Donor selection for patients with HLA-homozygous haplotypes in allogeneic hematopoietic stem cell transplantation.

Introduction

Allogeneic hematopoietic cell transplantation (HCT) is a potentially curative treatment for hematological malignancies, including leukemia and lymphoma, as well as bone marrow failure syndromes. The use of human leukocyte antigen (HLA)-matched donors is the standard approach in HCT because HLA compatibility is a major determinant of graft-versus-host disease (GVHD), graft failure, and survival [ 1 – 4 ]. HLA mismatches in the graft-versus-host (GVH) direction primarily increase GVHD risk, whereas mismatches in the host-versus-graft (HVG) direction may increase the risk of graft failure [ 5 – 7 ]. When HLA-matched donors are unavailable, alternative donors, including haploidentical related donors, HLA-mismatched unrelated donors, or cord blood (CB), are commonly used [ 8 – 10 ].

HLA-homozygous haplotypes occur with varying frequencies among ethnic groups, with an estimated global prevalence of 1–4% [ 11 – 14 ]. In recipients with HLA-homozygous haplotypes, a haploidentical donor who carries the recipient haplotype can result in mismatches limited to the HVG direction while maintaining complete matching in the GVH direction, potentially reducing GVHD risk. Previous studies have reported comparable outcomes between HLA-matched and haploidentical related donor transplantation in this population [ 15 ].

However, donor selection remains unclear when no suitable related donor is available. Therefore, we analyzed the Japanese national transplantation registry to compare transplantation outcomes across donor sources, including unrelated donors and CB, and to inform donor selection for patients with HLA-homozygous haplotypes.

Methods

Study design and patients

Clinical data on HCT recipients were collected by the Japanese Society for Transplantation and Cellular Therapy (JSTCT) and the Japanese Data Center for Hematopoietic Cell Transplantation (JDCHCT) using the Transplant Registry Unified Management Program [ 16 , 17 ]. This study was conducted in accordance with the Declaration of Helsinki and approved by the institutional review boards of Kyoto University Hospital and the Data Management Committees of JSTCT and JDCHCT. All patients provided written informed consent to participate in this study.

We included patients aged ≥ 16 years with hematologic malignancies who were homozygous at the allele level at HLA-A, -B, -C, and -DRB1 and underwent their first bone marrow (BM), peripheral blood stem cell (PBSC), or cord blood (CB) transplantation between 2000 and 2022. Donor sources were related donors, unrelated donors, and CB. Donors were classified as homo-to-homo (matched at HLA-A, -B, -C, and -DRB1) or hetero-to-homo (allele-level mismatches at HLA-A, -B, -C, and/or -DRB1 restricted to the host-versus-graft direction). We excluded patients with missing HLA-A, -B, -C, or -DRB1 typing for either the recipient or donor.

Endpoints

The primary endpoint was overall survival (OS). Secondary endpoints included relapse-free survival (RFS), cumulative incidence of relapse, non-relapse mortality (NRM), neutrophil engraftment, platelet engraftment, grade II-IV acute GVHD, grade III-IV acute GVHD, and chronic GVHD.

Definitions

OS was defined as the time from transplantation to death from any cause, with survivors censored at the last follow-up. RFS was defined as the time from transplantation to relapse or death from any cause. Neutrophil engraftment was defined as the first of 3 consecutive days with an absolute neutrophil ≥ 0.5 × 10⁹/L post-transplantation. Platelet engraftment was defined as the first of 7 consecutive days with a platelet count ≥ 20 × 10⁹/L without platelet transfusion post-transplantation. Acute and chronic GVHD were diagnosed and graded according to established criteria [ 18 , 19 ]. The cumulative incidence of chronic GVHD was evaluated among patients who were alive and relapse-free at day 100 post-transplantation.

Disease risk was classified as standard or advanced according to disease type, disease status, and established prognostic criteria. Standard stage included acute leukemias in complete remission, MDS with IPSS Low or Intermediate-1 risk (or IPSS-R Very Low, Low, or Intermediate risk), CML in chronic or accelerated phase, and lymphoproliferative disorders in complete or partial remission. All other conditions were classified as advanced stage.

Conditioning regimens were classified as myeloablative (MAC) when they included total body irradiation (TBI) > 8 Gy, oral busulfan ≥ 9 mg/kg, intravenous busulfan ≥ 7.2 mg/kg, melphalan > 140 mg/m 2 , or thiotepa ≥ 10 mg/kg. All other regimens were classified as reduced-intensity conditioning (RIC) [ 20 , 21 ]. Post-transplantation cyclophosphamide (PTCy) was defined as cyclophosphamide 40–50 mg/kg administered on days 3, 4, or 5 post-transplantation [ 22 , 23 ].

HLA typing and haplotype inference

HLA-A, -B, -C, and -DRB1 typing was performed at allele-level resolution (2-field resolution) for all patients and donors. Recipients were classified as having HLA-homozygous haplotypes when they were homozygous at all four loci (HLA-A, -B, -C, and -DRB1) at the allele level. Haplotypes were inferred based on HLA allele combinations using established haplotype frequencies in the Japanese population [ 11 ].

