mean everolimus trough levels were 6.9 ± 2.6 ng/mL at Week 2 postbaseline, 9.3 ± 3.1 ng/mL at Month 7 postbaseline and 8.9 ± 12.7 ng/mL at Month 11 postbaseline.

Patients in the CNI group (TAC: 58 patients, CsA: 49 patients) were treated for 241.1 ± 113.1 days (median: 309.0 days, range: 1.0–357.0 days) at mean doses of TAC and CsA of 6.6 ± 3.3 mg/day and 305.5 ± 101.9 mg/day, respectively.

Demographic and other baseline characteristics

Apart from the sex distribution with relatively more females in the everolimus group (43.8% vs. 30.6%), the treatment groups appeared to be balanced in terms of demographic and other baseline characteristics (Table 1). 10.8% of the ITT patients (CNI: 15.3%; everolimus: 6.3%) were positive for Hepatitis C, and 10.3% (equally distributed between groups) had hepatic malignancies.

Table 1.  Demographic and other baseline characteristics (ITT population)
 Everolimus N = 96CNI N = 98Total N = 194
  1. MELD = model for end-stage liver disease; SD = standard deviation.

  2. If frequency ≥5.0% in either treatment group.

Sex (n, %)
 Males54 (56.3)68 (69.4)122 (62.9)
 Females42 (43.8)30 (30.6)72 (37.1)
Age (years)
 Mean ± SD52.3 ± 9.952.9 ± 10.152.6 ± 10.0
 Median (range)54.0 (19.0–68.0)55.0 (19.0–68.0)54.0 (19.0–68.0)
BMI (kg/m2)
 Mean ± SD26.2 ± 4.426.4 ± 4.226.3 ± 4.3
 Median (range)26.1 (17.9–39.9)25.8 (18.3–38.3)25.8 (17.9–39.9)
Primary reason for Tx (n, %)
 Alcoholic cirrhosis28 (29.2)34 (34.7)62 (32.0)
 Hepatocellular carcinoma14 (14.6)23 (23.5)37 (19.1)
 Hepatitis B8 (8.3)8 (8.2)16 (8.2)
 Hepatitis C7 (7.3)8 (8.2)15 (7.7)
 Sclerosing cholangitis3 (3.1)10 (10.2)13 (6.7)
 Cryptogenic cirrhosis7 (7.3)4 (4.1)11(5.7)
 Primary biliary cirrhosis6 (6.3)2 (2.0)8 (4.1)
 Polycystic liver5 (5.2)3 (3.1)8 (4.1)
 Autoimmune hepatitis5 (5.2)1 (1.0)6 (3.1)
Time from liver transplant to randomization (days)
 Mean ± SD43.1 ± 11.342.5 ± 11.442.8 ± 11.3
 Median (range)42.0 (27.0–73.0)42.5 (25.0–66.0)42.0 (25.0–73.0)
Type of transplant (n, %)
 Complete liver85 (88.5)90 (91.8)175 (90.2)
 Split liver11 (11.5)8 (8.2)19 (9.8)
Donor characteristics (n, %)
 Heart-beating deceased donor91 (94.8)93 (94.9)184 (94.8)
 Non-heart-beating deceased donor1 (1.0)1 (1.0)2 (1.0)
 Living related donor2 (2.1)3 (3.1)5 (2.6)
 Living unrelated donor2 (2.1)1 (1.0)3 (1.5)
MELD score (points)
 Mean ± SD16.9 ± 8.716.4 ± 7.916.7 ± 8.3
 Median (range)15.0 (6.0–40.0)15.0 (6.0–40.0)15.0 (6.0–40.0)

Changes in renal function

Prior to Baseline, the 203 randomized patients showed an initial mean drop in CG-GFR from 98.9 ± 40.5 mL/min (last value before transplant) to 86.0 ± 28.7 mL/min (last value Baseline). At Baseline (i.e. after approximately 30 days of initial CNI treatment) CG-GFR values were similar between the treatment groups (Figure 3). The standard deviations indicated a broad interindividual variability at any measuring point. Mean CG-GFR over time showed a similar course until Week 4 postbaseline. Afterwards, the curves diverged with mean values in the everolimus group showing a continuous slight increase over time. The least square (LS) mean difference at Month 11 postbaseline was 2.9 mL/min in favor of everolimus (p = 0.46 in the ANCOVA, Table 2). Thus, the anticipated CG-GFR difference of 8 mL/min was not achieved. However, the sensitivity analysis performed in the PP population (LOCF) showed a LS mean difference for CG-GFR of 8.4 mL/min at Month 11 in favor of everolimus (p = 0.066).

