Randomized Controlled Trial of Sirolimus for Renal Transplant Recipients at High Risk for Nonmelanoma Skin Cancer


Scott B. Campbell, scottb_campbell@health.qld.gov.au


Sirolimus has antineoplastic effects and may reduce skin cancer rates in kidney transplant patients. This prospective, multicenter, randomized, open-label, controlled trial randomized 86 kidney transplant recipients (≥1 year posttransplant) with history of nonmelanoma skin cancer (NMSC) to continue calcineurin inhibitor (CNI) or convert to sirolimus. Patients were stratified by number of NMSC lesions (0–5, 6–20) in previous year. Primary end point was number of biopsy-confirmed new NMSC lesions per patient-year. Yearly NMSC rate was significantly lower with sirolimus (1.31 vs. 2.48 lesions/patient-year; p = 0.022). Squamous cell carcinoma occurred at a lower rate in the sirolimus versus CNI group (p = 0.038); basal cell carcinoma rate was similar in both. A lower proportion of patients receiving sirolimus developed new or recurrent NMSC (56.4% vs. 80.9%; p = 0.015) or new squamous cell carcinoma (41.0% vs. 70.2%; p = 0.006). No sirolimus patients and one CNI continuation patient experienced acute rejection. Incidence of treatment-emergent adverse events was similar between groups; however, discontinuation rates related to adverse events were significantly higher with sirolimus (46.2% vs. 0%; p < 0.001). In kidney transplant recipients with history of NMSC, conversion from CNI to sirolimus reduced rates of NMSC, without increasing acute rejection risk.


analysis of covariance


basal cell carcinoma


calcineurin inhibitor


Clinical Operation Randomization Environment


glomerular filtration rate


mammalian target of rapamycin


nonmelanoma skin cancer


squamous cell carcinoma


Advances in immunosuppression have considerably enhanced transplantation outcomes, with reduced acute rejection episodes and improved short-term graft and patient survival. However, improvement in long-term outcomes, including the risk for posttransplant malignancies with long-term immunosuppression, is still needed.

Compared with the age-matched general population, kidney transplant recipients experience greater rates of de novo and recurrent neoplasms, most commonly involving the skin (1). Nonmelanoma skin cancer (NMSC) accounts for more than 90% of posttransplant skin cancers (2–5). The course of squamous cell carcinoma (SCC) is more aggressive in immunosuppressed transplant recipients (6–8) and may occur from 65 to 250 times more frequently in transplant patients versus the general population, while basal cell carcinoma (BCC) occurs 10–20 times more often (9–11). Prevalence of cutaneous neoplasms increases with duration of immunosuppressive therapy (2,3,10–14).

The etiology of immunosuppression-related, posttransplant malignancy may be a combination of reduced immunosurveillance and the direct oncogenic potential of specific immunosuppressants (15). Reports associating reduction or discontinuation of immunosuppression, along with use of sirolimus, with decreased carcinogenesis (16–23) led to recommendations for immunosuppression reduction and switching to therapy with mammalian target of rapamycin (mTOR) inhibitors, when appropriate (24,25).

Sirolimus, an mTOR inhibitor, is an immunosuppressive drug with antitumor effects via inhibition of cancer cell proliferation, suppression of tumor angiogenesis and promotion of tumor cell apoptosis (26–28). Sirolimus exhibits its immunosuppressant effects via inhibition of T- and B-cell proliferation. The mechanism by which this inhibition occurs is also responsible for the inhibitory effects of sirolimus on cancer cell proliferation (28), prompting research of sirolimus-based therapy and posttransplant malignancy. Lower malignancy rates have been observed in transplant patients receiving sirolimus-based immunosuppression versus patients receiving calcineurin inhibitor (CNI) therapy (18,20,22,29,30). Most of these reports were based on retrospective or post hoc analyses.

This open-label, randomized, controlled trial is the first multicenter study prospectively designed to better characterize the above observations through comparing the rates of new NMSC lesions with conversion to sirolimus versus continued CNI therapy in stable kidney transplant recipients at high risk for developing further skin cancer.

Materials and Methods


This study, conducted in Australia, New Zealand and the United States, enrolled patients who developed NMSC (i.e. SCC or BCC) within 3 years and who underwent kidney transplant at least 1 year before enrollment. Patients were 18 years or older, on CNI-based regimens with cyclosporine or tacrolimus for at least 1 year prior, and on consistent immunosuppressive regimens for at least 1 month, with calculated Nankivell glomerular filtration rate (GFR) of at least 40 mL/min and proteinuria 500 mg or less per day. Patients with malignancies other than NMSC within 3 years of enrollment, history of NMSC with metastatic disease or excessive number of lesions (>20 in previous year) were ineligible. Patients receiving systemic retinoid therapy, field treatment with topical agents, photodynamic therapy, medium-depth chemical peels or laser resurfacing within 12 months before enrollment were excluded.