Statistical analysis

Patient characteristics were summarized using descriptive statistics. Categorical variables were compared using the chi-square test or Fisher’s exact test, as appropriate, and continuous variables were compared using the Kruskal–Wallis test. OS and RFS were estimated using the Kaplan–Meier method. Follow-up was calculated from transplantation to the date of last contact or death. The median follow-up time was estimated using the reverse Kaplan–Meier method. The cumulative incidences of relapse, NRM, engraftment, and GVHD were estimated using cumulative incidence curves [ 24 ]. Competing events were defined as follows: relapse for NRM; death without relapse for relapse; death without engraftment for neutrophil and platelet engraftment; and death without acute GVHD for acute GVHD. For chronic GVHD (evaluated among patients alive and relapse-free at day 100), competing events were death and relapse. Group differences were assessed using the log-rank test (OS/RFS) or Gray’s test (competing-risk endpoints) [ 25 ]. Cox proportional hazards models were used for OS and RFS, and Fine–Gray models were used for competing-risk endpoints (relapse, NRM, engraftment, and GVHD) [ 26 ]. The unrelated donor (URD) homo-to-homo group served as the reference group. Candidate covariates for multivariate models included patient characteristics (age, sex, CMV serostatus, performance status, hematopoietic cell transplantation–comorbidity index, and disease risk) and transplant-related factors (year of transplantation, conditioning intensity, and GVHD prophylaxis). Hematopoietic cell transplantation–comorbidity index and performance status were retained in the final stepwise models, confirming that these covariates were adjusted for in the primary analysis.

As baseline comparisons for the subsequent combined subgroup analyses, we first evaluated OS stratified by donor source alone and by donor HLA matching status alone. We then assessed potential heterogeneity of the HLA matching effect across donor sources by including an interaction term between donor source and donor HLA homozygosity in the Cox model for OS. Because interaction tests can be underpowered, we prespecified a two-sided P value < 0.10 as suggestive of effect modification. When this criterion was met, we presented analyses stratified by donor source and by HLA matching status.

Covariates were selected using a forward stepwise selection procedure (entry and retention criterion of P < 0.05), with donor source (the main exposure) forced into all models. Statistical significance was defined as a two-sided P < 0.05. All statistical analyses were performed using Stata version 16 (StataCorp., College Station, TX, USA) and EZR software version 1.68 [ 27 ].

Results

Patient characteristics

From the registry database, we identified 38001 patients aged ≥ 16 years with hematologic malignancies who underwent their first allogeneic HCT between 2000 and 2022 and had available HLA-A, -B, -C, and -DRB1 typing information. Among these, 724 patients were identified as having HLA-homozygous haplotypes. We excluded 33 cord blood (CB) units that included GVH direction mismatches. After applying these inclusion and exclusion criteria, 691 patients were enrolled (Fig. 1 ). Detailed patient and transplant characteristics are summarized in Table 1 . Compared with the URD homo-to-homo group, the URD hetero-to-homo and CB homo-to-homo groups included older recipients and higher proportions of patients with adverse baseline features, including higher HCT-CI and, for the CB homo-to-homo group, poorer performance status. These imbalances should be considered when interpreting subgroup differences in survival.

Regarding stem cell sources, 121 patients received grafts from related donors (RD) homo-to-homo, 76 from RD hetero-to-homo donors, 374 from URD homo-to-homo, and 22 from URD hetero-to-homo donors. Additionally, 40 and 58 patients received CB units classified as homo-to-homo and hetero-to-homo without GVH-direction mismatches, respectively. The median age at transplantation was 51 years (range, 16–76), with 26.0% of patients being older than 60 years. Underlying diseases included acute myeloid leukemia/myelodysplastic syndrome ( n = 364; 52.7%), acute lymphoblastic leukemia ( n = 127; 18.4%), chronic myeloid leukemia/myeloproliferative neoplasia ( n = 39; 5.6%), and lymphoma/multiple myeloma ( n = 161; 23.3%). A majority of patients ( n = 416; 60.2%) had standard-stage disease, 225 (32.6%) had advanced-stage disease, and the disease stage was unknown in 50 patients (7.2%). Most transplants ( n = 518; 74.9%) were performed in or after 2011, while 173 (25.0%) were performed between 2000 and 2010. Donor-specific antibodies (DSA) were detected in only three patients (0.4%). Regarding GVHD prophylaxis, 12 patients with RD hetero-to-homo transplants received PTCy.

Distribution of HLA mismatches in the hetero-to-homo groups

To characterize HLA mismatch patterns in the hetero-to-homo groups, we summarized the distribution of mismatches by the number of mismatched loci and by mismatched locus, stratified by donor source (Supplementary Table S1 ). URD hetero-to-homo transplants were predominantly single-locus mismatches (17/22, 77.2%), whereas multi-locus mismatches were more common in RD hetero-to-homo (3 loci: 19/76, 25.0%; 4 loci: 47/76, 61.8%) and CB hetero-to-homo transplants (3 loci: 32/58, 55.2%). Across donor sources, mismatches frequently involved HLA-DRB1 (RD 90.7%; URD 72.7%; CB 79.3%).