Figure 3A.

Course of mean: (A) CG-GFR from Randomization to Month 11 postbaseline (LOCF, ITT). (B) MDRD-GFR from Randomization to Month 11 postbaseline (LOCF, ITT). The figures show the course of mean CG-GFR (primary analysis) and MDRD-GFR (exploratory analysis) from the time of randomization (baseline = BL) to Month 11 postbaseline in the ITT population with last observation carried forward (LOCF). Vertical error bars indicate the respective standard deviations. Results of the ANCOVA for the LS mean difference in GFR at Month 11 postbaseline.

Table 2.  Comparison of GRF values at 11 months postbaseline
 Everolimus group ITT, N = 96 PP, N = 48CNI group ITT, N = 98 PP, N = 58
  1. CI = confidence interval; ITT = intent-to-treat; LOCF = last observation carried forward; LS = least square; PP = per-protocol; SE = standard error.

  2. *Primary (confirmatory) efficacy analysis; assuming a CG-GFR difference by 8 mL/min in favor of the everolimus regimen, a standard deviation of 20 mL/min, 80% power, and an alpha-level of 5%.

  3. Between-group difference (calculated as CNI group minus everolimus group) at month 11 post-baseline; results based on ANCOVA model.

CG-GFR in ITT, LOCF (mL/min)*
 Mean value at baseline83.5 ± 27.083.2 ± 29.1
 Mean value at month 11-postbaseline87.9 ± 32.084.1 ± 34.9
 Mean change from baseline4.4 ± 27.10.9 ± 29.2
 LS mean difference (SE)−2.923 (3.920)
 95%-CI/p-value for difference[−10.659; 4.814]/ p = 0.457
CG-GFR in PP, LOCF (mL/min)
 Mean value at baseline82.9 ± 21.683.8 ± 28.4
 Mean value at month 11-postbaseline92.4 ± 29.084.8 ± 29.0
 Mean change from baseline9.4 ± 21.11.0 ± 24.7
 LS mean difference (SE)−8.358 (4.495)
 95%-CI/p-value for difference[−17.286; 0.570]/ p = 0.066
MDRD-GFR in ITT, LOCF (mL/min)
 Mean value at baseline78.0 ± 27.474.9 ± 24.6
 Mean value at month 11-postbaseline80.3 ± 26.472.1 ± 24.5
 Mean change from baseline2.0 ± 23.2−2.8 ± 23.1
 LS mean difference (SE)−7.778 (3.338)
 95%-CI/p-value for difference[−14.366; −1.191]/ p = 0.021
MDRD-GFR in PP, LOCF (mL/min)
 Mean value at baseline80.7 ± 25.078.0 ± 23.0
 Mean value at month 11-postbaseline84.9 ± 21.675.6 ± 21.3
 Mean change from baseline3.7 ± 21.1−2.4 ± 20.5
 LS mean difference (SE)−7.411 (2.962)
 95%-CI/p-value for difference[−14.733; −0.089]/ p = 0.047

The ANCOVA using the MDRD formula (LOCF) showed a LS mean treatment difference of 7.8 mL/min (p = 0.021) in the ITT population and of 7.4 mL/min (p = 0.047) in the PP population both in favor of the everolimus regimen. Generally, results were similar when “data as observed” analyses were performed instead of the LOCF procedure.

Immunosuppressive efficacy

Graft loss and rejection episodes were generally infrequent, and both the comparison of the crude event rates and Kaplan-Meier estimates did not show any relevant group differences (see Figure 4 for various crude event rates).

Figure 4.