Study follow-up was to comprise a 2-year treatment phase plus a 1-month follow-up for adverse events. A blinded review of the data was performed 7 months after the last patient was randomized, because of low enrollment and a high early discontinuation rate. It was determined that there was minimal added power to detect a difference with continuing follow-up through 2 years versus limiting follow-up to 1 year, assuming that the rate of NMSC would continue as had been observed to that point. As a result, the study duration was amended to at least 1 year of treatment plus 1-month follow-up, with last-patient visit in January 2009.

The trial was sponsored by Wyeth Pharmaceuticals, which was acquired by Pfizer Inc. in October 2009. The authors and Wyeth representatives designed the study. Drs. Campbell, Walker, Russ and other study investigators collected data. Wyeth monitored the conduct of the study, performed statistical analyses, and held the data. The authors interpreted the data and collaborated in the preparation of the manuscript, supported by a professional medical writer provided by Pfizer Inc. All authors had access to the data, approved the article for submission, and assume responsibility for completeness and accuracy of the data and data analyses.

Institutional review boards at participating centers approved the study protocol, and the study was conducted in accordance with international standards of good clinical practice. All patients provided written informed consent before enrollment.


Screening and baseline evaluations were performed within 4 weeks prior to randomization using a computer-generated schedule. Patients were stratified by the number of new NMSC lesions in the previous 12 months (0–5 or 6–20) and randomly assigned (1:1) to CNI continuation or conversion to sirolimus. Patients were allocated to treatment according to a computer-generated Clinical Operation Randomization Environment (CORE) schedule. When the first site accessed the system, the first block of randomization codes were allocated to that site. The clinical enrollment system then assigned blocks of randomization codes in ascending sequential order each time a new site was activated or as an active site required an additional block. The block size was not revealed to investigators.

The CNI continuation group maintained their baseline therapy postrandomization, and doses of CNIs were adjusted or switched at the investigator's discretion. The sirolimus conversion group discontinued CNIs and started sirolimus on the same day with a single 6- to 12-mg loading dose, followed by 2–4 mg daily. Sirolimus dosing was adjusted to maintain whole-blood trough concentrations of 5–15 ng/mL, as measured by high-performance liquid chromatography (6–18 ng/mL as measured by immunoassay). Once sirolimus was at least 5 ng/mL, mycophenolate mofetil was decreased to 1.5 g/day or less, mycophenolate sodium to 1080 mg/day or less and azathioprine to 75 mg/day or less.

For both groups, mycophenolate mofetil, mycophenolate sodium or azathioprine were switched, decreased, temporarily withheld or discontinued as needed, based on clinical circumstances. Patients receiving corticosteroids at randomization were maintained on a minimum dose of 2.5 mg/day of prednisone (or equivalent). Corticosteroid withdrawal was prohibited. In patients not receiving corticosteroids at baseline, initiation of corticosteroids was permissible.


At baseline and at regular intervals, medical staff performed complete physical examinations and assessments of serum chemistries, lipids, liver function tests and blood counts. Study dermatologists examined patients at least every 3 months. All suspected NMSC lesions were to be biopsied and/or excised, with lesions reported at time of biopsy or treatment. Dermatologists were not blinded to treatment, as patients could be evaluated by both a dermatologist and transplant health care provider within a single clinical setting. Because the primary end point was limited to biopsy-confirmed NMSC lesions, pathologists, who were responsible for determining the nature of each lesion examined, were blinded to the immunosuppression treatment to maintain unbiased evaluation of the primary end point.

Patients who discontinued protocol-assigned treatment early, yet continued study participation, were evaluated at: (1) the time of treatment discontinuation; (2) 4 weeks postdiscontinuation and (3) at weeks 24, 52, 76 and 104 following randomization. These visits included examination by the study dermatologist.


The study was designed to have approximately 90% power to detect a mean difference of 1 new lesion, with two-sided alpha of 5%. Assuming a standard deviation of 2 lesions per year, a sample size of 90 patients per group was needed. These assumptions were based on previously published data that reported a mean accrual of NMSC to be 1.85 ± 3.84 tumors/person/year (31).

