Between January 2010 and December 2018, 1,284 KT procedures were performed in Seoul St. Mary’s Hospital, Seoul, Korea. Patients with a positive XM test, ABO-incompatible transplantation, concurrent kidney and other solid organ or hematopoietic stem cell transplantation, and those who were pediatric (<18 years) were excluded. In total, 1,027 KT recipients (629 LDKT and 398 DDKT recipients) were included. Patients whose HLA-DSA results were positive in the LSA assay but negative in the XM test were defined as the low-DSA group. Patients without HLA-DSA in both the LSA assay and the XM test were defined as the no-DSA group. Negative XM test results were confirmed using both CDC-XM and FCXM. Among the 1,027 patients, 89 had low-DSA, including 68 LDKT recipients and 21 DDKT recipients. Finally, there were 68 low-DSA and 561 no-DSA patients in the LDKT group and 21 low-DSA and 377 no-DSA patients in the DDKT group (Fig. 1). The median follow-up period was 53.5 months (interquartile range: 30.1–82.1 months). This study was performed in accordance with the Declarations of Helsinki (2013) and Istanbul (2008) and was approved by the Institutional Review Board of Seoul St. Mary’s Hospital (KC22RISI0380). The requirement for informed consent was waived because of the retrospective study design and the use of noninvasive procedures.

Our center’s immunological workup protocol has been reported previously [18]. All LDKT candidates underwent panel-reactive antibody (PRA) screening and XM tests as baseline tests. XM results were reported using T-/B-CDC-XM and T-/B-FCXM tests. In patients with positive PRA screening or XM test results, we investigated the presence of HLA-DSA using the LSA assay. The criterion for PRA positivity was initially 20% but was changed to 0% during the study period. Eighty-three patients with low-DSA (93.3%) and 277 patients without low-DSA (29.5%) had PRA values >20%. Patients on the DDKT waitlist were screened for PRA as a pretransplant workup. If the result was positive, the patient was tested for HLA-DSA. When a kidney became available for transplantation, an XM test was performed. After June 2010, T-FCXM and B-FCXM were performed in addition to T-/B-CDC-XM. FCXM negativity was confirmed by both T-FCXM and B-FCXM in 380 of 398 DDKT recipients. PRA screening was conducted using the Luminex assay (LABScreen mixed class I and II; One Lambda, Canoga Park, CA, USA). The results are presented as %PRA. The CDC-XM and FCXM tests were conducted using standard procedures [19, 20]. HLA-DSA was identified by the LSA assay using LABScreen single antigen HLA class I-combi and class II-group 1 kits (One Lambda). The criterion for positivity was an MFI value >1,000. In all recipients and donors, HLA typing was performed using LIFECODES HLA-A, B, C, DRB1, and DQB1 Sequence-Specific Oligonucleotide Typing Kits (Immucor Transplant Diagnostics, Stamford, CT, USA). If the LSA assay-detected anti-HLA antibody in the recipient corresponded to the HLA type of the donor, it was classified as HLA-DSA. HLA-DSA was classified into three groups based on the peak MFI value: strong, >10,000; moderate, 5,000–10,000; and weak, 1,000–5,000.

Our center’s desensitization protocol has been reported previously [13, 18, 21]. In brief, we used a stratified protocol based on baseline MFI values. The target of desensitization was negative conversion in the XM test for XM-positive patients and reduction of the HLA-DSA MFI value to<5,000. In patients with strong HLA-DSA, rituximab (RTX) was administered at a dose of 375 mg/m² (MabThera; Genentech, San Francisco, CA, USA) 2–3 weeks before transplantation. Plasmapheresis/immunoglobulin (PP/IVIG) therapy was initiated 13 days before transplantation and administered every other day. Immunosuppressant treatment was initiated seven days before transplantation in these patients. In patients with moderate HLA-DSA, RTX was administered, as mentioned above. PP/IVIG therapy was administered every other day for seven days before transplantation. Immunosuppressants were initiated two days before transplantation. In patients with weak HLA-DSA, only RTX at a dose of 375 mg/m² was administered 7–10 days before transplantation when the PRA values were >50%.