Baseline associations of donor source and HLA matching with overall survival

The median follow-up among survivors was 5.11 years (interquartile range, 2.23–9.26 years). We first evaluated outcomes according to donor source and HLA matching status as baseline comparisons for the subsequent combined subgroup analyses. The unadjusted 5-year OS rates were 50.6% (95% confidence interval (CI), 42.6%–58.0%) in the RD group, 53.7% (95% CI, 48.3%–58.7%) in the URD group and 33.8% (95% CI, 23.5%–44.4%) in the CB group (Fig. 2 A).

In multivariate analysis, the CB group showed a trend toward higher mortality risk compared with the URD group (adjusted hazard ratio [HR], 1.30; 95% confidence interval CI, 0.95–1.77, P = 0.097) (Supplementary Table S2 ).

The unadjusted 5-year OS rates were 51.9% (95% CI, 47.3%–56.3%) in the homo-to-homo group and 44.1% (95% CI, 35.4%–52.5%) in the hetero-to-homo group (Fig. 2 B). Multivariate analysis showed no significant differences in mortality risk between the groups (Supplementary Table S3 ).

We next asked whether HVG-direction mismatches are associated with inferior survival across donor sources. We then examined whether the combination of donor source and HLA matching category identified clinically distinct risk groups and which transplantation outcomes contributed to the observed differences in survival. In the Cox model for OS, the interaction between donor source and HLA matching category met the prespecified criterion for possible effect modification (cord blood × hetero-to-homo: HR 0.50, P = 0.077). We therefore evaluated outcomes according to combined donor source and HLA matching subgroups. The results of the univariate analyses are presented in Supplementary Table S4 .

Outcomes across combined donor source and HLA matching subgroups

The unadjusted 5-year OS rates were 51.9% (95% CI, 41.7%–61.1%) in the RD homo-to-homo group, 48.7% (95% CI, 35.5%–60.6%) in the RD hetero-to-homo group, 54.3% (95% CI, 48.0%–59.0%) in the URD homo-to-homo group, 42.2% (95% CI, 19.3%–63.6%) in the URD hetero-to-homo group, 25.6% (95% CI, 11.0%–42.0%) in the CB homo-to-homo group, and 38.7% (95% CI, 25.0%–52.0%) in the CB hetero-to-homo group (Fig. 3 A). Although the unadjusted OS rates in the RD hetero-to-homo and CB hetero-to-homo groups were lower than those in the URD homo-to-homo group, multivariate analysis showed no significant differences between the URD homo-to-homo group and the RD hetero-to-homo or CB hetero-to-homo groups. The CB homo-to-homo group showed a higher mortality risk compared with the URD homo-to-homo group (adjusted HR, 1.71; 95% CI, 1.11–2.64; P = 0.015). The URD hetero-to-homo group showed a trend toward inferior OS compared with the URD homo-to-homo group, although this was not statistically significant (adjusted HR, 1.77; 95% CI, 0.97–3.22; P = 0.061) (Table 2 ). To address potential era effects, we performed a sensitivity analysis restricted to transplants performed from 2011 to 2022. In this cohort, the CB homo-to-homo subgroup was not statistically significantly associated with inferior OS compared with the URD homo-to-homo group (adjusted HR, 1.38; 95% CI, 0.86–2.23; P = 0.174) (Supplementary Table S5 ).

The unadjusted 5-year RFS rates were 44.3% (95% CI, 34.6%–53.6%) in the RD homo-to-homo group, 36.2% (95% CI, 24.6%–47.9%) in the RD hetero-to-homo group, 49.6% (95% CI, 44.1%–54.8%) in the URD homo-to-homo group, 32.1% (95% CI, 12.4%–53.9%) in the URD hetero-to-homo group, 7.3% (95% CI, 1.3%–20.5%) in the CB homo-to-homo group, and 35.1% (95% CI, 22.5%–48.0%) in the CB hetero-to-homo group (Fig. 3 B). Multivariate analysis showed lower RFS in the CB homo-to-homo group than in the URD homo-to-homo group (adjusted HR, 1.94; 95% CI, 1.32–2.86; P = 0.001) (Table 2 ).

All models were adjusted for variables that met the entry criterion ( P < 0.05) in stepwise selection, with donor source forced to remain in all models. Cox proportional hazards models were used for OS and RFS. Fine-Gray competing-risk regression models were used for NRM, relapse, engraftment, and GVHD outcomes.

