In this multicenter, retrospective study, we assessed all consecutive adult KT recipients between July 2018 and July 2022 from nephrology and KT departments of Grenoble (France) and Bologna (Italy). The study included all patients with histological signs of caAMR or MVI + DSA-C4d- according to the 2022 Banff classification and that received intravenous TCZ at a dose of 8 mg/kg monthly and/or subcutaneous at the dose of 162 mg/15 days. Patients presenting de novo or recurrence of glomerulopathy, acute tubular necrosis were excluded. In Grenoble, surveillance kidney biopsies were performed 1-year after starting TCZ.

Patients were monitored for renal function (eGFR calculated using the CKD-EPI formula), urine albumin‐to‐creatinine ratio (UACR) (g/g), adverse and severe adverse events from treatment, graft- and patient-survival rates, and DSA levels every 3 months. Patient were all followed at least 1 year following MVI treatment initiation and data were collected at last available follow-up.

TCZ was given off-label in the absence of validated therapeutic alternative. Regarding retrospective data assessment, all patients signed an informed consent form. The study was performed in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments. All medical data were collected from Grenoble database [CNIL (French National committee for data protection) approval number 1987785v0] and from Bologna KT Center informatics repository for clinical data (protocol number 244/2023/Sper/AOUBo).

In both centers, all patients received induction with antithymocyte globulins or Basiliximab, methylprednisone (500 mg), and started peri-operatively with 1 g of mycophenolate mofetil or 720 mg mycophenolic acid.

Maintenance immunosuppression after KT consisted of tacrolimus or cyclosporine associated with mycophenolate mofetil/mycophenolic acid or everolimus. In Grenoble, steroids were rapidly tapered until withdrawal at month-3 except for highly sensitized patients (PRA more than 75%), patients who had experienced a previous acute rejection, patients with IgA nephropathy, or patients with circulating DSAs. In Bologna, steroids were rapidly tapered and maintained at prednisone 5 mg per day.

MVI-positive biopsies, including caAMR and MVI + DSA-C4d- were diagnosed on biopsies performed for clinical indications, i.e., rising serum creatinine and/or de novo proteinuria and/or de novo DSA detection. C4d staining was performed for all biopsies. The kidney-graft sample was processed for light microscopy by fixing in alcohol-formol-acetic acid (AFA, fluid and embedded in paraffin). Specimens were stained with trichrome blue, hematoxylin eosin and safran, periodic acid-schiff reagent, C4d staining, and immunofluorescence studies, including direct immunofluorescence for immunoglobulins, immunoglobulin light chains, C3 and C1q fractions of the complement. The results of biopsies were all re-assessed according to the 2022 Banff classification of rejection [10].

DSA were monitored at the time of each kidney biopsy, i.e., at MVI diagnosis and at 1-year post treatment. Screening of HLA Class I and II DSA in recipients was performed on sera using a Luminex platform with two different single antigen bead assay kits in the HLA department of Grenoble and of Bologna (Luminex Single Antigen assay, Immucor, Norcross, GA, United States and LABScreen, One Lambda Inc, Canoga Park, CA, United States) Screening and single-antigen assessment were performed every 6 months post-transplant and systematically at the time of a kidney biopsy. The retained mean fluorescence intensity (MFI) values corresponded to the manufacturers background corrected MFI value. Positivity for the Luminex analysis was defined as an MFI > 500.

IL-6 was assayed on frozen sera at the time of MVI diagnosis (before any treatments) and at 1-year post treatment initiation. In Grenoble, IL-6 quantitative dosage in sera was performed using a Lumipulse G600II system (Fujirebio, Tokyo, Japan).

The primary endpoint was the eGFR trajectory before and after TCZ initiation. To calculate and compare the trajectories of eGFR before and after TCZ, a mixed linear regression model with random intercepts was employed. The model included an interaction term between the time variable and the period variable, which allowed for the evaluation of differences in the trajectories between the “Before” and “After” periods. Five patients were diagnosed MVI-positive within the first-year post-transplantation. For the calculation of GFR trajectory, while most of studies remove patients from with end-stage renal disease from the eGFR trajectories, we assigned an eGFR value of 10 mL/min/1.73 m²for patients who experienced graft loss during follow-up. This value was chosen to reflect a realistic approximation of kidney function at the time of progression to end-stage kidney disease while maintaining the continuous nature of the variable. This approach avoids the introduction of extreme imputation values (e.g., 0 mL/min/1.73 m² of eGFR), which could disproportionately skew the trajectory analysis, particularly for patients already in advanced stages of chronic kidney disease at baseline.

Secondary endpoints were the one-year death-censored graft survival rate, proteinuria (g/g of creatininuria) and DSA. Six patients were not classified because of a follow-up shorter than 1-year and were excluded of the analysis. Thirty-three patients did not reach 2-year of follow-up and were excluded for the 2-year eGFR trajectories analysis. Evolution of histological biopsies was analyzed only in the Grenoble-Alpes hospital where patients had follow-up biopsies at 1-year.

