Abstract
The optimal long-term regimen for immunosuppression for kidney transplant recipients is unknown. We conducted a randomized trial involving 150 kidney transplant recipients to compare tacrolimus/sirolimus, tacrolimus/mycophenolate mofetil (MMF), and cyclosporine/sirolimus. All patients received daclizumab induction and maintenance corticosteroids. Median follow-up was 8 yr post-transplant. Acute rejection (AR) occurred significantly less often among those treated with tacrolimus/MMF (12%) than among those treated with tacrolimus/sirolimus (30%) or cyclosporine/sirolimus (28%). Mean estimated GFR was consistently higher in the tacrolimus/MMF arm, especially after controlling for donor age in a multivariable model during the first 36 mo (P ≤ 0.008). The rate of dying with a functioning graft was significantly higher among those treated with tacrolimus/sirolimus (26%) than among those treated with tacrolimus/MMF (12%) or cyclosporine/sirolimus (4%). We did not observe significant differences in actuarial graft survival at 8 yr post-transplant between the groups. Patient noncompliance seemed responsible for 45% (13/29) of observed graft failures, with 11 of these occurring after 36 mo. Significantly more viral infections, protocol violations, and need for antilipid therapy occurred among patients receiving sirolimus, but we did not observe differences between the groups with regard to infections requiring hospitalization or new-onset diabetes. Taken together, these results suggest that maintenance therapy with tacrolimus/MMF is more favorable than either tacrolimus/sirolimus or cyclosporine/sirolimus.
Calcineurin inhibitors (CNIs), tacrolimus (TAC), and cyclosporine microemulsion (CSA), are potent immunosuppressive agents used in lowering the incidence of acute rejection (AR) following solid organ transplantation.1–2 However, these medications can contribute to long-term graft loss through their intrinsic and irreversible nephrotoxic effects.3–4 Over the past 10 to 15 yr, sirolimus (SRL) and mycophenolate mofetil (MMF), two adjunctive non-CNI and non-nephrotoxic drugs,5–11 have been used in an attempt to lower concomitant maintenance CNI levels, possibly reducing graft attrition. SRL and MMF have antiproliferative effects on vascular smooth muscle cells, perhaps decreasing arterial intimal thickening, angiogenesis, and less well-defined causes of chronic allograft injury (CAI).12–14 We thus embarked on a prospective, randomized study of 150 patients to determine preferred long-term regimens combining these agents, including interim analyses performed at 1 and 3 yr post-transplant.15–17 In the present report, we compare, at a median follow-up of eight years post-transplant, the efficacy/safety of TAC in decreasing maintenance dosage combined with SRL versus MMF (TAC/SRL versus TAC/MMF), and CSA in decreasing maintenance dosage combined with SRL (CSA/SRL) in adult primary-kidney-transplant recipients.
RESULTS
Patient Demographics and Early Outcomes
Distributions of selected baseline characteristics by treatment arm (Table 1) show no significant differences for most characteristics. However, by chance, recipients ≥45 yr of age were more common in TAC/SRL (68%), followed by TAC/MMF (58%), and, lastly, CSA/SRL (44%; P = 0.05).
Distributions of selected baseline characteristics by treatment arm
Four patients (2.7%) never had functioning kidneys; three GFs occurred during the first week post-transplant (primary nonfunction: one in TAC/SRL, one in CSA/SRL; renal vein thrombosis: one in TAC/MMF). One additional GF occurred in CSA/SRL at day 12 due to thrombotic microangiopathy (without rejection). In total, 4% (6/150) of patients had DGF.