The primary end point, the number of new biopsy-confirmed NMSC lesions per patient per year, was calculated by summarizing the total number of new lesions reported over the observation period and standardizing it to an annual rate. The between-group difference in standardized rates was compared using a Poisson regression model adjusted by baseline NMSC stratum.

Primary efficacy analysis was based on the intent-to-treat population, comprising all randomly assigned patients receiving at least one dose of study medication for the duration of study follow-up, whether or not they remained on assigned therapy. As prespecified in the protocol, some analyses were performed on the on-therapy population, comprising all patients remaining on assigned therapy up to discontinuation of that therapy.

Secondary efficacy end points included time to first new biopsy-confirmed NMSC lesion and number of lesions recurring at the site of a previously treated lesion. Time to first new biopsy-confirmed NMSC lesion was graphically displayed with Kaplan-Meier curves and compared using the Cox proportional hazards method adjusted for baseline NMSC stratum.

In addition, calculated creatinine clearance (Nankivell method), serum creatinine, and urine protein:creatinine ratios were assessed. Analysis of covariance (ANCOVA) was used to compare GFR and serum creatinine change from baseline, with baseline as the covariate. Patients’ protein:creatinine ratios were summarized by each scheduled visit, and the nonparametric Wilcoxon rank sum test was used to compare between-group differences. Graft loss, death and biopsy-confirmed acute rejection were summarized. Safety end points included the incidence of infection, wound healing complications and other malignancies.


Among 87 patients randomized from September 2005 to October 2007, one was randomized in error (Figure 1). Of the remaining 86 patients, 39 converted to sirolimus and 47 remained on CNIs. Overall, 54 patients (62.8%) discontinued treatment prior to 2 years: 31 (79.5%) receiving sirolimus and 23 (48.9%) on CNIs (p = 0.004). Ten (25.6%) sirolimus patients and 19 (40.4%) CNI patients (p = 0.174) discontinued owing to early study termination. The most common reason for discontinuation was AEs in the sirolimus group (46.2% vs. 0% in the CNI group; p < 0.001).

Figure 1.

Study flow diagram.*p = 0.001. CNI, calcineurin inhibitor.

Mean length of follow-up for the intent-to-treat population (i.e. on- and off-therapy periods) was 1.68 and 1.74 years for the sirolimus and CNI groups, respectively (p = 0.127). Because of the higher rate of early treatment discontinuations in the sirolimus group, mean length of follow-up for on-therapy patients was significantly shorter in that group (0.95 vs. 1.62 years; p < 0.001).

Patient characteristics

Baseline demographics were similar between groups (Table 1). The mean time posttransplant to randomization was approximately 9 years. The majority of patients (84%) were in the stratum of 0–5 lesions in the 12 months prior to enrollment. The mean numbers of BCC and SCC lesions in the previous 12 months were balanced across treatment groups. Corticosteroid use and concomitant use of mycophenolate mofetil, mycophenolate sodium and azathioprine were also similar between groups. At screening, mean (SD) cyclosporine and tacrolimus trough levels in the group of patients randomized to convert to sirolimus were 102.4 (67.7) ng/mL and 6.9 (2.8) ng/mL, respectively. Mean (SD) sirolimus trough concentrations ranged from 7.1 (2.2) ng/mL to 12.6 (8.5) ng/mL over the course of the study and were maintained within the protocol-specified target range (Table 2). Although target trough concentrations were not specified for cyclosporine A and tacrolimus, mean (SD) trough concentrations throughout the study were maintained between 83.8 (38.6) ng/mL and 96.9 (55.4) ng/mL for cyclosporine A and between 5.8 (2.2) ng/mL and 8.0 (5.1) ng/mL for tacrolimus.

