A Comparative Effectiveness Analysis of Early Steroid Withdrawal in Black Kidney Transplant Recipients

Abstract

Background and objectives

There is continued debate whether early steroid withdrawal is safe to use in high–immunologic risk patients, such as blacks. The goal of this study was to use comparative effectiveness methodology to elucidate the safety of early steroid withdrawal in blacks with kidney transplants.

Design, setting, participants, & measurements

Our cohort study used United Network of Organ Sharing data including all adult black kidney transplant recipients from 2000 to 2009 followed through 2014. Propensity score matching was used to equalize baseline risk between continued steroid and early steroid withdrawal groups. Interaction terms were used to assess if the effect of early steroid withdrawal on outcomes varied by baseline and post-transplant factors. Of 26,582 eligible black patients with kidney transplants (5825 [21.9%] with early steroid withdrawal), 5565 patients with early steroid withdrawal were matched to 5565 blacks on continued steroid use.

Results

Black patients with early steroid withdrawal had similar risk of graft loss (hazard ratio, 0.98; 95% confidence interval, 0.92 to 1.04; P=0.42) and lower risk of death (hazard ratio, 0.91; 95% confidence interval, 0.84 to 0.99; P=0.02), primarily driven by a late mortality advantage (>4 years post-transplant). Delayed graft function, cytolytic induction, tacrolimus, and mycophenolate significantly modified the effect of early steroid withdrawal on outcomes (P<0.05). Acute rejection rates were slightly higher in the continued steroid group (13.0% versus 11.3%, respectively; P<0.01), but this was not associated with graft or patient survival.

Conclusions

Overall, early steroid withdrawal in black kidney transplant recipients was not associated with graft loss but seemed to be associated with better long–term patient survival. Early steroid withdrawal in blacks not receiving cytolytic induction, tacrolimus, and mycophenolate or those with delayed graft function was associated with higher risk of graft loss and death.

Introduction

Because of advances in medical management and immunosuppression, graft survival for kidney transplant recipients has dramatically improved over the past 40 years (1). In 1977, the average graft lasted only 2–3 years (2). The most recent Scientific Registry of Transplant Recipients annual report shows that the average graft half–life is 11.3 years, thus increasing by nearly fivefold over this time period (1). However, graft survival is still suboptimal, with the predominant causes of late graft loss being attributable to medication–related safety issues, including nonadherence and adverse drug reactions (3,4). To address the latter issue, transplant programs have implemented and tested protocols that seek to minimize or eliminate the long-term use of immunosuppressant medication with considerable toxicities (5,6).

Corticosteroids are considered a cornerstone to immunosuppressant regimens but carry the burden of considerable chronic toxicities (7). Starting in the 1990s, transplant programs began using more potent immunosuppression regimens to use early steroid withdrawal (ESW), hoping to minimize the detrimental long–term adverse drug reactions associated with chronic corticosteroids (1,7). Two randomized, controlled trials (RCTs) showed that ESW can be potentially used in a safe manner, with similar rates of graft loss. There may be added benefit of intermediate improvement in hyperglycemia, hypertension, weight gain, and bone health, although demonstration of long-term benefit is lacking (8,9). ESW is used across a significant number of transplant programs, increasing from 4.3% in 1998 to 32.4% in 2012 (1). However, it is important to note that the use of corticosteroid withdrawal must be balanced with the potential of increased risk of acute rejection and chronic allograft nephropathy.

There continues to be controversy over ESW’s long-term safety within high–immunologic risk populations (6,7). Blacks are considered such a group, because the literature has clearly shown that they have a number of important factors that increase immunologic risks (10). These include lower rates of living donors, socioeconomic disadvantages, greater HLA mismatches, higher capacity to metabolize immunosuppressant medications, and polymorphisms that increase immunoreactivity (10–13). Additionally, the aforementioned RCTs that assessed ESW safety contained disproportionately low numbers of blacks (8,9). Thus, the safety of ESW in black recipients has not been established; some literature shows that it may increase graft dysfunction and potentially, graft loss (14). The objective of this study was to use a large national dataset of kidney transplants and novel comparative effectiveness techniques to elucidate the long-term safety of ESW in black recipients.