Relapse and NRM according to donor source and HLA matching

The unadjusted 5-year cumulative incidence of NRM was 19.2% (95% CI, 12.1%–27.5%) in the RD homo-to-homo group, 9.7% (95% CI, 3.7%–19.3%) in the RD hetero-to-homo group, 18.6% (95% CI, 14.6%–22.9%) in the URD homo-to-homo group, 21.3% (95% CI, 6.0%–42.8%) in the URD hetero-to-homo group, 34.1% (95% CI, 17.9%–51.1%) in the CB homo-to-homo group, and 24.1% (95% CI, 13.6%–36.3%) in the CB hetero-to-homo group (Fig. 4 A). Multivariate analysis showed no significant differences in NRM between the groups (Table 2 ). Causes of death are listed in Table 3 . Relapse or progression was the most common cause of death in all groups.

The unadjusted 5-year cumulative incidence of relapse was 36.5% (95% CI, 27.6%–45.4%) in the RD homo-to-homo group, 54.1% (95% CI, 41.4%–65.1%) in the RD hetero-to-homo group, 31.9% (95% CI, 27.0%–36.8%) in the URD homo-to-homo group, 46.6% (95% CI, 21.9%–68.1%) in the URD hetero-to-homo group, 58.6% (95% CI, 38.8%–73.9%) in the CB homo-to-homo group, and 40.8% (95% CI, 27.4%–53.8%) in the CB hetero-to-homo group (Fig. 4 B). Multivariate analysis showed higher relapse risks in the RD hetero-to-homo and CB homo-to-homo group than in the URD homo-to-homo group (adjusted HR, 1.71; 95% CI, 1.17–2.48; P = 0.005 and adjusted HR, 1.81; 95% CI, 1.15–2.84; P = 0.010, respectively) (Table 2 ).

Impact of donor source and host-versus-graft mismatch on engraftment

The cumulative incidence of neutrophil engraftment at day 28 was 96.7% (95% CI, 90.8%–98.8%) in the RD homo-to-homo group, 92.1% (95% CI, 82.7%–96.5%) in the RD hetero-to-homo group, 94.9% (95% CI, 92.1%–96.7%) in the URD homo-to-homo group, 86.4% (95% CI, 59.8%–95.9%) in the URD hetero-to-homo group, 85.0% (95% CI, 68.6%–93.2%) in the CB homo-to-homo group, and 77.6% (95% CI, 64.3%–86.4%) in the CB hetero-to-homo group (Fig. 5 A).

Multivariate analysis showed that the RD homo-to-homo group had a higher neutrophil engraftment rate than the URD homo-to-homo group (HR, 1.47; 95% CI, 1.17–1.85; P = 0.001). The CB homo-to-homo and CB hetero-to-homo groups showed lower engraftment rates than the URD homo-to-homo group (adjusted HR, 0.65; 95% CI, 0.50–0.84; P = 0.001 and adjusted HR, 0.51; 95% CI, 0.40–0.66; P < 0.001, respectively) (Table 2 ). In a direct comparison between the CB homo-to-homo and CB hetero-to-homo groups, we found no significant differences in neutrophil or platelet engraftment.

The cumulative incidence of platelet engraftment at day 60 was 85.0% (95% CI, 77.1%–90.3%) in the RD homo-to-homo group, 77.6% (95% CI, 66.3%–85.6%) in the RD hetero-to-homo group, 85.2% (95% CI, 80.8%–88.6%) in the URD homo-to-homo group, 61.9% (95% CI, 36.7%–79.5%) in the URD hetero-to-homo group, 62.5% (95% CI, 45.5%–75.7%) in the CB homo-to-homo group, and 63.8% (95% CI, 49.8%–74.8%) in the CB hetero-to-homo group (Fig. 5 B). Multivariate analysis showed that the cumulative incidence of platelet engraftment at day 60 was higher in the RD homo-to-homo group compared with the URD homo-to-homo group (adjusted HR, 1.36; 95% CI, 1.04–1.77; P = 0.024). The CB homo-to-homo, and CB hetero-to-homo groups showed lower platelet engraftment rates than the URD homo-to-homo group (adjusted HR, 0.48; 95% CI, 0.36–0.64; P < 0.001 and HR, 0.45; 95% CI, 0.35–0.57; P < 0.001, respectively) (Table 2 ).

Acute GVHD and chronic GVHD across donor source and host-versus-graft matching subgroups

The unadjusted cumulative incidence of grade II–IV acute GVHD within 100 days post-transplantation was 27.5% (95% CI, 19.8%–35.7%) in the RD homo-to-homo group, 17.3% (95% CI, 9.7%–26.8%) in the RD hetero-to-homo group, 35.7% (95% CI, 30.9%–40.7%) in the URD homo-to-homo group, 31.8% (95% CI, 13.7%–51.7%) in the URD hetero-to-homo group, 32.5% (95% CI, 18.5%–47.3%) in the CB homo-to-homo group, and 36.2% (95% CI, 24.0%–48.6%) in the CB hetero-to-homo group (Fig. 6 A).

Multivariate analysis showed that the RD homo-to-homo group and RD hetero-to-homo group had lower risks of grade II–IV acute GVHD than the URD homo-to-homo group (adjusted HR, 0.55; 95% CI, 0.36–0.83; P = 0.005 and adjusted HR, 0.45; 95% CI, 0.24–0.79; P = 0.007, respectively) (Table 2 ).