Table 1.  Baseline characteristics
 Sirolimus (N = 39)Calcineurin Inhibitor (N = 47)Total (N = 86)
Mean age (years) ± SE59.1 ± 1.559.0 ± 1.259.0 ± 1.0
Sex, n (%)
 Male31 (80)34 (72)65 (76)
 Female8 (20)13 (28)21 (24)
Race, n (%)
 Asian1 (3)01 (1)
 White38 (97)47 (100)85 (99)
Mean height (cm) ± SE174.1 ± 1.5172.6 ± 1.5173.3 ± 1.1
Mean weight (kg) ± SE85.4 ± 2.880.3 ± 2.382.6 ± 1.8
Current transplant, n (%)
 Primary transplant35 (90)39 (83)74 (86)
 Retransplantation4 (10)8 (17)12 (14)
Time after transplant, months
 Mean (SD)115.0 (57.6)109.6 (55.1)112.0 (56.0)
Concomitant use of other immunosuppressants, n (%)
 Corticosteroids29 (74.4)36 (76.6)65 (75.6)
 Azathioprine8 (20.5)15 (31.9)23 (26.7)
 Mycophenolate mofetil24 (61.5)24 (51.1)48 (55.8)
 Mycophenolate sodium1 (2.6)4 (8.5)5 (3.8)
Stratification group—n (%) in previous 12 months
 0–5 lesions33 (85)39 (83)72 (84)
 6–20 lesions6 (15)8 (17)14 (16)
Nonmelanoma skin cancer lesions in previous 12 months, mean (SD; min – max)
 Basal cell carcinoma0.6 (0.7; 0–2)0.8 (1.2; 0–4)0.7 (1.0; 0–4)
 Squamous cell carcinoma2.0 (2.3; 0–9)1.7 (1.8; 0–10)1.8 (2.0; 0–10)
Table 2.  Patient trough levels, intent-to-treat population
Time PointSirolimus Mean (SD), ng/mLCNI Mean (SD), ng/mL
Week 2n = 37n = 28n = 13
 11.8 (7.1)96.9 (55.4)6.8 (2.6)
Week 4n = 34n = 29n = 11
 8.7 (3.6)93.8 (50.8)7.3 (3.3)
Week 8n = 35n = 29n = 13
 9.9 (4.9)91.0 (47.5)8.0 (5.1)
Week 24n = 24n = 28n = 13
 10.3 (5.6)96.7 (58.1)6.1 (2.2)
Week 52n = 18n = 26n = 13
 7.9 (2.9)83.8 (38.6)6.0 (2.1)


In total, 86 biopsy-confirmed new NMSC lesions occurred in the sirolimus group versus 203 in the CNI group. Sirolimus conversion was associated with a significantly lower rate of new biopsy-confirmed NMSC lesions per patient-year versus CNI continuation (1.31 vs. 2.48, p = 0.022; intent-to-treat; Table 3). Overall, a greater proportion of patients in the sirolimus conversion group were lesion-free during the study (43.6% vs. 19.1%; p = 0.015).

Table 3.  Summary of efficacy measures; intent-to-treat and on-therapy populations
End PointsSirolimus (N = 39)Calcineurin Inhibitor (N = 47)p-Value1
  1. 1Poisson regression was used to model NMSC counts using the years in study as an offset. The generalized estimated equations approach was used to estimate parameters and compare treatment differences.

  2. ITT = intent-to-treat; OT = on-therapy.

Nonmelanoma skin cancer
 Rate of biopsy-confirmed new lesions (number per patient-year)
  Overall, ITT1.312.480.022
 Patients with ≥1 new or recurrent biopsy-confirmed lesion, n (%)
  Overall, ITT22 (56)38 (81)0.015
  OT14 (36)35 (75)<0.001
Squamous cell carcinoma
 Rate of biopsy-confirmed new lesions (number per patient-year), ITT0.881.710.038
 Rate of biopsy confirmed new lesions (number per patient-year), OT0.951.680.145
 Patients with ≥1 new biopsy-confirmed lesion, n (%)
  Overall, ITT16 (41)33 (70)0.006
  OT10 (26)31 (66)<0.001
Basal cell carcinoma
 Rate of biopsy-confirmed new lesions (number per patient-year), ITT0.430.770.104
 Rate of biopsy-confirmed new lesions (number per patient-year), OT0.410.820.095
 Patients with ≥1 new biopsy-confirmed lesion, n (%)
  Overall, ITT14 (36)24 (51)0.163
  OT8 (21)23 (49)0.007

More than two-thirds of all new NMSCs were SCC lesions. The proportion of patients with at least 1 new biopsy-confirmed SCC lesion was significantly lower in the sirolimus conversion group (16 [41.0%] vs. 33 [70.2%] patients; p = 0.006; Table 3). In each group, all but one of the SCC lesions were well or moderately differentiated. Most were in situ; of those that were invasive, fewer than 10% were associated with perineural invasion. The proportion of patients with invasive SCC was significantly lower in the sirolimus conversion group versus CNI continuation (10 [25.6%] vs. 24 [51.1%]; p = 0.018). Fourteen (35.9%) sirolimus conversion patients and 24 (51.1%) CNI continuation patients reported at least one new BCC lesion (p = 0.163). When standardized to yearly rates (Table 3), new SCC lesions occurred at a lower rate per patient-year in the sirolimus conversion (0.88) versus CNI continuation group (1.71; p = 0.038), whereas the rates of new BCC lesions were not significantly different between groups (0.43 vs. 0.77; p = 0.104).

