Advances in immunosuppression have facilitated increased use of steroid-avoidance protocols in pediatric kidney transplantation. To evaluate such steroid avoidance, a retrospective cohort analysis of pediatric kidney transplant recipients between 2002 and 2009 in the United Network for Organ Sharing database was performed. Outcomes (acute rejection and graft loss) in steroid-based and steroid-avoidance protocols were assessed in 4627 children who received tacrolimus and mycophenolate immunosuppression and did not have multiorgan transplants. Compared to steroid-based protocols, steroid avoidance was associated with decreased risk of acute rejection at 6 months posttransplant (8.3% vs. 10.9%, p = 0.02) and improved 5-year graft survival (84% vs. 78%, p < 0.001). However, patients not receiving steroids experienced less delayed graft function (p = 0.01) and pretransplant dialysis, were less likely to be African-American and more frequently received a first transplant from a living donor (all p < 0.001). In multivariate analysis, steroid avoidance trended toward decreased acute rejection at 6 months, but this no longer reached statistical significance, and there was no association of steroid avoidance with graft loss. We conclude that, in clinical practice, steroid avoidance appears safe with regard to graft rejection and loss in pediatric kidney transplant recipients at lower immunologic risk.
Steroids have traditionally played a central role in immunosuppression for pediatric kidney transplant recipients, preventing T cell activation and thereby protecting allografts from immune-mediated damage (1). Harmful sequelae of prolonged steroid therapy are well known and include cataracts, diabetes, hypertension, infection and cardiovascular morbidity (2). Furthermore, of particular importance in the pediatric population is the growth-suppressive effect of long-term steroid exposure (3,4). Recently, there has been an increasing interest in the development of posttransplant immunosuppression protocols which limit or avoid exposure to steroids in order to minimize these undesired side effects.
Several trials of complete steroid avoidance (5–9) or early steroid withdrawal (10–12) in pediatric patients have been published with encouraging results, with graft survival and incidence of acute rejection similar to steroid-based protocols. These studies, however, have been mostly limited to single center experiences. Furthermore, only historical controls (often treated prior to the routine use tacrolimus and mycophenolate) were used for comparison.
We sought to more comprehensively investigate the effect of steroid avoidance on graft outcomes by using data collected by the United Network for Organ Sharing (UNOS) on all pediatric kidney transplant recipients from 2002 to 2009, an era characterized by the predominant use of tacrolimus and mycophenolate as integral parts of maintenance immunosuppression. Our primary aims were to evaluate the association of steroid-avoidance protocols with graft survival and development of acute rejection within 6 months of transplant. We hypothesized that after controlling for confounding variables, steroid avoidance would not be associated with poor graft outcomes.
This study was a retrospective cohort analysis of all pediatric transplant recipients from 2002 to 2009 using data from the Organ Procurement and Transplantation Network (OPTN) database. All data were provided by OPTN. The database contains immunosuppression regimens at the time of hospital discharge and the most recent information updated from follow-up visits, which occur 6 months after transplant and on the transplant anniversary for every living patient with a functioning graft. Mandatory information includes graft status (functioning or failed), patient status, date of graft failure, and if there were any episodes of acute rejection in the follow-up period. Most recent serum creatinine values are also recorded for each follow-up encounter. In total, there were 6920 children less than 18 years of age who received a kidney transplant from 2002 through 2009. We excluded those patients with multiorgan transplants (n = 470) and those not receiving both tacrolimus and mycophenolate at hospital discharge (n = 1823), which resulted in a final cohort of 4627 patients. Of the excluded patients not receiving tacrolimus and mycophenolate, only 109 patients were discharged without maintenance steroids and were prescribed a combination of one antimetabolite and one calcineurin inhibitor. Our cohort was divided into two groups, those patients who were discharged on maintenance steroids and those who were discharged without maintenance steroids.