Mortality and incidence of clinically relevant, graft-related events at Month 11 postbaseline (ITT). BPAR, biopsy-proven acute rejection.The composite endpoint “efficacy failure” consisted of BPAR, graft loss, death, or lost to follow-up from any reason. Data within bars show the crude event rates within each treatment group, p-values above each bar pair indicate the respective results for the between-group comparison (Fisher's exact test).

Safety profile

Altogether 35 patients died prior to randomization. More patients in the everolimus group than in the CNI group experienced treatment-emergent AEs (TEAEs) requiring permanent discontinuation (29.7% vs. 13.7%) or adjustment/transient interruption of study drug treatment (27.7% vs. 12.7%). The majority of events leading to premature treatment termination in the everolimus group were associated with infections (9 patients, 8.9%) or leukopenia/thrombocytopenia (7 patients, 6.9%). The incidence of nonfatal SAEs was slightly higher in the everolimus group (66.3% vs. 56.9%), while the rates of fatal events (4 in each group) were identical. Six deaths (3 in either treatment group) were associated with sepsis/infections; one death in each group was due to cardiac failure without underlying infection.

Adverse events of special interest are summarized in Table 3. TEAEs related to arterial hypertension, hyperlipidemia, anemia, leukopenia, and infections occurred more frequently in the everolimus group, while no relevant differences were noted for diabetes mellitus. No cases of HAT were reported during the first year of study, and only one patient had an interstitial lung disease (ILD)-related event documented (severe pneumonitis in the everolimus group). Likewise, cases possibly related to wound healing impairment were rare.

Table 3.  Selected TEAEs of special interest (safety population)
SOC Preferred termEverolimus N = 101 n (%)CNI N = 102 n (%)Total N = 203 n (%)
  1. All SOCs and preferred terms listed in this table were tested for group differences using two-sided Fisher's exact test (nonadjusted). This symbol denotes treatment group differences with an exploratory p-value of p ≤ 0.05.

  2. aAll preferred terms within this SOC with an incidence of ≥5.0% in any treatment group are included.

  3. bSelection of preferred terms possibly associated with wound healing impairment.

  4. cSelection of preferred terms possibly associated with diabetes mellitus or hyperlipidemia.

  5. dEVR group: carcinoma in situ of skin, lipoma, melanocytic naevus, and EBV-associated lymphoproliferative disorder (1 patient each); CNI group: adenocarcinoma, basal cell carcinoma, bladder transitional cell carcinoma, gastric cancer, hepatic neoplasm malignant recurrent, malignant melanoma in situ, metastases to bone, metastases to lung, metastases to lymph nodes and squamous cell carcinoma (1 patient each).

  6. eApart from four cases of “moderate” proteinuria (EVR group), all remaining cases were regarded as “mild” (i.e. no severe or serious proteinuria occurred). Only one case (EVR group) led to premature discontinuation of study treatment.

  7. fSelection of preferred terms probably associated with arterial hypertension.

