We performed a longitudinal single-center retrospective study, which included kidney recipients diagnosed with ABMR, according to the Banff 2017 classification. Concretely, we have identified kidney recipients who received a treatment for ABMR from January 1, 2004 to December 31, 2019 (including a combination of PE, IVIg, and rituximab) in the database of Renal Transplant Unit at Hospital Clinic de Barcelona; then biopsies at ABMR diagnosis were reanalyzed according to the criteria specified by Banff (2017). Recipients who received a multivisceral transplant, and those with transplant glomerulopathy (TG) in the initial biopsy, cg ≥ 1 in the Banff histopathological classification, were excluded (10).

Demographic, clinical, biochemical, histopathological, and immunological data were evaluated for both the donor and the recipient. Clinical characteristics, maintenance immunosuppression, and ABMR treatment were analyzed at the diagnosis and follow-up period. Charlson comorbidity index (CCI) was also assessed at ABMR diagnosis (11).

Infections that required hospitalization at least 48 h within the first year after ABMR diagnosis were recorded and described in relation to clinical variables.

The study was performed according to the Declaration of Helsinki principles and approved by the Hospital Research Ethics Committee. Study registration: Agencia Española de Medicamentos y Productos Sanitarios (AEMPS): 14566/RG 24161. Study code: UTRINM-2017-01.

The decision to perform a renal biopsy and patient treatment was based on the clinical judgment at ABMR diagnosis, including biopsies due to impaired renal function and protocol biopsies (at 3 or 12 months after kidney transplantation). Active ABMR was diagnosed and categorized according to the Banff criteria of 2017 (10). The day when the biopsy was diagnosed was considered as the date of active ABMR diagnosis. Immunologically, donor-specific antibodies (DSAs) were tested using Single Antigen Bead Test (LIFECODES^® Single Antigen, Immucor, Georgia, US). In our Center, the Single Antigen Beads Test has been used since 2001. However, the criteria to consider an allele positive (the MFI over 1,500 and 4 times higher than the Lowest Reactive Antigen of the same locus) changed in 2017. Since 2017, the criterion is that the MFI is >750 and the cut-off for the ratio [MFI/LRA (Lowest Ranked Antigen)] is specific for each individual bead (12).

Our primary outcome was death-censored kidney graft survival at 1 year after ABMR diagnosis and at the follow-up period, between patients with and without persistent inflammation in the control biopsy, either microvascular (MVI, defined as g + ptc ≥ 1) or tubular (t ≥ 1). Follow-up biopsies after ABMR treatment were performed according to the physician's criteria. Secondary outcomes were defined as kidney graft function at 6 months from the ABMR treatment and kidney graft function at the last follow-up. Kidney graft function was assessed by serum creatinine (SCr), estimated glomerular filtration rate (eGFR, according to the Chronic Kidney Disease Epidemiology Collaboration equation), and urine protein to creatinine ratio (UPCR) (13, 14). Early ABMR was defined as that which occurred within 6 months from the kidney transplant, while late ABMR was defined as that occurring after 6 months from the kidney transplant (15).

Patient survival was defined as the last day of follow-up or the date of death. Kidney graft failure was defined as one of the following: return to dialysis or re-transplantation. Response to ABMR treatment was defined as improvement or stabilization of eGFR at 6 months compared to eGFR at ABMR diagnosis.

PE was performed in Cobe Spectra or Spectra Optia separators (Terumo BCT, Lakewood, CO, USA) using 5% albumin (Albutein® 5%, Grífols, Spain) as a replacement solution. One plasma volume was exchanged in each session (16).

Data are presented as mean (SD) for parametric variables, and median [interquartile range (IQR)] for the non-parametric ones. The corresponding tests used were the t-test, McNemar Test, Wilcoxon test, Chi-Square, and ANOVA as appropriate.

Kaplan–Meier was used to estimate graft survival and compared using the log-rank test. For the survival analysis throughout the follow-up, we used the Cox regression model, and the logistic regression model was used for the 1-year survival analysis using IBM SPSS Statistics 26.0 (SPSS, Inc; Chicago, Illinois) software for Windows. All the tests were two-tailed, and the significance level was defined as a P-value < 0.05.