AR, CAI, and GF
Overall, 35 patients (24%) have developed AR, with 25 (17%) episodes occurring during the first 36 mo and ten (7%) episodes thereafter (Table 2). AR occurred in 15, 6, and 14 patients in Groups TAC/SRL, TAC/MMF, and CSA/SRL, respectively; actuarial estimates ±SE of the percentage developing AR through 96 mo post-transplant were 34 ± 7%, 14 ± 5%, and 31 ± 7%, respectively (Figure 1). Using the intent-to-treat approach, differences in AR rate among the three study groups were noted within the first 36 mo (P = 0.05), but no differences existed in AR rate beyond 36 mo post-transplant (P = 0.95). As reported previously,15,17 the AR rate during the first 12 mo post-transplant was significantly higher for CSA/SRL in comparison with Groups TAC/SRL and TAC/MMF combined (P = 0.03), whereas the AR rate during months 13 to 36 post-transplant was significantly higher for TAC/SRL in comparison with Groups TAC/MMF and CSA/SRL combined (P = 0.01). Thus, the AR rate during the first 36 mo post-transplant,17 and throughout the study, has favored TAC/MMF in comparison with TAC/SRL and CSA/SRL combined (P = 0.02 and 0.03, respectively; Figure 1). Seventeen AR (including 12/15 grade IB and higher) cases were treated with antilymphocyte therapy: six in TAC/SRL, five in TAC/MMF, and six in CSA/SRL (Table 2).
Incidence of I. biopsy-proven acute rejection (AR), II. GF (censoring deaths and never functioning grafts), and III. GF with compliance (censoring deaths, never functioning grafts, and GFs attributed to noncompliance) by treatment arm, using the intent-to-treat approach1
Freedom-from-AR by treatment arm (intent-to-treat).
The multivariable (Cox model) test for AR rate of the observed favorable effect of TAC/MMF versus Groups TAC/SRL and CSA/SRL combined, after controlling for the significantly unfavorable effects of non-Caucasian race/ethnicity (P = 0.0004), younger recipient age (continuous variable, P = 0.0006), and donor age ≥50 yr (P = 0.02), yielded P = 0.005 (Table 3). Low- and high-risk subgroups for AR incidence were defined according to patients having zero or one (n = 71) versus two or three (n = 79) unfavorable characteristics (non-Caucasian race/ethnicity, recipient age <50 yr, and donor age ≥50 yr). Only four of 71 low-risk patients developed AR, trending in favor of TAC/MMF (0/18) versus TAC/SRL and CSA/SRL combined (4/53; P = 0.26, logrank test). Figure 2 shows that, in the high risk subgroup, a significantly more favorable AR rate was observed in TAC/MMF, with an actuarial estimate at 96 mo ±SE of 22 ± 8% versus 59 ± 8% for TAC/SRL and CSA/SRL combined (P = 0.003, logrank test).
Results of various multivariable models testing the observed favorable effect of group TAC/MMF vs. groups TAC/SRL and CSA/SRL combined
Freedom-from-AR Comparison of TAC/MMF versus TAC/SRL and CSA/SRL arms combined, among higher-risk patients (those having ≥2 of non-Caucasian race/ethnicity, recipient age <50 yr, and donor age ≥50 yr; intent-to-treat). Note that separate Kaplan-Meier curves for the TAC/SRL (n = 23, 13 events) and CSA/SRL (n = 24, 12 events) arms would overlay on top of the curve shown for the two arms combined.
A total of ten patients (7%) developed a second AR episode. This rate was not significantly different by treatment group, with three, three, and four such events occurring in TAC/SRL, TAC/MMF, and CSA/SRL, respectively (P = 0.94, logrank test). CAI was observed in 29, 23, and 26 patients in Groups TAC/SRL, TAC/MMF, and CSA/SRL, respectively (P = 0.57), with mild-to-moderate or higher-grade injury noted in 12, 8, and 9 patients, respectively (P = 0.66).