Continuous variables are presented as means ± standard deviations (SD) or as medians with quartiles [Q1–Q3] in cases of high dispersion. Qualitative data are reported as the numbers of patients/events and percentages. The Wilcoxon or the Kruskal--Wallis tests were used for continuous variables, the chi-squared test was used for categorical data. For paired categorical data we used the Stuart-Maxwell test. For graft survival analyses, a multivariate generalized logistic-regression model was run. A two-sided p-value of <0.05 was considered statistically significant. Statistical analyses were conducted using R statistical software [29].

Between July 2018 and July 2024, 64 patients presented histological lesions of MVI on a for cause kidney biopsy performed after a median period of 58 [16–130] months post-transplantation. Patients’ age was 48.6 ± 14 years; female/male ratio was 0.4. Most of KT were deceased donor transplants (83.3%). The baseline characteristics of the cohort in both categories of MVI (i.e., caAMR and MVI + DSA-C4d-) are reported in Table 1. Chronic glomerulopathy lesions (cg score of ≥1) were present in 53.1% of patients, glomerulitis lesions (g score of ≥1) in 81.2%, and peritubular capilaritis (ptc score of ≥1) in 82.3% of patients. C4d staining was positive in 26.5% of patients. DSA were positive in 40 of patients (62.5%). eGFR at the time of biopsy was similar in caAMR and MVI + DSA-C4d- groups (41.1 ± 15 versus 37.9 ± 17, p = 0.378 respectively). We also compared the patients in both centers (Supplementary Table S1). Patients from Grenoble versus Bologna were significantly younger (44.8 versus 53.3 years, p = 0.02), received more often antithymoglobulin (100% versus 44.4%, p < 0.001), had less DSA at the time of biopsy (48.7% versus 84%, p = 0.01) and had earlier MVI diagnosis (median 35 months post-KT versus 154 months, p < 0.001).

eGFR at the time of biopsy (before TCZ treatment) was 39.4 ± 16 mL/min/1.73 m². The eGFR 1-year before TCZ treatment was 54.0 ± 19 mL/min/1.73 m² and at 1 year after TCZ treatment was 40.8 ± 18 mL/min/1.73 m². Among thirty-one patients (43%) with a follow-up of 2 years post-treatment, last eGFR was 43.6 ± 18 mL/min/1.73 m².

The linear mixed-effect model of the eGFR trajectory showed a significant difference between the pre- TCZ and post- TCZ period: -1.2 ± 0.2 vs. +0.03 ± 0.2 mL/min/1.73 m²/month, respectively; p = 0.001 (Figure 1A). When considering only the patients with 2 years of follow-up, eGFR trajectory decreased from −1.2 ± 0.2 mL/min/1.73 m²/month before TCZ versus −0.15 ± 0.1 mL/min/1.73 m²/months after 2 years TCZ, p < 0.001 (Figure 1B).

eGFR trajectory was then evaluated according to the Banff 2022 classification. In caAMR phenotype patients, the eGFR trajectory significantly decreased from −0.9 ± 0.3 mL/min/1.73 m²/month before TCZ to −0.2 ± 0.3 mL/min/1.73 m²/month after TCZ, p = 0.038), as shown in Supplementary Figure S1A. In MVI + DSA-C4d- patients, there was also a significant improvement of eGFR trajectory after TCZ introduction (−1.9 ± 0.5 mL/min/1.73 m²/month before TCZ versus +0.5 ± 0.5 mL/min/1.73 m²/month after TCZ, p = 0.007) (Supplementary Figure S1B).

In both centers, the improvement of eGFR trajectory post TCZ was significant: -1.4 ± 0.5 mL/min/1.73 m²/month before TCZ versus +0.1 ± 0.2 mL/min/1.73 m²/month after TCZ, p = 0.004 in Grenoble and 0.6 ± 0.3 mL/min/1.73 m²/month before TCZ versus −0.5 ± 0.2 mL/min/1.73 m²/month after TCZ, p = 0.03 in Bologna (Supplementary Figure S2).

In the whole cohort, albuminuria/creatininuria ratio declined from 1.0 ± 1.3 g/g to 0.6 ± 0.7 at 1-year after TCZ (p = 0.033; Figure 2A).

