Differential Effect of Graft-versus-Host Disease on Survival in Acute Leukemia according to Donor Type

Allogeneic hematopoietic cell transplantation (HCT) is a potentially curative therapy for adults with acute leukemia. The anti-leukemic activity of allogeneic HCT depends on both the intensity of conditioning regimen and the strength of the graft-versus-leukemia (GVL) effect. It is difficult to evaluate independently the meaningful GVL effect because this effect may occur without the occurrence of clinical graft-versus-host disease (GVHD). Indeed, the GVL effect is based on the phenomenon that patients with GVHD had a lower risk of leukemia relapse compared with patients without GVHD in previous studies (1, 2). Although the incidence and severity of acute and chronic GVHD are associated with donor type and graft source (3–6), the strength of GVL effects between donor type and graft source is unclear.

Several studies have suggested that the GVL effect might be enhanced by HLA-mismatched HCT, such as cord blood transplantation (CBT; refs. 7–10). CBT has unique characteristics, such as a lower incidence and severity of GVHD but does not compromise the GVL effects in spite of less stringent criteria for HLA matching (3–6). Indeed, recent studies have shown that relapse rate was significantly lower after CBT than HCT from matched sibling donors (MSD) or matched unrelated donors (MUD; refs. 3, 11). Therefore, the sensitivity of the GVL effects may differ between donor type and graft source. To clarify the effect of varying severities of acute and chronic GVHD on transplant outcomes, we performed a retrospective analysis using a nationwide Japanese database for a large cohort of adult patients with acute leukemia treated with allogeneic HCT from 8/8 allele-MSD, 8/8 allele-MUD, or unrelated cord blood (UCB).

This retrospective study was performed by the Donor/Source Working Group, the Adult Acute Myelogenous Leukemia (AML) Working Group, and the Adult Acute Lymphoblastic Leukemia (ALL) Working Group of the Japan Society for HCT (JSHCT). The clinical data were obtained by the Transplant Registry Unified Management Program (TRUMP) of the Japanese Data Center for HCT (JDCHCT), covering almost all (>300) HCT centers in Japan. Each HCT center is required to report annual follow-up data of consecutive HCTs (12, 13). The inclusion criteria consisted of patients' ages between 16 and 65 years with AML and ALL who received their first allogeneic HCT from MSD, MUD, or single-unit UCB between 2007 and 2017 in Japan, who achieved neutrophil engraftment, and who survived at least 60 days without relapse. We excluded patients that lacked data of survival and GVHD status and those who were conditioned with antithymocyte globulin (ATG). Finally, 6,548 patients were eligible for this study. This retrospective study was performed after approval by the institutional review board of the Institute of Medical Science, The University of Tokyo, where this study was carried out (30–71-B0128), and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from each patient.

The primary objective of this study was to clarify the effect of acute and chronic GVHD on leukemia-free survival (LFS) for each donor type. Secondary objectives were to clarify the effect of acute and chronic GVHD on overall survival (OS), hematological relapse, and non-relapse mortality (NRM) for each donor type.

The diagnosis and severity of acute and chronic GVHD were assessed by the treating physicians at each center according to consensus criteria (14, 15). The degree of HLA matching was based on allele levels for HLA-A, HLA-B, HLA-C, and HLA-DRB1 loci for HCT from MSD and MUD or antigenic levels for HLA-A, HLA-B, and HLA-DR loci HCT from single-unit cord blood. Myeloablative conditioning (MAC) was defined as total body irradiation with fractionated doses totaling ≥8 Gy, oral busulfan (BU) doses of ≥9 mg/kg, or intravenous BU doses of ≥7.2 mg/kg (16). Poor-risk cytogenetics was defined according to the National Comprehensive Cancer Network Guidelines for AML (17). Poor-risk cytogenetics for ALL included t(9;22), t(4;11), t(8;14), t(14;18), hypodiploid, and complex karyotypes according to the criteria described previously (18). LFS (inverse of treatment failure) was defined as the time from the date of HCT until death, relapse, or survival. OS (inverse of overall mortality) was defined as the time from the date of HCT until death or survival. Relapse was defined as hematological leukemia recurrence. NRM was defined as death without leukemia recurrence after HCT.

