Fludarabine-Treosulfan (FT10) Conditioning in Older AML Patients Undergoing Haploidentical Transplantation: A Prospective Italian Study With External Comparison to the Haplo-UK Reduced-Intensity Conditioning Cohort

Background: Allogeneic haematopoietic stem cell transplantation (allo-HSCT) remains the only curative strategy for high-risk acute myeloid leukemia (AML), but its applicability in elderly and comorbid patients is limited by conditioning-related toxicity. Haploidentical HSCT (haplo-HSCT) with post-transplant cyclophosphamide (PTCy) has expanded access to transplantation, yet the optimal reduced-intensity conditioning (RIC) regimen in this setting remains undefined. Treosulfan-based conditioning combined with fludarabine (FT10) has demonstrated a favourable toxicity profile in matched donor transplantation, but prospective data in the haploidentical platform are scarce. Methods: We conducted a prospective, multicentre feasibility study across eight transplant centres, enrolling patients with intermediate- or high-risk AML (ELN-2017), aged ≥65 years, with haematopoietic cell transplantation-comorbidity index (HCT-CI) ≥2, lacking an HLA-identical donor and undergoing FT10-based haplo-HSCT between June 2019 and December 2023. Conditioning consisted of treosulfan (30 g/m²) and fludarabine (150 mg/m²), followed by GVHD prophylaxis with PTCy, cyclosporine and mycophenolate mofetil. Primary endpoint was cumulative incidence of nonrelapse mortality (NRM). Transplant outcomes were descriptively compared with an external reduced-intensity conditioning cohort from the prospective phase II Haplo-UK study. Results: Thirty-five AML patients were included in the FT10 cohort (median age 69 years); 58% were transplanted in complete remission (CR) with negative MRD, 22% in CR with positive MRD and 20% with active disease. Median neutrophil and platelet engraftment occurred at days +14 and +22, respectively, with full donor chimerism achieved in 94% of evaluable patients and one case of graft failure (2.8%). Cumulative incidence of NRM was 18% at day +100 and 28% at 1 year, with infections representing the leading cause of nonrelapse death. Outcomes were strongly influenced by disease status at transplant: 1-year OS and LFS were 79% and 70% in patients transplanted in CR/MRD-negative, compared with 42% and 38% in those transplanted with active disease. Acute grade II-IV and chronic GVHD occurred in 13% and 40% of patients, respectively. When restricted to patients transplanted in CR, survival and relapse outcomes were broadly comparable to those observed in the Haplo-UK cohort, despite the substantially older age and higher comorbidity burden of FT10 patients. Conclusions: In this prospective, real-world multicentre study, FT10-based haploidentical HSCT with PTCy proved feasible and effective in elderly, comorbid AML patients. These findings suggest that FT10 represents a viable conditioning option for a highly vulnerable population traditionally excluded from transplant and support further validation in larger comparative studies.