We performed a retrospective study including all KTRs (January 2012 to September 2019) from two Paris (France) area transplant units (Tenon and Mondor hospitals) with at least one preformed Class I or Class II DSA and with a mean fluorescence intensity (MFI) above 3,000 between 3 months before transplantation and the day of transplantation. Immunodominant DSA (iDSA) was defined as the DSA with the highest pre-KT MFI. KT was only performed if complement-dependent cytotoxicity (CDC) crossmatch for IgG was negative on the day of transplantation, but IgM CDC crossmatch was not a contraindication to perform KT. Patients who underwent ABO-incompatible transplantation or combined multiorgan transplantation were excluded. Follow-up ended December 1st 2020.

All patients received an induction treatment combining methylprednisolone (500 mg on the day of transplantation), rabbit-ATG (Thymoglobulin 1.5 mg/kg over four or 5 days, depending on center), mycophenolate mofetil (2–3 g/day), tacrolimus (target trough level 8–12 ng/mL during the first 3 months) and four intravenous immunoglobulin (IVIg) (Clairyg or Privigen, 2 g/kg) post-transplant courses (once every 3–4 weeks). Patients in the intensive group additionally received one rituximab 1,000 mg dose and six PE sessions (60 mL/kg, 100% plasma) after transplantation. The choice of standard or intensive induction regimen therapy was based on the center and the evaluation of the nephrologist before KT. Intensive induction protocol was based on previous studies with high immunological risk patients (13).

Maintenance therapy consisted of prednisone (20 mg/day during first month, followed by tapering of 2.5 mg every 2 weeks, reaching 10 mg/day at 3 months), calcineurin inhibitor (tacrolimus or ciclosporin with target trough level of 7–9 ng/mL and 150 ng/mL from 3 to 6 months after KT respectively, and 5–7 ng/mL and 100–150 ng/mL thereafter), and mycophenolate mofetil (2–3 g/day).

In case of T-cell mediated rejection, patients received 500 mg of methylprednisolone during 3 days, followed by prednisone 20 mg/day. Antibody mediated rejections were treated with methylprednisolone (500 mg during 3 days), PE, and IVIg according to the modified Marrakech-protocol (14).

Clinical and biological data were collected retrospectively and anonymously from computerized medical records.

Luminex assay was used both for screening and single antigen flow beads (SAFB) identification of anti-HLA abs directed against HLA Class I and Class II antigens (LSM12 and LSA kits, One Lambda, Canoga Park, CA) in Saint-Louis Hospital Immunology Laboratory (Paris, France). Pre-transplant follow-up consisted of screening every 3 months when sera were available, and one Class I + Class II SAFB per year, with additional testing when screening positivity increased above doubling for the highest bead ratio in at least one Class. The day of transplant serum was tested in a SAFB assay. Post-transplant follow-up only relied on SAFB testing for all sera shipped to the laboratory. SAFB positivity threshold was set at a normalized MFI>500 according to the baseline formula calculated using Fusion software, after subtraction of the minimum MFI value for the corresponding locus to account for the non-specific binding observed, e.g., in the presence of IVIg. The DSA nature of the detected antibody was assigned at the antigenic level for antigens represented by a single bead or when all the beads for a given antigen were positive. For antigens with at least one negative bead, DSA was assigned when the bead bearing the donor allele was positive, the donor allele being either deduced from the emergency SSP typing (low resolution Olerup until end of 2016, Linkage Biosciences, thereafter) or retrospective high definition SSO typing (One Lambda), or when DNA was not available, deduced from common haplotypes using the HaploSTATS tool. Retained DSA MFI value was the average for the positive beads corresponding to the donor antigen. All sera were ethylenediaminetetraacetic acid (EDTA)-treated pre-testing since mid-September 2015, and for this study, retesting was performed with EDTA for anterior sera suspected of undergoing complement interference.

The kidney allograft biopsies were fixed in FAA (a solution of alcohol, formalin, and acetic acid), and subsequently embedded in paraffin. The biopsy sections (4 μm thick) were stained with periodic acid-Schiff, Masson’s trichrome, Jones methenamine silver and hematoxylin and eosin. The allograft paraffin-embedded kidney biopsies were scored and graded according to the international Banff 2017 classification for kidney allograft transplantation by trained transplant pathologists (DB, AM). C4d staining was not performed with the same technique in the two centers (immunochemistery or immunofluorescence), and was therefore not included in the statistical analysis.