Patients, diseases, and transplant characteristics between groups were compared using a χ² test or Fisher's exact test for categorical variables and the Kruskal–Wallis test for continuous variables. The probabilities of LFS and OS were calculated using the Kaplan–Meier method, with a conditional landmark analysis, which could be used to remove guarantee-time bias (19). The probabilities of relapse and NRM were calculated on the basis of cumulative incidence curves to accommodate competing risks, with a conditional landmark analysis. The competing risk for relapse was NRM, whereas the competing risk for NRM was relapse. Univariate analyses were performed using the log-rank test for OS and LFS and Gray's test for relapse and NRM.

The Cox proportional hazards regression model was used to estimate hazard ratios (HR) with a 95% confidence interval (CI) for treatment failure (1-LFS), overall mortality (1-OS), relapse, and NRM in the multivariate analyses, by treating the development of GVHD as a time-dependent covariate. To evaluate the HRs for relapse and NRM, patients who experienced NRM or relapse were censored. When evaluating the effect of GVHD on transplant outcomes, acute GVHD and chronic GVHD were separately analyzed. The effect of acute GVHD was evaluated by comparing no acute GVHD versus grade I–II acute GVHD versus grade III–IV acute GVHD for patients who survived at least 60 days without relapse. The effect of chronic GVHD was evaluated by comparing no chronic GVHD versus limited chronic GVHD versus extensive chronic GVHD for patients who survived at least 100 days without relapse. The following variables other than GVHD were considered in the multivariate analysis: age (16–49 years vs. 50–65 years), sex (male vs. female), performance status (0–1 vs. 2–4), disease (AML vs. ALL), cytogenetics (other than poor vs. poor), disease status at HCT (CR vs. non-CR), conditioning regimen (MAC vs. RIC), GVHD prophylaxis (calcineurin inhibitors plus methotrexate vs. others), and transplant year (2007–2012 vs. 2013–2017). When evaluating the effect of chronic GVHD on transplant outcomes, the prior history of acute GVHD (none vs. grade I–II vs. grade III–IV acute GVHD) was included in the multivariate analysis. To adjust for multiple testing for each outcome in univariate and multivariate analysis, P < 0.0166 (0.05/3) was considered statistically considered with the Bonfferoni correction, and P < 0.0041 (0.0166/4) was considered statistically considered for subgroup analysis. P values between 0.0166 or 0.0041 and 0.05 were considered to have a marginal significance.

In all these analyses, we separately evaluated three distinct cohorts based on donor type (MSD, MUD, and UCB). All P values were two-sided, and the data were analyzed by EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan; ref. 20), a graphical user interface for the R 4.0.4 software program (R Foundation for Statistical Computing).

The characteristics of patients, diseases, and transplant procedures are summarized in Table 1. Among 6,548 patients, 1,322, 2,429, and 2,797 patients received allogeneic HCT from MSD, MUD, and UCB, respectively. The median age at HCT was 42 years, 47 years, and 49 years in MSD, MUD, and UCB recipients, respectively (P < 0.001). The proportions of recipient cytomegalovirus seropositive status (P = 0.004), HCT-CI≥3 (P < 0.001), and PS≥2 (P < 0.001) were different among donor types. MUD transplants were less frequently performed in recent years (P < 0.001). A higher proportion of UCB recipients received RIC regimen (P < 0.001), and calcineurin inhibitors+mycophenolate mofetil-based GVHD prophylaxis (P < 0.001). UCB recipients were more likely to have AML (P < 0.001) and a non-CR status (P < 0.001).

In the entire cohort, the median follow-up period for survivors after HCT was 41 months (range, 2–138 months). The probability of LFS at 3 years was 55% (95% CI, 52%–58%), 58% (95% CI, 56%–60%), and 53% (95% CI, 51%–55%) after HCT from MSD, MUD, and UCB, respectively (Supplementary Fig. S1A). In the multivariate analysis, the hazard risk of treatment failure was significantly lower in MUD (HR, 0.88; P = 0.022) or UCB (HR, 0.86; P = 0.007) recipients compared with MSD recipients (Supplementary Table S1). The probability of OS at 3 years was 63% (95% CI, 60%–66%), 64% (95% CI, 62%–66%), and 58% (95% CI, 56%–60%) after HCT from MSD, MUD, and UCB, respectively (Supplementary Fig. S1B). In the multivariate analysis, there was no significant difference in overall mortality between MSD recipients and both MUD (HR, 0.94; P = 0.30) and UCB (HR, 0.96; P = 0.53) recipients (Supplementary Table S1).