Patients in “steroid-avoidance protocols” were defined as those patients discharged without maintenance steroids. Baseline characteristics, including demographic information and immunologic risk factors, were compared between the two groups. Covariates included recipient age, gender, race, year of transplantation, donor source (living or deceased), human leukocyte antigen (HLA) mismatch status, panel reactive antibody (PRA) of > 20%, dialysis prior to transplant, history of prior kidney transplant, delayed graft function (DGF, defined as the need for dialysis within one week of transplant), cold ischemia time > 24 hours, donor age and use of induction agent. Our outcomes were episodes of acute rejection occurring within 6 months posttransplant and graft survival time. Acute rejection was defined as any clinically treated and reported episode. Graft survival time was defined as return to dialysis or retransplant, with patients censored who died with a functioning transplant. Graft function was also assessed using the 2009 Schwartz equation (13) to estimate glomerular filtration rate (GFR) in those patients with serum creatinine reported at discharge and at one and 5 years posttransplant. One-year creatinine values were included if they were reported within 30 days of the 1-year transplant anniversary, and 5 year values if they were within 90 days of the 5-year transplant anniversary.
Statistical analyses were performed using SAS 9.2 statistical software. For descriptive statistics, categorical variables were reported as percentages, and continuous variables reported as median and interquartile range. In univariate analyses, the Wilcoxon rank–sum test and t-test were used for continuous variables and chi-square testing for categorical variables. Graft survival was examined with Kaplan–Meier curves using the log-rank test. Patients lost to follow-up were censored at their last recorded visit. Multivariable logistic regression and Cox proportional hazards regression analysis were performed using the independent variable of interest (steroid use) along with all variables associated with acute rejection (p < 0.1) and graft loss in univariate analyses. Variables with a p-value of <0.05, selected using backward elimination, were then considered statistically significant and included in final regression models. Odds ratio of acute rejection and hazard ratios of graft loss were reported for each independent predictor along with Wald 95% confidence intervals.
Demographic and clinical characteristics were compared between children who received steroid-avoidance and steroid-based protocols (Table 1). Patients receiving steroid-avoidance protocols tended to be non-African American patients who were less sensitized and received a first kidney transplant from a living donor. In addition, they were more likely to receive induction with a lymphocyte-depleting agent and less likely to receive interleukin-2 receptor antibodies. Age difference was statistically different (with steroid avoidance prescribed for younger patients), although the median age was the same in both groups. As shown in Figure 1, the use of steroid-avoidance protocols increased during the study period, with 8.7% and 37.0% of patients treated with steroid avoidance in 2002 and 2009, respectively. There were no differences in gender, HLA match status, donor age, or cold ischemia time. In total, 27.6% of our patient cohort received steroid avoidance. At the first follow-up visit which occurred after 90 days posttransplant (median 182 days), 14.9% of patients who were initially receiving steroid avoidance were on steroid therapy. Similarly, 12.5% of patients receiving steroids were weaned off steroids at follow-up (median 183 days).
Table 1. Demographic and clinical characteristics—steroid based versus steroid avoidance
Received steroids (n = 3351)
Steroid avoidance (n = 1276)
Data presented as median (interquartile range) or percent.
Comparisons were made using the chi-square test for binary variables and Wilcoxon rank–sum test for continuous variables.
Cold ischemia time includes only those patients who received a deceased donor kidney transplant.
1Lymphocyte depleting agents included antithymocyte globulin, OKT3 and alemtuzumab.
2Interleukin-2 receptor antibodies included daclizumab and basiliximab.
324 subjects with missing data.
4 1363 subjects with missing values.
549 subjects with missing data.
6257 patients with missing values.
13 (8, 16)
13 (7, 16)
Lymphocyte depleting agent1
IL-2 receptor antibody2
Donor type (deceased)
HLA match category3
PRA > 20%4
Prior kidney transplant
Dialysis prior to transplant5
Delayed graft function
Cold ischemia time > 24 h6
27 (19, 38)
28 (20, 38)
Median transplant year
2006 (2004, 2008)
2007 (2005, 2008)
Patients receiving steroid-avoidance protocols were less likely to have an episode of acute rejection at 6 months after transplant. Specifically, at 6 months 90/1089 (8.3%) patients developed acute rejection in the steroid-avoidance group compared to 301/2765 (10.9%) of patients who received maintenance steroids (p = 0.02). 14.7% of steroid-avoidance patients did not have rejection data reported at 6 months, compared to 17.5% in protocols with maintenance steroids at discharge, which was statistically significant (p = 0.02).