From January 2004 to December 2019, 116 kidney transplant recipients were diagnosed with an active ABMR (Figure 1). In 97 patients (83.6%), the biopsy was performed by indication due to impaired renal function, while the other 19 patients (16.4%) underwent a protocol biopsy at 3 or 12 months after transplantation. Table 1 summarizes demographical and clinical data at ABMR diagnosis. Most of them were men (59.5%), with a mean age at the rejection of 50.8 ± 14 years, and a median follow-up from ABMR diagnosis of 33.5 [62.7] months. The median duration of dialysis was 4 [4] years. About 76% of the donors were deceased donors. Up to 48% of patients had received a previous kidney transplant, and 14% presented a positive cytometry crossmatch at transplantation.

Regarding the characteristics of rejection (Table 1), the median time from transplantation to active ABMR was 27.5 [287.7] days, with 70.7% (82) of the ABMR episodes diagnosed within 6 months after transplantation and which were considered as early ABMR [median of 16 (21.5) days from transplant]. Twenty-seven of the patients required dialysis after the diagnosis of ABMR.

All patients were treated for ABMR with a combination of corticosteroids, PE, IVIg, and rituximab. The active ABMR treatment protocol consists of a combination of five sessions of PE, IVIG 200 mg/kg every 2 PE, and two rituximab doses. However, in this cohort, 15 patients (12.9%) were not treated with rituximab by concerns of the treating physician because of a perception of increased risk of infections. Also, five patients (4.3%) did not receive PE treatment for problems related to vascular access. One plasma volume was exchanged in each session with a median of 5 [1] sessions.

The global response to ABMR treatment was 54.3%, with a significant increase in eGFR at 6 months after treatment and at the end of the follow-up period (P = 0.003). Table 2 summarizes the changes in eGFR. Patients with an early ABMR [median 16 (21.5) days] have a better response than those with a late ABMR [median 25.9 (40) months], 67 vs. 23.5% for early and late ABMR, respectively, odds ratio (OR) 0.15 [95% CI 0.06–0.38], P < 0.001. Overall graft failure at 1 year and throughout the follow-up was 32.8 and 38.8%, respectively. Death-censored graft failure for the same time points was 25.9 and 31%, respectively. The presence of DSA at diagnosis was not associated with worse graft survival (P = 0.15).

The treating physicians decided follow-up biopsy after ABMR treatment in 90 patients. Demographic, clinical, biochemical and immunological characteristics from patients with and without follow-up biopsy have shown in Supplementary Table 1, whereas histological parameters in biopsy at ABMR diagnosis are presented in Supplementary Table 2. The prevalence of living donors, previous kidney transplants, and human leukocyte antigen (HLA) sensitization were higher in the follow-up of patients who underwent biopsy. There were no significant differences in death-censored graft survival, where P = 0.31 between patients with or without follow-up biopsy.

Histopathological findings at ABMR diagnosis and follow-up biopsies are summarized and compared in Table 3.

The presence of tubulitis at ABMR diagnosis was associated with an increased risk of graft loss at 1 year (OR 1.79 [95% CI 1.05–3.06], P = 0.03) and at follow up [HR 2.10 (95% CI 1.04–4.26), P = 0.04]. The combination of interstitial fibrosis and tubular atrophy (IFTA) and the coexistence of a T Cell-mediated rejection (TCMR) were associated with an increased risk of graft loss at follow-up [HR 1.62 (95% 1.09–2.40), P = 0.02 and HR 2.48 (95% CI 1.07–5.76), P = 0.03 for IFTA and TCMR, respectively; Table 4A].

A follow-up biopsy after ABMR treatment was performed in 90 patients (77.6%), with a median time from the treatment of 2 [4] months. In general, there was a significant decrease in inflammation and an increase in chronicity parameters between the biopsy at diagnosis and the follow-up biopsy (Table 3).

Although 46 out of 90 rebiopsied patients (51%) had responded clinically to ABMR treatment, about 64 (71%) follow-up biopsies presented a persistent MVI ≥ 1, and in 45 (50%), MVI ≥ 2 was detected. Twenty-six out of 46 patients with improvement in graft function had persistence of MVI ≥ 1. There were no statistically significant differences in baseline characteristics between patients with and without persistent MVI (Supplementary Table 3). There were also no significant histological differences in the biopsy at ABMR diagnosis between patients with and without the persistence of microvascular inflammation, except for a greater severity of MVI in patients with persistent MVI (Supplementary Table 4).