To evaluate the predictors of TCZ response, patients whose eGFR did not decrease (trajectory ≥0) between baseline and 1-year post treatment were defined as “responders” to TCZ, whereas patients whose eGFR decreased after 1-year were defined as “non-responders.” According to this definition, twenty-five (37.8%) patients were classified as “responders” and 41 (62%) patients were “non-responders”. In univariate analysis the factors that resulted to be significantly associated with TCZ response were younger age (OR = 0.97 [0.96–0.98], p = 0.035), lower chronic glomerulopathy lesions in the initial biopsy (OR = 3.45 [1.14–11.9], p = 0.035), and the MVI + DSA-C4d- phenotype (OR = 3.58 [1.24–10.8], p = 0.019). In the multivariate analyses, younger age (OR = 0.9 [0.91–0.99], p = 0.022), lower score of chronic glomerulopathy (OR = 4.49 [1.29–18.9], p = 0.025) and MVI + DSA-C4d- phenotype (OR = 5.25 [1.55–20.71], p = 0.01) remained significantly associated with the 1-year response to TCZ (Table 2).

At 1-year post-TCZ, patient survival was 98.4% and graft survival was 93.75%. During the follow up, one patient died (due to gastric cancer 48 months after TCZ start and one patient discounted TCZ, documented by a 4-fold increase over the baseline of AST and ALT blood levels. At last-follow-up visit post-TCZ, patient survival was 98.4% and graft survival was 85.9%.

The patients who lost their graft during the first year were significantly older compared to the rest of the cohort (61.4 ± 15 vs. 47.9 ± 14 years-old, p = 0.016), were diagnosed for MVI after a longer period post-transplantation (16.4 years [12–20] versus 4.8 [1–10.3] years, p = 0.003) and had a higher UACR at the time of biopsy (1.5 ± 0.9 vs. 0.9 ± 1.3 g/g, p = 0.024). There was no difference in the initial histological severity between patients who developed graft loss and the others. All patients who experienced transplant failure had received their kidney grafts from deceased donors, while 9.4% of grafts of the rest of the cohort were from living donors, a difference that did not reach statistical significance (p = 0.180). At last follow-up visit of 19.7 [13–32] months, 15 patients (20.8%) had lost their graft after a mean time of 13.4 ± 10 months.

At the beginning of TCZ treatment, the median MFI of iDSA was 9,537 [1,426–15,075] and, at 1-year post treatment, decreased to 7,250 [3,100–14,975] (p = 0.001) (Figure 2B). Forty (62.5%) patients had at least a positive circulating DSA before TCZ treatment and, at 1-year post treatment, 28 (38.9%) of patients had a positive DSA (p < 0.001) (Figure 2C). Most of those DSA were class II (84.3%). Patients referred as “responders” were more often DSA negative (72.5% versus 45.8%, p = 0.03), but the average MFI of the iDSA was not statistically different between “responders” and “non-responders”: 3,400 [1,265–19,175] versus 10,000 [2,912–13,900] respectively, p = 0.648. There was no statistical difference for DSA presence or iDSA MFI between patients who lost their graft versus those with a functioning graft at 2-year post- TCZ.

Serum specimens were available for 25 patients before and at 1-year post-treatment. In these patients, IL-6 levels significantly increased at 1 year after treatment initiation (Supplementary Figure S3). Baseline IL-6 levels were tendentially higher (although not significant) the “non-responders” than in the “responders” (7.1 ± 10 versus 5.9 ± 3.1 pg/mL, p = 0.118). Lastly, there was no association between serum IL-6 levels at baseline and graft loss (not shown).

All patients of Grenoble had follow-up kidney biopsies. Figure 3 shows the evolution of g scores, ptc scores, g + ptc scores, c4d staining and chronic glomerulopathy lesions (cg) between the first biopsy with microvascular inflammation (MVI) diagnosis and 1-year post TCZ treatment. After 1-year in the TCZ group, a significant decrease was observed in the g score (p = 0.014) but not in the ptc, g + ptc, c4d and cg scores. Within the g-score, 29.3% had a score of 3 before TCZ vs. 22% at 1-year post-TCZ, p = 0.032.

Within the cohort, 17 patients (26.6%) received TCZ subcutaneously during the first year of treatment. The evolution of eGFR post TCZ was followed in patients converted to sub-cutaneous TCZ within the first year and those who remained on intravenous infusion. In patients converted to sub-cutaneous injections, eGFR trajectory during the first year was – 0.21 ± 0.46 mL/min/1.73 m²/month, not statistically different from patients that received only intravenous TCZ: +0.15 ± 0.29 mL/min/1.73 m²/month, p = 0.521 (Supplementary Figure S4). Sub-cutaneous administration was not associated with a reduced response to TCZ (20.8% in the group of responder vs. 30% in the non-responder patients, p = 0.421).

Thirteen patients (20.3%) discontinued TCZ either because of histological or clinical stabilization (n = 7, 10.9%) or because of side effects (n = 6, 9.3%) at the end of follow-up (2 discontinued during the first year). The side effects that resulted in TCZ discontinuation were infections in five patients (peritonitis, cryptococcosis disease, CMV disease, tuberculosis and campylobacter infection), hepatotoxicity (1) within the first year and a DRESS (Drug reaction with eosinophilia and systemic symptoms syndrome) within the second year post-TCZ.