Adverse Outcomes of Tacrolimus Withdrawal in Immune–Quiescent Kidney Transplant Recipients

Abstract

Concerns about adverse effects of calcineurin inhibitors (CNIs) have prompted development of protocols that minimize their use. Whereas previous CNI withdrawal trials in heterogeneous cohorts showed unacceptable rates of acute rejection (AR), we hypothesized that we could identify individuals capable of tolerating CNI withdrawal by targeting immunologically quiescent kidney transplant recipients. The Clinical Trials in Organ Transplantation-09 Trial was a randomized, prospective study of nonsensitized primary recipients of living donor kidney transplants. Subjects received rabbit antithymocyte globulin, tacrolimus, mycophenolate mofetil, and prednisone. Six months post-transplantation, subjects without de novo donor-specific antibodies (DSAs), AR, or inflammation at protocol biopsy were randomized to wean off or remain on tacrolimus. The intended primary end point was the change in interstitial fibrosis/tubular atrophy score between implantation and 24-month protocol biopsies. Serially collected urine CXCL9 ELISA results were correlated with outcomes. The study was terminated prematurely because of unacceptable rates of AR (4 of 14) and/or de novo DSAs (5 of 14) in the tacrolimus withdrawal arm. Positive urinary CXCL9 predated clinical detection of AR by a median of 15 days. Analyses showed that >16 HLA-DQ epitope mismatches and pretransplant, peripheral blood, donor–reactive IFN-γ ELISPOT assay results correlated with development of DSAs and/or AR on tacrolimus withdrawal. Although data indicate that urinary CXCL9 monitoring, epitope mismatches, and ELISPOT assays are potentially informative, complete CNI withdrawal must be strongly discouraged in kidney transplant recipients who are receiving standard-of-care immunosuppression, including those who are deemed to be immunologically quiescent on the basis of current clinical and laboratory criteria.

Current immunosuppression for kidney transplant recipients most commonly includes a calcineurin inhibitor (CNI), an antiproliferative agent, and corticosteroids.^1,2 With the introduction of cyclosporin in the 1980s, it became clear that routine use of CNIs decreased acute rejection (AR) rates and resulted in improvement in short-term outcomes compared with the prior era.^1,3 These improvements have not, however, been associated with similar improvements in long-term graft survival.^4,5 The adverse effects of CNIs, including drug–induced renal parenchymal fibrosis, allograft dysfunction, and cardiovascular morbidity among others, have raised concerns that CNIs may contribute to poor long-term outcomes and have resulted in a desire to develop immunosuppression protocols that avoid, withdraw, or minimize their use.^6–8

Published studies of CNI withdrawal in unselected cohorts of kidney transplant recipients taking standard three–drug immunosuppression indicate that elimination of CNI increases the risk of AR,^6,7,9–16 which can potentially precipitate a fibrogenic process that contributes to graft failure. Some of these previous trials targeted patients who are apparently low risk as defined by relatively limited conventional clinical criteria, including living donor source, human leukocyte antigen (HLA) matching (particularly at class II loci), and lack of HLA sensitization.^11,14–16 Although these previously performed studies confirmed high rates of AR after CNI withdrawal, despite the selection criteria, the studies also suggested that a subset of these clinically low–risk transplant recipients can be safely withdrawn from CNI without negative consequences to the patient or graft. If it were possible to prospectively identify the subset of individuals capable of tolerating CNI withdrawal using objective and reproducible histologic and immunologic criteria, it would permit targeting CNI withdrawal to only those most likely to benefit from the intervention.

We hypothesized that, by selecting immunologically quiescent, low–risk kidney transplant recipients and treating them with T cell–depleting induction therapy, it would be possible to identify accurately the specific subset of subjects capable of tolerating CNI withdrawal (leaving patients on MMF and steroids) and as a consequence, improve their long-term graft histology and function. To this end, we report the results of the Clinical Trials in Organ Transplantation-09 (CTOT-09) Trial, in which nonsensitized, primary living donor kidney allograft recipients were given induction therapy with rabbit antithymocyte globulin and if stable at 6 months after transplant, randomized to undergo tacrolimus (TAC) withdrawal or remain on TAC. Both groups received mycophenolate mofetil and prednisone. On the basis of our previous work showing that measurements of urinary chemokines, including CXCL9 protein, are useful biomarkers to diagnose AR,^17,18 we reasoned that urinary CXCL9-based early diagnosis of AR would guide reinstitution of TAC without negative long-term consequences on graft outcome (or function).

