Eligible patients were all adults who underwent ABO incompatible living donor kidney transplantation with a baseline anti-A and/or anti-B IgG isoagglutinin titer ≤1:8. Transplants were performed between October 2012 and December 2022. Follow-up was completed until October 31, 2023.

Our objectives were to compare: i) graft and patient survival -defined as survival free from graft loss or death-according to the use or absence of pre-transplant apheresis in patients with low baseline IgG isoagglutinin titers (≤1:8); ii) the incidence of immunological complications - ABMR, T cell-mediated rejection [TCMR], borderline rejection- and iii) the occurrence of infectious complications between groups. We also aimed to identify risk factors associated with graft and patient survival.

Clinical data were retrospectively extracted from medical records—both paper and electronic—using local hospital systems: Sined and DxCare (Strasbourg), Orbis (Necker), Sined (Rouen), Millenium and DIVAT database (Nantes), DxCare and R@N (Bordeaux). Isoagglutinin titers were provided by the French Blood Establishment (EFS). We recorded rituximab administration (date and dose), induction therapy (anti-thymocyte globulin (ATG) or basiliximab), and maintenance immunosuppression at hospital discharge (tacrolimus, cyclosporine, mycophenolate mofetil [MMF], corticosteroids). The number and type of apheresis sessions (PE, IA, or double filtration plasmapheresis [DFPP]) performed pre- and post-transplant were documented. The use of intravenous immunoglobulin (IVIg) in the post-transplant period was also recorded. Presence of preformed or de novo donor specific antibodies (DSA) was recorded considering a MFI threshold of 1000 (Luminex assay, Lab. One lambda, in all centers).

Anti-A and/or anti-B IgG and IgM titers were measured at baseline (prior to desensitization), on the day before or the day of transplantation (pre-transplant titer), and post-transplant at week 1, month 1, month 3, and month 12, as well as at the time of any suspected rejection. For final analysis, only IgG titers were considered. Techniques used for isoagglutinin measurement varied by center: tube hemagglutination at Necker, gel hemagglutination at Rouen, Strasbourg, Bordeaux and Nantes.

Rituximab (a single dose of 375 mg/m²) was administered before transplantation (day −30) in all but four patients, according to local protocols. Apheresis sessions were performed according to local procedures. It was systematically applied in Strasbourg (9 patients, treated with 3 sessions) and in the majority of patients in Necker (24/28 patients treated with 1–13 sessions; mean 3.3 sessions). In contrast, apheresis was generally not used in Rouen (6/9 patients without apheresis), Bordeaux (9/11 patients without apheresis), or Nantes (15/18 patients without apheresis).

The apheresis techniques used were as follows: Glycorex immunoadsorption (n = 5) and plasma exchange (n = 4) in Strasbourg; plasma exchange in Necker (n = 23), except for one patient with preformed DSA treated with immunoadsorption (IA); plasma exchange in Bordeaux (n = 2) and Nantes (n = 3); and immunoadsorption (IA, Globaffin®) in Rouen (n = 3). Additional details regarding local procedures are provided in Supplementary Table S1.

Delayed graft function (DGF) was defined as the need for at least one dialysis session during the first post-transplant week. Follow-up variables included serum creatinine and glomerular filtration rate according to CKD EPI formula at 3, 12, and 36 months, graft function status at end of follow-up, date of return to dialysis (if applicable), patient death (date and cause: cardiovascular, oncologic, or infectious), and biopsy-proven rejection episodes (classified according to Banff criteria). The presence of de novo DSA, including MFI and class, and isoagglutinin titer at the time of rejection, were also recorded. ABMR was defined as follows: immediate cortical necrosis consistent with hyperacute rejection occurring within hours after transplantation, early thrombotic microangiopathy, or histological findings consistent with ABMR according to the Banff classification on graft biopsy performed later post-transplant. Early ABMR was defined as ABMR rejection occurring during the first post-transplant month.

For infectious complications, the type and date of each event were collected. In cases of recurrent infections of the same type, only the first occurrence was dated. Events occurring after graft loss were not included in the analysis.