Twenty-nine patients developed GF (excluding the four grafts that never functioned; (Table 2): ten in TAC/SRL, nine in TAC/MMF, and ten in CSA/SRL (P = 0.98). Causes of GF were CAI (n = 4), polyoma viral infection (n = 1), recurrent membranoproliferative glomerulonephritis (MPGN; n = 1), and noncompliance (n = 4) in Group TAC/SRL; CAI (n = 1), recurrent MPGN (n = 1), recurrent SLE (n = 1), and noncompliance (n = 6) in Group TAC/MMF; and acute rejection (n = 1), CAI (n = 2), thrombotic microangiopathy (n = 1), polyoma viral infection (n = 2), recurrent FSGS (n = 1), and noncompliance (n = 3) in Group CSA/SRL. Of note, three GFs due to Polyoma viral infection occurred with TAC/SRL and CSA/SRL. In addition, 45% (13/29) of GFs were attributed to patient noncompliance (consistently not taking their prescribed immunosuppression); 11 of 13 occurred beyond 36 mo post-transplant (note: only two first episodes of AR that developed in these 13 patients, one at 51 mo for a patient in TAC/MMF and one at 77 mo for a patient in CSA/SRL, were determined to have occurred after the patients had become noncompliant). If these 13 GFs were censored (freedom-from-GF-with-compliance), then a nonsignificant trend for fewer GFs occurring in TAC/MMF (3 events) versus Groups TAC/SRL and CSA/SRL combined (13 events) was observed (P = 0.25, Figure 3). Thus, while fewer patients developed GF-with-compliance in Group TAC/MMF, its favorable AR rate did not translate into a concomitant significantly lower GF rate.
Freedom-from-graft failure-with-compliance by treatment arm (intent-to-treat).
Renal Function
Consistently favorable renal function over time was observed for patients in TAC/MMF (Table 4), especially when comparing the geometric mean serum creatinine and mean eGFR for TAC/MMF versus the average of Groups TAC/SRL and CSA/SRL at months 12, 24, and 84 (P = 0.03, 0.02, and 0.04 for serum CR; P = 0.02, 0.005, and 0.09 for eGFR, respectively). No statistical differences in proteinuria between groups at 36 and 84 mo post-transplant were observed; however, the trend was lower in TAC/MMF (data not shown).
Renal function by treatment arm
Stepwise linear regression found older donor age (continuous variable) to be significantly associated with lower eGFR at each time point (P ≤ 0.0002, Table 3). T-tests of the observed favorable effect of TAC/MMF on eGFR mean in comparison with the average of TAC/SRL and CSA/SRL group means, after controlling for the donor age effect, yielded significant results at 12, 24, and 36 mo post-transplant (P = 0.0009, <.0001, and 0.008, respectively), and borderline results at months 60 and 84 (P = 0.02 each). Mean eGFR for each group over time, stratified by donor age (< and ≥37.5 yr, the median), appears in Table 4. The most favorable impact of Group TAC/MMF appears at 24 mo post-transplant, with its mean eGFR±SE being 70.6 ± 5.7 ml/min for donor age <37.5 yr versus 57.2 ± 3.8 and 60.4 ± 3.5 for Groups TAC/SRL and CSA/SRL, respectively, and 56.6 ± 3.3 ml/min for donor age ≥37.5 yr versus 49.1 ± 3.6 and 42.4 ± 2.3 for Groups TAC/SRL and CSA/SRL, respectively.