Results

Clinical Outcomes

In total, 52 subjects were enrolled, 47 subjects were transplanted, and 21 subjects were randomized (Table 1). The reasons for failure to reach randomization were withdrawal of consent (n=8) (Figure 1), ineligibility for randomization (n=14), lost to follow-up (n=1), and study randomization procedures stopped (n=3).

Table 1.

Characteristics of study subjects

Characteristics	Total Transplanted (n=47)a	Total Randomized (n=21)	Randomized		P Value
			Control (n=7)	TAC Withdrawal (n=14)
Donor age, yr
n	46	21	7	14	0.54
Mean (SD)	47.7 (10.74)	45.2 (11.31)	47.4 (11.12)	44.1 (11.65)
Median	48	45	45	46
Minimum, maximum	24, 66	24, 63	34, 63	24, 58
Donor sex
Men	12 (25.5)	10 (47.6)	4 (57.1)	6 (42.9)	0.66
Women	34 (72.3)	11 (52.4)	3 (42.9)	8 (57.1)
Donor race
White	38 (80.9)	17 (81.0)	7 (100)	10 (71.4)	0.31
Black or African American	3 (6.4)	1 (4.8)	0	1 (7.1)
Asian	3 (6.4)	3 (14.3)	0	3 (21.4)
American Indian or Alaska Native	0	0	0	0
Unknown or not reported	2 (4.3)	0	0	0	NA
Recipient age, yr
n	47	21	7	14	0.79
Mean (SD)	50.9 (12.68)	46.0 (12.52)	44.9 (8.51)	46.5 (14.38)
Median	51	47	41	48
Minimum, maximum	24, 76	24, 73	37, 61	24, 73
Recipient sex
Men	35 (74.5)	15 (71.4)	7 (100)	8 (57.1)	0.06
Women	12 (25.5)	6 (28.6)	0	6 (42.9)
Recipient race
White	39 (83.0)	17 (81.0)	7 (100)	10 (71.4)	0.31
Black or African American	4 (8.5)	1 (4.8)	0	1 (7.1)
Asian	3 (6.4)	3 (14.3)	0	3 (21.4)
American Indian or Alaska Native	1 (2.1)	0	0	0
Unknown or not reported	0	0	0	0	NA
HLA mismatch
n	46	21	7	14	>0.99
Mean (SD)	3.5 (1.50)	3.4 (1.29)	3.4 (1.27)	3.4 (1.34)
Median	4	3	3	4
Minimum, maximum	1, 6	2, 6	2, 6	2, 6
<3	13 (27.7)	6 (28.6)	1 (14.3)	5 (35.7)	0.61
≥3	33 (70.2)	15 (71.4)	6 (85.7)	9 (64.3)
CMV status (donor, recipient)
+, +	13 (27.7)	5 (23.8)	0	5 (35.7)	0.40
+, −	8 (17.0)	3 (14.3)	1 (14.3)	2 (14.3)
−, +	7 (14.9)	4 (19.0)	2 (28.6)	2 (14.3)
−, −	17 (36.2)	8 (38.1)	3 (42.9)	5 (35.7)
Not done, +	1 (2.1)	1 (4.8)	1 (14.3)	0	NA
Missing, −	1 (2.1)	0	0	0	NA

P values comparing the two treatment arms result from t tests, Fisher’s exact tests, or Cochran-Mantel-Haenszel tests of general association. CMV, cytomegalovirus; NA, not applicable.

^aOne donor for a transplanted but not randomized recipient did not provide any donor characteristic data.

Figure 1.

Tacrolimus withdrawal resulted in adverse outcomes in a highly-selected, low-risk study population. Consort diagram depicting the outcomes of the 52 enrolled subjects, 47 of whom underwent transplantation. Twenty-one of 47 transplanted patients met the criteria for randomization. Outcomes of the 14 randomized to Tacrolimus withdrawal arm and 7 randomized to the control arm are delineated at the bottom of the diagram. ACR: acute cellular rejection, DSA: donor specific antibody. DR and DQ refer to the HLA locus to which the DSA was reactive.

