Patients randomized, assigned to study treatment, lost to follow-up, and discontinued treatment before 5 years after transplantation in the original trial and patients included in the long-term analysis of transplant outcomes in the national registry.
Time to allograft failure from any cause including death (A) and allograft failure censored for patient death (B) in patients randomized to withdraw corticosteroids (no steroids) or continue corticosteroids (steroids).
The outcomes of trial participants were similar to that of contemporary registry patients in patients treated with and without corticosteroids. The P values shown in black compare time to event in trial and registry patients treated with corticosteroids, while the P values shown in gray compare trial and registry patients treated without corticosteroids.
eTable 1. Factors associated with risk of allograft loss from any cause and allograft loss censored for death in trial participants (multivariable Cox regression model)
eTable 2. Association of Corticosteroid Withdrawal with long-term transplant outcomes: Combined analysis of Trial Participants and Contemporary Registry Patients
eFigure 1. Time to allograft failure from any cause including death and the time to allograft failure censored for patient death among patients who remained on the trial assigned treatment at five years after transplantation
eFigure 2. Time to death at any time after transplantation including deaths that occurred after failure of the allograft
eFigure 3. Time to death with patient follow-up censored at time of graft failure
eFigure 4. Time to allograft failure from any cause including death in patients randomized to withdraw corticosteroids (no steroids) or continue corticosteroids (steroids) in patient sub-groups
eFigure 5. Time to allograft failure censored for patient death with a functioning allograft in patients randomized to withdraw corticosteroids (no steroids) or continue corticosteroids (steroids) in patient sub-groups
eFigure 6. Identification of Contemporary Transplant Recipients in the Organ Procurement and Transplant Network Registry who would have met trial enrolment criteria
Customize your JAMA Network experience by selecting one or more topics from the list below.
Woodle ES, Gill JS, Clark S, Stewart D, Alloway R, First R. Early Corticosteroid Cessation vs Long-term Corticosteroid Therapy in Kidney Transplant Recipients: Long-term Outcomes of a Randomized Clinical Trial. JAMA Surg. 2021;156(4):307–314. doi:10.1001/jamasurg.2020.6929
Do long-term kidney transplant outcomes differ in patients treated with and without maintenance corticosteroids?
In a randomized clinical trial that allocated 385 patients to maintenance immunosuppressive treatment with tacrolimus and mycophenolate mofetil with or without corticosteroids, there was no difference in kidney allograft survival between treatment groups during the median follow-up of 15.8 years after transplant.
Corticosteroids may not be necessary as part of a calcineurin-based multiple drug immunosuppressive regimen in kidney transplant recipients.
The complications of corticosteroids make the inclusion of these drugs in immunosuppressive protocols for kidney transplant patients undesirable. However, cessation of corticosteroids is associated with a higher risk of short-term rejection, and the long-term outcomes of patients withdrawn from corticosteroids remain uncertain.
To compare long-term kidney transplant outcomes of patients randomized to continue or withdraw corticosteroids.
Design, Setting, and Participants
This prospective multicenter randomized double-blind placebo-controlled trial was conducted between November 1999 and December 2002 with linkage to a mandatory national registry with validated outcome ascertainment until June 8, 2018. The study included 28 kidney transplant centers in the United States, including 386 low– to moderate–immune risk adult recipients of a living or deceased donor kidney transplant without delayed graft function or short-term rejection in the first week after transplant. Analyses were intention to treat. Analysis began September 2018 and ended June 2019.
Patients were randomized to receive tacrolimus and mycophenolate mofetil with or without corticosteroids 7 days after transplant.
Main Outcomes and Measures
Kidney allograft failure from any cause including death and allograft failure censored for patient death defined by the requirement for long-term dialysis or repeat transplant.
