The last version of the protocol is available in Supplementary Material. A randomized controlled, non-blinded multicenter superiority trial was conducted at 15 centers in France between February 12, 2015, and July 25, 2019. Among the centers, 11 sites enrolled patients including Paris (three sites), Lyon, Nantes, Toulouse, Montpellier, Suresnes, Créteil, Strasbourg, and Amiens. Eligible patients were randomized in a 1:1 ratio to receive bortezomib or to the standard-of-care group using a centralized web-based randomization system (CleanWeb^TM; Telemedicine Technologies SAS, France). Randomization occurred after AMR was confirmed by an inpatient hospital transplant biopsy, eligibility criteria were checked, and informed consent was obtained. Randomization was stratified according to the center and type of rejection, clinical or subclinical. AMR was classified as clinical if the biopsy was performed for graft dysfunction or new-onset proteinuria, and subclinical if performed as a screening biopsy or in response to the detection of dnDSA.

Patients were included if they met the following criteria: i) aged at least 18 years, ii) had received a first or second ABO-compatible kidney transplant either from a deceased donor after brain death or circulatory death, or from a living donor, performed at least 90 days earlier, iii) presence of one or more dnDSAs, iv) evidence of microvascular inflammation (sum of glomerular inflammation (g score) and peritubular capillary inflammation (ptc score), g + ptc ≥2), and chronic glomerulopathy (cg score <3) on graft biopsy, and v) provided prior written informed consent. Female patients of childbearing potential were required to have a negative pregnancy test on the day of enrollment, and all participants (male and female) were required to use at least one effective contraceptive method before the initiation of treatment, throughout treatment and for the duration of the study. Patients with concomitant borderline lesions or T-cell-mediated rejection, corresponding to mixed rejection phenotypes, were eligible for inclusion.

Exclusion criteria were as follows: i) presence of a DSA before or at the time of transplantation, ii) patients receiving a third or fourth kidney transplant, iii) receiving combined transplantation with a non-renal organ, iv) previous history of AMR during the current transplantation, v) severe graft dysfunction, defined as an estimated GFR (eGFR) < 20 mL/min/1.73 m², vi) severe histological lesions of chronic AMR, defined by the cg score of 3/3, vii) severe neuropathy or hematologic disturbances (platelet count <100 × 10⁹/L or absolute neutrophil count <1,000 × 10⁶/L), viii) active uncontrolled infection, chronic active hepatitis B or C, or HIV infection, ix) severe hepatic impairment or abnormal liver function tests, and x) acute diffuse infiltrative pneumonitis or pericardial involvement. Patients unable to understand or comply with the study protocol, as well as those enrolled in another therapeutic trial or within its exclusion period, were also excluded.

All patients initially received five plasma exchange sessions within 10 days of inclusion. During this period, maintenance immunosuppression was adjusted, if necessary, to include prednisone (10 mg/day), mycophenolate mofetil (500 mg twice daily) or mycophenolate sodium (360 mg twice daily), and tacrolimus with target trough levels of 6–10 ng/mL (Table 1).

Patients in the bortezomib group received two cycles of intravenous bortezomib, each consisting of four injections (1.3 mg/m² every 3 days), administered at three-week intervals. In addition, these patients received oral dexamethasone (20 mg) on each day of bortezomib administration and four IVIg infusions (2 g/kg body weight) at three-week intervals. Patients in the standard-of-care group received dexamethasone according to the same schedule as the bortezomib group (20 mg orally, four doses every 3 days, repeated after 3 weeks) along with four IVIg infusions (2 g/kg body weight) at three-week intervals. The planned follow-up duration was 12 months (Table 1).

All patients received prophylaxis with cotrimoxazole and valaciclovir for 6 months after inclusion to prevent Pneumocystis jirovecii, herpes simplex virus, and varicella-zoster virus infections. The bortezomib treatment was temporarily discontinued in the event of toxicity including neutropenia, thrombocytopenia, abnormal liver function tests and/or neuropathy. Permanent discontinuation was required in cases of voluntary withdrawal by the patient, severe irreversible toxicity and/or terminal graft dysfunction requiring return to dialysis.