Of 21 randomized subjects, 7 subjects were assigned to the control arm, and 14 subjects were assigned to TAC withdrawal. Clinical characteristics did not differ between groups (Table 1). No deaths or graft losses occurred in any of the randomized subjects during the period of follow-up. We were not able to assess the effect of the intervention on the primary end point of interstitial fibrosis/tubular atrophy (IF/TA; on 2-year graft biopsy), because the study was terminated prematurely by the Data Safety Monitoring Board (DSMB) because of absence of equipoise on the basis of predetermined stopping rules (Figure 2, Supplemental Table 1).

Figure 2.

ACR and/or de novo DSA developed at the time of, or rapidly after, stopping Tacrolimus. Clinical outcomes of the 21 randomized subjects. Each line represents a timeline for each of 7 control subjects and 14 subjects with TAC withdrawal (separated by a bold line). Time of initiating TAC withdrawal (black x), completion of TAC withdrawal (green x), time of clinically evident, biopsy–proven cellular rejection (red arrowhead), and time of detection of de novo DSA (blue arrowhead) are shown. One of seven control subjects (below the dotted line) was retrospectively noted to have a DSA before randomization that was not detected on the 6-month prerandomization sample but was detected on day 315 postrandomization. This subject was not considered to have developed a de novo DSA during the randomization period.

Thirteen of fourteen subjects in the TAC withdrawal arm completed withdrawal (Figures 1 and 2), but only six subjects remained stable off TAC for the duration of follow-up. All 14 subjects were taking at least 1500 mg/d MMF (or the equivalent dose of mycophenolic acid) at the time of TAC withdrawal. Eight subjects in the TAC withdrawal arm were restarted on full-dose TAC, because they reached protocol–driven end points. Three subjects developed clinically evident, biopsy–proven acute cellular rejection (ACR; one subject, Banff IIA; two subjects, borderline rejection as determined by local and/or central pathology reads) after randomization. A fourth subject developed clinically evident, mixed ACR (Banff IIA) plus antibody-mediated rejection (AMR; C4d positive with endarteritis) along with a de novo anti-DQ donor–specific antibody (DSA) (Supplemental Table 2). Three of these clinically recognized rejection episodes occurred within 15 days of completing TAC withdrawal; the fourth episode of rejection occurred on day 78 after completion of withdrawal. Rejection episodes were treated per center practice, but renewal of TAC was mandated.

Four additional subjects in the TAC withdrawal arm developed new DSAs after randomization without clinical evidence of kidney dysfunction (Figure 2, Supplemental Table 2). In one subject, the DSA was first detected with the subject taking 1 mg/d TAC (serum level of 2 μg/L). In the other three subjects, the DSA developed within 3 months after eliminating TAC. All of the DSAs in the TAC withdrawal arm were anticlass II and reactive to either donor HLA-DQ or DR alleles (Supplemental Table 2). No anticlass I DSAs were detected.

Clinically driven biopsies were performed in two of three subjects with TAC withdrawal who developed DSAs. In one subject, an initial indication biopsy showed borderline ACR with focal moderate capillaritis (ptc2); a follow-up biopsy 1 month later showed significant microvascular inflammation with moderate transplant glomerulitis (g2) and severe peritubular capillaritis (ptc3), which was highly suspicious for AMR. In the second subject, the indication biopsy showed borderline ACR, mild peritubular capillaritis (ptc1), and mild IF/TA.

Zero of seven subjects in the control arm developed ACR or AMR. In one control subject without DSA in the 6-month prerandomization sample, analysis of a stored sample obtained by the site investigator 33 days post-transplant (not protocol driven) and retrospectively tested after randomization revealed a de novo HLA-DQ DSA that was present before randomization (Supplemental Table 2). An anti-HLA antibody of the same specificity was detected on day 315 after randomization. Only one subject in the control arm developed de novo DSA postrandomization (at 365 days after randomization) (Figure 2). HLA-DQ DSA that developed de novo and after randomization occurred earlier and trended toward higher frequency in the TAC withdrawal arm (5 of 14) versus control arm (1 of 6; P value was NS).