The primary outcome was the incidence of biopsy-proven allograft rejection (BPAR). Patients with clinically diagnosed rejections without biopsies were excluded. Secondary outcomes were changes in allograft function measured as estimated glomerular filtration rate (eGFR), death-censored allograft survival rate, patient survival rate, and posttransplant infections. Allograft rejection was diagnosed using the latest version of the Banff classification at the time of biopsy, which is updated every 2 years. BPAR was classified as T cell-mediated rejection (TCMR) and ABMR according to histological scores [22]. Mixed rejection was not classified separately. Both ABMR and TCMR were added to the rejection events. Borderline change was not considered to be BPAR. Serum creatinine levels were measured at 3, 6, 12, 24, and 36 months after transplantation. The eGFR for each concordant time was assessed using the Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) equation [23]. Graft failure was defined as a return to dialysis dependence or retransplantation. In the analysis of death-censored graft failure, patients who died with a functioning transplant were censored at the time of death. An infection episode was defined as the occurrence of an infectious event requiring hospitalization. Cytomegalovirus (CMV) infection was defined as CMV DNAemia and/or any organ involvement by CMV [24]. BK virus (BKV) infection was defined as BKV DNAemia with DNA titer ≥10⁴ copies/mL and/or biopsy-proven BK polyomavirus-associated nephropathy (BKPyAN) [25]. Infection-free transplant survival was defined as the time from transplantation to the first infection episode.

Continuous variables are presented as means±SDs or medians with interquartile ranges. Categorical variables are presented as frequencies and percentages. Continuous variables were analyzed using the Student’s t-test or Mann–Whitney U test. The chi-square test or Fisher’s exact test was used to compare categorical variables. Point biserial correlation was used to confirm the correlation between XM positivity and HLA-DSA MFI values. Survival curves were generated using the Kaplan–Meier method. Groups were compared using the log-rank test. Predictors of ABMR were explored using multivariate logistic regression analysis. Clinical parameters showing significant differences (P<0.05) in univariate analysis or known to cause ABMR were fitted into the multivariate model. We selected patient sex, low -DSA, donor type, retransplantation, anti-thymocyte globulin (ATG) induction, and desensitization as predictors. PRA values (%) or high PRA values (>50%) were excluded from the model because of high variance inflation factors, suggesting multicollinearity with low-DSA. Multivariate Cox regression analyses with backward selection were used to investigate independent predictors of infectious episodes. Cox models were built considering patient age, sex, cold ischemic time, low-DSA, desensitization, and ATG induction. A linear mixed model was used to compare the changes in allograft function over time. All missing data were censored from the last follow-up date. For all tests, a two-tailed P<0.05 indicated statistical significance. All data were analyzed using SPSS software version 24 (IBM Corporation, Armonk, NY, USA).

Table 1 shows the clinical and immunological characteristics of patients with low-DSA in the LDKT and DDKT subgroups. In both donor-type subgroups, the low-DSA group comprised more females, retransplanted patients, and individuals with higher PRA values. There were more patients with class I HLA-DSA than with class II HLA-DSA. A small percentage of patients had both class I and II HLA-DSA. HLA-DSA strength was presented as the peak MFI value. HLA-DSA class and strength did not differ between the LDKT and DDKT subgroups (Supplemental Data Table S1). Patients with low-DSA received ATG more frequently as induction therapy and more frequently underwent desensitization therapy than patients without low-DSA. All DDKT recipients underwent dialysis prior to transplantation. Recipient or donor age did not differ according to low-DSA in LDKT or DDKT recipients. Additionally, among patients with low-DSA, desensitization was performed more frequently in LDKT recipients than in DDKT recipients (76.4% vs. 66.7%, P=0.003). Most DDKT recipients (14/21, 66.7%) received RTX alone as desensitization therapy. However, in the LDKT subgroup, 26 (38.2%) of 68 recipients received RTX and 26 (38.2%) of 68 received additional PP/IVIG (Supplemental Data Table S2).

Fifty-three patients had a positive XM result, and 40 of these 53 patients had HLA-DSA. However, HLA-DSA was not identified by the LSA assay in 13 patients. We analyzed the correlation between XM positivity and HLA-DSA MFI in 129 patients who were identified as having HLA-DSA (Fig. 2). The peak MFI value of HLA-DSA was significantly higher in the CDC-XM-positive group (7,255 [6,070–12,737], N=22) than in the CDC-XM-negative group (4,009 [2,127–7,270], N=107, P<0.001). The peak HLA-DSA MFI was significantly higher in FCXM-positive patients (7,218 [3,495–12,087], N=39) than in FCXM-negative patients (3,997 [2,089–6,796], N=90, P<0.001).