Of 385 patients, 191 were assigned to withdraw from corticosteroids (mean [SD] age, 46.5 [12.1] years), and 194 patients were assigned to continued corticosteroids (mean [SD] age, 46.3 [12.6] years). The median (interquartile range) follow-up time was 15.8 (12.0-16.3) years after transplant. The adjusted hazard ratios of allograft failure from any cause including death was 0.83 (95% CI, 0.62-1.10; P = .19) and for allograft failure censored for patient death was 0.78 (95% CI, 0.52-1.19; P = .25) and did not differ between the patients assigned to withdraw from corticosteroids vs assigned to continued corticosteroids. Results were consistent in a per-protocol analysis among 223 patients who continued the trial-assigned treatment of corticosteroid withdrawal (n = 114) or corticosteroids (n = 109) through at least 5 years after transplant. The outcomes of trial participants in either treatment group did not differ from similarly treated contemporary registry patients who met trial eligibility criteria and were treated with the same immunosuppressive drugs.
Conclusions and Relevance
Long-term corticosteroids may not be necessary as part of a calcineurin-based multiple drug immunosuppressive regimen in low– to moderate–immune risk kidney transplant recipients.
Kidney transplant recipients require ongoing immunosuppressive drug treatment to prevent rejection. The most common immunosuppressive drug regimen includes 3 drugs, a calcineurin inhibitor (cyclosporine or tacrolimus), mycophenolic acid or azathioprine, and corticosteroids (ie, prednisone).
The adverse effects of corticosteroids have prompted attempts to eliminate these drugs from treatment regimens. Studies using cyclosporine demonstrated that corticosteroid withdrawal was associated with an increased risk of short-term rejection1 and late allograft failure.2,3 The development of tacrolimus and antibody induction prompted renewed interest in eliminating corticosteroids.4 The most rigorous trial of corticosteroid withdrawal was reported in 2008 by Woodle and colleagues.5 This double-blind placebo-controlled trial enrolled 386 low– to moderate–immune risk deceased and living donor transplant recipients between 18 and 70 years of age in 28 transplant centers in the United States between November 1999 and December 2002. Trial participants were randomized to withdraw corticosteroids 7 days after transplant (n = 191) or continue maintenance corticosteroids (n = 195) at a dose of 5 mg per day by 6 months after transplant. Patients in both groups received induction immunosuppression with either a T-cell–depleting or T-cell–nondepleting antibody and maintenance tacrolimus and mycophenolate mofetil. After 5 years, there was no difference in the composite primary end point of death, allograft loss censored for patient death, or moderate to severe short-term rejection, but there was a higher rate of biopsy-confirmed rejection in patients withdrawn from corticosteroids (34 of 191 [17.8%]) compared with patients who continued corticosteroids (21 of 195 [10.8%]). Kidney function was similar between the groups, but weight gain, serum triglyceride levels, and insulin-requiring diabetes were lower in patients withdrawn from corticosteroids. Subsequently, single-center studies have reported acceptable 10- and 15-year outcomes in patients withdrawn from corticosteroids compared with historical controls.6,7
Despite this evidence, only 30% of transplant recipients are treated with corticosteroid withdrawal.8 Factors contributing to continued use of corticosteroids include concerns about long-term safety and uncertainty about the modest metabolic benefits of corticosteroid withdrawal.9 In contrast, proponents of corticosteroid withdrawal hypothesized that longer follow-up might be required for the metabolic benefits to affect outcomes.10-12
A 2016 meta-analysis highlighted the risk of rejection and concluded that the long-term consequences of corticosteroid withdrawal remain unclear because prospective long-term studies have not been conducted.13 Given further trials are unlikely, we determined the long-term outcomes of participants in the trial reported by Woodle and colleagues5 by linking trial data with the Organ Procurement and Transplant Network (OPTN) registry. Quiz Ref IDThe study objectives were to verify the reported trial outcomes, to determine the long-term outcomes of the trial participants in the OPTN registry, and to determine the external validity by comparing the outcomes of trial participants with contemporary patients in the OPTN registry who likely would have been eligible for the trial and received similar immunosuppressive treatment.