All statistical analyses were performed using R software (version 4.3.2) and appropriate packages. Quantitative variables with a symmetrical distribution were summarized as mean ± standard deviation (SD), while non-normally distributed quantitative variables and ordinal data were presented as median [first quartile; third quartile]. Categorical variables were expressed as counts (percentage). A p-value <0.05 was considered statistically significant. Group comparisons were conducted using Fisher’s exact test for categorical variables and the Kruskal–Wallis test for continuous variables. Survival analyses were performed using a composite endpoint defined as “graft loss or patient death.” Event-free survival was estimated using Kaplan–Meier curves, and differences between groups were assessed with the log-rank test. For the analysis of rejection events, cumulative incidence functions were estimated using the Aalen-Johansen estimator, accounting for graft loss or death as competing events. Between-group comparisons were performed using Gray’s test. Risk factors associated with ABMR, biopsy-proved rejection episodes (BPAR: ABMR + TCMR), and a composite endpoint of graft loss or death were investigated using Cox proportional hazards models. Cox regression results are reported as hazard ratios (HR) with 95% confidence intervals (95% CI). Comparisons of HRs to the null value (HR = 1) were assessed using Wald tests. Explanatory variables were first assessed in univariate analyses including all patients with available data. A multivariable Cox model was then constructed, incorporating variables that were statistically associated with the outcome in univariate analysis and those considered clinically relevant (p ≤ 0.2). In addition, we performed a LASSO-Cox regression to assess the association of the covariables with death and graft failure. The regularization coefficient was selected using 5-fold cross-validation as the one that optimized the C-statistics [REF glmnet]. Stability of variable selection was assessed by bootstrapping the whole procedure (1000 samples) then computing the proportion of samples for which each variable was selected by the LASSO-Cox regression.

A total of 78 patients with baseline anti-A and/or anti-B IgG isoagglutinin titers ≤1:8 were included (see characteristics in Table 1). The cohort was predominantly male (69%). Preemptive transplantation was performed in 45% of cases. Prior to transplantation, 58% of patients (n = 45) were sensitized and 6 (7.7%) had also preformed DSA with a median sum of MFI of 1400 (range 1000–6500). Blood group O was the most common among recipients (46%). Donors were predominantly female (71%), with blood group A being the most frequent (63%), followed by group B (26%) and AB (12%). The most common donor-recipient combination was from a group A donor to a group O recipient, accounting for 36% of cases. Most patients (94%) received anti-CD20 monoclonal antibody therapy prior to transplantation.

Among the 41 patients (53%) who did undergo apheresis, PE alone was most commonly used (n = 31), followed by IA (n = 6), and a combination of PE and IA (n = 3). One patient received a combination of PE and DFPP. On average, patients who underwent apheresis received 3 sessions (range: 2–5). In contrast, pre-transplant apheresis was not performed in 37 patients (47%).

The main baseline differences between patients who did and did not undergo pre-transplant apheresis are summarized in Table 1. Of note, one of the 37 patients with a baseline IgG ≤1:8 who did not undergo apheresis showed an increase in IgG titer to 1:16 on the day of transplantation. Briefly, patients in the no-apheresis group were less often treated with IVIG and MMF (p < 0.01 and 0.046 respectively).

In the whole cohort, patient survival after transplantation was 99.5% at 1 year, 97% at 3 years, and 95.5% at 5 years and graft survival was 94.3% at 1 year, 90% at 3 years, and 86% at 5 years.

A total of eight patients experienced delayed graft function and one patient displayed primary non-function (Table 2). Only 4 patients showed a IHG rebound (defined as a post-transplant IgG titer ≥1:16) in the first post-transplant month. During follow-up, we recorded 22 rejection episodes which occurred in 21 patients, including 14 ABMR, 3 TCMR, and 5 borderline rejections. Among them, 7 patients developed early acute ABMR, most cases occurring immediately after transplantation. Their characteristics are described in Supplementary Table S2. During follow up, 11 patients developed de novo DSA, after a median delay of 15.3 months (range 0.5–83 months). Cumulative incidence of ABMR was 12% at 1 year, and 19% at 3 and 5 years (Figure 1A). Ten patients lost their graft (see Supplementary Table S3) and 6 died. Median follow-up time after transplantation was 4.29 years [2.22, 6.72].

Comparison of outcomes according to pre-transplant apheresis status is shown in Table 2. In the group without apheresis, one patient experienced graft loss due to cortical necrosis at the time of transplantation, and four patients developed hyperacute rejection with thrombotic microangiopathy. Consistent with these findings, there was a trend toward a higher incidence of DGF in patients who did not receive apheresis compared to those who did (19% vs. 5%, p = 0.077). In 5 out of 8 patients with DGF (four in the no-apheresis group and one in the apheresis group), DGF was attributed to thrombotic microangiopathy and/or hyperacute rejection. The evolution of isoagglutinin titers during the first post-transplantation year according to the groups is depicted in Supplementary Figure S1 and showed a rebound in 3 patients of the apheresis group and 1 in the no-apheresis group. Post transplant apheresis were performed in 5 patients of each group, according to local center procedures, due to isoagglutinin rebound (n = 3) and/or rejection episodes (n = 7).