Materials and Methods

Study Design and Patients

This was a cohort study using national data acquired through an institutional review board–approved data use agreement with the United Network of Organ Sharing transplant registry. This registry dates back to 1987 and contains all transplants performed within the United States, including baseline donor and recipient characteristics and clinical outcomes. The registry is merged with the Social Security Death Master File to obtain accurate information regarding patient mortality (15). This study focused on adult (≥18 years of age) black kidney transplant recipients between January 1, 2000 and December 31, 2009; follow-up was through December of 2014. Pediatrics; nonblacks; and those with graft loss, death, or lack of follow-up ≤1 month were excluded along with those with missing data or patients who received a nonrenal transplant.

Outcomes

The primary outcome was graft loss defined as a return to dialysis, retransplantation, or death. Additional outcomes included acute rejection, death, and last–reported serum creatinine. Acute rejection was defined as either biopsy-proven or clinically suspected and treated events.

Exposure Variables

ESW was defined as withdrawal of steroids by discharge from the hospital for the transplant event. We also did a sensitivity analysis, defining ESW as the withdrawal of steroids within 90 days of transplant. Additional baseline and perioperative variables included recipient demographics (age, sex, body mass index [BMI], insurance, and waiting list time), comorbidities (diabetes, previous transplant, hypertension, and cardiovascular disease), immunologic risks (HLA mismatches and current panel reactive antibody [PRA] levels), donor (donor age, living donor, donor race, donor sex, expanded criteria donor, and donor after cardiac death) and perioperative transplant characteristics (cold ischemic time, warm ischemic time, baseline induction therapy, baseline maintenance immunosuppression, and delayed graft function), and transplant year.

Statistical Analyses

To minimize the effect of confounding on the outcomes, the comparative effectiveness technique of propensity score matching was used. This was conducted using logistic regression to estimate the propensity to receive ESW in each patient on the basis of the included covariates, with nearest neighbor approach without replacement and a maximum caliper set at 0.1 in a one to one fashion. To ensure a successful match, detailed postmatch balance statistics were assessed, including the distribution of propensity scores and the absolute standardized differences in covariate means (Supplemental Figures 1–3).

Baseline and perioperative characteristics were compared between the unmatched and matched groups using standard statistical methods: the chi-squared test for categorical data and either the t test or Mann–Whitney U test for continuous data. Time to event outcomes (graft loss and death) were compared using Cox regression. An assessment of whether the effect of ESW on outcomes differed through moderation was conducted by using interaction terms. Those variables showing statistically significant interaction led to stratified analyses. We also conducted a sensitivity analysis by defining ESW as withdrawal within 90 days of transplant to compare outcomes with the withdrawal by discharge (results are in Supplemental Figures 4 and 5). Statistical analyses were conducted using SPSS, v22.0 (IBM SPSS, Chicago, IL). A two–sided P value of <0.05 was considered statistically significant. Propensity score matching and assessment were conducted using R (The R Foundation; https://www.r-project.org/) through the SPSS R-plugin and a macro developed by Felix Thoemmes (16,17).

Results

In total, 171,104 kidney transplants were performed between January of 2000 and December of 2009. Of these, 8293 were excluded for age; 2123 were excluded for receiving a nonrenal transplant; 125,349 were excluded for nonblack race; 7731 were excluded for missing data; and 1026 were excluded for having graft loss, death, or lack of follow-up within 1 month of transplant (Figure 1), leaving 26,582 available for analysis. Of these, 5825 (21.9%) were identified as having ESW (withdrawal of steroids at discharge). The baseline characteristics for the cohort are displayed in Supplemental Table 1, showing significant differences for the vast majority of variables, including age, sex, BMI, insurance, history of diabetes, preemptive status, retransplant status, donor age, donor type, donor race, waiting time, current PRA, peak PRA, cold ischemic time, baseline immunosuppression, and the mean propensity score (0.30±0.15), versus continued steroid (CS; 0.20±0.12; P<0.001). The propensity score matching was conducted using the variables listed in Table 1; 5565 (97.5%) ESW black transplant recipients were matched to 5565 patients with CS, with the resulting cohorts having very similar baseline characteristics and mean propensity scores of 0.29±0.15 in the CS group and 0.31±0.15 in the ESW group.