The immunological characteristics of HLA-DSA were compared between LDKT and DDKT recipients according to the strength, class, and specificities of HLA-DSA. Fig. 3 presents the distribution of immunodominant HLA-DSA in the patients. MFI values or HLA-DSA specificities were missing for four patients. Overall, 46 (54.1%) patients had class I HLA-DSA, and 39 (45.9%) patients had class II HLA-DSA as the immunodominant antibody. MFI values were higher in class II HLA-DSA than in class I HLA-DSA (5,333 [2,243–7,898] vs. 3,407 [2,092], P=0.063). In both donor-type groups, HLA-DSA was more frequently of class I than of class II. However, the proportion of each HLA-DSA specificity differed according to donor type. Anti-HLA-B antibody (26/66, 39.4%) in class I HLA-DSA and anti-HLA-DR antibody (18/66, 27.3%) in class II HLA-DSA were common in LDKT recipients. DDKT recipients had anti-HLA-A antibody (6/19, 31.6%) in class I HLA-DSA and anti-HLA-DQ antibody (7/19, 36.8%) in class II HLA-DSA as the most frequent HLA-DSA (P=0.030).

The strength of HLA-DSA differed according to its class and specificity. Strong HLA-DSA were mostly class II, especially anti-HLA-DQ antibody. All strong HLA-DSA in DDKT recipients was anti-HLA-DQ antibody. Class II HLA-DSA accounted for half of moderate HLA-DSA in LDKT. In DDKT, it accounted for 66.7%, all of which were anti-HLA-DQ antibody. In weak HLA-DSA, class I antibodies had a higher proportion than class II antibodies. Anti-HLA-B antibody was the most common in LDKT, and anti-HLA-A antibody was the most common in DDKT. Anti-HLA-DQ antibody accounted for a small proportion of weak HLA-DSA.

As shown in Fig. 4, the development of overall BPAR, which was divided into TCMR and ABMR, was identified in the first year of transplantation. In the total cohort, the overall BPAR and TCMR rates did not significantly differ between the low-DSA and no-DSA groups. However, the incidence of ABMR was significantly higher in the low-DSA group than in the no-DSA group (13.5%, 12/89 vs. 2.5%, 23/938, P<0.001). In subgroup analysis, the overall BPAR rate did not significantly differ between low-DSA and no-DSA groups in LDKT or DDKT. The TCMR rate did not significantly differ according to baseline low-DSA in LDKT or DDKT. ABMR was more frequent in the low-DSA group than in the no-DSA group in both LDKT (11.8%, 8/68 vs. 2.5%, 14/561, P=0.001) and DDKT (19%, 4/21 vs. 2.4%, 9/377, P=0.003) recipients. The results of univariate and multivariate analyses correlating patient, donor, and immunological factors that can be associated with the occurrence of ABMR are shown in Table 2. Pretransplant low-DSA was the only independent predictor of ABMR development in the total cohort (OR: 5.820, 95% CI: 2.100–16.140) and in DDKT recipients (OR: 9.600, 95% CI: 1.790–51.560), whereas it had only a marginal impact in LDKT recipients (OR: 3.760, 95% CI: 0.990–14.260).

Allograft function did not differ between the low-DSA and no-DSA groups in the total cohort and LDKT and DDKT recipients until 36 months after KT (Supplemental Data Fig. S1). There was no significant difference in the death-censored allograft survival rate in the total cohort and LDKT and DDKT recipients according to the presence of low-DSA (Supplemental Data Fig. S2). Among patients who experienced ABMR in the first year after transplantation, one of 12 low-DSA and eight of 23 no-DSA patients eventually lost graft function. Treatment of ABMR, including steroid pulse, PP/IVIG, RTX, and bortezomib, was performed according to clinical decision. Rejection was the leading cause of allograft failure, except for one patient who lost his graft due to BKPyAN (Supplemental Data Table S3). In total, 48 patients (48/1,027, 4.7%) died, including one patient with low-DSA who died by suicide. There was no significant difference in the patient survival rate according to the presence of low-DSA in the total cohort and in LDKT recipients (Supplemental Data Fig. S3). As no patients died in the DDKT group, statistical analysis could not be performed.

In total, 567 (55.2%) cases of infectious complications occurred during the follow-up period. The incidence of BKV infection was higher in the low-DSA group than in the no-DSA group of LDKT recipients (23.6% vs. 13.4%). The incidence of infections caused by pathogens other than BKV did not significantly differ between the low-DSA and no-DSA groups in the total cohort and LDKT and DDKT recipients (Supplemental Data Table S4). In the LDKT group, the infection-free survival rate significantly differed between the low-DSA and no-DSA groups (P=0.041). However, such a difference was not observed in the total cohort and DDKT recipients (Supplemental Data Fig. S4). The results of univariate and multivariate Cox regression analyses predicting posttransplant infections are presented in Supplemental Data Table S5. Neither low-DSA nor desensitization was identified as a predictor of posttransplant infection.