DWFG, Graft Loss, and Death
Results of the intent-to-treat analysis for DWFG, graft loss (death uncensored), and death (Table 5) show significant differences among the three groups regarding the hazard rates of DWFG and any death (P = 0.01 and 0.002), particularly beyond 36 mo post-transplant (P = 0.008 and 0.0007). In fact, hazard rates of DWFG and any death were significantly higher for TAC/SRL compared with Groups TAC/MMF and CSA/SRL combined (P = 0.006 and 0.0002 overall; P = 0.002 and 0.0002 beyond 36 mo, respectively). While fewer deaths were observed in CSA/SRL versus TAC/MMF (three versus seven), patient survival was not significantly different between these two groups (P = 0.16). Causes of DWFG were cancer (n = 6), cardiovascular event (n = 4), infection (n = 2), and other (n = 1) in Group TAC/SRL; cardiovascular event (n = 4) and infection (n = 2) in Group TAC/MMF; and cardiovascular event (n = 1) and other (n = 1) in Group CSA/SRL. Six observed cancer deaths (all in Group TAC/SRL) included metastatic Kaposi's sarcoma (n = 1), pancreatic cancer (n = 1), lung cancer (n = 2), esophageal cancer (n = 1), and metastatic squamous cell skin carcinoma (n = 1), with five of six deaths occurring after 36 mo post-transplant. Table 5 also shows that if these six observed cancer deaths are censored, then the hazard rates of DWFG and any death due to noncancer causes were not significantly different among the three groups (P = 0.22 and 0.07, respectively). These results remained unchanged even after controlling in multivariable Cox models for the significantly unfavorable effects of older recipient age and having pretransplant diabetes mellitus (not shown).
Incidence of death with a functioning graft (DWFG), DWFG due to noncancer causes, graft loss (death uncensored), any death, and death due to noncancer causes by treatment arm, using the intent-to-treat approach1
In addition, no significant differences in the hazard rate of graft loss by treatment group were observed, with 24/50, 16/50, and 14/50 patients in Groups TAC/SRL, TAC/MMF, and CSA/SRL experiencing graft loss throughout the study (P = 0.17, Figure 4). Actuarial graft survival at 5 and 8 yr post-transplant was 70 ± 6% and 55 ± 7% in Group TAC/SRL, 71 ± 7% and 66 ± 7% in Group TAC/MMF, and 80 ± 6% and 71% ± 6% in Group CSA/SRL, respectively.
Graft survival by treatment arm (intent-to-treat).
Infections
Percentages of patients with infection requiring hospitalization were similar across treatment groups: 40% (20/50) in TAC/SRL, 30% (15/50) in TAC/MMF, and 42% (21/50) in CSA/SRL (P = 0.41). Most common types were urinary tract infection (9/50, 4/50, 6/50, respectively; P = 0.32) and pneumonia (6/50, 2/50, 4/50, respectively; P = 0.34). Herpes Zoster infection occurred in eight patients (four in TAC/SRL, one in TAC/MMF, and three in CSA/SRL); Epstein-Barr viral infection occurred in five patients (three in TAC/SRL, two in CSA/SRL); Polyoma viremia occurred in seven patients (four in TAC/SRL, three in CSA/SRL); and CMV infection occurred in four patients (two in TAC/MMF, two in CSA/SRL). Thus, a trend for less viral infections developing in Group TAC/MMF was observed (3/50) in comparison with Groups TAC/SRL (10/50) and CSA/SRL (10/50) combined (P = 0.02).
NODAT and Dyslipidemia
While the hazard rate of developing NODAT (excluding 28 patients with pretransplant diabetes) was not significantly different among the three groups (P = 0.37), a slightly smaller percentage developed NODAT in Group TAC/MMF (19%,7/36) in comparison with Groups TAC/SRL (32%,13/41) and CSA/SRL (31%,14/45) combined (P = 0.16). NODAT occurred by 36 mo in 28/34 cases; insulin was used in 23/34 cases.
With regard to dyslipidemia, mean cholesterol and triglyceride levels were consistently lower in TAC/MMF in comparison with TAC/SRL and CSA/SRL (data not shown), with significant differences throughout the follow-up period. Significantly higher percentages of patients in Groups TAC/SRL and CSA/SRL required antilipid medication in comparison with TAC/MMF. For example, the percentage on antilipid medication at 12, 36, and 84 mo post-transplant was 56% (27/48), 73% (30/41), and 55% (16/29) in TAC/SRL; 18% (8/45), 30% (13/44), and 39% (11/28) in TAC/MMF; and 87% (40/46), 84% (37/44), and 79% (26/33) in CSA/SRL (P < 0.0001 at 12 and 36 mo; P = 0.007 at month 84).