Between-group differences in time to first biopsy-confirmed new NMSC lesions were observed (p = 0.047; Figure 2A). Median time to first new lesion was 380 days for the sirolimus conversion group versus 163 days for the CNI continuation group. Median time to first biopsy-confirmed SCC lesion (Figure 2B) was approximately 750 days in the sirolimus conversion group versus 189 days in the CNI continuation group (p = 0.012). Time to first new BCC lesion (Figure 2C) was not different between groups (p = 0.22; median days, sirolimus conversion: nonestimable; CNI continuation: 682).

Figure 2.

Kaplan-Meier estimates of new biopsy-confirmed lesion-free survival for nonmelanoma skin cancer in the intent-to-treat population for nonmelanoma skin cancer (A), squamous cell carcinoma (B) and basal cell carcinoma (C).

For lesions recurring at previously treated lesion sites, no treatment difference was observed overall (sirolimus conversion, 0.107 lesions per patient-year; CNI continuation, 0.134; p = 0.748). Metastatic disease related to NMSC occurred in 1 (2.6%) and 3 (6.4%) patients in the sirolimus conversion and CNI continuation groups, respectively (p = 0.425).

Graft loss (including death) occurred in 2 patients in the sirolimus conversion group and 1 patient in the CNI continuation group (p = 0.59). Two deaths (1 in each group) and a single episode of acute rejection (in the CNI group) occurred in the study. Based on Nankivell-calculated GFR for the intent-to-treat population, no significant difference in renal function was observed between groups at 6, 12 or 24 months. Adjusted mean change from baseline GFR for the on-therapy population was +3.88 and −1.82 at 12 months in the sirolimus conversion and CNI continuation groups, respectively (p = 0.029, ANCOVA), and +6.64 and +0.57 at 2 years, respectively (p = 0.06). Median urine protein:creatinine ratio ranged from 0.11–0.14 in the sirolimus conversion group and 0.08–0.12 in the CNI continuation group and did not differ significantly between groups, except at week 104 (0.14 vs. 0.12, respectively; p = 0.03, Wilcoxon).


Table 4 lists treatment-emergent AEs occurring in more than 10% of patients. No significant between-group difference was observed in the number of patients with at least 1 AE (p = 0.067); however, significantly higher rates of acne, albuminuria, diarrhea, epistaxis, mouth ulceration, peripheral edema, rash and pneumonitis were noted in the sirolimus conversion group. No significant difference was noted between groups in the incidence of any serious AE (sirolimus, 15 [38.5%] patients; CNI, 21 [44.7%]). Two deaths occurred during the study; 1 (2.6%) patient receiving sirolimus died of severe biventricular heart failure and 1 (2.1%) patient on a CNI died of respiratory failure.

Table 4.  Treatment-emergent adverse events occurring in at least 10% of patients in either group
Adverse Event, n (%)Sirolimus (n = 39)Calcineurin Inhibitor (n = 47)p-Value1
  1. 1Fisher's exact test (2-tail).

Any38  (97)40  (85)0.067
Diarrhea16  (41)6  (13)0.006
Peripheral edema16  (41)3  (6)<0.001
Rash13  (33)1  (2)<0.001
Mouth ulceration13  (33)0<0.001
Cough increased8  (21)6  (13)0.388
Acne7  (18)1  (2)0.021
Albuminuria7  (18)00.003
Pain6  (15)5  (11)0.536
Hypertension6  (15)4  (9)0.501
Pneumonitis6  (15)00.007
Arthralgia5  (13)6  (13)1.000
Gout5  (13)3  (6)0.459
Hyperlipidemia5  (13)2  (4)0.237
Headache5  (13)1  (2)0.087
Delayed wound healing4  (10)3  (6)0.697
Chest pain4  (10)2  (4)0.404
Dyspnea4  (10)2  (4)0.404
Anemia4  (10)2  (4)0.404
Epistaxis4  (10)00.039

AEs leading to discontinuation of assigned treatment, but not necessarily from the study, occurred in 18 (42.6%) patients receiving sirolimus. Most frequent AEs causing discontinuation were pneumonitis (4 [10.3%] patients), diarrhea (2 [5.1%]) and decreased tolerance (2 [5.1%]). No discontinuations owing to AEs occurred in the CNI continuation group.


The development of new NMSC following renal transplantation is a major concern, and in some cases can be life threatening. The majority of new NMSC are SCCs, with a preponderance of BCCs prior to transplantation being reversed posttransplantation (13). Among kidney transplant recipients, approximately 75% of those who had at least one skin cancer pretransplant will develop an average of 16–20 NMSC lesions posttransplant, with a median of 6 months to onset (2,13).

