Enhanced long-term outcomes with third-party unrelated cord blood co-infusion in a 122-patients cohort undergoing second allo-HSCT for AML in remission

Abstract Background While allo-HSCT is a curative approach for patients with hematologic malignancies, among them with relapse, graft failure, or GVHD may necessitate a second transplant, which is often limited by poor outcomes and increased toxicity.Co-infusion of third-party donor-derived stem cells, particularly unrelated umbilical cord blood (UCB), has widely recognized as a supportive strategy to enhance transplant efficacy and safety, yet evidence about this in the setting of second allo-HSCT is limited,its impact on second transplant success rate, the incidence of GVHD, and long-term survival will be discuseed in this study. Aims This study assessed UCB co-infusion in relapsed acute myeloid leukemia (AML) patients receiving second allo-HSCT in complete remission (CR). Methods A total of 122 AML patients in CR undergoing a 2nd ATG/G-CSF–based allo-HSCT between 2012 and 2024 were retrospectively analyzed. Patients were stratified into two groups based on the use of third-party UCB: haplo-cord cohort (n=94) and haplo cohort (n=28). All patients received standard GVHD prophylaxis (CSA, MMF, and sMTX) and were followed longitudinally through January 1, 2025. Results The study included 122 AML patients (72 males, 50 females) with a median age of 34 years (range, 1–61). Median disease duration was 17.5 months; median time from first allo-HSCT to relapse was 5.5 months; and the median interval between transplants was 13 months. Before second transplant, 76 were MRD-negative and 46 MRD-positive. Conditioning regimens included TBI/CY (n=75) or BU/CY (n=47). Donor types were haploidentical (n=106) or matched unrelated (MUD, n=16). The median infused cell doses were 6.7 × 10⁸ (range, 3.87–20.01) mononuclear cells/kg and 5.5 × 10⁶ (range, 1.7–20.29) CD34⁺ cells/kg. Median time to neutrophil and platelet engraftment was 19 (range, 8–23) and 27 days (range, 7–63), respectively. Primary graft failure occurred in 11 patients (9.0%). The median follow-up was 12.3 (range, 0.1–147.3) months overall and 23.2 (range, 2–147.3) months among survivors. A total of 52 patients died, 92.3% within 1 year. The haplo-cord group showed a higher 2-year OS (57.8% vs. 43.6%, p=0.064) and significantly better 2-year LFS (59.3% vs. 39.8%, p=0.038). NRM was significantly lower with haplo-cord (29.7% vs. 52.7%, p=0.032), as was infection-related mortality (23.9% vs. 48.9%, p=0.020). Relapse incidence was lower in the haplo-cord group (14.9% vs. 34.9%), though not significant (p=0.564). Day-100 cumulative incidence of grade II–IV and III–IV aGVHD did not differ significantly between groups (24.5% vs. 28.8%, p=0.719; 14.9% vs. 21.6%, p=0.460). Moderate-to-severe cGVHD was numerically lower with haplo-cord (21.7% vs. 31.5%, p=0.248). The 180-day incidence of CMV reactivation was higher in the haplo-cord group (52.6% vs. 36.2%, p=0.231), while EBV rates were comparable (16.1% vs. 18.1%, p=0.749). Multivariable Cox and competing risks models included variables with p<0.2 in univariate analysis or established clinical relevance. MRD positivity was independently associated with inferior OS (HR, 1.55; p=0.00), LFS (HR, 1.57; p=0.00), increased RI (HR, 3.06; p=0.02), and higher NRM (HR, 2.37; p=0.01). UCB co-infusion was independently associated with improved OS (HR, 1.56; p=0.01), LFS (HR, 1.63; p=0.00), and reduced NRM (HR, 3.58; p=0.00). Higher CD34⁺ cell doses (>6 × 10⁶/kg) were associated with better OS (HR, 1.55; p=0.01), LFS (HR, 1.51; p=0.01), and lower NRM (HR, 2.76; p=0.02). No variable independently predicted grade III–IV aGVHD. However, haploidentical donors were associated with increased risk of moderate-to-severe cGVHD compared to MUD donors (HR, 4.45; p=0.01). Conclusion Third-party UCB co-infusion during second allo-HSCT in AML patients in CR was independently associated with improved survival and reduced non-relapse and infection-related mortality without lowering aGVHD incidence, supporting its potential utility pending further prospective validation.