Immunosuppressive Drug Doses and Trough Levels
Mean drug doses and trough levels by treatment group are shown in Tables 6A and 6B. Mean Methylprednisolone doses were higher for CSA/SRL in comparison with the average of Groups TAC/SRL and TAC/MMF throughout the first year and thereafter (P = 0.03, 0.01, and 0.03 at 12, 60, and 84 mo post-transplant, respectively). Mean MMF doses in Group TAC/MMF were lower than the target dose (2000mg/d) by 30% to 45%. Mean TAC doses over time were nearly identical for patients in TAC/SRL and TAC/MMF, yet the mean TAC trough level in Group TAC/MMF was higher at month 6 (8.85 ± 0.60 versus 7.40 ± 0.32 in TAC/SRL; P = 0.04) and remained slightly higher through 72 mo post-transplant. Mean TAC, CSA, and SRL trough levels in all groups stayed within the planned target ranges, with the exception of a lower than planned mean SRL trough level for Group TAC/SRL at month 6 (4.41± 0.59). While mean SRL dose was higher in Group TAC/SRL during the first 12 mo post-transplant and roughly equivalent between SRL groups thereafter, mean SRL trough levels trended higher in Group CSA/SRL (with statistical significance, P = 0.01, at months 6 and 36).
A. Mean immunosuppressive drug dosages by treatment arm
B. Mean immunosuppressive drug trough levels by treatment arm
Protocol Violation
The protocol violation rate was significantly higher in Groups TAC/SRL and CSA/SRL combined in comparison with TAC/MMF (P < 0.0001, Figure 5); percentages experiencing a protocol violation were 68% (34/50) in TAC/SRL, 18% (9/50) in TAC/MMF, and 78% (39/50) in CSA/SRL (Table 7). Patients discontinued SRL in TAC/SRL or CSA/SRL, 46% (46/100), at a higher rate than patients discontinued MMF in TAC/MMF, 10% (5/50; P < 0.0001). Reasons for discontinuing SRL included proteinuria/edema/body pain (15/100), infection/other clinical complication (12/100), AR/CAI (6/100), wound-healing issues (4/100), and other (9/100). Reasons for discontinuing MMF included GI symptoms (3/50) and leukopenia (2/50). Patients discontinued their assigned CNI at a significantly higher rate in CSA/SRL, 56% (28/50), in comparison with Groups TAC/SRL and TAC/MMF combined, 16% (16/100; P < 0.0001), and somewhat higher in TAC/SRL, 24% (12/50), in comparison with TAC/MMF, 8% (4/50; P = 0.04). Reasons for discontinuing CSA included AR/CAI (11/50), CNI toxicity (10/50), and other (7/50). Reasons for discontinuing TAC included CNI toxicity (9/100), neuropathy (3/100), and other (4/100). A total of 12/50 patients in Group CSA/SRL were switched from CSA to TAC, mostly due to AR/CAI. Among 90 patients who were alive with functioning grafts at 84 mo post-transplant, the percentage free-of-any protocol violation was 38% (11/29), 79% (22/28), and 21% (7/33) in Groups TAC/SRL, TAC/MMF, and CSA/SRL, respectively (P < 0.0001).
Freedom-from-protocol violation by treatment arm (intent-to-treat).