The trial publication provides a trial description.5 The trial enrolled primary deceased and living donor kidney transplant recipients aged 18 to 70 years, who were at low to moderate immune risk (defined by peak panel reactive antibody <50% and pretransplant panel reactive antibody <25%). Patients were randomized 1:1 with stratification by self-reported race (African American or not because short-term rejection and risk of allograft failure are higher in African American individuals) and donor type (living or deceased) at transplant. Randomization was based on a permuted design with block sizes of 6 within each clinical site by a central service that generated the allocation sequence and maintained the allocation code. Patients without short-term rejection or delayed graft function (defined by a minimum 30% reduction in pretransplant serum creatinine level within 7 days) started the trial treatment protocol on day 7 after transplant. Patients, investigators, study personnel, and those assessing outcomes remained blinded throughout the 5-year study. The study was approved by the University of British Columbia Research Ethics Board. Written informed consent and all trial data were obtained by the transplant center and submitted to the trial sponsor, Astellas Pharma.
The primary outcome was a composite of patient death, graft loss, or moderate to severe short-term rejection determined for 5 years after transplant. The trial was designed to determine if continuation of corticosteroids was superior to corticosteroid withdrawal, assumed a 10% event rate in patients who continued receiving corticosteroids, and had 80% power to detect a 10% increase in the composite outcome in the intervention group with an α error of 5% (1 sided).
The OPTN registry has captured all transplants in the United States and allows ascertainment of the outcomes of allograft failure censored for death (defined as a return to dialysis or repeat transplant) and death through linkage to national data on the treatment of long-term dialysis patients provided by the United States Renal Data System and the Social Security Death Master File. Astellas Pharma provided a copy of the trial data set including patient identifiers to the United Network of Organ Sharing, which maintains the OPTN registry. United Network of Organ Sharing staff identified the study participants in the OPTN registry using the following identifiers: transplant center, transplant date, recipient date of birth, recipient initials, recipient sex, donor type (living or deceased), donor age, and donor sex. A deidentified data set including a unique variable identifying trial participants in the registry was then provided to investigators for analysis. The trial outcomes of patient death and allograft failure censored for death were verified and additional outcomes in participants lost to trial follow-up were identified in the registry. Quiz Ref IDThe long-term outcomes of trial participants were determined in the OPTN registry until 2018 according to the intention-to-treat principle. Because the OPTN registry does not contain longitudinal information on immunosuppressive medications, a per-protocol analysis was performed among patients who continued taking the trial-assigned treatment 5 years after transplant as recorded in the trial data set. External validity was assessed by comparing the outcomes of trial participants with those among contemporaneous transplant recipients in the OPTN registry who would have met trial eligibility criteria and received the same immunosuppressive treatments as trial participants.
The time to allograft failure from any cause including death and allograft failure censored for death were determined by treatment group and in subgroups using the Kaplan-Meier method. Group differences were compared with the log-rank test.
The relative hazard of allograft failure from any cause including death and allograft failure censored for death in patients randomized to withdraw corticosteroids compared with patients who continued corticosteroids was determined using a Cox proportional hazards multivariable analyses including adjustment for differences in established determinants of transplant outcome including patient age, sex, race, history of diabetes, donor type, donor age, panel reactive antibody, the number of mismatches (0-6) at the human leukocyte antigen A, B, and DR gene loci, and type of induction immunosuppression were undertaken.
Similar methods were used to compare the outcomes of trial participant with those of registry patients. Cox multivariable regression models including trial participants and patients in the OPTN registry were performed including an interaction term for corticosteroid use by trial participation to determine if the association of corticosteroids with outcomes differed in trial patients compared with registry patients. Model assumptions were tested with plots of the log of the negative log of the estimated survival density function vs the log of survival time, and no violations were identified. Results were considered statistically significant for P value less than .05. There were no missing data in the trial data set. Analyses were conducted in R version 3.4.4 (R Foundation) with the survival package. Analysis began September 2018 and ended June 2019.