The time between kidney transplantation and graft biopsy was similar between patients with and without inflammation persistence in the follow-up biopsies (P = 0.23). The presence of DSA or the DSA title was similar between the two groups (P = 0.09). The presence of preformed DSA was associated with early rejection (P = 0.007; OR 5.1, CI95% 1.4–18.1). However, de-novo DSA was not associated with a higher risk of graft loss or persistence of MVI in the follow-up biopsies.

Death-censored kidney graft survival was significantly lower in those patients with persistent MVI (Figure 2A; Log-Rank, P = 0.002). Patients with glomerulitis (g) or peritubular capillaritis (ptc) had significantly lower death-censored kidney graft survival than patients without g or ptc at follow-up biopsy (Figures 2B,C).

In the Cox Regression analysis, the persistence of acute inflammatory lesions in the follow-up biopsies [g, ptc, MVI, tubulitis, acute tubular necrosis (ATN), and thrombotic microangiopathy (TMA)], significantly increased the risk of kidney graft loss throughout the follow-up (Table 4B). Also, the presence of TMA and ATN was associated with 1-year graft loss.

Here a multivariate analysis has many statistical limitations due to the number of events. Therefore, a bivariate analysis was performed instead of a multivariate analysis. After the adjusted analysis for the different histological characteristics, the persistence of MVI in the follow-up biopsies remains a risk factor for graft loss (Supplementary Tables 5, 6).

There was no specific retreatment protocol, and therapy was the decision of the treating physicians. Twenty-four out of 64 (37.5%) patients with persistent MVI in follow-up biopsies were retreated with a combination of rituximab (5 patients) and PE (19 patients). Death-censored kidney graft survival was significantly higher for those patients with persistent MVI in control biopsy who were retreated (95 and 73% vs. 75 and 50% at 1 and 5 years after retreatment, respectively, P = 0.04; Figure 3). In the Cox regression analysis, ABMR retreatment for those patients with persistent MVI was associated with a lower risk of graft loss [HR 0.40 (95% CI 0.16–0.99), P =0.048].

Presence of tubulitis (t) was observed in 19% (17) of follow-up biopsies, from which 13 had t grade ≥ 1 without meeting the active or chronic TCMR diagnosis criteria. Patients with t ≥ 1 in follow-up biopsies exhibited worse kidney graft survival rates at 1 year (66 and 44%, P = 0.02) (Figure 4). Moreover, the persistence of tubulitis in follow-up biopsies significantly increased the risk of kidney graft failure [HR 2.88 (95%CI 1.24–6.69), P = 0.01; Table 4B and Supplementary Table 6].

The presence of TG [HR 2.24 (95% CI 1.33–3.77), P = 0.002], and IFTA [HR 1.62 (95% CI 1.14–2.32), P = 0.008) in follow-up biopsies were associated with an increased risk of kidney graft failure at follow-up. TG was also associated with an increased risk of 1-year kidney graft loss [OR 4.72 (95% CI 1.27–17.61), P = 0.02; Table 4B).

Given the different responses to treatment according to the time from transplantation to rejection, we analyzed the histological characteristics associated with graft loss adjusted for time to rejection (early/late) in the biopsy at ABMR diagnosis and in the follow-up biopsy. Results were shown in Supplementary Table 7. In the time-adjusted analysis, the presence of tubulitis or concomitant cellular rejection in the biopsy at ABMR diagnosis was associated with graft loss. In the follow-up biopsy, the persistence of MVI> 1, ATN, or TMA and the appearance of chronic lesions (cg, IFTA) were associated with graft loss.

During the year after ABMR treatment, 101 infections required hospital admission at least 48 h in 57 patients, which were supposed to be the infection rate of 0.87 infections/treated patient. The presence of the comorbidities of the recipients was associated with an increased risk of infectious complications with admission requirement, with CCI ≥ 4 associated with an increased risk of infections [OR 4.2 (95%CI 1.79–9.81), P = 0.01].

The potential association between the total immunosuppression received and infections was analyzed. The following items were not associated with the development of infections: time from transplantation to ABMR treatment (P = 0.72), induction with thymoglobulin (P = 0.31), induction with rituximab (P = 0.7), any previous kidney transplantation (P = 0.58), and ABMR re-treatment (P = 0.31).