Incidence of protocol violations by treatment arm1
AR While Staying on the Protocol
Lastly, since significantly fewer patients in Group TAC/MMF switched out of their assigned treatment, it is possible that its significantly lower AR rate may be due to the observed greater protocol violation rates in the other two groups rather than TAC/MMF itself. Thus, in an attempt to address this issue, an analysis of AR while patients stayed on the protocol was performed. Specifically, the number of patients who developed AR while staying on the protocol was eight, five, and ten in Groups TAC/SRL, TAC/MMF, and CSA/SRL. If patients were censored following protocol violation, then the hazard rate of developing AR (23 events) significantly favored TAC/MMF versus Groups TAC/SRL and CSA/SRL combined (P = 0.01, as compared with the intent-to-treat, P = 0.03). Similarly, the multivariable (Cox model) test of TAC/MMF (versus TAC/SRL and CSA/SRL combined), after controlling for non-Caucasian race/ethnicity, younger recipient age, and donor age ≥50 yr, was more significant (P = 0.001) than its corresponding intent-to-treat result (P = 0.005).
DISCUSSION
Before May 2000, our standard immunosuppression regimen in primary kidney transplantation included daclizumab induction and maintenance with TAC, MMF, and corticosteroids.8 This prospective, randomized trial of 150 patients was therefore undertaken with two initial safety/efficacy goals: (1) to compare the newer agent SRL versus MMF in combination with a CNI, and (2) to simultaneously evaluate using lower trough levels for TAC (and CSA) than previously considered at this center. Our ultimate goal was to identify the long-term regimen providing the lowest rates of AR, adverse events, and graft loss. With a median follow-up of 8 yr post-transplant, graft survival was not significantly different among the three study groups. Yet despite fewer deaths observed in CSA/SRL versus TAC/MMF (although nonsignificant), the major differences observed in this study indicate that TAC/MMF performed best overall, having significantly lower rates of AR, renal dysfunction, dyslipidemia, and protocol violation, with trends for lower viral infection and GF-with-compliance rates. Most of these important differences, while maintained throughout 8 yr of follow-up, were already indicated by 1 to 3 yr post-transplant.
Throughout the study, AR incidence in Group TAC/MMF was only 12% (6/50); 6% (3/50) occurring during the first 36 mo post-transplant. This favorable outcome compares with observed higher rejection rates for Groups TAC/SRL and CSA/SRL, 30% (15/50) and 28% (14/50), respectively, and this difference clearly became significant once the multivariable effects of other prognosticators (race/ethnicity, recipient age, and donor age) were controlled. Although randomization of 50 patients per treatment arm only allowed reasonable statistical power for detecting moderate-to-large differences between groups, statistical testing after controlling for other prognosticators in a multivariable model would likely increase precision (and statistical power), as seen with the analysis for AR and eGFR. In addition, the comparison of AR rate favoring TAC/MMF remained statistically significant while patients stayed on the protocol, indicating that the favorable effect of TAC/MMF on AR correlates directly with the assigned immunosuppressive regimen (TAC/MMF).
The observed significantly higher AR incidence in the TAC/SRL arm during months 13 to 36 post-transplant might be a result of subtherapeutic SRL trough levels, particularly with a mean value <5 ng/ml at 6 mo post-transplant. In fact, in other randomized trials comparing AR incidence between TAC/MMF and TAC/SRL, three studies reported no significant difference,18–21 and one study reported a trend for lower AR incidence in the TAC/SRL arm.22
Significantly favorable renal function for TAC/MMF over TAC/SRL was reported in three of four randomized trials reported by other centers,18–21 along with significantly fewer adverse events such as dyslipidemia,18–21 elevated diastolic BP,19,21 proteinuria,21 and discontinuance of MMF versus SRL.18–21 In addition, in a Scientific Renal Transplant Registry (SRTR) analysis of approximately 45,000 patients, significantly favorable death-censored graft survival was reported for TAC/MMF versus TAC/SRL (and CSA/SRL).23 Thus, evidence provided by our longer-term results are consistent with those reported in most studies, suggesting that, despite its several shortcomings, TAC/MMF is currently the most optimal maintenance regimen in kidney transplantation.