Infections and infestationsa 74 (73.3) 61 (59.8) 135 (66.5)
Cytomegalovirus infection 8 (7.9)11 (10.8)19 (9.4)
Nasopharyngitis 15 (14.9)15 (14.7)30 (14.8)
Oral herpes 5 (5.0)0 (0.0)5 (2.5)
Pneumonia 7 (6.9)5 (4.9)12 (5.9)
Sinusitis 6 (5.9)1 (1.0)7 (3.4)
Urinary tract infection 16 (15.8)13 (12.7)29 (14.3)
Wound infection 7 (6.9)2 (2.0)9 (4.4)
“Serious” infections and infestationsa 27 (26.7) 19 (18.6) 46 (22.7)
Pneumonia 5 (5.0)2 (3.9) 8 (3.9)
Blood and lymphatic system disordersa† 38 (37.6) 25 (24.5) 63 (31.0)
Anemia 19 (18.8)11 (10.8)30 (14.8)
Leukopenia 21 (20.8)10 (9.8)31 (15.3)
Thrombocytopenia 8 (7.9)7 (6.9)15 (7.4)
General disorders and administration site conditionsa 50 (49.5) 39 (38.2) 89 (43.8)
Impaired healing 0 (0.0)1 (1.0)1 (0.5)
Injury, poisoning and procedural complicationsb 38 (37.6) 36 (35.3) 74 (36.5)
Incisional hernia 12 (11.9)10 (9.8)22 (10.8)
Wound complication 2 (2.0)4 (3.9)6 (3.0)
Wound dehiscence 0 (0.0)1 (1.0)1 (0.5)
Wound hemorrhage 1 (1.0)0 (0.0)1 (0.5)
Metabolism and nutrition disordersc 52 (51.5) 44 (43.1) 96 (47.3)
Diabetes mellitus 4 (4.0)8 (7.8)12 (5.9)
Hypercholesterolemia 23 (22.8)11 (10.8)34 (16.7)
Hyperglycemia 1 (1.0)3 (2.9)4 (2.0)
Hyperlipidemia 12 (11.9)2 (2.0)14 (6.9)
Hypertriglyceridemia 6 (5.9)3 (2.9)9 (4.4)
Type 2 diabetes mellitus 1 (1.0)0 (0.0)1 (0.5)
Neoplasms benign, malignant and unspecified (incl cysts and polyps)d 4 (4.0) 9 (8.8) 13 (6.4)
Nervous system disorders 38 (37.6) 35 (34.3) 73 (36.0)
Renal disordersa 21 (20.8) 25 (24.5) 46 (22.7)
Proteinuriae† 10 (9.9)2 (2.0)12 (5.9)
Renal failure 3 (3.0)7 (6.9)10 (4.9)
Vascular disordersf 26 (25.7) 23 (22.5) 49 (24.1)
Hypertension 20 (19.8)14 (13.7)34 (16.7)
Hypertensive crisis 1 (1.0)1 (1.0)2 (1.0)

The Kaplan–Meier curves for time to infection diverged from Day 56 onwards (Figure 5). Infections emerging between Day 56 and 112, however, were usually not severe and no distinct pattern in terms of type of these infections could be observed.

Figure 5.

Kaplan–Meier survival plot for the occurrence of infections (Safety population). This figure shows the Kaplan–Meier plot for the occurrence of infections during the course of the study (with Day 0 as start of randomization). Kaplan–Meier estimates at Month 11 postbaseline were 79.5% and 68.3% in the everolimus group and CNI group, respectively (p = 0.050 in the log-rank test).


The primary endpoint analysis in our study using the CG-GFR failed to prove superiority of the everolimus regimen in terms of the expected mean GFR difference of 8 mL/min determined at Month 11 postbaseline in the ITT population, while the observed difference of 8.4 mL/min in the exploratory PP analysis reached at least borderline significance (p = 0.066). When using the MDRD formula, the GFR differences were associated with exploratory p-values of <5% in both the ITT and PP population. When this study was designed, the Cockroft-Gault formula (which also incorporates body weight), was the most widely used formula for GFR estimation and therefore used for the endpoint definition. Meanwhile, it has been suggested that—while all formulas for calculating GFR have considerable limitations in LTx patients—calculation by the MDRD formula appears to be more appropriate in LTx patients (15) and, in fact, most of the current studies assessing renal function in LTx patients are using the MDRD formula. Overall, we conclude that there was a certain beneficial effect of everolimus on renal function in our study, although not as marked as anticipated based on the sample size estimation and recently published experience from other studies.

Generally, published data of RCTs employing everolimus early after LTx are currently rather sparse with limited comparability across studies. Levy et al. (16) conducted a Phase II RCT with 119 patients, who were treated after LTx with fixed everolimus doses of 0.5, 1 or 2 mg bid, or placebo, along with CsA and corticosteroids. CG-GFR initially decreased in all groups and then remained stable from Month 1 onwards, with no relevant differences among the groups at Month 12. The incidence of the composite endpoint (treated BPAR, graft loss, death or lost to follow-up) at Month 12 was similar for all 3 doses of everolimus (53.5% to 71.4%) and placebo (56.7%), but markedly higher than in our study, since all events were counted from the time of LTx, that is including the critical initial phase following transplantation. AEs were similar across groups, and overall, study results indicated that everolimus in combination with CsA had an acceptable safety and tolerability profile in LTx patients.