The unadjusted cumulative incidence of grade III–IV acute GVHD within 100 days post-transplantation was 10.0% (95% CI, 5.5%–16.2%) in the RD homo-to-homo group, 1.4% (95% CI, 0.1%–6.5%) in the RD hetero-to-homo group, 10.9% (95% CI, 8.0%–14.4%) in the URD homo-to-homo group, 4.5% (95% CI, 0.3%–19.5%) in the URD hetero-to-homo group, 5.0% (95% CI, 0.9%–15.1%) in the CB homo-to-homo group, and 5.3% (95% CI, 1.4%–13.3%) in the CB hetero-to-homo group (Fig. 6 B). Multivariate analysis showed that the RD hetero-to-homo group had lower risks of grade III–IV acute GVHD than the URD homo-to-homo group (adjusted HR, 0.10; 95% CI, 0.01–0.78; P = 0.028) (Table 2 ). In the sensitivity analysis excluding patients who received PTCy ( n = 12 in the RD hetero-to-homo group), the RD hetero-to-homo group continued to show significantly lower risks of grade II-IV acute GVHD and grade III-IV acute GVHD compared with the URD homo-to-homo group (adjusted HR, 0.50; 95% CI, 0.28–0.89, P = 0.019 and adjusted HR, 0.12; 95% CI, 0.01–0.93; P = 0.043, respectively).

The unadjusted 5-year cumulative incidence of chronic GVHD was 35.4% (95% CI, 26.7%–44.1%) in the RD homo-to-homo group, 29.8% (95% CI, 19.7%–40.5%) in the RD hetero-to-homo group, 36.4% (95% CI, 31.3%–41.5%) in the URD homo-to-homo group, 24.8% (95% CI, 7.1%–48.0%) in the URD hetero-to-homo group, 16.4% (95% CI, 6.4%–30.4%) in the CB homo-to-homo group, and 21.1% (95% CI, 11.5%–32.7%) in the CB hetero-to-homo group (Fig. 6 C). For chronic GVHD, multivariate analysis showed no statistically significant differences between the URD homo-to-homo group and any other groups (Table 2 ).

Impact of haplotype on transplant outcomes

The most frequent haplotype was "A*24:02-B*52:01-C*12:02-DRB1*15:02" ( n = 380) (Table 4 ). Detailed patient and transplant characteristics are summarized in Supplementary Table S6 .

To assess whether haplotype differences explained relapse risk, we compared transplantation outcomes among the RD ( n = 67), URD ( n = 220), and CB homo-to-homo groups ( n = 30) matched for this haplotype. The unadjusted 5-year cumulative incidence of relapse was 38.5% (95% CI, 26.1%–50.7%) in the RD homo-to-homo group, 34.8% (95% CI, 28.3%–41.4%) in the URD homo-to-homo group and 59.6% (95% CI, 37.4%–76.2%) in the CB homo-to-homo group (Fig. 7 A).

Multivariate analysis showed that the risk of relapse was significantly higher in the CB homo-to-homo group than the URD homo-to-homo group (adjusted HR, 1.97; 95% CI, 1.20–3.23; P = 0.007) (Table 5 ).

The unadjusted 5-year RFS rates were 42.1% (95% CI, 28.9%–54.8%) in the RD homo-to-homo group, 46.6% (95% CI, 39.8%–53.7%) in the URD homo-to-homo group and 3.8% (95% CI, 0.3%–16.3%) in the CB homo-to-homo group (Fig. 7 B). The CB homo-to-homo group showed lower relapse-free survival compared with the URD homo-to-homo group (adjusted HR, 2.21; 95% CI, 1.44–3.41; P < 0.001) (Table 5 ).

Discussion

We used a Japanese registry to evaluate outcomes after allogeneic HCT in patients with HLA-homozygous haplotypes according to donor source and HLA compatibility. In multivariate analysis, OS in the RD homo-to-homo, RD hetero-to-homo, and CB hetero-to-homo groups did not differ significantly from that in the URD homo-to-homo reference group. In contrast, the CB homo-to-homo group was associated with inferior overall survival, whereas the URD hetero-to-homo group showed a trend toward inferior overall survival that did not reach statistical significance. Because this was a retrospective observational study, the findings described below should be interpreted as associations rather than causal relationships. Furthermore, RFS was lower in the CB homo-to-homo group than in the URD homo-to-homo group. It should be noted that the inferior overall survival observed in the CB homo-to-homo and URD hetero-to-homo groups may be partially influenced by baseline imbalances in patient characteristics. As shown in Table 1 , the CB homo-to-homo group represented the oldest population (median age, 63.5 years), and both groups had higher proportions of patients with HCT-CI scores of ≥ 3 (20.0% and 31.8%, respectively) compared with the reference URD homo-to-homo group (12.3%). The URD hetero-to-homo group also had a higher median age (58.5 years) and a higher proportion of patients receiving reduced-intensity conditioning (54.5%) compared with the URD homo-to-homo group (median age 49 years; reduced-intensity conditioning, 33.8%), which may reflect a selection bias toward older or frailer patients in this subgroup.