A large percentage of patients in TAC/SRL and CSA/SRL did not remain on their assigned regimens for this study. A recent SRTR analysis of nearly 61,000 patients transplanted during 1998–2002 similarly reports that less than half of all recipients remain on their discharge regimens beyond 4 yr post-transplant, with a significantly greater percentage of TAC/MMF recipients remaining on their regimen in comparison with TAC/SRL and CSA/SRL.24 As there is always a learning curve with newer agents, it is possible that the longer experience of many centers in using MMF versus SRL at that time (including our center) may have unintentionally contributed to these differences in protocol adherence.
It is also noteworthy that 45% (13/29) of GFs observed in this longer-term study were attributed to patient noncompliance, with 11/13 occurring beyond 36 mo post-transplant. Noncompliance as a cause of GF may become more prominent as immunosuppression trials achieve longer-term follow-up. Future studies with larger cohorts to identify multivariable predictors of patient noncompliance may better help physicians to address (minimize) this problem. While a perfect correlation between AR incidence and subsequent GF risk clearly does not exist, there was, on the other hand, an expected (but nonsignificant) trend of fewer GFs-with-compliance occurring in the TAC/MMF arm due to fewer patients developing AR.
Lastly, the ELITE-Symphony Study randomized trial results at 12 mo post-transplant found a lower AR rate, better renal function and graft survival, fewer adverse events, and less discontinuation of a study drug due to an adverse event in TAC/MMF versus SRL/MMF (each comparison being statistically significant). However, the impact of these differences were smaller by 3 yr post-transplant.25–26
In conclusion, the excellent outcome in the TAC/MMF arm, as observed in the ELITE-Symphony Study,25–26 was achieved using reduced trough levels for TAC, with mean levels ranging from 8–10 ng/ml initially and 5–8 ng/ml beyond the first year post-transplant. While the combination of TAC/MMF (with or without steroid avoidance) is currently the most widely used maintenance regimen in kidney transplantation,27 nephrotoxicity due to long-term use of TAC is still a potential problem, and new immunosuppressive strategies designed to dramatically improve long-term graft survival are still required.
CONCISE METHODS
As previously reported,15–16 between May 2000 and December 2001, 150 recipients between 14 to 78 yr of age, of either deceased donor(DD) or non-HLA identical living donor (LD) first kidney transplants, were randomized immediately before transplantation into one of three study groups (the center institutional review board approved the protocol; patients gave written informed consent before enrollment): TAC/SRL, TAC/MMF, and CSA/SRL. In each arm, the CNI was not started until renal function had improved (serum creatinine concentration (Cr) <4 mg/dl absent dialysis). TAC was instituted at a dose of 0.1 mg/kg twice daily, with an initial target trough level (during the first 2 mo post-transplant) of 10 ng/ml, then lowered to 6–10 ng/ml between 3 to 6 mo post-transplant and 4–8 ng/ml thereafter. CSA was initiated at 5 mg/kg twice daily with an initial target trough level of 200–250 ng/ml, then lowered to 100–200 ng/ml thereafter. In both SRL arms, a loading dose of 4mg of SRL was given on the evening after surgery, and then daily, with a target trough level 6–10 ng/ml. Planned MMF dosing was 1 g twice daily, maintained as tolerated. All patients received daclizumab induction, maintenance corticosteroids, and nonimmunosuppressive adjunctive therapy, as previously reported.8,15–17 A protocol violation was defined as the patient discontinuing one or both of the planned maintenance drugs for at least 1 yr or the addition of a nonplanned maintenance drug. Noncompliance was defined as the recipient consistently failing to take his/her immunosuppressive medications as prescribed. All patients were documented as intent-to-treat.