Results of multivariate logistic regression analysis are shown in Table 2. After adjusting for all factors associated with rejection, steroid-avoidance protocols did not reach significance, although they still tended toward less episodes of acute rejection (odds ratio = 0.79, p = 0.07).
Table 2. Estimated odds ratios for factors associated with acute rejection at 6 months
Unadjusted OR (95% CI)
Adjusted OR (95% CI)
DGF = delayed graft function; OR = odds ratio.
Age (per 1-year increase)
1.06 (1.04, 1.08)
1.05 (1.03, 1.08)
1.57 (1.21, 2.03)
1.37 (1.05, 1.79)
2.82 (2.05, 3.86)
2.57 (1.86, 3.54)
0.74 (0.58, 0.94)
0.79 (0.62, 1.02)
Effect of acute rejection on graft survival
An episode of acute rejection occurring within 6 months of transplant was associated with decreased graft survival in patients receiving both steroid-based and steroid-avoidance protocols (p < 0.001, Figures 2A and B). Death-censored graft survival in children receiving steroids and experiencing acute rejection was 92.2% at 1 year and 61.8% at 5 years compared to 99.0% and 82.2% in those without acute rejection. Similarly, in steroid-avoidance protocols, graft survival was 92.1% at 1 year and 64.5% at 5 years in those with acute rejection and 98.8% and 86.6% at 1 and 5 years in children without an episode of acute rejection.
Overall graft survival
Patients receiving steroid-avoidance protocols demonstrated improved overall death-censored graft survival, as displayed in Figure 3. Graft survival was 97.5% at 1 year and 84.3% at 5 years in steroid-avoidance protocols compared to 97.0% and 78.0% in children receiving steroids (p < 0.001). The unadjusted hazard ratio for steroid-avoidance protocols was 0.72 (p < 0.001). In multivariate analysis controlling for factors associated with graft loss (Table 3), steroid-avoidance protocols tended to have prolonged survival with a hazard ratio of 0.84, although this no longer reached statistical significance (p = 0.08). Because steroid-avoidance protocols were associated with less acute rejection, we added acute rejection within 6 months as a covariate to the multivariate analysis. With this addition, the hazard ratio for steroid avoidance was 0.93 (p = 0.51).
Table 3. Estimated hazard ratios for factors associated with death-censored graft survival
Unadjusted HR (95% CI)
Adjusted HR* (95% CI)
Adjusted HR including AR (95% CI)
*PRA > 20% was statistically significant in the multivariate analysis not including AR; however, due to the large amount of missing values (1363 patients) it was not included. Removal from the model did not significantly change adjusted HR of steroid avoidance (HR = 0.85 with PRA included as a covariate).
AA = African American race; DGF = delayed graft function; HR = hazard ratio; AR = acute rejection.
Age (per 1-year increase)
1.11 (1.09, 1.13)
1.10 (1.08, 1.12)
1.10 (1.08, 1.13)
0.73 (0.63, 0.85)
0.75 (0.64, 0.87)
0.75 (0.63, 0.89)
Non AA race
2.19 (1.87, 2.56)
1.79 (1.52, 2.11)
1.94 (1.61, 2.34)
1.89 (1.61, 2.21)
1.39 (1.18, 1.65)
1.54 (1.27, 1.86)
6/6 HLA mismatch
0/6 HLA mismatch
3.68 (2.07, 6.54)
2.74 (1.54, 4.89)
2.35 (1.25, 4.44)
1–5/6 HLA mismatch
0/6 HLA mismatch
2.15 (1.24, 3.73)
2.09 (1.21, 3.63)
1.79 (0.98, 3.26)
1.73 (1.43, 2.09)
1.31 (1.08, 1.60)
2.47 (1.97, 3.09)
1.94 (1.54, 2.43)
AR in 6 months
No AR in 6 months
2.90 (2.36, 3.56)
2.51 (2.04, 3.09)
0.72 (0.60, 0.88)
0.84 (0.69, 1.02)
0.93 (0.75, 1.15)
Overall graft function
Baseline GFR was available in 90.6% of our cohort (n = 4192) and did not differ between patients receiving steroid-avoidance and steroid-based protocols, with estimated mean GFR values of 71.8 ± 35.2 and 72.3 ± 38.8 ml/min/1.73 m2, respectively (p = 0.67). At 1 year, 56.1% of children (n = 2598) had GFR data available, and graft function was slightly improved in children receiving steroid avoidance (71.2 ± 25.3 mL/min/1.73 m2 vs. 68.1 ± 24.3 mL/min/1.73 m2, p < 0.01). Five-year mean GFR values were available for only 18.2% of children (n = 842) and were 60.7 ± 23.5 mL/min/1.73 m2 in steroid-avoidance patients and 58.6 ± 20.1 mL/min/1.73 m2 in children receiving steroids, which was not significant (p = 0.25).