The cumulative incidence of relapse at 3 years was 34% (95% CI, 31%–37%), 28% (95% CI, 26%–30%), and 30% (95% CI, 29%–32%) after HCT from MSD, MUD, and UCB, respectively (Supplementary Fig. S1C). In the multivariate analysis, the hazard risk of relapse was significantly lower in MUD (HR, 0.79; P < 0.001) or UCB (HR, 0.80; P < 0.001) recipients compared with MSD recipients (Supplementary Table S1). The cumulative incidence of NRM at 3 years was 11% (95% CI, 9%–13%), 15% (95% CI, 13%–16%), and 17% (95% CI, 15%–18%) after HCT from MSD, MUD, and UCB, respectively (Supplementary Fig. S1D). In the multivariate analysis, there was no significant difference in NRM between MSD recipients and both MUD (HR, 1.18; P = 0.07) and UCB (HR, 1.10; P = 0.31) recipients (Supplementary Table S1).

In the univariate analysis with a conditional landmark analysis at 100 days, the probability of LFS was significantly lower in patients who developed grade III–IV acute GVHD compared with those who did not develop acute GVHD after HCT from MUD (P < 0.001; Fig. 1A–C). The probability of OS was significantly lower in patients who developed grade III–IV acute GVHD compared with those who did not develop acute GVHD after HCT from MSD (P = 0.011), MUD (P < 0.001), and UCB (P < 0.001; Fig. 1D–F). Interestingly, the probabilities of LFS and OS were significantly higher in patients who developed grade I–II acute GVHD compared with those who did not develop acute GVHD only after HCT from UCB (P < 0.001 for LFS, P < 0.001 for OS; Fig. 1C and F).

In the multivariate analysis, when no acute GVHD was used as the reference group among each donor type, development of grade I–II acute GVHD was significantly associated with lower treatment failure (P < 0.001) and overall mortality (P < 0.001) in patients after HCT from UCB but not in other donor types. The development of grade III–IV acute GVHD was significantly associated with higher treatment failure in patients after HCT from MUD (P < 0.001), and higher overall mortality in patients after HCT from MUD (P < 0.001), and UCB (P < 0.001; Fig. 1G and H).

In the univariate analysis with a conditional landmark analysis at 100 days, the cumulative incidence of relapse was significantly lower in patients who developed grade III–IV acute GVHD compared with those who did not develop acute GVHD after HCT from UCB (P < 0.001), but not MSD and MUD (Fig. 2A–C). The cumulative incidence of NRM was significantly higher in patients who developed grade III–IV acute GVHD compared with those who did not develop acute GVHD for each donor type (P < 0.001; Fig. 2D–F). Interestingly, the cumulative incidence of relapse (P < 0.001) was significantly lower in patients who developed grade I–II acute GVHD compared with those who did not develop acute GVHD only in patients after HCT from UCB (Fig. 2C).

In the multivariate analysis, when no acute GVHD was used as the reference group among each donor type, development of grade III–IV acute GVHD was associated with lower relapse in patients after HCT from MSD (P = 0.007), and UCB (P < 0.001). The development of grade III–IV acute GVHD was significantly associated with higher NRM in patients after HCT from MSD (P < 0.001), MUD (P < 0.001), and UCB (P < 0.001). Interestingly, development of grade I–II acute GVHD was also associated with significant lower relapse (P < 0.001) and NRM (P < 0.001) only in patients after HCT from UCB (Fig. 2G and H).

We also separately evaluated the effect of grade I and II acute GVHD on transplant outcomes in multivariate analysis using acute GVHD as time-dependent covariates. When no acute GVHD was used as the reference group among each donor type, development of grade I and II acute GVHD was also significantly associated with lower treatment failure, overall mortality, relapse, and NRM in patients after HCT from UCB but not in other donor types (Supplementary Fig. S2).

The causes of death according to the severity of acute GVHD for each donor type are summarized in Supplementary Table S2A. The most common cause of death was relapse in patients who did not develop acute GVHD and did develop grade I–II acute GVHD for each donor type. Only among UCB recipients, relapse (P = 0.002) and pulmonary complications (P = 0.011) were a less common cause of death in patients who developed grade I–II acute GVHD compared with those who did not develop acute GVHD. GVHD was a more common cause of death in patients who developed grade III–IV acute GVHD compared with those who did not develop acute GVHD, regardless of the donor type.