Patients were followed for the incidence of biopsy-proven AR, graft loss, other adverse events, and death. Delayed graft function (DGF) was defined as the requirement for dialysis during the first week post-transplant. Renal function was determined by serum Cr and estimated GFR (eGFR) using the abbreviated MDRD formula.28 AR was defined as a rise of 0.3mg/dl or greater from the nadir Cr, accompanied by a confirmatory kidney transplant biopsy within 24 h of initiation of antirejection therapy. Grading of AR and CAI (interstitial fibrosis/tubular atrophy) was performed according to the Banff classification.29 Graft loss was determined as the time of reestablishment of long-term dialysis or death. New onset diabetes after transplantation (NODAT) was defined as the use of insulin or oral antihyperglycemic agents for at least 30 d in patients without a preoperative history of diabetes mellitus.15–16 All infections and postoperative complications requiring hospitalization were documented.
Statistics
Unless specified otherwise, all statistical analyses were performed using an intent-to-treat approach. Standard one-way ANOVA (F-tests) and t-tests were used in comparing mean values. Percentages were compared across study groups using the Pearson (uncorrected) chi-square test. Arithmetic mean ± standard error (SE) was reported, except for variables that were highly skewed toward larger values, in which case geometric means and corresponding SEs were reported, with comparisons based on log-transformed values. For drug dose, whole blood drug trough level, serum Cr, eGFR, and proteinuria variables, sufficient individual patient data through 84 mo post-transplant were available. Incidence of first AR, CAI, graft failure (GF; censoring deaths and never functioning grafts), graft loss (death uncensored), death with a functioning graft (DWFG), death, protocol violation, and NODAT were compared by the logrank test, with time-to-failure curves generated using the Kaplan-Meier method. With the exception of patient survival, all clinical outcomes were censored (patients not followed) beyond the date of GF. Thus, the analyses of renal function (serum Cr and eGFR) at various times post-transplant were based on comparing patients who were still alive with functioning grafts at those times. While imputation of an arbitrarily chosen low value for eGFR (say, 10 ml/min) was considered for patients who previously experienced GF,25–26 the results were quite similar to those found without using imputation and will therefore not be shown.
With three study groups, six planned multiple comparisons were performed for each clinical outcome variable—three pairwise comparisons and three comparisons of one study group versus the other two groups combined. For instance, it was of interest to perform overall comparisons of MMF versus SRL (Group TAC/MMF versus Groups TAC/SRL and CSA/SRL combined) and TAC versus CSA (Groups TAC/SRL and TAC/MMF combined versus Group CSA/SRL). To adjust for these six planned multiple comparisons (not statistically independent), a modified Bonferroni approach (simple rule of thumb) was utilized: Any of the six specified multiple comparisons would require a P-value of 0.01 or less to be statistically significant (versus the traditional 0.05 level).
Furthermore, while randomization was performed to ensure unbiased comparisons among the study groups, 50 patients per study arm only allowed good statistical power for detecting moderate-to-large differences. Multivariable analysis of various outcome variables, adjusting for the influence of other significant prognosticators, was therefore undertaken with the goal of increasing precision (and, thus, statistical power) of the treatment group-specific comparisons. Specifically, stepwise Cox regression was utilized for time-to-failure outcomes; stepwise linear regression (analysis-of-covariance) was utilized for other outcomes such as eGFR.
Median follow-up among 84 patients who were known to be alive with a functioning graft as of the last follow-up date, May 1, 2009, was 96 (range: 88 to 107) months (i.e., 8 yr). An additional 12 patients were lost to follow-up and alive with functioning grafts at their lost-to-follow-up times (range: 39 to 96 mo). Finally, in each of the five figures presented (Figures 1 to 5), patients were censored at the earliest of the following times (if they occurred): graft failure, death, lost-to-follow-up, and May 1, 2009 (last follow-up date).
Disclosures
The authors of this manuscript have conflicts of interest to disclose as described by the Journal of the American Society of Nephrology. It should be noted that Astellas Pharma US, Inc., has provided partial salary support (with a grant of $75,000) for the completion of this study to the Kidney and Kidney/Pancreas Transplant Program at the University of Miami. Representatives of Astellas had no direct involvement in this study, and none of the coauthors have any financial interest in Astellas.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- Copyright © 2011 by the American Society of Nephrology