Third Allogeneic Hematopoietic Stem Cell Transplantation: Feasibility and Efficacy of a Single Institution

Relapse after second allograft (HCT) leaves few curative options. Third allografts (TP3) are considered, but evidence is limited and heterogenous. We evaluated the feasibility and efficacy of TP3 at our institution. We retrospectively analyzed 36 patients who underwent TP3 (1996-2024). All diseases, donor types and conditioning strategies were included. Primary endpoint was overall survival (OS); secondary endpoints were relapse-free survival (RFS) and 1-year cumulative incidence of relapse (CIR) and non-relapse mortality (NRM). Survival was estimated using Kaplan-Meier methods. Median age was 36 years (range; 7-71 years). Diagnoses included AML/MDS/BPCDN (58%, n=21), ALL 13% (n=5), MPAL (n=1), lymphoproliferative (n=2), myeloproliferative (MPD) (n=5) and graft failure (n=2). Two had confirmed donor-derived (dd) AML/MDS >1 year post TP2. Donor types included MUD(n=16), MSD(n=10), haploidentical(n=8), MMUD(n=1) and MMRD(n=1). 28/36(77.8%) changed donor between TP2 and TP3. Median interval between TP2 and TP3 was 516 days(range; 41-2122). 14/36(39%) were in CR/CRi, and the remaining 47% in active disease (NR/PD/SD); 1 had aplasia. Engraftment occurred in 94%; CR/CRi in 81% of evaluable patients. Median OS was 400 days (13.1 months), range; 8d-16.26y). Estimated OS at 12 and 24 months was 52.5% and 35%. Median RFS was 169 days (5.6 months). At 1 year, CIR was 28% and NRM 30.7%, plateauing after 6 months, with later attrition dominated by relapse. Both ddMDS/AML had prolonged survival (81.4,63.3 months). At median follow up of 11 years, 29/36 patients died (relapse-related 59%, non-relapse: GVHD,n=4; pulmonary toxicity,n=2; graft failure,n=2; liver failure,n=1; infection,n=3 [8d, 334d and 13.4y). At last follow up, seven (19%) patients were alive, indicating a durable survivor subset with median survival of 5.5 years (durations:0.6, 3.3, 5,3, 5.5, 11.1, 11.3, 16.3 years). Transplant indications were AML/MDS(n=4), graft failure(n=1), ddAML/MDS(n=1) and relapsed ALL(n=1); median TP2>TP3 interval was 531 days(149-1858) with 5/7 (71%) >365 days. 6/7 changed donors; all but one were in CR pre-TP3. Favorable outcomes were suggested with TP2-TP3 interval >12mo(OS 634 vs 110d, 64.6% vs 37.5%), donor change (477 vs 132d; 56.7% vs 37.5%) and MUD vs other donors (756 vs 110 days), though subgroup findings are hypothesis-generating in this small cohort. TP3 is feasible, but carries substantial risk of early NRM and relapse-driven failures. Nonetheless, long-term survival is achievable in a carefully selected subset, typically those with chemosensitive disease and a long disease-free interval post TP2, particularly when transplanted in remission, for non-relapse indications (graft failure, dd-leukemia) and with donor change (often MUD). Candidate and donor selection is critical, and warrants validation in larger, contemporary series.