Continuous variables were expressed as the mean ± SD. Categorical variables were reported as numbers and percentages. The intensive and standard groups were compared using the Mann-Whitney and Fisher exact tests for continuous and categorical variables, respectively. In survival analyses, Fine and Gray models were fitted, using death or loss of allograft function as a competitive event. To study the impact of day-0 DSA on ABMR occurrence, univariate models stratified on treatment regimen was used. Due to a log-linear type of association, sum of day-0 MFIs and MFI of day-0 iDSA were log-transformed. To study the impact of treatment regimen on the occurrence of ABMR, a multivariate model was fitted, using log(sum of day-0 MFIs) and the number of previous kidney transplantations (0 vs. 1 or more) as covariates. These covariates were chosen given their known prognostic value (based on the medical literature) on the risk of ABMR occurrence following transplantation. p < 0.05 was considered statistically significant and all tests were two-sided. Descriptive statistics were generated using Prism-6 (GraphPad). Survival analyses were performed using R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) and package cmprsk.

The study was conducted in accordance with the ethical guidelines of the Assistance Publique – Hôpitaux de Paris. No institutional review board approval was necessary at the time of the study as it was a retrospective study involving no intervention. The study was conducted according to the ethical standards of the 2000 Declaration of Helsinki as well as the Declaration of Istanbul 2008.

The two centers performed 1,457 kidney transplantations between 1st January 2012 and 1st September 2019. Of these, 60 hypersensitized patients were included in the study. Thirty-six patients received a standard induction and 24 patients an intensive induction including PE and rituximab. Fifteen patients in the standard group also received a limited number of PE (1.5 ± 2.1 sessions) during the post-transplant period. Table 1 shows patients’ initial characteristics. Fifty-two percent (n = 31) of the patients were women and 93.3% (n = 56) had prior history of sensitization. Sixty two percent of the patients (n = 37) underwent at least one previous transplantation (standard group: n = 18 [50%]; intensive group: n = 19 [79.2%], p = 0.1), and 71% of women (n = 22) had at least one pregnancy before KT. Both groups were similar regarding mean recipient age (52.2 ± 12.2 years), donor age (57.5 ± 15.7 years), cold ischemia time (mean: 17.4 ± 4.7 h), pre-transplant iDSA MFI level (intensive group: 8,435 ± 4,574; standard group: 8,935 ± 5,726; p = 0.93) and mean number of DSA (intensive group: 2.5 ± 1.2; standard group: 2.7 ± 2.0; p = 0.97). The iDSA was Class II in 36 patients (60%). The mean time on the waiting list was 1,660 ± 1,058 days (intensive group: 1,701 ± 1,130 days; standard group: 1,023 ± 1,632 days; p = 0.8), and the mean calculated panel reactive antibody (cPRA) was 80.3% ± 31.4% (intensive group: 79.7% ± 29.9%; standard group: 80.7% ± 32.8%; p = 0.9). Day 0 CDC IgM crossmatch was positive in eight patients (13.5%) (intensive group: n = 3 [12.5%]; standard group: n = 5 [13.9%]). Mean follow-up was 52.4 ± 28.8 months in the intensive group and 36.9 ± 28.4 months in the standard cohort (p = 0.03).

A total of 37 biopsy-proven rejection (BPR) episodes occurred in 24 patients during the follow-up (intensive group: n = 11 patients; standard group: n = 13 patients, p = ns), with 12 patients experiencing more than one BPR (intensive group: n = 5; standard group: n = 7). Twenty-four BPR (64%) occurred during the first year post-transplant (Table 2). The proportion of patients with BPR during the first year was not significantly different between the two groups (n = 8 [33.3%] in the intensive group and 12 [33.3%] in the standard group; p = 1).

The most frequent type of rejection was acute AMR, representing 71% of BPR in the first year. 78.5% of these BPR were AMR in the standard group and 60% in the intensive group (p = 0.39). AMR was associated with MFI of iDSA (HR = 2.07 [IQR: 1.04–4.1], p = 0.037), and with the sum of MFI at day 0 (HR = 1.92 [IQR: 1.12–3.29], p = 0.017), but not with the number of DSA at day 0 (HR = 1.2 [IQR: 0.964–1.50], p = 0.1). AMR-related microvascular inflammation tended to be more severe in the standard cohort, with a mean glomerulitis score of 0.8 ± 0.8 in the intensive group vs. 1.9 ± 1.0 in the standard group (p = 0.05), without any difference in peritubular capillaritis (1.07 ± 1.15 in the standard group and 1.14 ± 0.9 in the intensive group (p = 0.89)). Of note, three patients in the intensive group had a T-cell mediated rejection diagnosis during the first year. Histological BPR features are detailed in Table 2.

BPR was diagnosed earlier after transplantation in the standard group, as shown in Figure 1A (79 ± 158 days in the standard group vs. 211 ± 188 days in the intensive group (p = 0.005)). Mean time to AMR diagnosis was also shorter in the standard (75 ± 115 days) compared to the intensive group (220 ± 209 days, p = 0.03). However, global survival without rejection was not significantly different in the multivariate analysis (HR of BPR in the intensive group = 0.794 [0.34–1.9], p = 0.6).