Previous reports of sirolimus-based immunosuppression have examined the incidence and time to onset of NMSC lesions in both long-term and de novo transplant recipients. These trials, which mainly evaluated malignancy either retrospectively, as secondary end points, or as a part of longer-term follow-up, reported lower rates of malignancy in transplant patients receiving sirolimus-based immunosuppression when compared with patients receiving CNI therapy. A single-center, prospective, randomized trial of 44 patients (mean duration of immunosuppression, 230 months) showed that switching to sirolimus was superior to continuation of previous immunosuppression (e.g. azathioprine, cyclosporine, tacrolimus or mycophenolate) on the development of premalignant skin dysplasia (23). Furthermore, switching to sirolimus-based immunosuppression resulted in a significantly lower proportion of patients who developed histologically confirmed NMSC (6.3% vs. 47.1%; p = 0.017; 23). Additionally, another large study of de novo kidney transplant recipients examined cyclosporine withdrawal at 3 months with ongoing sirolimus maintenance therapy versus continued cyclosporine with sirolimus. Results showed evidence of reduced incidence of nonskin malignancy (9.6% vs. 4.0%; p = 0.032) and prolonged median time to first skin carcinoma (491 vs. 1126 days; p = 0.007) with CNI-free sirolimus maintenance therapy through 5 years posttransplantation (20). In the recently published CONVERT trial (N = 555), renal allograft recipients who converted to a sirolimus-based, CNI-free regimen experienced significantly fewer skin carcinomas (2.2% vs. 7.7%; p < 0.001) through 2 years postconversion versus those who continued on CNIs (22). Similarly, a review of five multicenter studies examining the 2-year incidence of malignancy in renal transplant recipients found that sirolimus-based therapy resulted in reduced skin cancer rates (18).

The results presented herein are the first from a prospective, multicenter, randomized controlled trial designed to specifically evaluate skin malignancy. This trial examined the incidence of new NMSC lesions in stable kidney transplant recipients at high risk for developing skin cancer (i.e. history of NMSC in prior 3 years), comparing continued CNI therapy versus conversion from a CNI to sirolimus. The principal finding was that conversion to a sirolimus-based regimen demonstrated a significantly lower rate of NMSC, with prolonged time to first lesion, compared with continuation of CNI therapy.

In order to evaluate the end point of malignancy in a timely manner, the patient population studied in this trial comprised an enriched, high-risk patient group (i.e. those developing NMSC within 3 years of enrollment and who were not receiving systemic retinoid therapy or topical field treatment for NMSC lesions within 1 year of enrollment). This was based upon previously published data that showed high rates of NMSC recurrence within a relatively short period of time in renal transplant recipients who had at least 1 NMSC prior to their first transplant (13). In addition, the number of lesions has been associated with severity of disease, with time to diagnosis of second and third lesions becoming progressively shorter (32). Therefore, a follow-up period of 1 to 2 years within this study was anticipated to have a sufficient number of NMSCs develop to be analyzed for the primary end point.

During the study, a high discontinuation rate in the sirolimus group limited the duration of the on-therapy follow-up period, although most of those patients remained in the study with regularly scheduled follow-up to be included in the ITT analysis. The observed high rate of early discontinuations related to AEs in the sirolimus group is commonly seen in similar sirolimus conversion trials. Numerous factors may contribute to this, including the abruptness of the conversion to sirolimus, the use of loading doses and blood concentrations that were higher than those utilized in current practice (an artifact of the date of study execution), possible patient-related factors (e.g. desire to revert to original immunosuppression due to newly emergent AEs). In contrast, patients in the CNI group remained on their longstanding regimen such that even if they experienced AEs, they were unlikely to change their regimen.

The potential for unequal discontinuation rates, given the nature of the patient population and significance of the intervention, was anticipated when developing the study. Therefore, per protocol, analysis of the primary end point was based on time-normalized data (i.e. annualized incidence rates, or number of lesions per patient-year). In addition, analysis by both the ITT (primary analysis) and on-therapy (secondary analysis) populations were both prespecified prior to study initiation to prospectively address any potentially limiting factors. As noted previously, despite high discontinuation rates, most patients receiving sirolimus remained in the study and underwent required dermatologic evaluations at periodic intervals, thus allowing for the potential identification of further lesion development during the off-therapy period. Therefore, sirolimus patients had a similar follow-up period as CNI patients by intent-to-treat analysis, which demonstrated a significantly lower rate of NMSC in the sirolimus group. There were significantly more NMSC-free patients in the sirolimus groups compared with CNI by ITT analysis (p = 0.015) and by on-therapy analysis (p < 0.001). The loss of statistical significance on the primary end point in the on-therapy population is likely explained by the difference in duration of the on-therapy period (p < 0.001). While a shorter on-therapy period for the sirolimus arm might conceivably cause less precision on the NMSC estimate, the NMSC reduction remained similar (46% reduction on-therapy vs. 47% reduction ITT).

