Peripheral Blood Stem Cell Versus Bone Marrow Graft for Non-T-Depleted Haploidentical Transplantation with Post-Transplant Cyclophosphamide in Patients with Secondary Acute Myeloid Leukemia in First Complete Remission: A Study from the ALWP/EBMT

Background: Secondary acute myeloid leukemia (sAML) is a distinct type of high risk AML with inferior outcome. Haploidentical stem cell transplantation (HaploSCT) with post-transplant cyclophosphamide (PTCy) is an encouraging therapeutic option for sAML (J Hematol Oncol 2023, 16(1):106). However, in the context of sAML, it is still unknown whether peripheral blood (PB) or bone marrow (BM) is the best graft source. In view of the high relapse incidence (RI) accompanying sAML, PB grafts, potentially with a stronger graft-versus-leukemia (GVL) effect, may be beneficial.

Methods: The study aim was to compare PB vs. BM as a graft source for HaploSCT with PTCy performed between 2010 and 2022 in patients (pts) with sAML in first complete remission (CR1). Statistical tests included a multivariable analysis (MVA) using a Cox proportional-hazards regression model for main outcomes.

Results: A total of 554 pts were included, 418 with PB and 136 with BM grafts. Median follow-up was 2.8 years (interquartile range [2.5 – 3]) and 4.1 years [3.4 - 5.3], respectively. The median year of the transplant was 2019 (range, 2011-2021) in pts with PB and 2017 (range, 2010-2022) in those with BM grafts (p<0.001). Pts with PB grafts were older, with a median age of 61.7 (range 19.1-75.6) vs. 59.2 (19.3-75.7) years, (p=0.008). and 60% and 57%, were male, respectively, (p=0.48). In 91% of pts, the antecedent hematological disease was myelodysplastic/myeloproliferative disorder with no difference between the groups. The cytogenetic risk was categorized as intermediate (61% vs. 62%), adverse (38% vs. 37%), and favorable (1% in both) for pts receiving PB vs. BM grafts, respectively (p=0.96) (data missing for 118 pts). Karnofsky performance status (KPS), the frequency of both pt and donor cytomegalovirus (CMV) seropositivity as well as female donor to male pt combination did not differ between the two groups. Time from diagnosis to HaploSCT was significantly shorter for the PB vs. BM cohorts, being 4.8 (range 11.1-17) vs. 5.3 (1.8-16.2) months (p=0.019). A lower percentage of HaploSCT with PB compared to BM grafts received myeloablative conditioning; 39% and 57%, respectively (p<0.001), with thiotepa/busulfan/fludarabine being the most frequent regimen for both groups (38% vs. 69%). PTCy-based graft-versus-host disease (GVHD) prophylaxis was combined with cyclosporine A (CSA)/mycophenolate mofetil in 63% vs. 69% and CSA/tacrolimus in 34% vs. 31% of transplants, respectively. Cumulative incidence of absolute neutrophil count >0.5 x 10⁹/L at day 60 was 94.9% vs. 90.2%, respectively. In MVA acute GVHD grades II-IV, III-IV, and extensive chronic GVHD (cGVHD) did not differ significantly between pts receiving PB vs. BM grafts with a 2-year estimation of 31% vs. 22.2% (hazard ratio [HR]= 1.27; p=0.27), 10.7% vs. 6.9% (HR= 1.41; p=0.37) and 11.2% vs. 7.6% (HR= 1.4; p=0.36), respectively, while total cGVHD 2-year estimation was lower with BM vs. PB grafts with 33.7% vs. 22.9% (HR= 1.41; p=0.02). Other HaploSCT outcomes also did not differ significantly between pts receiving PB vs. BM grafts. The HR for 2-year non-relapse mortality (NRM) was 1.25 (95% CI 0.8-1.94, p=0.326), and for 2-year RI it was 0.85 (95% CI 0.55-1.33, p=0.48). RI was the main cause of death in BM (51% of pts) and NRM in PB transplants (50% of pts). The HRs for 2-year leukemia-free survival (LFS), overall survival (OS), and GVHD-free, relapse-free survival (GRFS) were 1.04 (95% CI 0.76-1.42, p=0.81), 1.00 (95% CI 0.72-1.39, p=0.1) and 1.13 (95% CI 0.85-1.5, p=0.42), respectively. Adverse-risk cytogenetics was a poor prognostic factor for RI, LFS, OS, and GRFS and both higher age (per 10 years) and lower KPS were poor prognostic factors for OS and GRFS. Lower KPS was also a poor prognostic factor for NRM and LFS. Adverse-risk cytogenetics and low intensity conditioning were poor prognostic factors for RI. In addition, age (per 10 years) and female donor to male pt combination were associated with increased risk of total cGVHD.

Conclusions: In this retrospective analysis of HaploSCT with PTCy for sAML in CR1, comparing PB vs. BM grafts, we observed similar transplantation outcomes except for lower incidence of total cGVHD with BM grafts. Notably, HaploSCT was able to rescue ~60% of the pts with this devastating leukemia. We did not observe lower RI with PB grafts, being ~22% at 2 years. Hopefully, with the recently approved novel agents for sAML it will be possible to further improve outcomes.