Rejections episodes were more frequent in the no-apheresis group (40.5% vs. 14.6%, p = 0.01). Early ABMR episodes were also more frequent in the no-apheresis group (14% vs. 4.9%), although the difference did not reach statistical significance (p = 0.25). Of note, one of the two early ABMR episodes in the apheresis group occurred in a patient with high preformed donor-specific antibodies (DSA) with a cumulated MFI >6000. Kaplan–Meier analysis similarly demonstrated a trend toward a higher incidence of ABMR episodes in patients who did not receive apheresis (p = 0.073, Figure 1B). Evolution of glomerular filtration rate in both group during the study is depicted in Figure 2 and was not different between groups.

A significant difference in event-free survival—defined as survival free from graft loss or all-cause mortality—was observed comparing patients who underwent pre-transplant apheresis and those who did not (p = 0.021), with inferior outcomes in the no-apheresis group (Figure 3). At 3 years, event-free survival (death or graft loss) was 90% in the apheresis group and 79% in the no-apheresis group. At 5 years, it was 86% in the apheresis group and 68% in the no-apheresis group. When event-free survival was analyzed according to baseline IHG titer strata, no significant differences were observed between groups (Supplementary Figure S2). Subsequent analysis based on the induction regimen revealed that patients receiving thymoglobulin (n = 44) demonstrated superior event-free survival compared to those treated with Basiliximab (n = 34) (Figure 4A). When stratifying both apheresis status and induction therapy, the subgroup receiving pre-transplant apheresis in combination with thymoglobulin induction exhibited the most favorable outcome (p = 0.026) (Figure 4B).

When death and graft loss were analyzed separately, a higher number of events was observed in the group that did not undergo pre-transplant apheresis, but these differences did not reach statistical significance (p = 0.07 for mortality and p = 0.19 for graft loss, Supplementary Figure S3). Precise causes of graft loss are provided in Supplementary Table S3. When comparing outcomes based on induction therapy, graft loss was significantly less frequent among patients who received thymoglobulin compared to those who received basiliximab (p = 0.012, Supplementary Figure S4), while no significant difference was observed in mortality between the two groups (p = 0.61, data not shown).

Factors significantly associated with improved patient and graft survival in univariate analysis were the use of pre-transplant apheresis (HR = 0.28, 95% CI 0.091–0.89, p = 0.03) and thymoglobulin induction therapy (HR = 0.28, 95% CI 0.096–0.81, p = 0.019). Conversely, delayed graft function and early ABMR were linked to an increased risk of graft loss and patient death (HR = 3.3, 95% CI 1.1–10, p = 0.041 and HR = 6.9, 95% CI 2.1–22, p = 0.0012, respectively). The presence of preformed donor-specific antibodies was not statistically associated with outcomes in this cohort. These findings are summarized in Table 3. In multivariable Cox regression analysis, pre-transplant apheresis was independently associated with improved patient and graft survival (HR = 0.31, 95% CI 0.10–0.99, p = 0.049). Conversely, the occurrence of early ABMR (within the first month post-transplant) was significantly associated with an increased risk of patient death and graft loss (HR = 5.18 95% CI 1.58–17.01, p = 0.0007). These results are summarized in Table 4. The variables selected by the LASSO model corroborate those identified in the multivariate Cox model (Supplementary Figure S5).

Regarding rejection episodes, we observed a trend toward an association between pre-transplant apheresis and ABMR or BPAR (p = 0.08 and 0.09, respectively), in both univariate and multivariate analyses (Supplementary Tables S4, S5).

No differences were observed between groups with or without pre-transplant apheresis considering occurrence of all infections. The probability of viral, bacterial and fungal infections at 1- and 3-year post-transplantation according to apheresis group are described in Table 5. There were more fungal or parasitic infections in the no apheresis group (p = 0.047) −1 microsporidiosis vs. 2 pneumocystis pneumonia, 2 candidiases and 3 cryptosporidioses-, and a trend toward a higher incidence of CMV infection among patients who underwent apheresis (p = 0.056; Supplementary Figure S6). Presumed BKV nephropathy incidence was not different between the two groups.