The primary endpoint was a composite of two conditions assessed at 12 months: i) a >50% decrease from baseline in the mean fluorescence intensity (MFI) of the idDSA and, ii) stabilization or improvement of histological AMR lesions on the 12-month biopsy, defined as a difference in MVI score, defined as g + ptc, ≤1, and in cg score <1 compared with baseline. This composite endpoint was chosen to capture both immunological and histological responses to treatment.

Secondary endpoints included: i) each component of the primary endpoint analyzed separately, ii) in patients with multiple DSA, relative change from baseline in the MFI of each DSA and in the cumulative MFI, iii) change in DSA MFI after 6 months, iv) histological changes between baseline and 12-month biopsies, including acute and chronic lesions according to the latest Banff criteria, v) changes over 12 months in proteinuria (24-h excretion and urine protein/creatinine ratio), serum creatinine and eGFR (Modification of Diet in Renal Disease (MDRD) formula), and vi) patient and graft survival at 12 months. Safety was assessed throughout treatment and during the 12-month follow-up, with a focus on hematologic, hepatic, neurologic (including neuropathy) and infectious adverse events.

For histopathological assessment, all available baseline/screening and 12-month kidney biopsy slides were centrally reviewed at the pathology laboratory of Hôtel Dieu de Nantes by an expert renal pathologist blinded to treatment allocation. Slides were assessed using standard histological stains, including hematoxylin–eosin–saffron, periodic acid–Schiff, Masson’s trichrome, and silver staining, and lesions were graded according to updated Banff classification at the time of reading. Kidney biopsies were subsequently reclassified according to the 2022 Banff criteria using an automated method [15]. DSA were centrally assessed using Luminex single antigen (Labscreen One Lambda, Thermofisher) by immunologists blinded to treatment allocation. Investigators at each participating center collected clinical and transplant-related data, including sex, weight, rank of transplantation, donor type, dialysis vintage, local histological readings, local pre- and post-transplant HLA antibody assays, and concomitant immunosuppressive therapies.

The sample size calculation assumed primary endpoint rates of 5% in the standard-of-care group and 25% in the bortezomib group. With a two-sided alpha of 0.05% and 80% power, the planned sample size was 50 patients per group, or 100 patients in total. The planned inclusion period was 24 months. Despite multicenter recruitment and a 30-month extension of the inclusion period, the target sample size was not reached because of slow recruitment, reflecting the limited number of patients who met all eligibility criteria and were willing to participate. Recruitment was therefore closed at the end of the extended inclusion period after 60 patients had been randomized.

The primary endpoint was compared in the intention-to-treat (ITT) population with missing data considered as treatment failures. A sensitivity analysis was also performed in the per-protocol (PP) population. Secondary endpoints were analyzed in the ITT population for the two components of the primary endpoint, and on available data (complete-case analysis) for the other endpoints. All statistical tests were two-tailed with a significance level of 5%.

Baseline clinical and biological characteristics were described by group. Quantitative variables were expressed as mean ± standard deviation (SD) or median [Q1:Q3] and qualitative variables as numbers and percentages (%). Comparisons of the primary endpoint and other categorical variables between the two groups were performed using Chi-squared or Fisher’s exact tests, as appropriate. Quantitative variables were compared using Student’s t-test or Wilcoxon’s rank-sum test when distributional assumptions were not met. Relative risks (RR) and 95% confidence interval (95% CI) were estimated using log-binomial regression models. Longitudinal outcomes were analyzed using linear mixed-effects models. Secondary analyses were considered exploratory and no adjustment for multiple comparisons was performed. Given the limited sample size, secondary endpoints were analyzed on available data without imputation. All analyses were performed with R software (www.r-project.org).

This study was conducted in accordance with the Declaration of Helsinki. Approval was obtained from the Comité de protection des personnes Ile de France II (CPP) (Reference: 2014-03-01 RBM) and the Agence Nationale de Sécurité du Médicament et des Produits de Santé (ANSM) (Reference: 140368A-11). This trial was registered at ClinicalTrials.gov (NCT02201576) and an independent Data Safety Monitoring Board (DSMB) regularly reviewed the data to ensure patient safety and monitor the progress of the study.