The first report of complete early CNI withdrawal and conversion to everolimus in patients with LTx was recently published by Masetti et al., who performed a single-center Phase II RCT (17). A total of 78 LTx patients were randomized (2:1) to be converted to everolimus (N = 52) after 10 days of CsA treatment (overlapping up to Day 30) or to receive ongoing CsA treatment (N = 26) with or without mycophenolate mofetil (MMF), depending on the presence of chronic kidney disease. Patients with renal failure (cGFR <29 mL/min) were excluded. Mean MDRD-GFR values were similar at randomization but significantly different from Month 1 to Month 12 onwards (everolimus: 87.6 ± 26.1 mL/min on everolimus vs. 58.2 ± 17.9 mL/min on CNI, p < 0.001). Efficacy failure rates (everolimus: 25.0%; CNI: 30.8%) and infection rates were similar between the treatment groups, and no relevant safety issues were found. While these GFR results are very promising, it should be taken into account that this was a single-center study with a limited number of patients. In addition, the definitions of the analyzed populations and the handling of missing values are not clearly comprehensible from the information provided in the publication. Irrespective of these methodological differences, the studies reported by Levy et al. and Masetti et al. indicated—in accordance with our study results—that CNI reduction or elimination on concomitant everolimus treatment is feasible, safe, and most likely associated with beneficial effects on renal function.

The timing of the conversion, however, seems to play a decisive role for the balance between graft protection, renal protection, and adverse effects related to over-immunosuppression. In our study, patients in the everolimus arm, in whom CNI was intended to be completely withdrawn, did not show an increased risk of acute rejection episodes, since acute rejection rates occurring after randomization in the everolimus and CNI arms were 17.7% and 15.3%, respectively. Interestingly, this finding is in contrast to the observations made in a recent, three-armed, everolimus Phase III study H2304 (NCT00622869). In this study (18), a markedly increased rate of BPAR was observed after discontinuation of TAC at 4 months after LTx and introduction of everolimus 30 days after LTx (19.9% vs. 4.1% in the low TAC/everolimus arm and 10.7% in the TAC control arm). This complication was possibly associated with the abrupt TAC cessation at Month 4, whereas the CNI-weaning regimen employed in our study obviously provided better protection from rejection. On the other hand, a moderately increased rate of minor infections (and thus patients prematurely terminating study treatment due to infections) was observed in the everolimus arm compared with the CNI arm. Rates of serious infections or pneumonia, however, were similar. The increase of (minor) infections beyond Day 56 in the everolimus group might be due to more intense immunosuppression as a result of the relatively high preplanned trough levels for everolimus and/or the overlapping exposure to both everolimus and CNIs in the reduction phase lasting up to Week 8 postbaseline (= Day 56) as per protocol.

These findings suggest that the overlapping weaning regimen employed in our present study obviously provided sufficient graft protection, but may have been associated with a slight over-immunosuppression resulting in a transiently increased infection rate, while the more abrupt discontinuation of TAC in study H2304 was obviously associated with a temporary deficit in immunosuppressive potency.

In terms of optimal preservation of renal function, it might be best to start the CNI withdrawal as early as possible. However, in order to prevent potential HAT cases associated with mTOR introduction at the time of transplantation, we initiated the conversion not before 30 days after LTx, while Masetti et al. (17) introduced everolimus in addition to CsA as early as 10 days after LTx with a stop of CsA after an overlapping period of another 20 days. Besides the positive effects on renal function, which were clearly more pronounced than in our study, this early regimen resulted in a very low incidence of rejection episodes of 5.7%, but was obviously paid with a markedly high rate of wound healing complications, for example incisional hernias affecting 46.1% patients compared to 26.9% of patients in the CsA maintained arm. Patients in our study, however, who were converted at a later time point had low incidences of hernias in both the everolimus and CNI group (11.9% and 9.8%, respectively). Nevertheless, the time of conversion should not be unnecessarily postponed, since the very early regimen employed by Masetti et al. yielded better renal effects, and reports from late conversion to everolimus in renally impaired patients suggest that the reduced renal function remains intractable (12,19,20). In conclusion, these findings underline the need for further clinical studies to identify the best weaning regimen accomplishing the optimum balance between transplant protection, renal protection, and prevention of undue infections and wound healing problems.