We have demonstrated that steroid-avoidance protocols were not associated with worse graft outcomes in pediatric kidney transplant recipients receiving tacrolimus-based immunosuppression regimens which included mycophenolate. Two meta-analyses of adult randomized controlled trials of steroid withdrawal and complete avoidance protocols conducted prior to 2000 demonstrated an increased risk of graft failure and acute rejection (14,15). These meta-analyses included, however, studies conducted during a period of time when cyclosporine was the mainstay of immunosuppressive therapy. With the advent of tacrolimus, improved graft outcomes and decreased rates of rejection have been demonstrated both in adult and pediatric populations (16,17). This heralded the successful implementation of steroid withdrawal in clinically stable pediatric kidney transplant recipients (18,19) followed by the development of steroid-avoidance protocols (5–9). Furthermore, a recent Cochrane meta-analysis of adult randomized controlled trials demonstrated that there was no increased risk of acute rejection or graft loss in steroid-sparing regimens that used tacrolimus for baseline maintenance immunosuppression (20). The findings of our study further support the safety of steroid-avoidance protocols in appropriate pediatric kidney transplant recipients.
Our results confirm the findings of several smaller, single center trials of steroid-avoidance protocols in children in which graft function, graft survival and episodes of acute rejection were comparable to steroid-based protocols (5–9). Our study is the first, however, to include multiple centers and to control for known risk factors for poor outcome in pediatric recipients. Furthermore, we included only patients who received tacrolimus-based immunosuppressive regimens, while many previous studies compared steroid-avoidance protocols to historical controls who received different regimens, possibly confounding any encouraging results.
In unadjusted analysis, steroid-avoidance protocols were associated with improved 1-year graft function and death-censored graft survival, which underscores important differences in children participating in steroid-avoidance strategies. Specifically, steroid-avoidance protocols were less likely to include African-Americans, recipients of grafts from deceased donors, and children with a history of DGF or pretransplant dialysis, all known risk factors for poor outcome (21) that were associated with graft loss in the multivariate analysis. Perhaps most importantly, children in steroid-avoidance protocols experienced fewer episodes of acute rejection, a major risk factor for decreased graft survival that was confirmed in our cohort (22,23). In multivariate analysis controlling for acute rejection, the association of steroid avoidance with improved outcomes was thus no longer significant. The results of this study, therefore, demonstrate the impact of confounding variables when evaluating the effect of steroid use on graft outcome. Furthermore, it suggests that retrospective studies which do not account for important variables known to affect graft outcome may be inherently biased.
Improved 1-year creatinine-estimated GFR in our cohort should also be interpreted with caution, as adult and pediatric randomized controlled trials have demonstrated equivalent renal function in steroid-sparing regimens compared to steroid-based protocols (3,24,25). In addition to confounding variables, creatinine assay variability between groups (Jaffe reaction vs. enzymatic techniques) may have contributed to the small difference in GFR estimation. Steroid-avoidance patients continued to have slightly improved renal function at 5 years, although the small sample size at this time point likely contributed to the lack of statistical significance.
Interestingly, HLA match status was not associated with steroid avoidance. The percentage of children receiving steroid-avoidance protocols was between 26% and 28% among all HLA match categories, suggesting that recipients of poorly matched allografts are equally eligible for steroid-avoidance protocols. In contrast, several studies have included elevated PRA values among the exclusion criteria for steroid avoidance (3,6,11) which is consistent with the association of lower PRA values with steroid avoidance observed in our cohort.