Figure 1.

Consort diagram displaying how the study cohort and comparator groups were created. HR, hazard ratio.

Table 1.

Baseline and perioperative characteristics compared between the matched early steroid withdrawal and continued steroid groups

Variable	Continued Steroid, n=5565	Early Steroid Withdrawal, n=5565
Median transplant year (IQR)	2007 (2004–2008)	2006 (2005–2008)
Mean recipient age, yr ±SD	49.5±12.6	49.8±12.7
Women, %	38.6	38.7
Recipient BMI, kg/m2 ±SD	28.3±5.5	28.3±5.7
Primary insurance Medicare/Medicaid, %	73.8	73.2
History of diabetes, %	33.4	34.1
Preemptive transplant, %	9.3	10.0
Retransplant, %	6.9	7.0
Mean donor age, yr ±SD	39.3±15.7	39.4±15.7
Living donor, %	17.2	17.4
Donor ECD, %	16.1	16.3
Donor DCD, %	8.9	9.1
Black donor, %	31.9	31.9
Woman donor, %	42.3	42.6
Median waiting time, yr (IQR)	1.8 (0.7–3.3)	1.7 (0.7–3.3)
Median HLA mismatches (IQR)	5 (3–5)	4 (3–5)
Median current PRA, % (IQR)	0 (0–7)	0 (0–7)
Median peak PRA, % (IQR)	2 (0–19)	2 (0–18)
Median cold ischemic time, h (IQR)	16.8 (9.4–23.3)	16.0 (8.0–23.5)
Immunosuppression, %
IL-2 RA induction therapy	14.6	12.8
Cytolytic induction therapy	67.4	68.1
Tacrolimus	83.1	82.6
Cyclosporin	9.1	7.8
MMF/MPA	83.5	81.5
Azathioprine	0.9	0.5
mTOR inhibitor	11.5	11.2
Delayed graft function, %	23.4	22.6
Propensity score	0.29±0.15	0.31±0.15

IQR, interquartile range; BMI, body mass index; ECD, expanded criteria donor; DCD, donor after circulatory death; PRA, panel reactive antibody; RA, receptor antibody; MMF, mycophenolate mofetil; MPA, mycophenolic acid; mTOR, mammalian target of rapamycin.

Primary and secondary outcomes are displayed in Table 2. Acute rejection rates were marginally higher at 6 months (9.3% versus 8.1%; P=0.03), 1 year (10.3% versus 9.3%; P=0.09), and the end of follow-up (13.0% versus 11.3%; P<0.01) in the CS group compared with the ESW group, respectively. The estimated graft survival rates were similar at all time points between the CS and ESW groups (Figure 2, upper panel, Table 2) (P=0.42). For mortality, ESW was associated with reduced risk of death (hazard ratio [HR], 0.91; 95% confidence interval, [95% CI], 0.84 to 0.99; P=0.02); this was primarily seen at 4 years post-transplant (Figure 2, lower panel).

Table 2.

Clinical outcomes compared between the continued steroid and early steroid withdrawal groups

Outcome	Continued Steroid, %	Early Steroid Withdrawal, %	P Value
Acute rejection
6 mo	9.3	8.1	0.03
1 yr	10.3	9.3	0.09
Overall	13.0	11.3	<0.01
Graft survival
3 mo	99.2	99.2	0.42
1 yr	96.0	96.1	0.42
3 yr	88.0	88.2	0.42
5 yr	79.9	80.2	0.42
10 yr	60.5	60.9	0.42
Patient survival
3 mo	99.3	99.3	0.02
1 yr	96.9	97.2	0.02
3 yr	90.9	91.7	0.02
5 yr	85.0	86.3	0.02
10 yr	63.5	66.3	0.02
Median last reported SrCr, mg/dl (IQR)	2.5 (1.6–5.1)	2.5 (1.6–5.3)	0.55

SrCr, serum creatinine; IQR, interquartile range.