Both HCT-CI and performance status were retained in the final stepwise models and thus directly adjusted for in the multivariate analyses. Nevertheless, residual confounding from these and other unmeasured baseline conditions cannot be excluded.

Relapse risk was higher in the RD hetero-to-homo group than in the URD homo-to-homo group. The reasons for this difference remain unclear and may reflect multiple factors, including donor and recipient characteristics and transplant strategies. In addition, our findings were unchanged in a sensitivity analysis excluding patients who received PTCy, suggesting that the observed relapse signal was not driven by PTCy use. Because only 15.7% (12/76) of patients in the RD hetero-to-homo group received PTCy, we were underpowered to evaluate the independent effect of GVHD prophylaxis on relapse in this subgroup. Although prior studies have raised the possibility of an association between PTCy-based prophylaxis and relapse, larger studies are needed to clarify whether GVHD prophylaxis contributes to relapse risk in patients with HLA-homozygous haplotypes [ 28 , 29 ].

We also observed a higher relapse risk in the CB homo-to-homo group than in the URD homo-to-homo group. A previous study reported an increased relapse risk after CB transplantation with complete haplotype matching, particularly in pairs carrying the haplotype [ 30 ]. To assess whether our findings were driven by haplotype distribution, we performed a subgroup analysis restricted to pairs matched for the most frequent haplotype (A*24:02–B*52:01–C*12:02–DRB1*15:02). Relapse risk remained higher in the CB homo-to-homo group than in the URD homo-to-homo group. These findings suggest that the excess relapse observed in CB homo-to-homo transplantation is unlikely to be explained solely by a single high-risk haplotype. Several mechanisms may contribute to the higher relapse risk observed after CB homo-to-homo transplantation. CB transplantation is associated with delayed immune reconstitution, including impaired thymopoiesis and delayed functional T-cell recovery, which may limit early antileukemic immunity [ 31 ]. Under these conditions, extensive haplotype sharing may further attenuate alloreactive priming, potentially weakening graft-versus-leukemia effects despite acceptable GVHD rates. In addition, adult CBT studies suggest that transplantation outcomes depend on the balance between HLA disparity and cell dose, and that the optimal degree of HLA matching may vary across relapse risk groups [ 32 ]. Although current unit selection strategies focus on improving survival and engraftment by balancing HLA matching and cell dose, our findings, together with prior observations in haplotype-matched CBT, raise the possibility that maximal haplotype sharing may be associated with a higher risk of relapse in recipients with HLA-homozygous haplotypes, particularly in patients with intrinsically high relapse risk. The inferior OS in this group likely reflects the increased relapse burden. The distribution of HLA mismatches in the hetero-to-homo groups should also be considered when interpreting these findings (Supplementary Table S1 ). Mismatch burden was strongly linked to donor source, with RD, URD, and CB transplantations showing different distributions of mismatched loci. Accordingly, the present analysis could not disentangle the independent effect of specific mismatch patterns from that of donor source. This suggests that the unfavorable trend observed in the URD hetero-to-homo group may not be explained solely by the number of mismatched loci. However, because each subgroup was small, more detailed locus-specific analyses were not feasible.

Consistent with prior reports, CB transplantation was associated with lower neutrophil and platelet engraftment compared with other donor sources [ 33 ]. In our cohort, both CB homo-to-homo and CB hetero-to-homo groups showed lower engraftment compared with URD homo-to-homo, whereas we did not observe clear differences in engraftment between the two CB groups. Together, these findings suggest that, among CB transplantations without GVH direction mismatches, the presence of HVG direction mismatches may have limited impact on engraftment in patients with HLA-homozygous haplotypes. Previous studies evaluating mismatch direction in CB transplantation have reported inconsistent results. One study reported that HVG mismatches were associated with higher risks of engraftment failure and reduced survival compared with GVH-direction mismatches [ 5 ]. In contrast, another Japanese study found that isolated HVG mismatches did not affect engraftment or survival compared with bidirectional single-locus mismatches [ 34 ]. Differences in study design, cohort size, and the distribution of unidirectional versus bidirectional mismatches may partly explain these discrepancies. Our results support the observation that isolated HVG-direction mismatches do not necessarily compromise engraftment in this context.

The overall incidence of grade III–IV acute GVHD was low in this cohort of patients with HLA-homozygous haplotypes. Notably, the RD hetero-to-homo group showed a lower risk of severe acute GVHD than the URD homo-to-homo group, which may be influenced by GVHD prophylaxis strategies used in this setting, including PTCy or ATG [ 10 , 28 , 29 , 35 , 36 ]. No grade III–IV acute GVHD occurred among patients receiving PTCy or ATG in the RD hetero-to-homo group. However, the small sample size limits definitive conclusions.