Although steroid avoidance was not associated with increased risk of acute rejection and graft loss in multivariate analysis, the observational, nonrandomized design of this study prevents any strictly causal inferences about the benefit of steroids, especially in patients at higher immunologic risk. Two randomized trials in adult patients compared steroid-free or early avoidance regimens to triple immunosuppressive regimens including steroids, mycophenolate and tacrolimus and found no difference in graft survival. However, both studies demonstrated an increased incidence of biopsy-confirmed acute rejection in the absence of steroids (25,26). The only randomized trial of early steroid withdrawal in pediatric patients was the TWIST study, which compared steroid withdrawal in children receiving daclizumab, mycophenolate and tacrolimus to a control group receiving steroids, tacrolimus and mycophenolate (3). The authors found no significant difference in biopsy-proven acute rejection or graft survival between the experimental and control arms. However, this study was not powered to assess graft outcomes and was limited by follow-up of only 6 months. A major strength of our study is that it allowed for a minimum of 1-year follow-up, with an average follow-up of 3.4 years (interquartile range 1.8–4.9 years).
During the study period (2002–2009), the use of steroid-avoidance protocols increased from 8.7% in 2002 to a peak of 37% in 2009 in our cohort, rates similar to those reported in adult kidney recipients (27). Both open-label and randomized prospective studies of steroid-avoidance protocols have shown improvements in lipid profiles, blood pressure, and—most importantly—growth with steroid avoidance at 6 months to 1 year of follow-up (3,6). Therefore, the results of our study add to a body of literature which suggests that the risks of steroid avoidance are minimal and support the growing efforts to avoid harmful side effects associated with long-term use of steroids, especially in patients at low immunologic risk.
The strengths of our study include its large pediatric sample size (4627 patients), an extended period of follow-up and the comprehensive nature of the data obtained from the OPTN database, which contains every kidney transplant performed in the United States. Additionally, minority patients were well represented in our cohort, with 20% of patients African American and 24% Hispanic. Nonetheless, there are several limitations which deserve mention. First, the outcome of acute rejection was diagnosed clinically, and there was no requirement to have biopsy-proven acute rejection. Therefore, it is possible that this may have led to a misclassification bias, although we do not suspect that this would have significantly affected results. Second, outcomes in the OPTN database were recorded data from collection forms completed by medical professionals at discharge and at routine follow-up visits; therefore, a significant degree of attrition and lack of follow-up data were unavoidable. Furthermore, patients receiving maintenance steroids were less likely to have rejection at the 6-month interval reported. This may have been secondary to more vigilant follow-up practices for patients in steroid-avoidance protocols, either as a center-specific effect or due to preferential inclusion of patients with known compliance in steroid-avoidance protocols. Presuming patients lacking follow-up data are at increased risk for poor outcomes, this could only have artificially improved outcomes in children receiving steroids, which would not affect the overall conclusions of this study. Third, patients were categorized according to steroid prescription at discharge, although the use of steroids in approximately 13% of children had changed at their 6-month follow-up. In particular, it is tenable that for some patients steroids may have been weaned off shortly after discharge. Finally, and although we attempted to adjust for variables associated with graft outcomes, we could not control for center-specific variation or nonrepresented patient characteristics such as a history of previous compliance prior to transplant. It is probable that transplant centers implementing steroid-avoidance protocols have local practices that contributed to the improved outcomes in these patients, especially to the decreased risk of acute rejection.
In summary, this study demonstrates that steroid avoidance as clinically practiced in selected pediatric kidney transplant recipients is safe, not conferring an increased risk of acute rejection or graft failure. Although large randomized controlled trials to assess long-term risks of steroid avoidance may not be feasible, recent evidence does demonstrate that steroid avoidance improves growth and cardiovascular risk factors in children. Therefore, our results suggest that steroid-avoidance strategies are a reasonable approach in pediatric transplant recipients at low immunologic risk.
The authors have no funding sources to disclose. This work was supported in part by Health Resources and Services Administration contract 231-00-0115. The content is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government.
The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.