Figure 2.

Estimated survival curves compared between those with early steroid withdrawal (ESW) versus continued steroid (CS) for the outcomes of graft loss (upper panel) and death (lower panel). HR, hazard ratio.

The assessment of effect modification showed that four variables significantly varied the effect of ESW on graft loss and death (Figure 3, Table 3). For graft loss, in those who received cytolytic induction therapy, ESW was associated with a 10.2% lower risk of graft loss (HR, 0.90; 95% CI, 0.83 to 0.97), whereas ESW was associated with a 12.1% higher risk of graft loss in those who did not receive cytolytic induction (HR, 1.12; 95% CI, 1.02 to 1.24). In those who received baseline tacrolimus maintenance therapy, the risk of graft loss was 7.8% lower between the ESW and CS groups (HR, 0.92; 95% CI, 0.86 to 0.99), whereas in those who did not receive tacrolimus, the risk of graft loss was 19.6% higher in the ESW cohort (HR, 1.20; 95% CI, 1.05 to 1.36). mycophenolate mofetil/mycophenolic acid therapy and delayed graft function were also significant effect modifiers for graft loss (Table 3). Similar effect modification for these four parameters was also noted for mortality (Table 3).

Figure 3.

Graft survival curves for the outcome of graft loss comparing early steroid withdrawal (ESW) with continued steroid (CS) stratified by variables that were significant effect modifiers (cytolytic induction,tacrolimus, and delayed graft function). AA, black; UNOS, United Network of Organ Sharing.

Table 3.

Hazard ratios for the risk of graft loss and death in early steroid withdrawal versus continued steroid (reference) stratified by modifying variables and time periods

Outcome	Cytolytic Inductiona		Tacrolimusa		MMF/MPAa		Delayed Graft Functiona
	Yes	No	Yes	No	Yes	No	Yes	No
Graft loss (95% CI)	0.90 (0.83 to 0.97)	1.12 (1.02 to 1.24)	0.92 (0.86 to 0.99)	1.20 (1.05 to 1.36)	0.93 (0.86 to 0.99)	1.15 (1.01 to 1.30)	1.19 (1.06 to 1.33)	0.91 (0.85 to 0.98)
Death (95% CI)	0.80 (0.71 to 0.89)	1.11 (0.97 to 1.26)	0.83 (0.76 to 0.92)	1.22 (1.03 to 1.44)	0.81 (0.73 to 0.89)	1.26 (1.06 to 1.48)	1.05 (0.91 to 1.22)	0.86 (0.78 to 0.95)

MMF, mycophenolate mofetil; MPA, mycophenolic acid; 95% CI, 95% confidence interval.

^aP<0.05 for interaction between variable and early steroid withdrawal in Cox regression models.

Supplemental Figures 4 and 5 display the results of the sensitivity analysis, in which ESW was defined as withdrawal of steroids within 90 days of transplant. The results were similar to those of the analysis defining ESW at the time of discharge for the outcomes of acute rejection, graft loss, and effect modification. For mortality, there was a notable difference, in that ESW by 90 days was associated with higher mortality risk early post-transplant, with a survival benefit from 3 to 10 years post-transplant (nonproportionality of the hazard over time).

Discussion

The results of this study, which used a large national dataset and comparative effectiveness analysis techniques to provide insight into a contentious therapeutic dilemma in transplantation, show that ESW in black kidney transplant recipients may be safe and effective for graft survival in those who receive potent immunosuppression (cytolytic induction, tacrolimus, and mycophenolate) and do not develop delayed graft function. The use of cytolytic induction and potent maintenance immunosuppression was associated with a late mortality benefit in blacks who undergo ESW as long as they did not develop delayed graft function. Because more than one in five of all black kidney transplant recipients currently undergo ESW, these results provide clinically relevant and novel insights that can help providers develop patient–specific immunosuppressive regimens in blacks as a mechanism to potentially reduce the long–term deleterious effect of chronic corticosteroids without inducing increased risk of graft loss or death.