Among 197 patients randomized to corticosteroid withdrawal, 191 received the study-assigned treatment and were followed up in the trial for 5 years. All 191 patients were identified for long-term follow-up in the OPTN registry. Among 200 patients randomized to continue corticosteroids, 195 received the study-assigned treatment and 194 of these patients were identified for long-term follow-up in the OPTN registry (Figure 1). The characteristics of trial participants are shown in Table 1.
In the trial, 25 patients in the corticosteroid withdrawal group and 31 patients in the corticosteroid continuation group were lost to follow-up or withdrew consent during the 5-year trial follow-up period. All deaths (n = 24; 11 in corticosteroid withdrawal group and 13 in corticosteroid continuation group) and cases of allograft loss censored for death (n = 18; 11 in corticosteroid withdrawal group and 7 in the corticosteroid continuation group) reported in the trial were confirmed in the OPTN registry. Quiz Ref IDAdditional events during the trial were identified in the OPTN registry among trial participants who were lost to follow-up or withdrew consent. Among patients randomized to corticosteroid withdrawal, 6 additional patient deaths and 3 additional cases of death-censored graft loss were identified. Among patients who continued corticosteroids, 3 additional patient deaths and 9 additional cases of death-censored graft loss were identified. The additional events identified in the OPTN registry that were not included in the original trial report did not alter the trial conclusion of no difference in patient survival (174 of 191 [91%] in the corticosteroid withdrawal group; 179 of 195 [92%] in the corticosteroid continuation group; P = .65) or allograft survival censored for death (177 of 191 [93%] in the corticosteroid withdrawal group; 179 of 194 [92%] in the corticosteroid continuation group; P = .21) between the treatment groups 5 years after transplant.
The outcomes in the trial participants beyond 5 years after transplant were ascertained in the OPTN registry until June 2018. The median (interquartile range) of patient follow-up was 15.8 (12.0-16.3) years.
In the intention-to-treat analysis, the time to allograft failure from any cause including death and the time to allograft failure censored for death were not different in patients randomized to treatment with or without corticosteroids (Figure 2). Quiz Ref IDThe unadjusted hazard ratio (HR) of allograft loss from any cause in patients randomized to corticosteroid withdrawal was 0.84 (95% CI, 0.64-1.12; P = .23), compared with patients who continued corticosteroids, while the HR for allograft loss censored for death was 0.77 (95% CI, 0.52-1.16; P = .21). These hazards were similar in multivariable analysis (eTable 1 in the Supplement). Actual allograft half-lives defined as the time from transplant until allograft failure from any cause including death were 14.7 years among patients who continued corticosteroids and 17.5 years among patients who discontinued corticosteroids. Similar results were found in the per-protocol analysis (eFigure 1 in Supplement 2).
The time to death at any time after transplant (eFigure 2 in the Supplement) (P = .65) and time to death censored at time of allograft failure (eFigure 3 in the Supplement) (P = .41) were not significantly different between patients randomized to treatment with or without corticosteroids.
The time to allograft failure from any cause including death in patient subgroups is shown in eFigure 4 in the Supplement). As anticipated, living donor recipients, non–African American recipients, patients without rejection, and patients without diabetes had better outcomes. However, there was no statistically significant differences in allograft survival among these subgroups between patients treated with or without corticosteroids. Findings were similar for the outcome of allograft failure censored for death (eFigure 5 in the Supplement).
eFigure 6 in the Supplement shows the 3540 transplant recipients in the OPTN registry during the original trial who would have met trial eligibility criteria and received the same treatment as trial participants. The characteristics of contemporary registry patients grouped by whether they were (n = 3343) or were not (n = 197) treated with maintenance corticosteroids are shown in Table 2. Contemporary registry patients were similar to trial participants with the exception of receiving more nondepleting induction. During the median (interquartile) follow-up of 15.8 (9.4-16.8) years, the outcomes of trial participants treated with or without corticosteroids were similar to those of registry patients Figure 3.