Enrollment in the study was slower than expected, and despite extension of the enrollment period, fewer patients were enrolled and randomized than planned. Some patients were hesitant to enroll in a study late following their transplant. They were unwilling to have their stable immunosuppressive regimen changed because their concerns about rejection and graft function overrode that of malignancy. While the plan was to include 90 subjects per group, only 86 in total were randomized. Even though fewer patients were included, standard deviations in terms of NMSC rates were sufficiently smaller than anticipated that statistical significance based on the prespecified intent-to-treat population was still achieved. In addition, because more patients in the sirolimus group discontinued therapy earlier, the ability to detect real treatment effects is reduced using the intent-to-treat approach. The fact that a treatment effect was detected despite this bias against the detection of treatment effect leads us to believe that the effect is real.

During the conduct of the study, not all lesions were treated prior to randomization as was planned, evidenced by the number of lesions observed very early following randomization (Figures 2A and B). Centers reported that this occurred for a number of reasons, including scheduling issues or in some cases a patient having too many lesions to be treated at once. The preexistence of these lesions prompted additional sensitivity analyses, which were performed to exclude lesions at baseline, first 30 days, and first 90 days. The rate of new NMSC lesions and the proportion of patients with NMSC remained significantly lower in the sirolimus conversion group for each of the analyses (data not shown).

Of interest, no significant differences in GFR were noted in the ITT population at 6, 12 or 24 months. While this end point was secondary and the study was not powered to detect a between-group difference in GFR, this observation warrants mention. Between-group differences in the ITT population were at, or approached, statistical significance at weeks 2 (p = 0.05), 4 (p = 0.01) and 8 (p = 0.08). As discontinuations increased over the duration of the study, however, patients were converted back to CNI therapy, which may explain why differences at later time points were not observed in this population. When examining the on-therapy population, these differences were at or approached statistical significance at all time points except for weeks 12 and 24. It should be noted, however, that over time, patient numbers decreased, which limits the ability to draw any substantial conclusions regarding renal function in this study.

Previous studies with sirolimus have observed a similar reduction of malignancy risk in less-restrictive populations that were not selected based on previous NMSC history (18,20,23,33). For this study, in a population at high risk for further NMSC, a number of factors may be responsible for the significant reductions in annualized NMSC incidence rates. The withdrawal of CNIs after conversion to sirolimus may have contributed to these results, as CNIs and other immunosuppressants (e.g. azathioprine, corticosteroids) have been shown to exhibit prooncogenic effects through the upregulation of p53-oncogene and transforming growth factor-β (TGF-β) expression (34–36). Sirolimus, in addition to other mTOR inhibitors, is associated with interference of angiogenesis and cancer cell growth and progression in general, and may have played an important role (14,26,27).

In vivo molecular differences have been observed with NMSC lesions in renal transplant recipients compared with the general population, specifically in terms of p53 oncogene and transforming growth factor β-1 upregulation as well as activation of mTOR-regulating pathways (36). Furthermore, the presence of human papillomavirus (HPV) has been directly associated with NMSC development, particularly SCC (37). The role of mTOR inhibitors has been shown to extend outside of its cytostatic and antiproliferative properties. The potential for mTOR inhibitor-induced interference with viral replication of herpes viruses and retroviruses has been studied, and the conversion to sirolimus has also been reported to be effective in the treatment of HPV-induced viral warts in the posttransplant patient (38).

It is also interesting to note that the reduction in NMSC incidence was largely driven by a reduction in SCC rates compared with BCC rates, although the study was not powered to detect a difference in subtypes of NMSC. This is not surprising, given that most of the excess NMSCs post-renal transplantation are SCCs. In the posttransplant patient, SCCs demonstrate markedly more aggressive growth, dissemination and proliferation (24). The lower event rate of BCC (compared with SCC) in the context of this study duration and population may have contributed to the potential for Type 2 statistical error with regard to BCC.

