Impact of graft-versus-host disease on outcomes of cord blood transplantation according to HLA disparity and its prophylaxis type

Patients aged ≥ 16 years who received their first single-unit CBT for acute myeloid leukemia, acute lymphoblastic leukemia, and myelodysplastic syndrome between 2008 and 2020 and achieved neutrophil engraftment were retrospectively analyzed. Patients who received GVHD prophylaxis other than calcineurin inhibitor (CNI)–methotrexate (MTX) or CNI–mycophenolate mofetil (MMF) and those who received in vivo T-cell depletion, using components, such as anti-thymocyte globulin, were excluded. Furthermore, patients with HLA 7 and 8 allele mismatches in the HLA-A, -B, -C, and -DRB1 loci (HLA 1/8 match or 0/8 match) were excluded because of the small number of patients and difficulty of analysis. The analyzed data were obtained from the Transplant Registry Unified Management Program, which included data from the Japanese Society for Transplantation and Cellular Therapy and the Japan Cord Blood Bank Network [12‐14]. Informed consent was waived for patients who received transplants before 2017 because of the retrospective nature of this study; nevertheless, these patients were offered the opportunity to opt out. Informed consent was obtained from all other patients. This study conformed to the principles of the Declaration of Helsinki and was approved by the ethics committee of the Tohoku University School of Medicine.

HLA disparity was assessed based on the number of mismatches among eight alleles at the HLA-A, -B, -C, and -DRB1 loci. A regimen of total-body irradiation (≤ 8 Gy) and administration of busulfan (< 9 mg/kg) and melphalan (≤ 140 mg/m²) was defined as reduced-intensity conditioning [15]. For patients with insufficient data about doses of the conditioning regimen, we used clinician-reported conditioning intensity, such as myeloablative or reduced intensity. Adverse cytogenetic risk is defined as follows; A complex karyotype (three or more abnormalities), monosomal karyotype (two or more autosomal monosomies or one single autosomal monosomy with one or more structural abnormalities), −5/del(5q), −7/del(7q), 11q23 abnormality except for t(9;11), inv(3), t(3;3), t(6;9), and t(9;22) in acute myeloid leukemia, and abnormal chromosome 7 and complex karyotype (three or more abnormalities) in myelodysplastic syndrome. None of the chromosomal abnormalities in acute lymphoblastic leukemia were included in the adverse cytogenetic risk category.

Chi-square and Kruskal–Wallis tests were used to compare categorical and continuous variables in patient characteristics. The landmark method was used with univariate analysis to assess the impact of aGVHD on OS, relapse, and NRM, and the landmark was set at 90 d post CBT. Cumulative incidence of aGVHD, relapse, and NRM was evaluated by considering death as a competing risk, and Gray test and log-rank test were used to compare the cumulative incidences and OS, respectively. Multivariate adjustment for possible confounding variables using a cause-specific Cox proportional hazards regression model was used to assess the impact of aGVHD on the relapse rate, NRM, and OS. aGVHD development was considered a time-dependent covariate. The variables used in multivariate analysis are as follows: aGVHD I-II development, aGVHD III-IV development, age (< 55 or ≥ 55 years), karyotype risk (low/intermediate, adverse, or unknown karyotype), complete remission status at CBT, HCT-specific comorbidity index (HCT-CI) (< 3 or ≥ 3), performance status (PS) (0/1 or ≥ 2), recipient CMV serostatus, donor/recipient sex combination (female donor-to-male recipient, others, or unknown), number of transfused CD34 + cells (< 0.8 or ≥ 0.8 × 105/kg), conditioning regimen intensity (myeloablative or reduced-intensity), total body irradiation for conditioning, and transplantation year (< 2016 or ≥ 2016). EZR (version 1.61; Saitama Medical Center, Jichi Medical University, Saitama, Japan). Statistical significance was set at P-value < 0.05, except in the analysis of the interaction effect between GVHD prophylaxis and aGVHD or HLA disparity and aGVHD, where P-value < 0.1 was considered significant.