This study has several limitations. First, this was a retrospective, registry-based observational study, and the findings should be interpreted as associations rather than causal effects. Residual confounding related to donor selection and other unmeasured clinical factors may remain. Second, transplants were performed over a long period from 2000 to 2022, during which changes in transplantation practice and supportive care may have introduced era-related confounding. Although we adjusted for year of transplantation and performed a sensitivity analysis restricted to transplants performed from 2011 to 2022, subgroup-specific estimates changed in that analysis: the signal for inferior OS in the CB homo-to-homo subgroup was attenuated, whereas the URD hetero-to-homo subgroup remained associated with inferior OS. These findings indicate that the subgroup estimates were sensitive to era restriction and should be interpreted cautiously, particularly given the limited sample size. Third, sample sizes were small in some subgroups, particularly the URD hetero-to-homo subgroup and the CB subgroups, which limited statistical precision. Fourth, generalizability may be limited because the cohort was derived from a Japanese population. Finally, interpretation of the CB transplantation findings should be limited to cases without GVH-direction mismatches, because CB units with GVH-direction mismatches were excluded.

In this registry-based observational study, RD hetero-to-homo and CB hetero-to-homo transplantation were not significantly associated with inferior OS compared with URD homo-to-homo transplantation and may be considered possible alternatives in selected patients. However, these findings should not be interpreted as evidence of equivalence and require validation in contemporary prospective cohorts.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 88 KB)

Acknowledgements

We thank all physicians and data managers at the centers who contributed to data collection on transplantation from the Japanese Society for Transplantation and Cellular Therapy and the Transplant Registry Unified Management Program 2. In addition, we thank the members of the Data Management Committees of the JSTCT and TRUMP for their assistance.

Author contribution

NY, MI, and JK designed the study, and reviewed and analyzed the data. NY wrote the first draft of the manuscript. KK, FK, MH, MK, and SM contributed to data collection. All authors approved the final version of the manuscript.

Funding

This work was supported in part by JSPS KAKENHI grant number 24K11515 (JK), AMED grant number JP25ek0510046 (JK), and AMED grant number JP25ck0106989 (NY).

Data availability

The data supporting the findings of this study are available from the corresponding author, JK, upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Kanda J, Saji H, Fukuda T et al (2012) Related transplantation with HLA-1 Ag mismatch in the GVH direction and HLA-8/8 allele-matched unrelated transplantation: a nationwide retrospective study. Blood 119:2409–2416

2. Kanda J, Fuji S, Kato S et al (2014) Decision analysis for donor selection in stem cell transplantation-HLA-8/8 allele-matched unrelated donor vs HLA-1 AG mismatched related donor. Blood Cancer J 4:e263

3. Kanda Y, Kanda J, Atsuta Y et al (2013) Impact of a single human leucocyte antigen (HLA) allele mismatch on the outcome of unrelated bone marrow transplantation over two time periods. A retrospective analysis of 3003 patients from the HLA Working Group of the Japan Society for Blood and Marrow Transplantation. Br J Haematol 161:566–577

4. Robin M, Ruggeri A, Labopin M et al (2015) Comparison of unrelated cord blood and peripheral blood stem cell transplantation in adults with myelodysplastic syndrome after reduced-intensity conditioning regimen: a collaborative study from Eurocord (Cord blood Committee of Cellular Therapy & Immunobiology Working Party of EBMT) and Chronic Malignancies Working Party. Biol Blood Marrow Transplant 21:489–495

5. Stevens CE, Carrier C, Carpenter C, Sung D, Scaradavou A (2011) Hla mismatch direction in cord blood transplantation: impact on outcome and implications for cord blood unit selection. Blood 118:3969–3978

6. Kanda J, Ichinohe T, Fuji S et al (2015) Impact of HLA mismatch direction on the outcome of unrelated bone marrow transplantation: a retrospective analysis from the Japan Society for Hematopoietic Cell Transplantation. Biol Blood Marrow Transplant 21:305–311

7. Hurley CK, Woolfrey A, Wang T et al (2013) The impact of HLA unidirectional mismatches on the outcome of myeloablative hematopoietic stem cell transplantation with unrelated donors. Blood 121:4800–4806

8. Terakura S, Atsuta Y, Tsukada N et al (2016) Comparison of outcomes of 8/8 and 7/8 allele-matched unrelated bone marrow transplantation and single-unit cord blood transplantation in adults with acute leukemia. Biol Blood Marrow Transplant 22:330–338

9. Milano F, Gooley T, Wood B et al (2016) Cord-blood transplantation in patients with minimal residual disease. N Engl J Med 375:944–953

10. Bolaños-Meade J, Hamadani M, Wu J et al (2023) Post-transplantation cyclophosphamide-based graft-versus-host disease prophylaxis. N Engl J Med 388:2338–2348

11. Ikeda N, Kojima H, Nishikawa M et al (2015) Determination of HLA-A, -C, -B, -DRB1 allele and haplotype frequency in Japanese population based on family study. Tissue Antigens 85:252–259

