Graft-Versus-Leukemia Effect Is Independent from Graft-Versus-Host Disease Following Ptcy-Based HLA-Haploidentical Transplantation

Abstract

Background. The potential link between graft-versus-host disease (GVHD) occurrence and acute myeloid leukemia (AML) relapse in patients given human leukocyte antigen (HLA)-haploidentical allogeneic hematopoietic stem cell transplantation (Haplo-HCT) with post-transplant cyclophosphamide (PTCy)-based GVHD prophylaxis has remained debated. Here, we addressed this issue in a large cohort of AML patients with active AML at transplantation, a group of patients who mainly depends on graft-versus-leukemia (GVL) effects for tumor eradication. Methods. Population selection criteria for this study consisted of adults (>= 18 years old), first Haplo-HCT between 2010 and 2020 using PTCy-based GVHD prophylaxis, and primary refractory or relapsed AML (i.e. all patients had active disease at the time of transplant conditioning initiation). Twelve serial Landmark analyses from day 30 to day 365 were generated to assess the impact of GVHD on transplantation outcomes (dynamic landmarking). Results. A total of 803 patients met the population selection criteria. One hundred and thirty-three of them were excluded from the analyses because they died before achieving engraftment (n=90), received a second transplantation for graft failure (n=10) or because missing data in the registry on GVHD (n=31) or missing conditioning intensity (n=2). Therefore, the analyses were carried out in a cohort of 670 patients who were alive and had engrafted. Status at transplantation was primary refractory (55%), first relapse (35%) or second or more advanced relapse (10%). Conditioning was myeloablative in 289 patients. Two-year relapse and nonrelapse mortality incidences were 49% and 19%, respectively. Two-year leukemia-free survival (LFS) was 32%. There were no associations between acute nor chronic GVHD of any grade and lower relapse incidence. In contrast, each grade III-IV acute (HR=3.79, P<0.0001) and extensive chronic (HR=3.23, P<0.0001) GVHD correlated with higher nonrelapse mortality, translating to decreased LFS (HR=1.51, P=0.008 for grade III-IV acute and HR=1.90, P=0.0008, for extensive chronic GVHD, respectively). Conclusions. Our study suggests a dissociation of GvL effects from GVHD in patients with active AML given PTCy-based Haplo-HCT.