In multivariable analyses that combined trial patients and OPTN registry patients (eTable 2 in the Supplement), the HR for graft loss from any cause including death was 0.86 (95% CI, 0.70-1.08), while the HR for allograft loss censored for death was 1.06 (95% CI, 0.79-1.40) in corticosteroid withdrawal patients compared with patients treated with corticosteroids. Trial participants had a similar outcomes as OPTN registry patients (allograft failure from any cause: HR, 0.90; 95% CI, 0.73-1.10; P = .31; allograft failure censored for death: HR, 1.01; 95% CI, 0.75-1.34; P = .97). Furthermore, the interaction terms for corticosteroid use by trial participation were nonsignificant (allograft failure from any cause HR: 0.99; 95% CI, 0.70-1.41; P > .99; allograft failure censored for death: HR, 0.76; 95% CI, 0.46-1.25; P = .28), indicating that the association of corticosteroid use with either outcome did not vary significantly between trial participants and registry patients.
This study demonstrates the long-term safety of corticosteroid withdrawal in kidney transplant recipients after a median follow-up of 15.8 years after transplant by extending the follow-up of a rigorous randomized clinical trial of corticosteroid withdrawal. The 5-year duration of follow-up in the original trial is exceptional for an immunosuppressive trial in kidney transplant and was informed by an earlier Canadian multicenter trial of corticosteroid withdrawal in cyclosporine-treated patients that reported an increased risk of late allograft failure only after 3 years of transplant.3Quiz Ref ID Despite the rigorous design and 5-year follow-up of the original trial and evidence that patients preferentially favor discontinuation of corticosteroids over other immunosuppressant drugs, only 30% of patients with similar characteristics to trial participants who have undergone transplant since the trial publication have been treated with corticosteroid withdrawal.8,14 This is particularly concerning because the characteristics of trial participants are similar to those among the majority (approximately 65%) of kidney transplant recipients in the United States today.8
The reluctance to adopt corticosteroid withdrawal may in part be related to higher incidence of early short-term rejection. Short-term rejection is an important complication that is associated with an increased risk of allograft failure.15 In the original trial, there was no increase in moderate or severe short-term rejection events in patients withdrawn from corticosteroids, and the 7% higher incidence of biopsy-confirmed rejection in patients withdrawn from corticosteroids was due to histologically low-grade rejection events.5 Post hoc analysis revealed a higher incidence of long-term histological changes in patients randomized to corticosteroid withdrawal leading to concerns that this strategy might still be associated with inferior long-term outcomes. Understanding of the long-term implications of histologically low-grade T-cell–mediated rejection events that are easily treated with short courses of high-dose corticosteroids is evolving and has been informed by advances in molecular diagnostics and more recent trials with novel immunosuppressant drugs that have suggested no long-term sequelae of these rejection events.16-18 It is notable that trial participants who experienced rejection had inferior long-term allograft survival compared with those who did not, but there was no difference in survival based on treatment group assignment. The subgroup of trial participants with short-term rejection within 5 years after transplant (n = 55) is too small to understand the effect of the histological features of rejection or other patient- and treatment-related factors that may modify the association of rejection with long-term allograft survival.19
The time to allograft failure from any cause including death did not differ between trial participants randomized to treatment with or without corticosteroids. An increase in long-term patient survival might have been anticipated because the adverse effects of corticosteroids, including dyslipidemia, hyperglycemia, hypertension, and obesity, are risk factors for cardiovascular disease, which is the leading cause of death in transplant recipients. A number of studies have demonstrated improvements in cardiac risk factors in steroid-free protocols.10,20-23 The original trial publication reported some benefit in cardiovascular risk factors among patients allocated to corticosteroid withdrawal, including lower serum triglyceride levels, less requirement for insulin in those who developed new-onset diabetes, and less weight gain. However, there was no difference in blood pressure, high-density lipoprotein and low-density lipoprotein cholesterol, or new-onset diabetes between treatment groups.5 Unfortunately, the OPTN registry does not contain longitudinal information about posttransplant complications, and therefore we are unable to determine long-term differences in cardiovascular risk factors, cardiac events, or other long-term metabolic complications of corticosteroids between treatment groups.