The number of eligible cases was 4,619, including 1,429 HLA 8/8–6/8-matched cases, 1,596 HLA 5/8-matched cases, and 1,594 HLA 4/8–2/8-matched cases. Each group was further divided into groups that received CNI–MTX (MTX prophylaxis) and CNI–MMF (MMF prophylaxis) as GVHD prophylaxis. The clinical characteristics of these groups are shown in Table 1. Cumulative incidence of aGVHD development is shown in Figure S1. aGVHD developed before day 90 post CBT in most cases, and aGVHD III-IV development was higher with MMF prophylaxis than with MTX prophylaxis (16.4% vs 9.2%, P < 0.001).

The unadjusted OS for patients who developed aGVHD in each group stratified by HLA disparity and GVHD prophylaxis is shown in Figure S2A–C. With the HLA 8/8–6/8 match, the impact of aGVHD III-IV was similar regardless of the GVHD prophylaxis type used. However, with the HLA 5/8 and 4/8–2/8 matches, aGVHD III-IV development had a more detrimental effect on patients receiving MMF prophylaxis than on those receiving MTX prophylaxis. Multivariate analysis was therefore performed to compare the impact of aGVHD on OS between these groups. aGVHD I-II development had a favorable impact on OS compared to the absence of aGVHD, regardless of HLA disparity and GVHD prophylaxis type (Fig. 1A–C). aGVHD III-IV development worsened OS in the HLA 8/8–6/8-matched group that received CBT with MTX prophylaxis (hazard ratio [HR] 1.66, P = 0.021) (Fig. 1A), whereas no adverse impact on OS was noted in the HLA 5/8- and 4/8–2/8-matched groups using MTX prophylaxis (Fig. 1B and C). In the MMF prophylaxis group, the impact of aGVHD III-IV development was unfavorable, regardless of HLA disparity (Fig. 1A–C). These differences in the impact of aGVHD III-IV development on OS between HLA disparity and GVHD prophylaxis types were supported by the P-value for the interaction effect between aGVHD III-IV and GVHD prophylaxis type (Fig. 1A–C).

Bild vergrößern

The cumulative incidence of relapse and NRM in patients who developed aGVHD before day 90 post CBT in each group stratified by HLA disparity and GVHD prophylaxis type is shown in Figures S3 and S4. The incidence of relapse was the lowest with aGVHD III-IV development in the HLA 4/8–2/8-matched group. aGVHD III-IV development worsened NRM similarly in the HLA 8/8–6/8-matched groups that received MTX and MMF prophylaxes. In contrast, in the HLA 5/8- and 4/8–2/8-matched groups, aGVHD III-IV development with MMF prophylaxis increased NRM substantially; this increase was relatively small with MTX prophylaxis. We subsequently performed multivariate analysis to examine the effect of aGVHD I–II and III–IV onset on the incidence of relapse and NRM. As shown in Fig. 2A and B, the onset of aGVHD had no impact on relapse in patients with HLA 8–6/8 and HLA 5/8 matches who received MTX or MMF prophylaxis. In contrast, the onset of aGVHD I-II and III-IV in HLA 4/8–2/8-matched patients who received MTX prophylaxis significantly reduced the incidence of relapse (HR 0.62, P = 0.0019 and HR 0.44, P = 0.0086, respectively), while no effect was noted in those who received MMF prophylaxis (Fig. 2C). The difference in the impact of aGVHD III-IV between GVHD prophylaxis types in HLA 4/8–2/8-matched patients was demonstrated by analyzing the interaction effect.

Bild vergrößern

As shown in Fig. 3A–C, aGVHD I-II development improved NRM regardless of HLA disparity and GVHD prophylaxis type, although this effect was marginal in HLA 8/8–6/8-matched groups that received MMF prophylaxis. In contrast, aGVHD III-IV development worsened NRM in all groups (Fig. 3A–C). Analysis of the interaction effect between aGVHD III-IV and GVHD prophylaxis showed that the impact of aGVHD III-IV was similar to that of MTX and MMF prophylaxis in the HLA 8/8–6/8- and 5/8-matched groups (Fig. 3A and B), whereas a more unfavorable impact of aGVHD III-IV development was noted in the HLA 4/8–2/8-matched group that received MMF prophylaxis than in the group that received MTX prophylaxis, which was supported by the analysis of the interaction effect (Fig. 3C).

Bild vergrößern