To assess the effects of TAC withdrawal on long-term kidney function, we compared absolute 24-month eGFR and change in eGFR from 6 to 24 months between the TAC withdrawal and control groups in a secondary analysis (Figure 3). The analyses revealed no significant differences between groups. Additionally, the 24-month eGFR and 6- to 24-month change in eGFR for subjects with TAC withdrawal who developed de novo DSA and/or ACR (and in whom TAC was reinstituted) were not different from those of the subjects with TAC withdrawal who remained TAC-free (Figure 3, green symbols), and they did not differ from the controls.

Figure 3.

Tacrolimus withdrawal did not have an impact on kidney allograft function. Kidney function in the randomized cohort: (upper panel) 24-month eGFR and (lower panel) change in eGFR between 6 and 24 months are depicted for each randomized subject stratified by control versus TAC withdrawal groups. There were no significant differences in either eGFR or change in eGFR between those who failed TAC withdrawal (developed rejection and/or de novo DSA and required reinstitution of full-dose TAC; black circles) and those who remained off TAC for >18 months (green circles) or between those who failed TAC withdrawal and controls. P value is NS between groups for both analyses. M, month.

Urinary Chemokines and Rejection

A secondary goal of the study was to determine the use of serial urinary CXCL9 testing to diagnose and/or predict ACR during TAC withdrawal. Within 21 randomized subjects, urinary chemokine values obtained at 3 months post-transplant and the time of the 6-month protocol biopsy were all negative (below the detection limit of the assay). We observed seven episodes of confirmed positive CXCL9 assays in six subjects in the TAC withdrawal arm. For two of these subjects, positive test results were attributed to BK virus or systemic viral infection, and the urinary chemokine abnormalities were resolved with resolution of the underlying viral infection and without treatment for AR. The five other confirmed positive CXCL9 test results occurred in four subjects, each of whom developed ACR 11–108 days (median of 15 days) after the initial positive result (Figure 4). In one of these four subjects, an initial protocol biopsy prompted by a positive urinary CXCL9 showed no AR, but persistence of a positive urinary CXCL9 during the ensuing 3 months prompted a second biopsy that showed borderline cellular AR. Urinary CXCL9 data were not collected proximal to the two cases of graft inflammation that were performed to evaluate asymptomatic DSA (the subjects had been restarted on TAC, urinary chemokines assays were not mandated, and the biopsy was performed at the site PI’s discretion). All urinary CXCL9 assays were negative in six subjects fully withdrawn from TAC without ACR or DSA along with all of the chemokine assays in seven subjects in the control arm. Overall, a confirmed positive urinary CXCL9 result in the absence of coincident infection (which is readily and rapidly diagnosed using currently available, noninvasive testing strategies) was associated with higher incidence of cellular AR (P<0.01) with a sensitivity of 66.7% (95% confidence interval [95% CI], 22.3% to 95.7%), specificity of 100% (95% CI, 78.2% to 100%), PPV of 100 (95% CI, 39.8% to 100%), and NPV of 88.2% (95% CI, 63.6% to 98.5%). A similar association was observed within the subjects with TAC withdrawal (P=0.02) with a sensitivity of 66.7% (95% CI, 22.3% to 95.7%), specificity of 100% (95% CI, 63.1% to 100%), PPV of 100% (95% CI, 39.8% to 100%), and NPV of 80% (95% CI, 44.4% to 97.5%). Consistent with our previous results,¹⁷ measurements of urinary CXCL10 were not informative (data not shown).

Figure 4.

Urinary CXCL9 values were elevated prior to the clinical detection of ACR in study subjects undergoing Tacrolimus withdrawal. Individual timelines for six subjects in the TAC withdrawal arm with positive urinary CXCL9 test results that illustrate associations with biopsy-proven ACRs. The two subjects who had positive urine CXCL9 results but did not experience ACR (the first two subjects) were diagnosed with BK polyoma or systemic viral infections, no biopsies were performed, and serum creatinine remained unchanged. All urine tests in seven subjects in the control arm were negative for CXCL9.