Forty patients had a systematic kidney biopsy performed at 3 months post-transplantation (intensive group: n = 17 [70.8%]; standard group: n = 23 [63.9%]). Amongst these, 3 biopsies (7.5%) were also performed for cause (intensive group: n = 2, cause: control of anterior BPR; standard group: n = 1, cause: acute kidney injury). Glomerulitis score was significantly higher in the standard group (0.4 ± 0.8, vs. 0.06 ± 0.2) compared to the intensively treated patients (p = 0.04).

Both sets of patients experienced a significant decrease of iDSA MFI at 3 and 12 months when compared to day 0. In the standard group, mean MFI of iDSA dropped from 8,935 ± 5,726 at day 0, to 5,605 ± 5,620 at 3 months (−37%, p = 0.002) and remained stable thereafter at 5,348 ± 6,630 at 12 months (decrease of −40% when compared to day 0, p = 0.0003). In the intensive group, mean MFI of iDSA dropped from 8,435 ± 4,574 at day 0, to 4,878 ± 5,805 at 3 months (−42%, p = 0.002) and remained stable thereafter at 4,709 ± 5,645 at 12 months (−44% when compared to day 0, p = 0.004). There was no significant difference in iDSA MFI reduction and mean MFI at 3 and 12 months after transplantation between both groups (Table 2). Mean MFI of DSA did not significantly differ at 3 and 12 months. The number of patients with de novo DSA was similar in the two groups at 3 and 12 months after transplantation (Table 2).

Kidney function was significantly better in the intensive group at 1-month post-KT (estimated glomerular filtration rate (eGFR) in the intensive group: 50.9 ± 27.0 mL min⁻¹.1.73 m ² versus 31.0 ± 16.4 0 mL min⁻¹.1.73 m² in the standard group; p = 0.003), but this difference was not observed afterwards. Estimated glomerular filtration rate (eGFR) was 48.8 ± 19.1 mL min⁻¹.1.73 m² (CKD-EPI) and 47.6 ± 18.3 mL min⁻¹.1.73 m² (CKD-EPI) at 12 and 36 months after transplantation in the intensive group, and 37.5 ± 16.3 mL min⁻¹.1.73 m² (CKD-EPI) and 37.5 ± 9 mL min⁻¹.1.73 m² (CKD-EPI) in the standard group (p = 0.65 at month 12 and 0.09 at month 36) (Table 2). Proteinuria was also not significantly different at 3 or 12 months.

Mean eGFR was not different in the subgroup of patients who experienced BPR during first year post-transplantation (41.4 vs. 42.1 mL/min/1.73 m²; p = 0.91).

Global patient survival at the end of follow-up (1,316 ± 895 days) was 75% in the intensive group and 77% in the standard cohort, with six and eight deceased patients, respectively (p = 1). Death-censored graft survival at 12 months was better in the intensive group (100% in the intensive group vs. 80.6% in the standard group, p = 0.035), but this difference did not persist to the end of follow-up (83.3% in the intensive group and 63.9% in the standard group, p = 0.15) as shown in Figure 1B. Excluding death of the recipient, six patients, all from the standard group, lost their graft during the first year of follow-up. Mean time to graft loss was 80 ± 121 days. Causes of graft loss were acute AMR for one patient, reduction of immunosuppressive regimen in the context of severe infection for two patients, and acute peri-operative graft ischemia without evidence of macroscopic or histologic arterial or venous thrombosis in three patients.

A total of 14 deaths happened during follow-up (intensive group: n = 6, standard group: n = 8, p = 1). Infection was the most frequent cause of death, representing 57% of the total of deaths during follow-up (intensive group: n = 2, standard group: n = 6), the most frequent lethal pathogen being SARS-CoV-2 (n = 4, one in the intensive group and three in the standard cohort). Other deaths were due to cardiovascular events (n = 3, intensive group) and cancer (n = 1, standard group). The cause of death was not specified in two patients.

During first year of follow-up, 39 patients (65%) were diagnosed with at least one infection (intensive group: n = 16 [66.7%], standard group: n = 23 [63.9%], p = 1). For six of these patients (10% of total), hospitalization in an intensive care unit (intensive group: n = 1 [4.2%]; standard group: 5 [13.9%], p = 0.3) was required. The most frequent type of infection during the first year was pyelonephritis, representing 40.4% of infections. Mean time to first infection was 244 ± 351 days in the standard group and 215 ± 298 days in the intensive group (p = 0.76). There was no difference in the frequency of viremia at 3 and 12 months for cytomegalovirus, Epstein-Barr virus and BK virus.