Between February 12, 2015, and July 25, 2019, 60 patients were randomized. The planned sample size of 100 patients was not reached by the end of the extended recruitment period because of slower-than-anticipated enrolment related to the restrictive eligibility criteria. For the ITT analysis, 30 patients were allocated to each group. For the PP analysis, 20 patients were included in the bortezomib group and 19 in the standard-of-care group; one patient randomized to bortezomib did not receive the intervention and was therefore analyzed in the standard-of-care group for the PP analysis (Figure 1). Baseline characteristics are presented in Table 2. Patients in the bortezomib group were more frequently women (53.3% vs. 13.3%) and had a lower mean age (44.4 ± 13.8 vs. 51.3 ± 13.0 years). The median duration from transplantation to inclusion was 76 months in both groups. Notably, retransplantations occurred only in the bortezomib group (13.3%) (Table 2). A history of T-cell mediated rejection (TCMR) prior to inclusion was reported in one patient in the bortezomib group, and four in the standard-of-care group. Most patients received triple maintenance immunosuppression consisting of prednisone, ciclosporin or tacrolimus, and mycophenolate (Table 2). At inclusion, six patients in each group were not receiving a calcineurin inhibitor. AMR was more often subclinical than clinical for both groups (bortezomib group: 60.0% and standard-of-care group: 66.7%).

At inclusion, patients in the bortezomib group more frequently had two or more dnDSA than in the standard-of-care (60.0% vs. 41.3%), with a median of 2 [1:3] DSA vs. 1 [1: 2] in the standard-of-care group, and they also had a higher proportion of class I DSA than in the standard-of-care group (whether considering the idDSA or all the DSA) (Table 3). Both the idDSA MFI and the sum of all the DSA MFI per patient were higher for patients in the bortezomib group.

Concerning biopsy findings at inclusion, discrepancies were present between local and central readings of AMR lesions (Table 4). According to the central assessment, only 25 patients (83.3%) in the bortezomib group and 24 patients (80.0%) in the standard-of-care group had MVI score ≥2, which was used as an inclusion criterion by local readings. Chronic glomerulopathy (cg score ≥1) was only reported in seven patients (23.3%) in each group, explaining the low median [IQR] score of 0[0:0]. According to the 2022 Banff criteria reclassification, active AMR was present in 18 patients (60.0%) in each treatment group. Chronic active AMR was present in seven patients (23.3%) in the bortezomib group and six patients (20.0%) in the standard-of-care group; one additional patient (3.3%) in the standard-of-care group had chronic inactive AMR. The remaining patients were classified as having C4d staining with DSA, probable AMR, or no AMR. Associated TCMR lesions were present in nine patients (30.0%) in the bortezomib group and five patients (16.7%) in the standard-of-care group (Table 4).

In both the ITT and PP analyses, the primary endpoint rate did not significantly differ between groups. In the ITT analysis, the endpoint was achieved by 40.0% of patients in the bortezomib group and 33.3% in the standard-of-care group (RR = 1.20 [95% CI: 0.61 to 2.34], p = 0.59). Missing primary endpoint data were considered treatment failure in four patients in the standard-of-care group and one in the bortezomib group. In the PP analysis, the corresponding rates were 50.0% and 42.1% (RR = 1.19 [95% CI: 0.60 to 2.36], p = 0.62) (Table 5). No significant interaction was observed between treatment allocation and rejection type (clinical or subclinical; p = 0.86).

When separately analyzed, neither component of the combined endpoint was different between groups. After 12 months, a >50% reduction in idDSA MFI was observed in 56.7% of patients in the bortezomib group and 46.7% of patients in the standard-of-care group in the ITT analysis (p = 0.44), and 65.0% and 57.9% in the PP analysis (p = 0.65), respectively. Stabilization or improvement of histological lesions at month 12 was reached in 56.7% of patients in both groups in the ITT analysis (p = 1.00), and 68.4% and 66.7% of patients in the PP analysis, respectively (p = 0.91). Thus, 5/28 (17.8%, bortezomib) and 2/25 (8.0%, standard-of-care) of patients experienced a >50% decrease in idDSA MFI without improving or stabilizing MVI, and 5/28 (17.8%, bortezomib) and 7/25 (28.0%, standard-of-care) improved or stabilized MVI without significantly reducing idDSA MFI.