Epitope Mismatching and Risk of Developing DSAs

Emerging evidence suggests that the number of immunogenic epitope mismatches between HLA alleles of donor and recipient correlates with the risk of developing de novo DSA post-transplant.¹⁹ We retrospectively calculated epitope mismatches between donor and recipient in 21 randomized study subjects (including the control subject who developed an early DSA) and analyzed the relationships between mismatches and the development of DSA (Figure 5A). Using a predefined and published threshold of >16 DQ epitope mismatches to define high risk,¹⁹ positive DQ epitope load was associated with a higher incidence of de novo DSA among all randomized subjects (7 of 13 versus 0 of 8; P=0.02; sensitivity=100%; 95% CI, 59% to 100%; specificity=57.1%; 95% CI, 28.9% to 82.3%; PPV=53.9%; 95% CI, 25.1% to 80.8%; NPV=100%; 95% CI, 63.1% to 100%) and among subjects in the TAC withdrawal arm (5 of 8 [62.5%] versus 0 of 6; P=0.03; sensitivity=100%; 95% CI, 47.8% to 100%; specificity=66.7%; 95% CI, 29.9% to 92.5%; PPV=62.5%; 95% CI, 24.5% to 91.5%; NPV=100%; 95% CI, 54.1% to 100%). Epitope mismatches at the DR locus were not informative, because only two subjects developed de novo anti-DR DSA.

Figure 5.

Epitope mismatches and pretransplant ELISPOT results correlate with de novo DSA and AR. (A) Epitope mismatches at HLA-DQ loci. The bars depict the total numbers of randomized subjects with (left bar) >16 epitope mismatches and (right bar) ≤16 mismatches at HLA-DQ loci. The shaded portions of each bar (and accompanying percentages) depict those subjects in each group who developed a de novo post-transplant DSA. Mismatches at HLA-DR loci were not informative (not shown). (B) Results of pretransplant antidonor IFN-γ ELISPOT assays available in 15 of 21 randomized subjects. The bars depict the total numbers of randomized subjects with (left bar) positive pretransplant ELISPOT results (>25 per 300,000 peripheral blood mononuclear cells)²⁰ and (right bar) negative pretransplant ELISPOT results. The shaded portions of each bar (and accompanying percentages) depict those subjects in each group who developed an episode of AR and/or a de novo DSA in the post-transplant period.

Pretransplant Donor–Reactive Enzyme–Linked Immunosorbent Spot Testing and Post-Transplant Events

We analyzed the relationship between a positive, pretransplant donor–reactive IFN-γ enzyme–linked immunosorbent spot (ELISPOT) assay, reflective of donor–reactive T cell memory,^20–22 and post-transplant events in the randomized cohort; 15 of 21 randomized subjects had pretransplant donor–specific ELISPOT data available. Using a cutoff of 25 spots for positivity,^20,22,23 six of nine (67.7%) subjects with positive pretransplant ELISPOT results experienced ACR or developed de novo DSA post-transplant compared with zero of six (0.0%) subjects who were pretransplant ELISPOT negative (P<0.05; sensitivity=100%; 95% CI, 54.1% to 100%; specificity=66.7%; 95% CI, 29.9% to 92.5%; PPV=66.7%; 95% CI, 29.9% to 92.5%; NPV=100%; 95% CI, 54.1% to 100%) (Figure 5B). A similar, albeit nonsignificant trend was noted in the subjects with TAC withdrawal (five of seven ELISPOT⁺ [71%] developed ACR or de novo DSA versus zero of three [0%] ELISPOT⁻ who developed neither, P=0.17; data not shown). ELISPOT assays performed on samples collected monthly up to and including 6 months post-transplant (time of randomization) showed no significant differences between those who developed AR or DSA postrandomization versus those who did not (data not shown).

Discussion

In this trial, which was closed to enrollment after meeting predetermined stopping rules, 8 of 14 primary living donor kidney transplant recipients deemed to be low risk by conventional clinical criteria and immunologically quiescent by immune and histologic criteria developed AR and/or de novo DSA on withdrawing TAC. Previous studies by others have shown that AR³ and de novo anticlass II DSA,²⁴ which occur in the absence of CNI, can be associated with an elevated risk of late graft failure. Thus, although we were unable to assess the primary study end point because of early study termination, our findings support the conclusion that risks associated with AR and DSA outweigh any potential benefits of TAC withdrawal, even in this highly selected, low-risk cohort of living donor recipients treated with antithymocyte globulin induction.

