Progress in reprogramming failed graft-versus-leukemia immunity in acute myeloid leukemia relapse after allogeneic transplantation

Relapse of acute myeloid leukemia (AML) after allogeneic hematopoietic stem cell transplantation (allo-HSCT) affects 40–50% of recipients and remains the leading cause of posttransplant mortality; median overall survival is approximately 5 months and 1-year survival is below 20%. Conventional salvage—chemotherapy, immunosuppression withdrawal, donor lymphocyte infusion (DLI), hypomethylating agents, targeted agents, and second transplantation—benefits a minority and has improved only incrementally over two decades. Relapse in this setting reflects leukemia that has survived an engrafted allogeneic immune system. Documented escape mechanisms include genomic loss of the mismatched human leukocyte antigen (HLA) haplotype, epigenetic downregulation of class II HLA molecules, checkpoint ligand upregulation, microenvironmental suppression, and clonal evolution—each calling for targeted redirection of the graft-versus-leukemia (GVL) response rather than further nonspecific cytotoxicity. We propose a GVL Failure Framework that organizes posttransplant relapse into three biologically defined categories of immune escape, aligns matched cellular therapies with each, and overlays a separate clinical transplant-eligibility axis that sets the therapeutic goal. Several cellular strategies have matured: autologous and donor-derived chimeric antigen receptor T (CAR-T) cells targeting CD33, CD123, C-type lectin-like molecule 1 (CLL-1)/CD371, and CD117; armored CAR-T products secreting interleukin-18; CD83-directed CAR-T cells targeting blasts and alloreactive T cells simultaneously; HLA-DRB1-directed CAR-T and chimeric antigen receptor natural killer (CAR-NK) cells exploiting donor-recipient mismatch for leukemia specificity; and off-the-shelf CAR-NK platforms reprogramming innate rather than adaptive allogeneic immunity. We integrate the mechanistic rationale and early clinical evidence for each approach, propose a four-pathway treatment algorithm built around relapse-clone HLA typing before DLI, and outline the gaps that must close before cellular therapy enters standard practice.