In a post hoc analysis of patients with baseline idDSA MFI >15,000, the primary composite endpoint was achieved in 42.1% of patients in the bortezomib group and 8.3% in the standard-of-care group (p = 0.10). A >50% reduction in idDSA MFI was observed in 52.6% and 16.7% of patients, respectively (p = 0.065) (Supplementary Table S1).

In separate sensitivity analyses adjusted individually for baseline characteristics that were imbalanced between groups despite randomization (sex, transplantation rank, idDSA class, and baseline idDSA MFI), the association between treatment allocation and the primary endpoint remained non-significant (Supplementary Table S2).

Changes in serum creatinine, eGFR, and urine protein/creatinine ratio between baseline and 12 months did not significantly differ between patients in the bortezomib and standard-of-care groups (Table 5; Figure 2). No deaths occurred, and one patient in each group returned to dialysis.

For patients receiving bortezomib, the idDSA MFI decreased by 45.1% [-69.7: -6.1] after 6 months (n = 23), and by 60.2% [-80.6: -11.5] after 12 months (n = 28). Similar reductions were observed for patients in the standard-of-care group: 50.7% [-76.6: -6.6] after 6 months (n = 27; p = 0.92) and 60.0% [-81.3: -11.5] after 12 months (n = 27; p = 0.85) (Figure 3). Median relative change in the sum of all DSA MFI at 12 months was also comparable: 60.3% [-74.3: -27.6] in the bortezomib group (n = 28) and 57.6% [-81.8: -21.5] in the standard-of-care group (n = 27; p = 0.80).

Patients whose idDSA MFI dropped by at least half at 12 months (responders) had lower baseline MFI values than non-responders: median 16,080 vs. 22,582 in the bortezomib group and 2,996 vs. 22,054 in the standard-of-care group.

Median scores of histological lesions remained stable between baseline and 12 months in both groups (Table 6). Histological failure was due to the progression of MVI (Δg + ptc score ≥2) in three patients in each group and of transplant glomerulopathy (Δcg score ≥1) in eight patients in the bortezomib group and five patients in the standard-of-care group (Figure 3). One patient in each group had progression of both MVI and cg scores, while most with cg progression showed no or only mild worsening of MVI. Among patients in the ITT population with an MVI score ≥2 at baseline and available 12-month histological data, nine of 23 patients in the bortezomib group and 10 of 20 patients in the standard-of-care group showed a decrease in MVI score between baseline and 12 months. Complete MVI resolution, defined as an MVI score of 0 at 12 months, occurred in 3/23 patients (13.0%) and 2/20 patients (10.0%), respectively. In the PP analysis, complete MVI resolution occurred in 3/20 patients (15.0%) and 2/17 patients (11.8%), respectively. The cg score decreased in three patients in the bortezomib group and two patients in the standard-of-care group (Table 6; Figure 3).

In the bortezomib group, tubulitis with a Banff t score ≥1 was present in 29.6% of patients at baseline and 7.4% at 12 months, compared with 20.8% and 4.2%, respectively, in the standard-of-care group. Interstitial inflammation with a Banff i score ≥1 was present in 19.0% and 7.0% of patients in the bortezomib group and in 21.0% and 8.0% of patients in the standard-of-care group at baseline and 12 months, respectively. Interstitial fibrosis and tubular atrophy (IFTA) scores worsened in six patients in the bortezomib group and eight patients in the standard-of-care group, which was associated with a cg or MVI progression in two patients in each group. According to the 2022 Banff classification at the 12-month biopsy, 19/27 patients in the bortezomib group and 19/24 patients in the standard-of-care group still had active or chronic active AMR (Table 6).

Overall, 132 adverse events were reported in the bortezomib group and 118 in the standard-of-care group, with infections, neurological and gastrointestinal events being the most frequent in both groups. In the bortezomib group, 11 patients discontinued or did not complete the two planned treatment courses due to hematological, neurological or hepatic toxicity. Twenty-three patients in the bortezomib group (76.7%) experienced at least one serious adverse event compared to 14 (46.7%) in the standard-of-care group (p = 0.017), corresponding to 43 and 30 serious adverse events, respectively (Table 7). No significant between-group differences were observed in the proportions of patients experiencing adverse events or serious adverse events within individual System Organ Class categories. Infections, including urinary tract and viral infections were the most common serious adverse events in both groups.