In the univariate analysis with a conditional landmark analysis at 129 days, which was the median time of development of chronic GVHD, the probabilities of LFS and OS were significantly lower in patients who developed extensive chronic GVHD compared with those who did not develop chronic GVHD after HCT from MSD (P = 0.011 for LFS, P = 0.001 for OS) and MUD (P = 0.002 for LFS, P < 0.001 for OS; Fig. 3A, B, D, and E). In contrast, the probabilities of LFS and OS were significantly higher in patients who developed limited chronic GVHD compared with those who did not develop chronic GVHD after HCT from UCB (P = 0.010 for LFS, P = 0.001 for OS; Fig. 3C and F).

In the multivariate analysis, when no chronic GVHD was used as the reference group among each donor type, development of limited chronic GVHD was significantly associated with lower treatment failure (P < 0.001) and overall mortality (P < 0.001) only in patients after HCT from UCB (Fig. 3G and H).

In the univariate analysis with a conditional landmark analysis at 129 days, there were no significant differences of relapse incidence in patients who developed limited or extensive chronic GVHD compared with those who did not develop chronic GVHD for each donor type (Fig. 4A–C). Meanwhile, the cumulative incidence of NRM was significantly higher in patients who developed extensive chronic GVHD compared with those who did not develop chronic GVHD for each donor type (P < 0.001 for MSD, MUD, UCB; Fig. 4D–F). Interestingly, the cumulative incidence of NRM was significantly lower in patients who developed limited chronic GVHD compared with those who did not develop chronic GVHD only in patients after HCT from UCB (P < 0.001; Fig. 4F).

In the multivariate analysis, when no chronic GVHD was used as the reference group for each donor type, development of extensive chronic GVHD was associated with a significant lower relapse in patients after HCT from MUD (P < 0.001) and a lower relapse but with marginal significance in patients after HCT from MSD (P = 0.016). The development of extensive chronic GVHD was also significantly associated with higher NRM in patients after HCT from MSD (P = 0.011), MUD (P < 0.001), and UCB (P = 0.002). Interestingly, development of limited chronic GVHD was associated with a significant lower NRM (P < 0.001) and a lower relapse but with marginal significance (P = 0.021) only in patients after HCT from UCB (Fig. 4G and H).

The causes of death according to the severity of chronic GVHD for each donor type are summarized in Supplementary Table S2B. The most common cause of death was relapse in patients who did not develop chronic GVHD and did develop limited or extensive chronic GVHD for each donor type. Relapse was a less common cause of death in patients who developed limited chronic GVHD compared with those who did not develop chronic GVHD in MUD (P = 0.011) and UCB (P = 0.017) recipients. GVHD was a more common cause of death in patients who developed extensive chronic GVHD compared with those who did not develop chronic GVHD, regardless of the donor type.

We also evaluated the effect of GVHD on transplant outcomes stratified according to disease type and status, which could affect the effects of GVHD on transplant outcomes, for each donor type.

The reductive effects of grade I–II acute GVHD on treatment failure and overall mortality after UCB transplantation were commonly noted in almost all subgroups of patients with AML and ALL, and those in CR and with non-CR status at HCT. However, the reductive effects of grade I–II acute GVHD on NRM after UCB transplantation were noted only in subgroups of patients with AML. The reductive effect of grade I–II acute GVHD on relapse after UCB transplantation was noted in subgroups of patients with AML and ALL, and those in CR status (Supplementary Fig. S3).

The reductive effects of limited chronic GVHD on treatment failure and NRM after UCB transplantation were noted in some subgroups of patients with AML and those with non-CR status at HCT, but not in those with ALL and with CR status. The reductive effects of limited chronic GVHD on overall mortality after UCB transplantation were noted in almost all subgroups of patients with AML and those in CR and with non-CR status at HCT, but not in those with ALL. The reductive effects of extensive chronic GVHD on relapse were noted in patients with CR status after MUD transplantation (Supplementary Fig. S4).

Finally, we evaluated whether HLA disparity and GVHD prophylaxis regimen affected the beneficial effects of mild GVHD on transplant outcomes after CBT. The beneficial effects of grade I–II acute GVHD on LFS, OS, and relapse were noted only in the HLA 4/6–matched group, but not in the HLA 5–6/6–matched group (Supplementary Fig. S5A–S5D). The beneficial effects of limited chronic GVHD on LFS, OS, and NRM were also noted only in the HLA 4/6–matched group, but not in the HLA 5–6/6–matched group (Supplementary Fig. S5E–S5H). GVHD prophylaxis regimens did not affect the beneficial effects of mild GVHD on outcomes after CBT (Supplementary Fig. S6).

