Post-transplant cyclophosphamide in one-antigen mismarched unrelated donor transplantation versus haploidentical trasplantation in acute myeloid leukemia a study from the acute leukemia working party of the EMBT

Introduction. In the absence of an HLA-identical sibling or a matched unrelated donor, whether to prefer a Haploidentical (Haplo) or a one antigen mismatched unrelated donor (MMUD) for allogeneic hematopoietic cell transplantation (HCT) remains an unanswered question. Implementation of graft-versus-host disease (GVHD) prophylaxis with post-transplant cyclophosphamide (PTCY) was initially pioneered in the Haplo and then also extended to MMUD setting, resulting in low rates of GVHD. Methods. This was a retrospective study from the EBMT registry. Included were adults undergoing either Haplo- or MMUD-HCT for acute myeloid leukemia during the period 2010-2018 and who were in first or second complete remission at allo-HCT. Only patients receiving unmanipulated grafts with PTCY as GVHD prophylaxis were included. Ex vivo and in vivo T-cell depletion were exclusion criteria. Comparisons were made among three groups: MMUD-HCT with peripheral blood as stem cell source (PBSC; n=124); Haplo-HCT with bone marrow (Haplo-BM; n=560); Haplo-HCT with PBSC (Haplo-PB; n=769). Results. Patients in Haplo-PB were older (median age of 55 years versus (vs) 52 and 51 years in MMUD-HCT and Haplo-BM, respectively; p<0.01). Median time from diagnosis to transplantation was 6 months in all the groups. Class I (80%) were more frequent than class II mismatches (20%) in MMUD-HCT. A myeloablative conditioning regimen was used in 52%, 71% and 58% of patients in MMUD-HCT, Haplo-BM and Haplo-PB, respectively (p<0.01). The most frequently used regimen in MMUD-HCT was busulfan-fludarabine (43%), while a thiotepa busulfan-fludarabine regimen was the most frequently used in Haplo-BM (64%) and Haplo-PB (35%). A calcineurin inhibitor in association to mycophenolate mofetil were the most frequently adjuvant immunosuppressive agents. Median time for neutrophil engraftment was 20 days in MMUD-HCT and Haplo-BM and 19 in Haplo-PB, with a cumulative incidence (CI) of neutrophil engraftment at day 30 of 91%, 88% and 89% for MMUD-HCT, Haplo-BM and Haplo-PB, respectively. The CI of grade II-IV acute GVHD was lower in Haplo-BM (21% vs 33% and 34% in MMUD-HCT and Haplo-PB, respectively, p<0.01). Similarly, CI of grade III-IV acute GVHD was lower in Haplo-BM, being 6% vs 13% in both MMUD-HCT and Haplo-PB (p<0.01). Due to longer follow-up in Haplo-BM, outcomes were calculated after censoring at 2 years. CI of chronic GVHD of all grades and of extensive chronic GVHD were significantly lower in Haplo-BM (for chronic GVHD of all grades 26% vs 36% and 35% in MMUD-HCT and Haplo-PB, p=0.01; for extensive chronic GVHD 9% vs 16% and 13% in MMUD-HCT and Haplo-PB, p=0.03, respectively). No differences were observed in the CI of relapse, this being 20% in both MMUD-HCT and Haplo-PB and 23% in Haplo-BM (p=0.39). The CI of non-relapse mortality (NRM) was significantly lower in MMUD-HCT as compared to Haplo-HCT (11% versus 20% and 25% in Haplo-BM and Haplo-PB, respectively; p<0.01). We observed no differences in the probability of either overall survival (OS, 70% vs 62% and 60% in MMUD-HCT, Haplo-BM and Haplo-PB, p=0.12) or leukemia-free survival (LFS, 69% vs 56% and 55%, p=0.06) among the three groups. In Haplo-PB GVHD-free/relapse-free survival (GRFS) was lower (42% vs 49% in the other two groups, p<0.01). In multivariate analysis, Haplo-HCT was associated to a lower LFS and a higher NRM as compared to MMUD-HCT (reference group) (hazard ratio (HR) for LFS 1.57, 95% CI 1.06-2.33, p=0.02; 1.52 (95% CI 1.05-2.22, p=0.03 for Haplo-BM and Haplo-PB, respectively) (HR for NRM 2.25, 95% CI 1.16-4.33, p=0.02; 2.58, 95% CI 1.37-4.86, p<0.01 for Haplo-BM and Haplo-PB, respectively). On the other hand, Haplo-BM was associated to lower risk of grade II-IV (HR 0.64, 95% CI 0.42-0.96; p=0.03) and grade III-IV acute GVHD (HR 0.45, 95% CI 0.24-0.85; p=0.01) and also lower risk of chronic GVHD (HR 0.52, 95% CI 0.28-0.96; p=0.04). Of note, no differences in GRFS or OS were observed in mul