There are two multicenter RCTs that compare steroid withdrawal with CS under contemporary immunosuppression (antibody induction, cyclosporin or tacrolimus, and mycophenolate) within kidney transplant recipients. The first enrolled 386 patients (ESW, n=191; CS, n=195), withdrawing steroids at 7 days post-transplant. Of these, 76 were black, with 34 in the ESW (18%) and 42 in the CS (22%) groups. Although the 5-year outcomes showed similar rates of death–censored graft survival (94.2% ESW versus 96.4% CS) and patient survival (94.2% ESW versus 93.3% CS), biopsy–proven acute rejection rates were higher in the ESW group (17.8% versus 10.8%, respectively). Additionally, multivariable analysis using Cox regression showed that blacks had more than four times the risk of graft loss (HR, 4.31; 95% CI, 1.46 to 12.7) compared with nonblacks (9). However, likely because of small numbers, the authors did not subanalyze the effect of ESW on graft survival only within blacks. This study allowed for the use of noncytolytic induction therapy, which may have led to the higher rates of acute rejection in the ESW group. A second older study using cyclosporin and mycophenolate with or without induction and withdrawing steroids at 3 months showed higher rates of acute rejection at 12 months post-transplant in the ESW group (25% versus 15%), with similar rates of 1-year graft survival (95% versus 96%). However, this study was conducted in Europe, with no reported patients of African descent (8).

The Induction with Tacrolimus (INTAC) Study, a multicenter RCT that assessed the effect of various induction agents (alemtuzumab versus antithymocyte globulin in high-immunologic patients and basiliximab versus alemtuzumab in low-immunologic patients) on graft outcomes in patients on tacrolimus and mycophenolate, all with withdrawal of steroids at 5 days post-transplant in all patients, did not specifically compare ESW with CS. However, 99 of 474 patients enrolled were black, and they were all classified as high immunologic risk. Within the high–immunologic risk group, the rate of acute rejection was fairly low and similar between alemtuzumab and antithymocyte globulin (18% versus 15%). Acute rejection rates were significantly higher in the low-risk group that received noncytolytic induction compared with those who received alemtuzumab induction (22% versus 10%; P=0.003). To our knowledge, the authors did not analyze outcomes specifically within the black patients (18). However, these results, taken in context of the study by Woodle et al. (9) and the data for this analysis, show that cytolytic induction is needed when ESW is used, particularly in black patients.

There are a number of single-center studies that attempt to determine the effect of ESW on graft outcomes specifically within black recipients. A study from Drexel included 206 patients (103 black) and compared graft outcomes between blacks and nonblacks in those undergoing ESW. All patients received basiliximab induction, cyclosporin or tacrolimus, and mycophenolate or sirolimus as maintenance immunosuppression. The incidence of acute rejection was similar between blacks and nonblacks (16% versus 14%, respectively); however, rates of graft dysfunction, serum creatinine, and subclinical acute rejection were all significantly higher in blacks. Although the authors question the safety of ESW in blacks on the basis of these results, this analysis did not compare ESW with CS specifically in blacks (14). In another single–center study by Padiyar et al. (19) from Case Western University, the authors showed that blacks were at significantly higher risk of acute rejection (odds ratio, 3.33) compared with nonblacks when undergoing ESW. However, as with all previous analyses discussed, this study also did not specifically compare ESW with CS in blacks and did not show any differences in graft survival or death (19). Finally, in a number of nonrandomized studies, Hricik et al. (20,21) showed that steroid withdrawal was safe in 44 black kidney transplant recipients using tacrolimus and sirolimus maintenance therapy without induction; only two patients (6.7%) developed acute rejection, and at a mean of 14.3 months of follow-up, 90% remained off steroids (20,21). There are additional studies that compare ESW in blacks versus nonblacks for graft outcomes, but no studies specifically compare ESW with CS in blacks (6,22,23). Thus, the results of the data presented in our analysis provide novel information that suggests that ESW may be safe and effective in blacks under particular potent immunosuppression regimens and those without delayed graft function.