We evaluated the differential GVL effect associated with GVHD in acute leukemia based on donor types in a large cohort from a nationwide Japanese database. Our data showed that the reductive effects of grade I–II acute GVHD and limited chronic GVHD on treatment failure and mortality were commonly noted in patients receiving from UCB. Interestingly, the reductions of treatment failure and overall mortality were due not only to the reduction of relapse rate, but also the reduction of NRM in patients receiving from UCB. These data suggested that the development of mild GVHD could improve survival after CBT for acute leukemia.

Recent retrospective studies have demonstrated that relapse rate is similar between different donor types, such as MSD, 8/8MUD, and UCB in patients with acute leukemia (3–6). Although the incidence and severity of GVHD could be independent of donor type, the strength of GVT effects between donor type is limited (21–23). Ringdén and colleagues (21) demonstrated that the GVL effect from HLA-identical siblings is similar to those from MUDs for acute leukemia patients undergoing allogeneic HCT. Our study clearly demonstrated that a survival benefit of the GVL effect was commonly present only in CBT patients with mild GVHD, which is consistent with a previous report published by the JSHCT GVHD working group (24). Their report showed that the development of mild acute and chronic GVHD reduced the relapse rate after single CBT in previous cohort of adults with acute leukemia and myelodysplastic syndrome from 2000 to 2012 (24). All these data suggest that the GVL effects after CBT might be stronger than those after HCT from MSD or MUD. Although the exact mechanism of stronger GVL effects after CBT is unclear, this might be partly due to the enhancement of GVL effect by HLA-mismatched donor transplantation (25), because our data also demonstrated that the reductive effect of grade I–II acute GVHD on relapse was noted only in the HLA 4/6–matched group. Indeed, lower relapse rates after CBT were associated with an increased number of HLA mismatches (7, 8) and a specific HLA locus mismatch (9, 10), suggesting that the GVL effect after CBT is an attractive concept for acute leukemia.

The strength of the GVT effects associated with GVHD is also dependent on disease type (18, 26–33). Although some studies showed that ALL was more sensitive to GVT effects associated with GVHD than AML (26, 30, 33), several studies showed that AML was more sensitive to GVHD than ALL (28, 29). Meanwhile, the strength of the GVT effects according to disease burden at HCT is controversial (21, 28, 33–35). An early study showed that the reductive effect of GVHD on relapse was stronger in patients with advanced leukemia compared with those in CR (33), whereas recent reports have shown that the GVL effect was not present in advanced leukemia (21, 34). Indeed, our study showed that the reductive effect of mild acute GVHD on relapse was noted only in CBT patients with AML and ALL, and those in CR status. The marginal reduction of relapse by the development of mild chronic GVHD was noted only in CBT patients with AML. Importantly, the survival benefit of GVL effects was present only in CBT patients with mild GVHD. Recently, Milano and colleagues (3), reported a significant reduction of relapse after CBT but only in minimal residual disease (MRD)–positive groups compared with HCT from matched or mismatched unrelated donors, suggesting that the GVL effects after CBT might be induced by MRD-positive status. However, we included various methods to detect MRD at HCT from our registry-data but could not evaluate the association between MRD status and GVL effects. Therefore, the association between MRD status and stronger GVL effect after CBT is a future matter of investigation.