The external validity of the trial findings in the United States was assessed by comparison with 3540 contemporaneous patients in the OPTN registry who likely would have met the trial eligibility criteria including 197 patients withdrawn from corticosteroids. In a multivariable analysis including trial participants and registry patients, corticosteroid withdrawal was not associated with an increased risk of graft loss, trial participants had a similar hazard for graft loss as registry patients, and the association of corticosteroid withdrawal with allograft failure did not differ between trial participants and registry patients. Although these findings are reassuring, they are based on observational data and may be subject to residual confounding.
The study also informs the feasibility of using national registries to determine long-term outcomes of previously completed clinical trials and to support future trials in transplant. New strategies to increase clinical trials in transplant are needed because there has been limited improvement in long-term outcomes. Outcomes not captured in the trial publication were identified in the registry. Therefore, potential advantages of using national registries to support clinical trials include improved ascertainment of outcomes during the conduct of a trial as well as ascertainment of long-term outcomes after trial completion. To our knowledge, this approach has only been applied using the Australia and New Zealand Transplant Registry.24,25 Long-term outcome assessment in the OPTN registry is currently limited to allograft failure and patient death, and we were unable to determine outcomes such as rejection, kidney function, metabolic complications, cardiovascular events, and infections. The utility of the OPTN registry to support clinical trials could be enhanced by additional prospective data collection in trial participants. We were unable to determine long-term immunosuppressive drug treatment in the OPTN registry, and therefore our per-protocol analyses are based on trial data at 5 years after transplant and it is possible that changes in immunosuppression may have contributed to the similar outcomes observed between treatment groups.
Long-term differences in nonfatal outcomes such as cardiovascular disease, diabetes, infections and metabolic bone disease could not be determined between patients treated with and without maintenance corticosteroids.
In summary, corticosteroid withdrawal is not associated with an increased risk of long-term allograft failure from any cause including death or allograft failure censored for death in low– to moderate–immune risk kidney transplant recipients treated with induction immunosuppression tacrolimus and mycophenolic mofetil.
Corresponding Author: John S. Gill, MD, MS, St Paul’s Hospital, 1081 Burrard St, Providence Building Ward 6a, Vancouver, BC V6Z 1Y6, Canada (email@example.com).
Accepted for Publication: November 27, 2020.
Published Online: February 3, 2021. doi:10.1001/jamasurg.2020.6929
Author Contributions: Drs Gill and Clark had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Woodle and Gill contributed equally as co–first authors.
Concept and design: Woodle, Gill, First.
Acquisition, analysis, or interpretation of data: Woodle, Clark, Stewart, Alloway, First.
Drafting of the manuscript: Woodle, Gill, Clark, First.
Critical revision of the manuscript for important intellectual content: Woodle, Clark, Stewart, Alloway, First.
Statistical analysis: Clark, Stewart.
Obtained funding: Woodle.
Administrative, technical, or material support: Woodle, Stewart, Alloway, First.
Supervision: Woodle, Gill.
Conflict of Interest Disclosures: Dr Woodle reports grants from Astellas Pharma and grants and personal fees from Sanofi during the conduct of the study. Dr Gill reports grants and personal fees from Astellas Pharma outside the submitted work. Dr Alloway reports grants from Astellas Pharma during the conduct of the study. No other disclosures were reported.
Funding/Support: Dr Gill is supported by a foundation grant from the Canadian Institutes of Health Research.
Role of the Funder/Sponsor: The funding organization and original trial sponsor (Astellas) had on role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The interpretation and reporting of these data are the responsibility of the author(s) and in no way should be seen as official policy of the US government.
Additional Contributions: We wish to acknowledge the original trial sponsor, Astellas Pharma, who supported the original trial by making the original trial data set available, the transplant centers that participated in the original trial, and Aparna Sadavarte, MS, at United Network of Organ Sharing who identified the trial participants in the Organ Procurement and Transplant Network registry. No compensation was received outside their standard salary.