As intended by the study design, no problems of the hepatic artery were seen in our study. Likewise, the remaining safety analyses in our study did not point to an unduly increased risk when CNIs were withdrawn and converted to everolimus. As expected, both cytopenia events and hyperlipidemia occurred more frequently among patients on everolimus. Due to the higher incidence of these events, more patients in the everolimus group than in the CNI group terminated the study treatment due to AEs. Overall, a couple of expected side effects such as proteinuria occurred more frequently in the everolimus group, but these seemed not to reach a critical level.

Some potential limitations should be considered when interpreting the results of our study. There was a relatively high number of patients with premature withdrawal from the originally assigned study drug treatment (about 50% of the patients in the EVR group and about 40% of patients in the CNI group), thereby resulting in a reduced number of “true” observations on the originally intended treatment at Month 11 postbaseline and thus in a potential source for bias. The number of initial randomization failures (approximately 46%) was rather high, suggesting that only a selection of LTx patients in accordance with the eligibility criteria were studied, thereby limiting the generalizability of the study results. The numbers of patients in the ITT population and the PP population differed considerably in both treatment arms; however, the results observed in the ITT and PP population were generally similar, suggesting that study data were robust across the analysis populations.

Although our primary analysis formally failed to meet the protocol-defined superiority criterion, there was reasonable indication that the renal function was positively influenced by the employed everolimus regimen at least in the subgroup of patients completing the course of the study. In addition, our study showed that complete conversion from a CNI-based regimen to an mTOR-based regimen starting about 4 weeks after transplantation in patients with de novo LTx is acceptably safe and feasible without increasing the risk of graft complications, wound healing disturbances, or HAT. However, additional studies are warranted to confirm the nephroprotective effects of everolimus, to evaluate its renal-sparing properties in the long term, to determine the optimal conversion regimen following LTx, and to define the characteristics of those patients who are most likely to benefit from the conversion to everolimus.


The PROTECT study was funded by Novartis Pharma GmbH, Germany. Support for manuscript preparation was provided by Dr. Bernd Graulich, Winicker Norimed GmbH, Nuremberg, Germany.


Disclosure 1 (Commercial Organizations):

The manuscript preparation was funded by Novartis Pharma GmbH, Germany.

Support for manuscript preparation was provided by Dr. Bernd Graulich, Winicker Norimed GmbH, Nuremberg, Germany.

Disclosure 2 (Conflict of Interest):

Most of the authors of this manuscript have no conflicts of interest to disclose as described by the AJT. The following authors have reported conflicts of interest to disclose as described by the AJT.

  • 1Mr. Lutz Fischer: (a) received Speaker's fees, research funding, and travel grants as member of advisory boards from Novartis Pharma GmbH within the past 3 years; (b) received Speaker's fees and travel grants from Astellas Pharma GmbH within the past 3 years.
  • 2Ms. Martina Sterneck: (a) received travel grants from Novartis Pharma GmbH, Astellas Pharma GmbH, and Biotest AG within the past 3 years; (b) was member of an advisory board organized by Merck KGaA.
  • 3Mr. Frank Lehner: received research funds (consultancy and Speaker's fees) from Novartis Pharma GmbH, Astellas Pharma GmbH and Roche AG.
  • 4Mr. Juergen Klemphauer: received research funds (consultancy and Speaker's fees) from Novartis Pharma GmbH, Astellas Pharma GmbH, Genzyme Germany, Bristol-Myers Squibb GmbH & Co. KGaA and Roche AG.
  • 5Mr. Michael Hack: is employee of Novartis Pharma GmbH, Germany.
  • 6Mr. Stephan Ladenburger: is employee of Novartis Pharma GmbH, Germany.
  • 7Mr. Hans J. Schlitt: received research support, honoraria as speaker and/or as advisory board member from Novartis, Roche, Genzyme and BMS.