Six of fourteen subjects who underwent TAC withdrawal remained off TAC for 24 months with stable allograft function and without rejection or DSA. Although our findings confirm previous reports that selected individuals can safely tolerate CNI withdrawal,²⁵ our attempt to identify these individuals using standard clinical/epidemiologic factors indicative of low risk (first recipients of living donor kidneys with low PRA and absent DSA without inflammation on a 6-month surveillance biopsy) failed.

Our analyses suggest that subjects with low HLA epitope mismatches and negative pretransplant antidonor ELISPOTs are more likely to tolerate TAC withdrawal, but these observations require additional validation. Building on our previous work, in which we documented a 92% negative predictive value of urinary CXCL9 testing for AR,¹⁷ the findings from this trial suggest that serial post-transplant urinary CXCL9 testing is useful for diagnosing incipient ACR during immunosuppression withdrawal. Remarkably, no ACRs occurred in subjects with negative urinary CXCL9 testing. Our study results lend support to the use of urinary CXCL9 testing as a diagnostic tool in the context of a therapeutic intervention trial in kidney transplant recipients. Nonetheless, we contend that, in the absence of additional validation, the small number of observations obtained from this prematurely terminated trial does not yet support routinely using results of urinary CXCL9 testing for making clinical decisions to withdraw immunosuppression in kidney transplant recipients.

Importantly, we do not contend that any of the subjects exhibited evidence of tolerance, and we were not attempting to induce tolerance. Although putative molecular signatures of tolerance that theoretically could be used to identify patients capable of tolerating drug withdrawal have been described by others,^26,27 their use for this purpose remains unproved. Additional multicenter trials will also be required to assess and compare such molecular signatures with the biomarkers tested herein (among others) as guides for individualizing immunosuppression in patients with transplants.

Overall, the results of the CTOT-09 Trial provide insights that can help guide decision making regarding immunosuppression after kidney transplantation and that will aid in the design of future clinical trials. The risks of TAC withdrawal from patients taking conventional three–drug immunosuppression (CNI, MMF, and steroids) outweigh the putative benefits, even in immunologically quiescent, low-risk recipients. It seems logical to conclude that TAC withdrawal from patients deemed to be at higher risk also should be discouraged, at least in those maintained on MMF and steroids. In conjunction with HLA epitope analysis, post-transplant CXCL9 protein monitoring has the potential to guide therapeutic decision making for kidney transplantation. Additional prospective studies are needed to confirm the use of such biomarker–supplemented risk assessment strategies to facilitate safe minimization of immunosuppression.

Concise Methods

Study Design

The CTOT-09 Trial was a prospective, randomized trial sponsored by the National Institutes of Health (www.clinicaltrials.gov; NCT01517984). The protocol development team was led by D.E.H. and P.S.H. All authors vouch for data accuracy and completeness. Each site participated under the auspices of its Institutional Review Board. An independent, NIAID-appointed DSMB was responsible for periodic safety review and guided by predetermined protocol–defined stopping criteria.

Subjects

Subjects were eligible to participate if they were ≥18 years of age and the recipient of a primary living donor (related or unrelated), HLA–mismatched kidney transplant (Supplemental Table 1). All had to have peak flow–based panel reactive antibodies for classes I and II HLA antigens <30%, a negative T and B cell cross-match (flow cytometry), and a negative pretransplant DSA (see below). Subjects were enrolled before transplantation and eligible for randomization 6 months after transplant if the following criteria were met (Supplemental Table 3): absence of AR in the first 6 months, absence of de novo DSA at 6 months, absence of BK polyoma viremia, an MMF dose of >1500 mg daily, and absence of AR (including Banff borderline) on a 6-month protocol biopsy read by the Central Pathology Core (University of Manitoba; I.W.G.). Enrollment was targeted to 300 subjects, with 210 subjects randomized 2:1 to TAC withdrawal:TAC maintenance; both groups received MMF and prednisone. Only 47 subjects were enrolled, and 21 subjects were randomized before the study was terminated by the NIAID DSMB.