12. Enrich E, Campos E, Martorell L et al (2019) HLA-A, -B, -C, -DRB1, and -DQB1 allele and haplotype frequencies: an analysis of umbilical cord blood units at the Barcelona Cord Blood Bank. Hla 94:347–359

13. Chentoufi AA, Uyar FA, Chentoufi HA et al (2022) HLA diversity in Saudi population: high frequency of homozygous HLA alleles and haplotypes. Front Genet 13:898235

14. Watanabe M, Kanda J, Kojima H et al (2020) Wide availability of HLA-matched or a few loci-mismatched donors in the graft-vs-host direction among nonsibling first-degree relatives. Hla 95:543–554

15. Kanda J, Ikegame K, Fuji S et al (2016) Haploidentical and matched sibling donor hematopoietic cell transplantation for patients with HLA-homozygous haplotypes. Biol Blood Marrow Transplant 22:2031–2037

16. Atsuta Y, Suzuki R, Yoshimi A et al (2007) Unification of hematopoietic stem cell transplantation registries in Japan and establishment of the TRUMP System. Int J Hematol 86:269–274

17. Atsuta Y (2016) Introduction of transplant registry unified management program 2 (TRUMP2): scripts for TRUMP data analyses, part I (variables other than HLA-related data). Int J Hematol 103:3–10

18. Przepiorka D, Weisdorf D, Martin P et al (1995) 1994 consensus conference on acute GVHD grading. Bone Marrow Transplant 15(6):825–828

19. Sullivan KM, Agura E, Anasetti C et al (1991) Chronic graft-versus-host disease and other late complications of bone marrow transplantation. Semin Hematol 28:250–259

20. Bacigalupo A, Ballen K, Rizzo D et al (2009) Defining the intensity of conditioning regimens: working definitions. Biol Blood Marrow Transplant 15:1628–1633

21. Giralt S, Ballen K, Rizzo D et al (2009) Reduced-intensity conditioning regimen workshop: defining the dose spectrum. Report of a workshop convened by the center for international blood and marrow transplant research. Biol Blood Marrow Transplant 15:367–369

22. Sugita J, Kamimura T, Ishikawa T et al (2021) Reduced dose of posttransplant cyclophosphamide in HLA-haploidentical peripheral blood stem cell transplantation. Bone Marrow Transplant 56:596–604

23. Curtis DJ, Patil SS, Reynolds J et al (2025) Graft-versus-host disease prophylaxis with cyclophosphamide and cyclosporin. N Engl J Med 393:243–254

24. Gooley TA, Leisenring W, Crowley J, Storer BE (1999) Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med 18:695–706

25. Gray RJ (1988) A class of K-sample tests for comparing the cumulative incidence of a competing risk. Ann Stat 16:1141–1154

26. Jason P, Fine RJG (1999) A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 94:496–509

27. Kanda Y (2013) Investigation of the freely available easy-to-use software “EZR” for medical statistics. Bone Marrow Transplant 48:452–458

28. Atsuta Y, Sugita J, Nakamae H et al (2022) Comparable survival outcomes with haploidentical stem cell transplantation and unrelated bone marrow transplantation. Bone Marrow Transplant 57:1781–1787

29. Sugita J, Atsuta Y, Nakamae H et al (2022) Comparable survival outcomes with haploidentical stem cell transplantation and cord blood transplantation. Bone Marrow Transplant 57:1681–1688

30. Kanda J, Kawase T, Tanaka H et al (2020) Effects of haplotype matching on outcomes after adult single-cord blood transplantation. Biol Blood Marrow Transplant 26:509–518

31. Komanduri KV, St John LS, de Lima M et al (2007) Delayed immune reconstitution after cord blood transplantation is characterized by impaired thymopoiesis and late memory T-cell skewing. Blood 110:4543–4551

32. Atsuta Y, Kanda J, Takanashi M et al (2013) Different effects of HLA disparity on transplant outcomes after single-unit cord blood transplantation between pediatric and adult patients with leukemia. Haematologica 98:814–822

33. Yanada M, Konuma T, Yamasaki S et al (2020) Time-varying effects of graft type on outcomes for patients with acute myeloid leukemia undergoing allogeneic hematopoietic cell transplantation. Biol Blood Marrow Transplant 26:307–315

34. Kanda J, Atsuta Y, Wake A et al (2013) Impact of the direction of HLA mismatch on transplantation outcomes in single unrelated cord blood transplantation. Biol Blood Marrow Transplant 19:247–254

35. Kawamura K, Kanda J, Ohashi K et al (2020) Influence of HLA 1-3-locus mismatch and antithymocyte globulin administration in unrelated bone marrow transplantation. Ann Hematol 99:1099–1110

36. Penack O, Abouqateb M, Peczynski C et al (2024) PTCy versus ATG as graft-versus-host disease prophylaxis in mismatched unrelated stem cell transplantation. Blood Cancer J 14:45

Associated Data

Supplementary Materials

Supplementary file1 (DOCX 88 KB)

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author, JK, upon reasonable request.