It is not surprising that the effect of ESW on graft outcomes in blacks was substantially modified by the use of potent baseline immunosuppression. Black recipients are known to be at considerably higher immunologic risk compared with nonblack recipients. Studies have shown that blacks have more HLA polymorphisms (24), tend to be more likely to be presensitized to MHC antigens (25), have greater numbers of HLA mismatches (26), have immune hyper-responsiveness (27), and have a number of important cytokine polymorphisms (28,29), all of which place them at higher risk for acute allograft rejection and immunologically mediated graft loss. Our analysis supports this, because the effect of ESW on graft outcomes was significantly modified by the use of potent immunosuppression, including tacrolimus (compared with cyclosporin or no calcineurin inhibitor), mycophenolate, and cytolytic induction (compared with IL-2 receptor antagonists or no induction therapy). On the basis of this result and the results from the aforementioned INTAC Study (18), all of the available data strongly suggest that clinicians should use more potent immunosuppressive regimens (cytolytic induction with tacrolimus and mycophenolate maintenance therapy) in blacks when ESW is planned.

It is also reasonable to expect that other baseline factors that are strong immunologic risks, such as delayed graft function, would modify the effect of ESW on graft and patient outcomes within blacks (30,31). On the basis of this, transplant clinicians should use caution when attempting ESW in blacks who develop delayed graft function. It is noteworthy that a number of factors did not significantly affect the influence of ESW on outcomes in blacks, including recipient characteristics (age, sex, history of diabetes, BMI, waiting time, or insurance type), donor characteristics (living, deceased, expanded criteria donor, donor after circulatory death, age, sex, or race), and certain immunologic risk factors (current PRA or HLA mismatches). This should provide some level of comfort that ESW can potentially be safely used in blacks, regardless of these characteristics. The finding that current PRA did not substantially modify the effect of ESW on graft outcomes is of particular interest and may reflect that modern flow cytometry crossmatch techniques may minimize the direct effects of PRA on clinical outcomes (32–34).

In the sensitivity analysis, defining ESW within 90 days of transplant showed that ESW in blacks was associated with an increased risk of early death (within 3 months of transplant). This may be a reflection of a potential beneficial effect of early steroid use on mortality or represent prescribing bias, because ESW after discharge may have been used in patients because of clinical events (infection and cardiovascular events), which are not captured or controlled for in this analysis. However, within both ESW definition analyses, ESW was associated with a late mortality benefit (>4 years post-transplant), because cardiovascular disease, infection, and malignancy are the leading causes of death in kidney transplant recipients. These results suggest that ESW may be protective against these common causes of late death (35,36). However, because the true cause of death is not well documented in registry data, it is not feasible to determine a true causal relationship with ESW and mortality, and additional studies are warranted in this area.

The results of this analysis need to be taken in the context of its limitations. This study used registry data in a retrospective manner and thus, could be prone to confounding or bias. To limit this, we used propensity score matching to produce two cohorts that were remarkably similar in all baseline measurements and conducted sensitivity analyses to determine the influence of unmeasured prescribing bias. This analysis is also susceptible to error from entry by transplant centers, because it was not validated at the patient level. Furthermore, there are a number of important variables that may influence outcomes that could not be measured or included in this study, including immunosuppression dosing and drug levels, medication adherence, and additional baseline patient characteristics, particularly socioeconomic and social support measures (14,37,38). Given these limitations, this analysis should not be construed to provide the highest grade of evidence regarding the use of ESW in black kidney transplant recipients and should be interpreted in conjunction with previous studies, including the INTAC Study and other single–center analyses (18–23).

In conclusion, these results show that ESW in adult black kidney transplant recipients is likely safe and effective in those who receive cytolytic induction therapy and tacrolimus and do not develop delayed graft function. The results showing that ESW was associated with lower risk of late death require further validation and investigation into the potential etiologies.

Disclosures

None.