Subject Management and Immunosuppression

Subjects underwent a preimplantation biopsy and received induction therapy with rabbit antithymocyte globulin (ATG/thymoglobulin; cumulative dose of ≥4.5 mg/kg). TAC doses were adjusted to maintain trough levels of 8–12 ng/ml for the first 3 months and 5–8 ng/ml thereafter. The MMF dose (initially 1000 mg two times per day) was adjusted on the basis of subjects’ tolerance to the drug. Therapeutic drug monitoring using trough mycophenolic acid levels was mandated only if the initial dose was lowered. If the trough mycophenolic acid level was <1.9 ng/ml, the dose was titrated upward as clinically tolerated in an effort to achieve a dose of at least 750 mg two times per day. Methylprednisolone or its equivalent was administered perioperatively according to center practice. Oral corticosteroids were required and tapered not <5 mg/d prednisone by 3 months post-transplant. Surveillance for BK polyoma viremia and prophylaxis for cytomegalovirus and Pneumocystis were required and prescribed according to center practice. After randomization, immunosuppression remained unchanged in control subjects. In those randomized to TAC withdrawal, MMF (≥1500 mg/d) and prednisone were continued, but TAC was reduced by one third at initiation of taper, reduced by another one third after 1 month, and discontinued no longer than 4 months after randomization.

Renal biopsy was mandated for subjects with an unexplained deterioration of renal function (serum creatinine >30% over baseline). Beginning at randomization, TAC withdrawal arm subjects were monitored every other week for 4 months and then, less frequently for clinical signs of allograft rejection (Supplemental Table 4). At each designated study visit, serum samples were tested for de novo DSA and reported back to the site within 1 week. Subjects who developed a de novo DSA were restarted on TAC and treated as per the site’s standard of care. Urine samples were collected at each study visit, tested for chemokine CXCL9 and CXCL10 protein levels (see below), and reported to the site PI within 4 days. A positive test for either chemokine prompted repeat testing within 2 weeks. In the interim, investigators tested for urinary tract infection, cytomegalovirus, BK, and other viral infections. If the repeat urine remained chemokine positive (a confirmed positive) in the absence of infection, a protocol biopsy was mandated.

Clinical End Points

The study’s primary end point was the percentage of subjects in each arm with incremental changes in IF/TA scores >2 (on the basis of Banff chronicity scores) comparing a 24-month protocol biopsy with the preimplantation biopsy. Secondary end points (6, 12, 18, and 24 months) included incidence of AR, eGFR,²⁸ allograft and subject survival rates, and percentage of subjects with de novo post-transplant DSA.

Laboratory Studies

Urine ELISAs

ELISAs for CXCL9 and CXCL10 were performed at three core laboratories (University of Manitoba [P.N.], Cleveland Clinic [R.L.F.], and Icahn School of Medicine at Mount Sinai [P.S.H.]) under good laboratory practice compliance as previously described.¹⁷ A threshold of 200 pg/ml urine for either chemokine was considered positive.

Anti-HLA Antibody Determination

Anti-HLA antibodies were measured at the core laboratory at the University of Manitoba using Luminex LABScreen Single Antigen HLA Class I and Class II Antibody Detection.^17,24,29 Assignment of a DSA required an MFI>1000 and an appropriate donor–specific epitope pattern. Determination of donor specificity was made by joint review (P.N. and K.T.). After completion of intermediate/high-resolution molecular typing to define alleles, donor–recipient HLA epitope mismatches were determined using HLA Matchmaker.¹⁹

ELISPOT Assays

IFN-γ ELISPOT assays were performed as previously published³⁰ using standardized operating procedures at Icahn School of Medicine at Mount Sinai (P.S.H.). Recipient peripheral blood mononuclear cells were tested against donor B cells, and results were quantified by image analysis.

Statistical Analyses

Data are summarized using descriptive statistics for categorical (counts and percentages) and continuous (means and SDs) variables. Univariate comparisons were performed using Fisher’s exact test for categorical variables and t tests for continuous variables.

GFR was estimated using the Chronic Kidney Disease Epidemiology Collaboration equation.³¹ Serum creatinine data were complete through 24 months post-transplant for all 21 randomized subjects.

Disclosures

The work was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under Award Number U01-AI063594 (to P.S.H.). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.