Our study had several limitations. One limitation was the study was based on a retrospective registry-based data in Japan. Therefore, there could have been a selection bias and unavailable data, such as a mutation profile for leukemia (36, 37), HLA-DPB1 mismatch (10, 25), killer immunoglobulin-like receptors (KIR) ligand mismatch (38), and post-transplant maintenance therapy (39, 40), which could have affected the GVL effects or relapse rate after HCT. Second, the development of acute and chronic GVHD frequently overlapped, and the treatment intervention and response for acute and chronic GVHD could not be considered to have evaluated the GVL effects. Therefore, it was difficult to clarify the relevant effects of acute and chronic GVHD on relapse in our study. In addition, a beneficial effect of mild acute GVHD has been controversial. Previous studies showed a positive impact of grade I acute GVHD (28) and grade II acute GVHD (41, 42), but a negative impact of grade II acute GVHD (29). In our study, a beneficial effect of mild acute GVHD was noted with patients who developed each grade I and II acute GVHD only after HCT from UCB, as well as combined grade I–II acute GVHD, which was consistent with previous reports (18, 24). Third, the data for single-unit CBT alone were included in our study, because double-unit CBT was not approved for use outside of clinical trials during the study period in Japan. Several retrospective studies have shown that double-unit CBT mediated stronger GVL effects than single-unit CBT (43, 44), but this effect was not confirmed in prospective studies (45, 46). Moreover, the beneficial effects of survival on mild acute and chronic GVHD after CBT are independent of not only the lower relapse but also lower NRM, which is consistent with a previous report in Japan (24, 47). Although the exact mechanism of the reductive effect of mild GVHD on NRM after CBT is unclear, we speculate that development of mild GVHD may affect delayed or impaired immune reconstitution after CBT without excess of adverse effects, which could contribute to lower incidence of NRM. Fourth, recent studies have shown that the GVL effect from haploidentical donors is similar or superior to those from HLA-identical siblings for patients with acute leukemia undergoing allogeneic HCT (22, 23). However, haploidentical HCT data were not included in our study, due to the relatively small number of patients during the study period. Further studies are warranted to compare the GVL effects after CBT with those after haploidentical HCT. Fifth, racial differences could affect rates of acute and chronic GVHD not only in MRD recipients (48), but also UCB recipients (49). Because previous study showed that the risk of acute GVHD was significantly lower in the Japanese patients compared with the Caucasian patients (48). Therefore, it remains to be established whether our results can be applied to other racial groups.

In conclusion, our registry-based data demonstrated that the development of mild acute and chronic GVHD was associated with lower relapse and NRM after CBT, which contributed to the improvement of leukemia-free and OS for acute leukemia. These data indicated that the development of mild GVHD could improve survival after UCB transplantation for acute leukemia.

J. Kanda reports personal fees from Takeda Phamaceutical, Celgene, Novartis Pharma, Astellas Pharma, Chugai Pharmaceutical, Kyowa Kirin, Otsuka Pharmaceutical, Bristol Myers Squibb, JCR Pharmaceuticals, MSD, Daiichi Sankyo, Sanofi, Janssen Pharmaceutical, and Ono Pharmaceutical outside the submitted work. S. Hirabayashi reports personal fees and other support from Revorf Co., Ltd. outside the submitted work. N. Uchida reports personal fees from Chugai Pharmaceutical Co., Astellas Pharma Inc., Otsuka Pharmaceutical Co., Sumitomo Dainippon Pharma Co., and Novartis Pharma Inc. during the conduct of the study. M. Sawa reports personal fees from Chugai, Pfizer, Astellas, Nippon-Shinyaku, Ono, MSD, BMS, Kyowa Kirin, Asahi-Kasei, Novartis, Eisai, Otsuka, Sumitomo Dainippon, Sanofi, Takeda, Celgene, Mochida, Shire, and Mundiphama outside the submitted work. Y. Atsuta reports other support from AbbVie GK, Astellas Pharma Inc., Mochida Pharmaceutical Co., Ltd., Meiji Seika Pharma Co, Ltd., Chugai Pharmaceutical Co., Ltd., and Kyowa Kirin Co., Ltd. outside the submitted work. F. Kimura reports grants from Astellas, Takeda, Taiho, Ono, MSD, Eisai, Kyowa Kirin, Chugai, and Daiichi Sankyo, and grants and non-financial support from Bristol Myers Squibb outside the submitted work. No disclosures were reported by the other authors.

T. Konuma: Conceptualization, formal analysis, supervision, investigation, methodology, writing–original draft, project administration. J. Kanda: Resources, writing–review and editing. Y. Kuwatsuka: Resources, writing–review and editing. M. Yanada: Resources, writing–review and editing. T. Kondo: Resources. S. Hirabayashi: Resources, writing–review and editing. S. Kako: Resources. Y. Akahoshi: Resources. N. Uchida: Resources. N. Doki: Resources. Y. Ozawa: Resources. M. Tanaka: Resources. T. Eto: Resources. M. Sawa: Resources. S. Yoshioka: Resources. T. Kimura: Resources. Y. Kanda: Resources. T. Fukuda: Resources. Y. Atsuta: Resources, funding acquisition. F. Kimura: Resources, supervision.

The authors thank all of the physicians and staff at the hospital, the Japan Marrow Donor Program, and the cord blood banks who provided the clinical data to the Transplant Registry Unified Management Program of the Japanese Data Center for Hematopoietic Cell Transplantation. This work was supported in part by the Practical Research Project for Allergic Diseases and Immunology (Research Technology of Medical Transplantation) from Japan Agency for Medical Research and Development, AMED under grant 18ek0510023h0002.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.