Furthermore, although both SCC and BCC are categorized as NMSCs, the lesions have distinct mechanisms of proliferation, specifically related to mTOR pathways, which may account for different responsiveness to sirolimus. Genetic mapping of NMSC lesions has found distinct gene modulation differences between SCC and BCC (39). In addition, recent evidence using murine models of BCC have shown that Akt1-mTOR pathways have a significant role in the development of BCCs (40). However, use of sirolimus to reduce BCC growth in these models were not completely effective, pointing to additional Akt1-dependent (and mTOR-independent) pathways responsible for BCC growth and development (40). Based on these clinical data, it is possible that longer follow-up in a larger patient population may have produced more robust reductions in BCC incidence; in addition, the potential multipathway pathogenesis of BCC could play a role in a relatively diminished ability to impact BCC incidence.

The results of this and other studies examining the reduction of skin and nonskin malignancies suggest a benefit in converting to a sirolimus-based immunosuppressive regimen. The primary objective was achieved by demonstrating a significantly lower rate of NMSC with conversion to sirolimus in renal transplant patients at high risk for NMSC compared with rates in patients remaining on CNI-based therapies. Further studies to better characterize these findings and to understand the mechanisms involved are warranted.


We thank Michael J. McLaughlin, MD, Christine H. Blood, PhD, and Albert Balkiewicz, MSc, of Peloton Advantage, LLC, for editorial assistance with manuscript preparation, which was funded by Pfizer Inc. We also gratefully acknowledge Kristin Ingrassia and Melanie Anson of Wyeth Pharmaceuticals (acquired by Pfizer Inc in October 2009) for study coordination and data review.

The following lead investigators participated in this study:

Australia: Scott Campbell, Princess Alexandra Hospital, Brisbane, QLD; Josette Eris, Royal Prince Alfred Hospital, Camperdown NSW; Randall Faull, Royal Adelaide Hospital (RAH), Adelaide SA; John Kanellis, Monash Medical Centre, Clayton VIC; Philip O’Connell, Westmead Hospital, Westmead NSW; Bruce Pussell, Prince of Wales Hospital, Randwick NSW; Dwarakanathan Ranganathan, Royal Brisbane and Women's Hospital, Herston QLD; Graeme Russ, Queen Elizabeth Hospital (QEH), Woodville SA; Rowan Walker and Shlomo Cohney, Royal Melbourne Hospital, Melbourne VIC; New Zealand: Helen Pilmore, Auckland City Hospital, Grafton, Auckland; USA: S. Forrest Dodson, Rush University Transplantation Program, Chicago, IL; Amit Govil, University of Cincinnati Medical Center, Cincinnati, OH; Mysore S. Anil Kumar, Hahnemann University Hospital, Philadelphia, PA; Pamela R. Patton, University of Florida, Gainesville, FL; David Perkins, University of San Diego Medical Center, San Diego, CA; John Pirsch, University of Wisconsin, Madison, WI; Carlos Zayas, Piedmont Hospital Transplantation Services, Atlanta, GA.

The lead dermatologist at each participating center, in order of the above centers, included:

Australia: Greg Siller, Stephen Lee, Warren Weightman (RAH), Paul Curnow, Ian Hamann, Glenda Wood, Ivan Robertson, Warren Weightman (QEH), and Boon Tiong Tan; New Zealand: Denesh Patel; USA: Lady Christine C. Dy, Brett Coldiron, Herbert Allen, Anthony L. Aulisio, Terence C. O’Grady, Juliet Gunkel and David C. Olansky.


This study was sponsored by Wyeth Pharmaceuticals, which was acquired by Pfizer Inc in October 2009. Dr. Campbell reports that he has served on advisory boards for Wyeth/Pfizer, Novartis, Janssen-Cilag and Roche. He has also been a principal investigator or coinvestigator in multicenter clinical trials for the same companies, and has received travel assistance to transplantation meetings from Wyeth/Pfizer, Novartis and Janssen-Cilag. Dr. Walker has served or currently serves on advisory boards for Wyeth/Pfizer, Novartis, Janssen-Cilag and Roche. He has also been an investigator in multicenter clinical trials for these companies, and has received travel assistance to transplantation meetings from Wyeth/Pfizer, Novartis and Janssen-Cilag. Dr. Russ has received travel grants and honoraria from Wyeth/Pfizer and served on advisory boards. None of the authors listed above received financial support related to the development of this manuscript. Ms. Jiang and Dr. See Tai, at the time the study was conducted, were employees of Wyeth Pharmaceuticals, which was acquired by Pfizer Inc in October 2009. No other potential conflicts of interest relevant to this article were reported.